Keywords: Brain-computer/machine interface, Brain functional imaging - MEG, Human performance - Speech
Abstract: Decoding speech directly from the brain has the potential for the development of the next generation, more efficient brain computer interfaces (BCIs) to assist in the communication of patients with locked-in syndrome (fully paralyzed but aware). In this study, we have explored the spectral and temporal features of the magnetoencephalography (MEG) signals and trained those features with convolutional neural networks (CNN) for the classification of neural signals corresponding to phrases. Experimental results demonstrated the effectiveness of CNNs in decoding speech during perception, imagination, and production tasks. Furthermore, to overcome the long training time issue of CNNs, we leveraged principal component analysis (PCA) for spatial dimension reduction of MEG data and transfer learning for model initialization. Both PCA and transfer learning were found to be highly beneficial for faster model training. The best configuration (50 principal coefficients + transfer learning) led to more than 10 times faster training than the original setting while the speech decoding accuracy remained at a similarly high level.

Keywords: Brain-computer/machine interface, Brain functional imaging - EEG
Abstract: Virtual reality (VR) offers the potential to study brain function in complex, ecologically realistic environments. However the additional degrees of freedom make analysis more challenging, particularly with respect to evoked neural responses. In this paper we designed a target detection task in VR where we varied the visual angle of targets as subjects moved through a three dimensional maze. We investigated how the latency and shape of the classic P300 evoked response varied as a function of locking the electroencephalogram data to the target image onset, the target-saccade intersection, and the first fixation on the target. We found, as expected, a systematic shift in the timing of the evoked responses as a function of the type of response locking, as well as a difference in the shape of the waveforms. Interestingly, single-trial analysis showed that the peak discriminability of the evoked responses does not differ between image locked and saccade locked analysis, though it decreases significantly when fixation locked. These results suggest that there is a spread in the perception of visual information in VR environments across time and visual space. Our results point to the importance of considering how information may be perceived in naturalistic environments, specifically those that have more complexity and higher degrees of freedom than in traditional laboratory paradigms.

Keywords: Brain-computer/machine interface, Neural stimulation, Neural signal processing
Abstract: The accuracy of brain-computer interfaces (BCIs) is important for effective communication and control. The mu-based BCI is one of the widely used systems, of which the related methods to improve users' accuracy is still poorly studied. Here, we examined the way to enhance the mu-based BCI performance by rhythmic electrical stimulation on the ulnar nerve at the contralateral wrist at the alpha frequency (10 Hz) during the left- and right-hand motor imagery. Time-frequency analysis, spectral analysis, and discriminant analysis were performed on the electroencephalograph (EEG) data before and after the intervention of electrical stimulation in 9 healthy subjects. We found that the ERD/S on the somatosensory and motor cortex during left- or right-hand imagination was more obvious at the mu rhythm after intervention. Furthermore, average classification accuracy between left- and right-hand imagery significantly increased from 78.43% to 88.17% after intervention, suggesting that the electrical stimulation at alpha frequency effectively regulates the brain's mu rhythm and enhances the discriminability of the left-hand and right-hand imagination tasks. These results provide evidence that the electrical stimulation at the alpha frequency is an effective way to improve the mu-based BCI performance.

Keywords: Brain-computer/machine interface, Neurorehabilitation, Brain functional imaging - EEG
Abstract: Development of noninvasive brain-machine interface (BMI) systems based on electroencephalography (EEG), driven by spontaneous movement intentions, is a useful tool for controlling external devices or supporting a neuro- rehabilitation. In this study, we present the possibility of brain-controlled robot arm system using arm trajectory decoding. To do that, we first constructed the experimental system that can acquire the EEG data for the not only movement execution (ME) task but also movement imagery (MI) tasks. Five subjects participated in our experiments and performed four directional reaching tasks (Left, right, forward, and backward) in the 3D plane. For robust arm trajectory decoding, we propose a subject-dependent deep neural network (DNN) architecture. The decoding model applies the principle of bi-directional long short-term memory (LSTM) network. As a result, we confirmed the decoding performance (r-value: >0.8) for all X-, Y-, and Z-axis across all subjects in the MI as well as ME tasks. These results show the feasibility of the EEG-based intuitive robot arm control system for high-level tasks (e.g., drink water or moving some objects). Also, we confirm that the proposed method has no much decoding performance variations between ME and MI tasks for the offline analysis. Hence, we will demonstrate that the decoding model is capable of robust trajectory decoding even in a real-time environment.

Keywords: Brain-computer/machine interface, Neural signal processing
Abstract: Steady-State Visual Evoked Potentials (SSVEP) based Brain-Computer Interface (BCI) has achieved very high information transmission rate (ITR), but its portability and fundamental interactions with the surrounding environment were limited. The combination of Augmented Reality (AR) and BCI is expected to solve these problems. In this paper, we combined AR with the SSVEP-BCI to build a more portable and natural BCI system in Microsoft Hololens. We designed the AR-BCI system and studied the influence of different algorithms on the system performance. The analysis of SSVEP signals collected in AR environment shows that the extended filter bank canonical correlation analysis was better than task-related component analysis. The average recognition accuracy and ITR obtained by using EEG data of 1s, 1.5s, and 2s length were 87.7%、95.4%、97.6% and 64.6 bit/min、62.9 bit/min、55.6 bit/min, respectively. Compared with the existing AR-BCI studies, the ITR has been greatly improved in this study.

Keywords: Brain-computer/machine interface, Neural signal processing, Brain functional imaging - EEG
Abstract: Brain-computer interface (BCI) based on rapid serial visual presentation (RSVP) is an efficient information detection technology by detecting event related brain response evoked by target stimuli. In the protocol design of the RSVP-BCI a range of parameters could influence the task difficulty, which may result in the changes of mental workload for subjects. This paper focused on the presentation rate in the RSVP paradigm aiming to investigate its influence on mental workload, and the separability of brain states during RSVP tasks with different setup of presentation rate. 64-channel Electroencephalographic (EEG) data were recorded during RSVP tasks with three levels of presentation rate in ten healthy subjects. The results show that different presentation rates indeed contribute to significant differences on mental workload revealed by one-way repeated measures analysis of variance (ANOVA) on z-scored RSME. Higher presentation rate results in the significant decrease on both behavioral and single-trial recognition performance of target images. Classification results on three levels of mental workload show that the mean accuracy reaches 65.5% and the highest accuracy reaches 88.3%. This work implies that mental workload induced by different presentation rates during RSVP tasks could be accurately recognized, and provides a possible method to monitor the mental workload in the application areas of RSVP-BCI.

Keywords: Neural networks and support vector machines in biosignal processing and classification, Data mining and processing in biosignals, Nonlinear dynamic analysis - Biomedical signals
Abstract: Respiratory ailments afflict a wide range of people and manifests as conditions asthma and sleep apnea. Continuous monitoring of chronic respiratory ailments is seldom used outside the intensive care ward due to large size and cost of the monitoring system. While Electrocardiogram (ECG) based respiration extraction is a validated approach, adoption is limited by access to a suitable continuous ECG monitor. Recently, due to the widespread adoption of wearable smartwatches with in-built Photoplethysmogram (PPG), they are being seen as a viable candidate for continuous and unobtrusive respiration monitoring. Research in this domain, however, has been predominantly focussed on estimating respiration rate from PPG. In this work, a novel end-to-end deep learning network called RespNet is proposed to perform the task of extracting the respiration signal from a given input PPG as opposed to extracting respiration rate. The proposed network was trained and tested on two different datasets utilizing different modalities of reference respiration signal recordings. Also, the similarity and performance of the proposed network against two conventional signal processing approaches for extracting respiration signal was studied. The proposed method was tested on two independent datasets with a Mean Squared Error of 0.262 and 0.145 and Cross-Correlation coefficient of 0.933 and 0.931 for the respective datasets. These reported errors and similarity was found to be better than conventional approaches. The proposed approach would aid clinicians to provide comprehensive evaluation of sleep-related respiratory conditions and chronic respiratory ailments while being comfortable and inexpensive for the patient.

Keywords: Physiological systems modeling - Signal processing in physiological systems, Physiological systems modeling - Signals and systems, Physiological systems modeling - Multivariate signal processing
Abstract: The current non-invasive gold standard for the measurement of blood pressure (BP) is the oscillometric cuff at the upper arm, despite its known limitations. In particular, its poor adequacy with continuous monitoring and its measurement incommodity call for the development of simpler and more convenient solutions. Among these, solutions based on pulse wave analysis (PWA) and photoplethysmography (PPG) are of particular interest, due to their low-cost, strong patient compliance, and applicability in and out of clinical settings. In that context, we have recently disclosed a PPG-based PWA algorithm (oBPM) dedicated to the continuous monitoring of BP in patients undergoing induction of general anesthesia. As is standard with PPG-based BP monitoring techniques, an initial calibration procedure with a reference device is required to allow the estimation of absolute values of BP (in mmHg). However, due to their sensitivity to peripheral effects such as vasomotion, the applicability of PPG-based techniques is often limited by the constant need of re-calibration procedures, sometimes in matters of minutes. In the present study, we evaluated the long-term stability of the calibration for our algorithm by performing PPG measurements at irregular time intervals over a period of 3 months in 13 healthy volunteers. For each measurement, diastolic BP (DBP) was assessed by an oscillometric device and estimated by the oBPM algorithm. We found the calibration to remain stable over the entire 3-month period, with estimation errors remaining stable over time and complying with the ISO 81060-2:2018 standard. In addition, we verified – in 11 of our 13 subjects – the sensitivity of the oBPM algorithm to changes in DBP. This was done in a protocol involving static leg extension exercises. Excellent trending ability (average per-subject concordance rate of 97.7 ± 5.2 %, and correlation coefficient of 0.98 ± 0.02, p < 0.001) was found between cuff-derived DBP changes and our estimates.

Keywords: Physiological systems modeling - Signal processing in physiological systems, Neural networks and support vector machines in biosignal processing and classification
Abstract: Sleep quality has a significant impact on human mental and physical health. Detecting sleep-wake stages is of paramount importance in the study of sleep. The gold standard method for sleep-wake stages classification is the multi-sensors based polysomnography (PSG) systems, which is normally recorded in clinical settings. The main drawback of PSG is the inconvenience to the subjects and can hamper the normal sleep. This paper describes an automated approach for classifying sleep-wake stages using finger-tip photoplethysmographic (PPG) signal. The proposed system used statistical features of PPG signal and supervised machine learning models including K-nearest neighbors (KNN) and support vector machine (SVM). The models are trained using 80% events (3486 sleep-wake events) from the dataset and the rest 20% events (872 sleep-wake events) are used for testing. On the test events, cubic KNN, weighted KNN, quadratic SVM and medium Gaussian SVM show 69.27%, 70.53%, 71.33% and 72.36% overall accuracy respectively for predicting the sleep and wake stages. This result advocates that the statistical features of PPG are capable of recognizing the changes in physiological states. The KNN and SVM classifier adopt the statistical features from PPG signal to differentiate between the wake and sleep stages.

Keywords: Physiological systems modeling - Signal processing in physiological systems, Physiological systems modeling - Signals and systems, Signal pattern classification
Abstract: In this work, we present the feasibility to use a simpler methodological approach for the assessment of the short-term complexity of Heart Rate Variability (HRV). Specifically, we propose to exploit Pulse Rate Variability (PRV) recorded through photoplethysmography in place of HRV measured from the ECG, and to compute complexity via a linear Gaussian approximation in place of the standard model-free methods (e.g., nearest neighbor entropy estimates) usually applied to HRV. Linear PRV-based and model-free HRV-based complexity measures were compared via statistical tests, correlation analysis and Bland-Altman plots, demonstrating an overall good agreement. These results support the applicability of the simpler proposed approach, which is faster and easier-to-implement, making our approach eligible for portable/wearable devices and thus broadening the out-of-lab accessibility of autonomic indexes.

Keywords: Signal pattern classification, Data mining and processing - Pattern recognition, Data mining and processing in biosignals
Abstract: Photoplethysmography (PPG) provides a simple, convenient and noninvasive method to assess pulse oximetry. Several attempts have been made to use PPG also to estimate blood pressure and arterial stiffness. This paper attempts to assess obesity classes, age group, and hypertension classes using PPG measured from the finger. One set of features was derived from the normalized pulse width of PPG and the other from original PPG. The features were calculated based on the pulse decomposition analysis using five lognormal functions and the up-slope of the PPG pulse. Using kNN and SVM as classifiers, the results were validated using leave-one-out validation. Performances of both features sets have no significant difference, and the kNN outperformed the SVM. The best accuracies are 93%, 88%, and 92% for obesity (5 classes), age group (7 classes), and hypertension (4 classes) respectively. These three assessment targets have a strong relationship with arterial stiffness, therefore it also leads to a study about arterial stiffness using PPG. Width normalization to 1 second might affect some features points based on pulse decomposition analysis. This study also found that the up-slope analysis might give good indices when width normalization was employed. However, these findings still require more experiments to gain conclusions that are more comprehensive.

Keywords: Time-frequency and time-scale analysis - Time-frequency analysis
Abstract: Photoplethysmogram (PPG) has been used with great effect for predicting human vitals such as heart rate variability or blood oxygen saturation (mbox{SpO}_2), etc. The quality of PPG signal is affected mainly by noise, drift and motion artifacts. Although noise and drift are relatively easy to remove, motion artifacts pose a challenge. Classical techniques for motion artifact removal are time-frequency based. However, this affects the signal as the PPG signal and motion artifacts distortion lie in the same frequency band, making it difficult to remove the motion artifacts without affecting the signal. In this work, we propose a motion artifact removal technique in time domain, which is based on correcting individual pulses in the PPG signal, considering a global pulse average and a windowed local pulse average. We show the effectiveness of our approach both qualitatively as well as quantitatively.

Keywords: Optical imaging - Coherence tomography, Image analysis and classification - Machine learning / Deep learning approaches
Abstract: Optical coherence tomographic (OCT) images provide valuable information for understanding the changes occurring in the retina due to glaucoma, specifically, related to the retinal nerve fiber layer and the optic nerve head. In this paper, we propose a deep learning approach using Capsule network for glaucoma classification, which directly operates on 3D OCT volumes. The network is trained only on labelled volumes and does not attempt any region/structure segmentation. The proposed network was assessed on 50 volumes from 2 different scanners and found to achieve 0.97 for the area under the ROC curve (AUC). This is considerably higher than the existing approaches which are majorly based on machine learning and that rely on segmentation of the required structures from OCT. Our network also outperforms 3D convolutional neural networks despite the fewer network parameters and fewer epochs needed for training.

Keywords: Optical imaging - Coherence tomography, Image enhancement
Abstract: This paper introduces a new approach to single-image super-resolution in Optical Coherence Tomography (OCT) images. Retinal OCT images can be used to diagnose various diseases, not only peculiar to the eye, but also some systemic diseases. Nevertheless, as with any imaging modality, the acquired images suffer from degradation due to various causes. To overcome this and enhance image quality, Super-Resolution (SR) techniques are widely used. This work explores a convex regularization approach based on a multivariate generalization of the minimax-concave (GMC) scheme in a forward-backward splitting (FBS) scheme. Based on the assumption that sparse representations of OCT images are heavy-tailed, an alpha-stable dictionary is employed. This approach is implemented with overlapping and non-overlapping patches. Since the Point Spread Function (PSF) of the images used is generally unknown, it is estimated using a method originally proposed for ultrasound images. The algorithm is tested on OCT images of murine eyes. The results show that the proposed convex regularization method provides results that are competitive with the state-of-the-art. Indeed, significant deblurring and quality enhancement are achieved using the proposed algorithm and in most cases it provides the best results, both objectively and subjectively.

Keywords: Optical imaging - Coherence tomography, Image enhancement - Denoising, Image reconstruction and enhancement - Filtering
Abstract: Reducing speckle noise from the optical coherence tomograms (OCT) of human retina is a fundamental step to a better visualization and analysis in retinal imaging, as thus to support examination, diagnosis and treatment of many eye diseases. In this study, we propose a new method for speckle reduction in OCT images using the super-resolution technology. It merges multiple images for the same scene but with sub-pixel movements and restores the missing signals in one pixel, which significantly improves the image quality. The proposed method is evaluated on a dataset of 20 OCT volumes (5120 images), through the mean square error, peak signal to noise ratio and the mean structure similarity index using high quality line-scan images as reference. The experimental results show that the proposed method outperforms existing state-of-the-art approaches in applicability, effectiveness, and accuracy.

Keywords: Optical imaging - Coherence tomography, Image feature extraction
Abstract: This study has investigated the relationship of the retinal vasculature and the retinal thickness for Macular Edema (ME) subjects. Ninety sets of Fluorescein Angiography (FA) and Optical Coherence Tomography (OCT) of 54 participants were analyzed. Multivariate analysis using the binary logistic regression model was used to study the association between vessel parameters and retinal thickness. The results reveal retinal vessel feature i.e. fractal dimension (FD) as the most sensitive parameter to the changes in retinal thickness associated with ME. Thus, indicating a direct relationship between the retinal vasculature and retinal thickness which is caused due to neovasculature causing exudates, leakages, and hemorrhages, with applications for alternate modality for detection of ME.

Keywords: Optical imaging - Coherence tomography, Image classification
Abstract: Optical Coherence Tomography (OCT) technology enabled the experts to analyze coronary lesions from high-resolution intravascular images. Studies have shown the relationship between bifurcation regions and a higher occurrence of wall thickening and lesions in these areas. Some level of automation could benefit experts, since examining pullback frames is a laborious and time-consuming task. Although Convolutional Neural Networks (CNN) have shown promising results in classification tasks of medical images, we did not identify the use of CNN’s in IVOCT images to classify bifurcation regions in the literature. In this work, we evaluated a CNN architecture in the bifurcation classification task trained with IVOCT images from 9 pullbacks from 9 different patients. We used data augmentation to balance the dataset, due to the low amount of bifurcation-labeled frames. Our classification results are comparable to other works in the literature, presenting better result in AUC (99.70%).

Keywords: Image reconstruction and enhancement - Tomographic reconstruction, Image enhancement - Denoising, Optical imaging
Abstract: In this paper, we apply combination of sparse representations and a total variation for reconstruction of retinal optical coherence tomography (OCT) images. The OCT imaging is based on interferometry, therefore OCT images suffer from the existence of a high level of noise. Utilization of effective interpolation and denoising algorithms are necessary to reconstruct high-resolution OCT images, especially when the subsampling of data is done during acquisition. In this paper we take total variational and Morphological Component Analysis (MCA) techniques to reduce noise and interpolate missing data. Different over-complete dictionaries are constructed by using curvelet transform , wavelet transform or DCT which represent the texture and cartoon layers in B-scans. Comparative analysis of image interpolation is done by two combinations of dictionaries, which are (DCT+Curvelert) and (DWT+ Curvelert) transforms. Layered structure are more distinguished in reconstructed image with curvelet dictionary and texture are mostly detectable by wavelet or DCT.Evaluation are done both visually and in terms of different performance measures. Our simulation results show that the (DCT+Curvelert)combination preserve the texture of the image well and the (DWT+Curvelert) combination has better performance in structure preservation.

Keywords: Physiological monitoring - Modeling and analysis, Modeling and analysis
Abstract: Knowledge of the level of mental workload induced by any task is essential for optimizing load share among the operators. This helps in assessing the capability of the operators; besides, helping in allocation of tasks to the operators. Since a persistently high workload experienced by operators such as aircraft pilots and automobile drivers many times compromises their performance and safety. Despite the availability of various mental workload evaluation techniques such as heart rate variability, pupil dilation, saccades, etc., assessment of mental workload is still a challenging task. In this work, we aim to evaluate the workload of the operator involved in long duration tasks. For this, experiments have been carried out in a working environment which provides tasks to be done simultaneously, tasks with a pause or break in activity and cross-functional tasks. The experiment data is recorded continuously in different modes and analyzed in segments to show the change in mental workload. The artificial neural network (ANN) architecture classified the workload data with an accuracy of 96.6%. The brain connectivity analysis shows the efficacy of the proposed approach.

Keywords: Wearable sensor systems - User centered design and applications, Physiological monitoring - Modeling and analysis
Abstract: Emotions potentially have a significant impact on human actions and recognizing affective states is an effective way of implementing Brain-Computer Interface (BCI) systems which process brain signals to allow direct communication and interaction with the environment. In this paper, a real-time emotion recognition model was developed on the basis of physiological signals. A sensor fusion method is developed to detect human emotion by using data acquired from ElectroEncephaloGraphy (EEG) and ElectroDermal Activity (EDA) sensors. The proposed physiology-based emotion recognition system using a neural network was implemented and tested on human subjects and, a classification accuracy of 94% on three different emotions was achieved.

Keywords: Wearable sensor systems - User centered design and applications, Physiological monitoring - Modeling and analysis, Bio-electric sensors - Sensor systems
Abstract: Generic dry-contact ear-EEG allows for discreet, user-friendly, unobtrusive, cost-effective and convenient recordings of EEG in real-life settings. In this study we introduce a new generic earpiece design with larger internal ear electrode distances, resulting in an increased spatial coverage compared to previous generic earpiece designs. The signal quality of ear-Fpz, within-ear (the measuring and reference electrode located in the same ear) and cross-ear (the measuring electrodes located in one ear and the reference electrode in the opposite ear) electrode configurations of the developed generic earpiece was evaluated with auditory steady-state responses (ASSR) and compared to dry-contact cEEGrid. Ten subjects with different ear sizes were included. The recordings were performed in a sleep setup, where the subjects were lying on a bed and the effect of sleeping position (back vs. sides) was investigated. We found that the generic earpiece attained statistically significant ASSRs with ear-Fpz, within-ear and cross-ear electrode configurations. However, the dry-contact cEEGrid achieved significantly higher average ASSR signal-to-noise ratio (SNR) compared to the generic earpiece. Additionally, this study showed no significant difference between back and side positions for the ear-EEG.

Keywords: Sensor systems and Instrumentation, Physiological monitoring - Novel methods, Novel methods
Abstract: The importance of fetal surveillance during pregnancy is worldwide accepted since its peculiar ability to anticipate fetal distress under a variety of conditions. The novel frontier in the field of remote fetal monitoring relies on a continuous and everyday-monitoring of fetal wellbeing. As a consequence, fECG monitoring systems have seen a net increase in popularity in the recent years. In this paper, we propose a novel algorithm for the detection of fECG and we validated its performances by testing it on an open source collection of 75 annotated fECG traces. Our results show the reliability of the proposed methodology in extracting fECG and deriving an estimate of fHR.

Keywords: Wearable sensor systems - User centered design and applications, Physiological monitoring - Instrumentation
Abstract: Capacitively-coupled ECG (ccECG) and bioimpedance (ccBIOZ) measurements are highly sensitive to motion artefacts. This limits their use in real-life situations. This work presents an array-based system for the simultaneous acquisition of ccECG and ccBIOZ, together with a quality-based electrode scanning approach for ccECG. This allows to increase the time coverage of contactless measurements in real-life situations and reduces the impact of artefacts. This solution was evaluated on a car seat and a mattress prototype. Results show the benefit of this combined array and algorithm approach: for every body position the algorithm was able to find more than one electrode combination providing high-quality ccECG. Night-long recordings were also performed, resulting in a mean time coverage of 72.5%.

Keywords: Physiological monitoring - Modeling and analysis, Implantable sensors, Bio-electric sensors - Sensing methods
Abstract: This work describes how is possible the definition of the light hole or lumen in implanted stents affected by restenosis processes using the BioImpedance (BI) as biomarker. The main approach is based on that neointimal tissues implied in restenosis can be detected and measured thanks their conductivity and dielectric properties. For that, it is proposed a four-electrode setup for bioimpedance measurement. The influence of the several involved tissues in restenosis: fat, muscle, fiber, endothelium and blood, have been studied at several frequencies, validating the setup and illustrating the sensitivity of each one. Finally, a real example using a standard stent, has been analyzed for stable and unstable restenosis test cases, demonstrating that the proposed method is useful to stent obstruction test. Bioimpedance simulation test has been performed using electric physics module in COMSOL.

Keywords: Time-frequency and time-scale analysis - Wavelets, Time-frequency and time-scale analysis - Time-frequency analysis
Abstract: Adaptive sampling is an interesting alternative for biosignal acquisition, transmission and storage, however further processing of non uniform representations is still waiting for development. In this paper a direct non-uniform to time-scale (NUTS) transform is presented and applied to the ECG signal. Well accepted limits of bandwidth in particular sections of the ECG and established standards for the assessment of diagnostic quality help in evaluation of the influence the transform has to the diagnostic result. The transform uses a regular-grid Coiflet 5-th order nearly symmetric wavelet, but the novelty is a pointwise calculating of its correlation accordingly to non-uniform distribution of the electrocardio-gram samples. In tests with CSE Database files the proposed transform method yields not bit-accurate ECG signals, but the diagnostic results are more influenced by the non-uniform representation (for QRS mean deviation: +0.7 ms vs. original files) than by the transform itself (for QRS additionally: +0.6 ms) and all the results remain within the accuracy tolerance of the CEN industrial standard.

Keywords: Time-frequency and time-scale analysis - Wavelets, Physiological systems modeling - Signals and systems
Abstract: Automatic methods for the detection of characteristic points in electrocardiography signals support cardiologists in assessing the state of a patient’s cardiovascular system. In this work, we apply a general method for parameter estimation to the specific problem of QRS complex, P-, and T-wave delineation, i.e. the computation of their on- and offset points in time. As input the method expects a piecewise Gaussian derivative model that is potentially a good fit for the morphology of electrocardiography waves, but a thorough investigation is needed. The model parameters are estimated by substituting zero-crossings of the input signals’ scale-space representation into scale-dependent algebraic expressions and are further refined by fitting the model to the electrocardiography signal in a least-squares sense. Validating the results on the QT database and comparing to state-of-the-art algorithms shows smallest mean error for 3 out of 9 fiducial points and for the others only small differences to the respective best competitors.

Keywords: Time-frequency and time-scale analysis - Nonstationary processing, Physiological systems modeling - Signal processing in physiological systems, Physiological systems modeling - Signals and systems
Abstract: Brugada syndrome (BS) is a genetic pathology that might cause sudden cardiac death (SCD) in patients with a structurally normal heart. Repolarization disorders have been postulated as a potential substrate for triggering cardiac arrhythmia in BS, that usually occur at rest or during sleep. In this paper, we have characterized ventricular repolarization markers during sleep on patients suffering from BS. To this end, standard 12-lead ECG recordings were analyzed in a population of 110 BS patients (25 symptomatic). The QT and the T-wave peak to T-wave end intervals (respectively QT and Tpe) were assessed from lead V5. The linear relationship between these markers and the instantaneous heart rate period (RR interval) are determined during each hour and for the whole sleep period. From the models obtained, corrected QT and Tpe measures were then estimated for each patient at 60 beats/min (QT60 and Tpe60) and at the mean heart rate observed during the involved time interval (QTHR and TpeHR). Results show larger values for symptomatic patients in all markers, with significant differences with respect to the asymptomatic group in the case of Tpe (Tpe60: p = 0.0012; TpeHR: p = 0.0014). Moreover, the temporal profiles of these markers reveal major differences among BS subgroups during the last 3 hours of sleep, where symptomatic patients presented increased QT60=HR (p = 0.01) and Tpe60=HR (p < 0.001), as compared to the initial sleep hours. We conclude that BS patients present different repolarization properties according to their symptomatology, especially during the final stage of sleep.

Keywords: Time-frequency and time-scale analysis - Time-frequency analysis, Neural networks and support vector machines in biosignal processing and classification
Abstract: Automatic classification of abnormal beats in ECG signals is crucial for monitoring cardiac conditions and the performance of the classification will improve the success rate of the treatment. However, under certain circumstances, traditional classifiers cannot be adapted well to the variation of ECG morphologies or variation of different patients due to fixed hand-crafted features selection. Additionally, existing deep learning related solutions reach their limitation because they fail to use the beat-to-beat information together with single-beat morphologies. This paper applies a novel solution which converts one-dimensional ECG signal into spectro-temporal images and use multiple dense convolutional neural network to capture both beat-to-beat and single-beat information for analysis. The results of simulation on the MIT-BIH arrhythmias database demonstrate the effectiveness of the proposed methodology by showing an outstanding detection performance compared to other existing methods.

Keywords: Time-frequency and time-scale analysis - Empirical mode decomposition in biosignal analysis, Signal pattern classification, Nonlinear dynamic analysis - Biomedical signals
Abstract: This work proposes to study the fetal heart rate (FHR) signal based on information about its dynamics as a signal resulting from the modulation by the autonomic nervous system. The analysis is performed using the complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) technique. The main idea is to extract a set of signal features based on that technique and also conventional time-domain features proposed in the literature in order to study their performance by using a support vector machine (SVM) as a classifier. As a hypothesis, we postulate that by including CEEMDAN based features, the classification performance should improve compared with the performance achieved by conventional features. The proposed method has been evaluated using real FHR data extracted from the open access CTU-UHB database. Results show that the classification performance improved from 67,6% using only conventional features, to 71,7% by incorporating CEEMDAN based features.

Keywords: Time-frequency and time-scale analysis - Empirical mode decomposition in biosignal analysis, Signal pattern classification
Abstract: Systolic time intervals Pre-Ejection Period (PEP) and Left Ventricular Ejection Time (LVET) are widely used indicators of cardiac functions. While accurate assessment of them requires costly equipment such as echocardiography devices, a satisfactory estimation can be done by analyzing signals from simple accelerometer and microphone attached to human chest. This paper reports a study where heart rate and the systolic intervals were derived from phonocardiogram (PCG) and seismocardiogram (SCG) simultaneously. Both sensors, the microphone for PCG and the accelerometer for SCG were attached on the chest wall, close to sternum (PCG) and apex of the heart (SCG). The signals were acquired from 10 participants in a 33-minute laboratory protocol with synchronized ECG measurements. Both signals went through an identical processing path: band pass filtering, envelope extraction with Hilbert transformation and peak detection from the envelope signal. In heart rate estimation, PCG and SCG reached 84% and 93% accuracy, respectively. The systolic interval accuracy estimation was based on deviation analysis as the absolute reference values for PEP and LVET were not available. In PEP estimation, the average standard deviations during the rest periods of the protocol were 4 ms for PCG and 8 ms for SCG. In LVET estimation, the deviations were nearly 10 fold compared to PEP. However, the results show that both methods can be used for accurate heart rate estimation and with careful mechanical attachment also PEP can be accurately derived from both. Due to sharper envelope signal waveform, PEP estimation was more accurate with PCG than with SCG.

Keywords: Neural signal processing, Neural interfaces - Neuromorphic engineering
Abstract: Neuromorphic devices and architectures offer novel ways of data manipulation and processing, especially in data intensive applications. At a single device level, various forms of neuroplasticity have been emulated over the past years, mainly with inorganic devices. Although the field of organic-based neuromorphic devices and circuits is still at its infancy, organic materials offer attractive features for neuromorphic engineering, including operation in aqueous electrolytes, interaction with various ionic species, biocompatibility and their ability to be integrated in flexible and stretchable substrates. These characteristics of organic neuromorphic devices will introduce in the future neuro-inspired features in signal processing at the interface with biology.

Keywords: Neural interfaces - Implantable systems, Neural stimulation - Deep brain, Brain functional imaging
Abstract: New technologies to interface with the mammalian brain in vivo are allowing unprecedented investigation of functional connectivity of neural circuitry. Electrophysiology, exploited for decades, is now flanked with a number of optical, genetic, acoustic and magnetic tools which are leading to multimodal operation of implantable probes in the brain for both readout and control of neural activity. This paper focuses on minimally-invasive probes combining multiple approaches for interfacing with the deep brain. It shows how tapered optical fibers can be a great technology platform for combining multiple tools, working across multiple domains, for closed loop interaction with the deep brain.

Keywords: Neural interfaces - Implantable systems, Neural interfaces - Microelectrode technology, Neural interfaces - Tissue-electrode interface
Abstract: Developments in microelectronics, micromachining, information and communication technologies led to novel devices that caused huge societal changes in many areas of life. In medicine, application of electrically active implants to treat neurological disorders and neural diseases emerged but technology stayed far behind the complexity of consumer electronics. Possibilities, challenges and threats to transfer microsystems technologies into neural microimplants will be introduced and discussed with respect to technical and biological target specifications. Robustness and longevity are key factors in translation of first results from neuroscientific studies into human clinical trials. Optical stimulation and recording as well as monitoring of neurotransmitters are desirable features in modern neural probes. Examples of miniaturized electrode arrays on flexible substrates are presented for several applications and stability and longevity are discussed as well as the integration of materials for neuro-transmitter monitoring and components for optical stimulation.

Keywords: Sensory neuroprostheses - Somatosensory
Abstract: The somatosensory system intertwines numerous neural populations with distinct functionalities, whose specific neuromodulation would be beneficial for the alleviation of chronic pain. We report on our effort to engineer light delivery implant system to selectively study the somatosensory system.

Keywords: Brain functional imaging
Abstract: Neuroscience has a longstanding goal of large-scale imaging of neuronal activity at the whole mammalian brain level. Due to the large performance gap between highly invasive optical microscopy and whole-brain macroscopy of indirect neuronal activity focusing on haemodynamics and metabolism, simultaneous real-time and direct imaging of large scale neuronal activity remains difficult. In this work we present a new functional optoacoustic neuro-tomography (FONT) method for both ex vivo and completely non-invasive in vivo functional imaging of neuronal activity across mouse brain using genetically encoded calcium indicators. The approach enables whole-brain 3D snapshots in high resolution, mapping neuronal activity with a single optoacoustic excitation. It can further distinguish stimulus evoked slow-haemodynamic activity and real time calcium activity from strong hemoglobin background absorption. The method can bridge the gap between whole-brain macroscopy and localized optical microscopy via direct neuronal imaging at depths and spatiotemporal resolutions not covered by other neuroimaging modalities.

Keywords: BioMEMS/NEMS - Tissue engineering and biomaterials, Micro- and nano-technology, Nano-bio technology design
Abstract: High-density electrodes with the nano feature size greatly enhance resolution and specificity during intracortical microstimulation. In this viewpoint, we fabricated and developed high-density nanowire (NW) electrodes, ~ 2.45×109 / cm2 that could directly stimulate specific region of the cortex with low current amplitude in the range of 120-180 μA. The proposed nanowire electrodes will help expand the capabilities of microstimulation and extend the range of dysfunctions that can be treated using microstimulation technique.

Keywords: Microfluidic techniques, methods and systems, Micro- and nano-technology
Abstract: Exosomes are nanosized extracellular vesicles that play a significant role in cell-cell communication. Recently, there is significant interest in exosome-related fundamental research, especially subgroups of exosomes as potential biomarkers for cancer diagnosis and prognosis. In this paper, we report a new size selective isolation method via elastic lift force and nanomembrane filtration and demonstrated the liposome recovery rate of 92.5% from a mixture solution of 1 μm polystyrene beads, 100 nm liposomes and proteins as a proof of concept for exosome isolation. This single microfluidic platform offers an improved approach with short processing time (< 2 hours) and low cost, and shows potential broad applicability to cancer biomarker studies.

Keywords: Microfluidic techniques, methods and systems, Microfluidic applications
Abstract: Plasma skimming is the phenomenon whereby the discharge hematocrit is lower than feed hematocrit naturally occurring in the microvessels with Poiseuille flow. It has been studied in Poiseuille flow extensively. Besides, plasma skimming has also been observed and investigated in blood pumps due to its potential to prevent hemolysis by skimming blood cells out of the small gap. However, whether plasma skimming occurs in blood pumps with whole blood has not been verified. Additionally, the independent influence of rotational speed and gap size has not been clarified. Therefore, in order to lay the foundation of applying plasma skimming to the development of blood pumps and also investigate the influence of rotational speed and gap size on plasma skimming respectively, we designed a simplified geometric device which not only imitates the flow inside clearances of blood pumps, but also provides different rotational speed and gap size conditions. We first conducted the verification tests of plasma skimming using whole blood with an initial hematocrit of 44% and the gap size was varied from 10 µm to 240 µm with 10 µm interval. The plasma skimming was verified occurring when the gap was less than 70 µm at a rotational speed of 800 rpm. Since plasma skimming was confirmed, we employed 30% hematocrit blood and performed the following tests to evaluate the influence of rotational speed of 600 rpm, 700 rpm, and 800 rpm respectively. As a result, the hematocrit of sampled blood declined as the rotational speed increased from 600 rpm to 800 rpm. And there was the lowest hematocrit of 16% when the gap was adjusted to 50 µm gap size at 800 rpm. This study further promotes the possibility of applying plasma skimming to the blood pumps with higher hemocapability.

Keywords: Micro- and nano-sensors, Micro- and nano-technology, Microfluidic techniques, methods and systems
Abstract: DNA quantification and characterization are of critical importance in disease diagnosis and clinical analysis, while label-free technology greatly simplifies the sensing protocol as it eliminates the extra step for attaching the indicator to DNA strands. In this work, we present a novel label-free DNA detection methodology based on electrical frequency-dependent impedance. The impedance of DNA strands conjunct with streptavidin-coated magnetic beads was measured at 8 different frequencies using an electrical impedance sensor integrated on a chip. Different concentrations of 300 bp double-stranded DNA samples were used to validate our sensor. The minimum DNA amount that could be successfully detected was 0.77 ng (3.9 amol). Detecting DNA fragments using our sensor could be further reduced from currently 20 minutes to under 15 minutes.

Keywords: Nano-bio technology design, Micro- and nano-technology
Abstract: In the last few years, the use of nanoparticles for therapeutic applications has attracted the interest of many scientists, who are looking for effective methods to target nanoparticles linked to drugs directly to the diseased organs. Among them, magnetic targeting consists of magnetic systems (magnets or coils) which can impress high gradient magnetic fields and then magnetic forces on the magnetic nanoparticles. Despite some studies have reported an effective improvement in drug delivery by using this technique, there is still a paucity of studies able to quantify and explain the experimental results. In this scenario, “in silico” models allow to analyze and compare different magnetic targeting systems in their ability to generate the required magnetic field gradient for specific human targets. In this paper we then evaluated, by means of computational electromagnetics techniques, the attitude of various ad-hoc designed magnetic systems in targeting the heart tissues of differently aged human anatomical models.

Keywords: Biomimetic materials
Abstract: Although in vitro studies are frequent for the study of traumatic spine and spinal cord injuries, few include the spinal cord due to its prompt post-mortem decay. Several materials have been proposed to mimic the spinal cord behaviour, but none matched its mechanical properties under transverse compression, which is vital for the study of burst fractures and other injury mechanisms leading to spinal cord compression. In this study, a new material named Soma Foama 15 (Reynolds Advanced Material, USA) was used to manufacture three spinal cord surrogates at 3 mixing ratios of elastomer to catalyst (1:1, 2:1 and 3:1) and tested at three different strain rates (0.5, 5 and 50 .s-1). The mixing ratio 3:1 presents a mechanical behaviour comparable to that of the porcine spinal cord at each of these strain rates, making the surrogate a valid substitute up to 75 % of transverse compression.

Keywords: Health Informatics - Virtual reality in medicine, Health Informatics - Healthcare modeling and simulation, Health Informatics - Human factors (ergonomics) in health information systems
Abstract: Simulation is a powerful learning tool, as it allows gaining direct experience in a controlled and repeatable way. However, the simulation is effective when it is able to reproduce the real conditions and when the user feels him/herself immersed and present in the situation. With the aim of improving these critical points, we propose an immersive virtual reality system for first-aid handling. Specifically, we increase the visual realism of medical mannequins and the contextualization, and we add the touch feedback by mapping the real mannequin into its virtual representation. Moreover, the interaction is performed by using a virtual representation of the users own hands by allowing a more realistic execution of tasks. The results show a good accuracy in the mapping between the real and the virtual mannequin, and a high degree of presence for both the control group and the medical one. These results and the low values of simulator sickness reported during the experiment are a good starting point for the use of the proposed mixed reality system in simulation scenarios.

Keywords: General and theoretical informatics - Artificial Intelligence, General and theoretical informatics - Machine learning, Health Informatics - Health information systems
Abstract: Regular nutrient intake monitoring in hospitalised patients plays a critical role in reducing the risk of disease-related malnutrition (DRM). Although several methods to estimate nutrient intake have been developed, there is still a clear demand for a more reliable and fully automated technique, as this could improve the data accuracy and reduce both the participant burden and the health costs. In this paper, we propose a novel system based on artificial intelligence to accurately estimate nutrient intake, by simply processing RGB depth image pairs captured before and after a meal consumption. For the development and evaluation of the system, a dedicated and new database of images and recipes of 322 meals was assembled, coupled to data annotation using innovative strategies. With this database, a system was developed that employed a novel multi-task neural network and an algorithm for 3D surface construction. This allowed sequential semantic food segmentation and estimation of the volume of the consumed food, and permitted fully automatic estimation of nutrient intake for each food type with a 15% estimation error.

Keywords: General and theoretical informatics - Ontology, Health Informatics - Personal health records, Health Informatics - Health information system interoperability
Abstract: One of the most widely acknowledged standards in health informatics is HL7. HL7 FHIR® (Fast Healthcare Interoperability Resources) is a new HL7 standard for exchanging electronic health data. It builds upon previous HL7 data format standards, but also leverages more modern technical concepts and approaches, aiming to be more developer-friendly. We present a developed ontology that, not only represents the domain entities of a personal health record (PHR) focusing on tele-health and integrated care, but also stores the actual data as instances of the defined ontology classes. Inspired and based on HL7 FHIR we defined a methodology for representing FHIR data types and FHIR resources in OWL and we have extended or restricted the resources to match specific domain needs. Additionally, since HL7 FHIR is a developing standard, we present a methodology for maintaining backwards compatibility as the ontology is updated to match the latest definition of the standard. All the effort is represented as an OWL-DL ontology that is publicly available for reuse and extension.

Keywords: Public Health Informatics - Health risk evaluation and modeling, Health Informatics - Disease profiling and personalized treatment, Health Informatics - Personal health records
Abstract: Studies in healthcare have revealed the risk factors associated with the onset of stroke in an individual. Data mining techniques have been used to predict the occurrence of stroke based on these factors by using patients’ medical records. However, there has been limited use of electronic health records to study the inter-dependency of different risk factors of stroke. In this paper, we perform an analysis of patients’ electronic health records to identify the impact of risk factors on stroke prediction. We also provide benchmark performance of the state-of-art machine learning algorithms for predicting stroke using electronic health records.

Keywords: Health Informatics - Health data acquisition, transmission, management and visualization
Abstract: Fetal magnetoencephalography (fMEG) is a method to record human fetal brain signals in pregnant mothers. Nevertheless the amplitude of the fetal brain signal is very small and the fetal brain signal is overlaid by interfering signals mainly caused by maternal and fetal heart activity. Several methods are used to attenuate the interfering signals for the extraction of the fetal brain signal. However currently used methods are often affected by a reduction of the fetal brain signal or redistribution of the fetal brain signal. To overcome this limitation we developed a new fully automated procedure for removal of heart activity (FAUNA) based on Principal Component Analysis (PCA) and Ridge Regression. We compared the results with an orthogonal projection (OP) algorithm which is widely used in fetal research. The analysis was performed on simulated data sets containing spontaneous and averaged brain activity. The new analysis was able to extract fetal brain signals with an increased signal to noise ratio and without redistribution of activity across sensors compared to OP. The attenuation of interfering heart signals in fMEG data was significantly improved by FAUNA and supports fully automated evaluation of fetal brain signal.

Keywords: Sensor Informatics - Multi-sensor data fusion, Health Informatics - Quality of service, trust, security, Health Informatics - Technology and services for home care
Abstract: Digital watermarking has been widely recognized as an effective tool for embedment of auxiliary data in the host record. This paper presents a new method of watermarking using lead-to-lead difference of values in the baseline of the host electrocardiogram. The method starts with delineation of the baseline and uses Kirchoff voltage law or interpolation to predict any selected lead from the remaining ones. Next, the difference between the predicted and actual value is considered as noise and subjects to measurement of level and distribution in the time frame of baseline. The watermark with patient data or results of accompanying measurements is coded accordingly to mimic the noise. Replacement of the baseline noise with the watermark data ends the process. With 12-lead CSE files and respective reference borders of PQ and TP segments, the capacity of watermark achieved 3875 bits per second, while the diagnostic value of the ECG remains untouched.

Keywords: Other smart implanted systems
Abstract: During development and pathological processes, cells in the central nervous system (CNS) are highly motile. Despite the fact that cell motion is driven by forces, our current understanding of the physical interactions between CNS cells and their environment is very limited. We here show how nanometer deformations of CNS tissue caused by piconewton forces exerted by cells contribute to regulating CNS development and pathologies. In vitro, growth and migration velocities, directionality, cellular forces as well as neuronal fasciculation and maturation all significantly depended on substrate stiffness. Moreover, when grown on substrates incorporating linear stiffness gradients, glial cells migrated towards stiffer, while axon bundles turned towards softer substrates. In vivo time-lapse atomic force microscopy revealed stiffness gradients in developing brain tissue, which axons followed as well towards soft. Interfering with brain stiffness and mechanosensitive ion channels in vivo both led to similar aberrant neuronal growth patterns with reduced fasciculation and pathfinding errors. Importantly, CNS tissue significantly softened after traumatic injuries. Ultimately, mechanical signals not only directly impacted neuronal growth but also indirectly by regulating neuronal responses to and the availability of chemical guidance cues, strongly suggesting that chemical and mechanical signaling pathways are intimately linked, and that their interaction is crucial for neuronal development.

Keywords: Smart implanted neurostimulation systems
Abstract: The older the population grows, the more medical implants for various indication areas are required and the more often they have to be replaced during the course of therapy. Many research initiatives cope with these challenges. We describe the concept, requirements, and architecture of a low energy, multi-purpose fully implantable preclinical neurostimulator applicable for small rodents as well as humans, prospectively.

Keywords: Smart implanted neurostimulation systems
Abstract: Recent work in the field of neural interfaces shows a trend to include an increasing number of recording channels per ASIC. This imposes strict area and power limits on each individual channel, which typically aggravates noise issues. To maintain State-of-the-Art (SoA) performance, various architectures and design techniques have gained popularity, especially chopper-stabilization and bandwidth adaptation. This paper reviews the design trade-offs associated with such techniques.

Keywords: Smart implanted neurostimulation systems
Abstract: A computational model is presented for developing closed-loop control strategies for deep brain stimulation (DBS) for Parkinson’s disease. The computational model incorporates the electric field distribution during DBS, with antidromic and orthodromic activation of corticofugal projection neurons, generation of oscillatory activity within the cortico-basal ganglia network, and simulation of the subthalami nucleus local field potential (LFP). The results provide a proof-of-concept for PID-type control based on oscillatory beta-band LFP activity.

Keywords: Medical devices interfacing with the brain or nerves
Abstract: Glioblastoma (GBM) is the most common malignant brain tumor and accounts for 70% of all primary brain tumors in adults. Despite aggressive multi-modal therapy including surgery, radiation, and chemotherapy, the prognosis remains poor with a median survival of around 2 years. Tumor-treating fields (TTFields), is a new frontier in cancer therapy, and has been recently approved for the treatment of GBM. We will discuss recent neuroimaging advances including novel physiologic and metabolic neuroimaging techniques and their role in monitoring treatment related temporal characteristics and assessing response to this unique treatment modality.

Keywords: Medical devices interfacing with the brain or nerves, Clinical engineering, Computer modeling for treatment planning
Abstract: Measurements of the electric properties of human brain tumor tissue in the frequency range of 20-1000 kHz are reported. The measurements were performed on excised human tumor samples, acquired during surgery and measured immediately after dissection. Samples were acquired from over fifty patients with tumors of different histology and malignancy grade. The measurements reveal that the electric properties of brain tumors are highly heterogeneous and depend on histological class and malignancy grade. These results may allow more precise modelling of the interaction of electric fields with tumor tissue when modelling applications such as Tumor Treating Fields.

Keywords: Clinical engineering, Medical devices interfacing with the brain or nerves, Computer modeling for treatment planning
Abstract: Tubulin is a key globular protein which is used to assemble into cylindrical structures called microtubules present in all eukaryotic cells including glia. Microtubules have unusual conductive properties which have been studied using various techniques indicating their ability to propagate ionic signals over micrometer distances. We have measured electrical properties of microtubules in buffer solutions of different ionic strengths and tubulin concentrations. As a result of these measurements, the values of hitherto poorly understood conduction properties such as conductivity itself, capacitance and inductance have now been rather precisely determined. These parameter values compare reasonably well to earlier approximate theoretical predictions and allow for the construction of an accurate cable model of signal propagation along microtubules both in vitro and in vivo. Of particular importance is the ability to estimate the resonance of these bioelectric elements when exposed to external electric fields such as TTFields. Moreover, we can now determine how length dependence of microtubules, ionic composition of the buffer solution (or the cytoplasm) and even the geometry of the microtubule network affects the resonance condition. This information can be used to design optimal frequency, intensity and direction of the applied AC electric field in order to achieve highest sensitivity of the cellular system. Estimates of the energy dissipated by a single microtubule and an ensemble of microtubules in a cell will be provided and compared to the typical physiological energy production values. This indicates that Ohmic dissipation of heat energy via microtubules in an AC field exposed cell is a significant factor in the elucidation of the mechanism of action of TTFields. We also provide a frequency dependence of the effective electrical circuit composed of microtubules as a function of ionic concentration and microtubule numbers as well as their geometrical organization.

Keywords: Neural stimulation (including deep brain stimulation), Neuromodulation devices
Abstract: We present some results pertinent to direct quasistatic EM modeling of a group of cortical pyramidal neurons with boundary element fast multipole method.

Keywords: Models of therapeutic devices and systems
Abstract: Tumor Treating Fields (~200 kHz AC electric fields) kill tumor cells with no side effects on normal cells but their underlying mechanism is not yet known. We present the first complete theory of how TTFields may work, tying together 1) electromagnetic effects, 2) intra-cellular structures, and 3) signaling pathways. Key empirical results used in numerical analysis, molecular dynamics and finite element modeling implicate polarizable tumor cell sub-structures. We suggest how components of the theory may be proven, and present predictions with implications for increasing TTFields’ efficacy.

Keywords: Models of therapeutic devices and systems
Abstract: We used a computational model to predict the spatial and temporal variation of the temperature in the head during TTFields therapy. We found that the device must operate in an intermittent fashion to ensure transducers’ temperatures remain below 41°C. Underneath the warmest transducer the scalp reached 42 °C, the skull 39.5 °C, CSF 39 °C and the brain 38.2 °C. The thermal dose predicted in the brain may affect brain function, but this has not been observed in clinical trials.

Keywords: Cardiovascular, respiratory, and sleep devices - Diagnostics, Sleep - Obstructive sleep apnea
Abstract: A number of automated apnoea hyponoea index (AHI) prediction algorithms first divided the signal of interest into epochs, make a prediction for each epoch, determine the number of epoch predictions per hour and map this to an AHI value. An underlying assumption of this approach is a smooth relationship between the apnoea plus hypopnea duration and the AHI value. In this study we investigate this relationship to establish if this assumption impacts the final system. We compare two models: one which divides the duration by recording time, and the second which divides the duration by sleep time. Data for study was obtained from 200 scored overnight polysomnogram recordings. Our results show that the relationship is a power-law distribution with an exponent of 0.9 and a multiplicative noise term. Analysis of the variance of the noise term revealed that algorithms that scale the apnea duration by the recording time will have an inherent 37% error in the AHI estimate, while algorithms that scale by sleep time will have an inherent 25% error in the AHI estimate. An ROC analysis of the duration based models at the clinically important values of AHI 5 and 15 revealed an area under the ROC of greater than 0.96 suggesting that the epoch approach has little impact on the ability of the system to correctly identify normal and apnoeic patients.

Keywords: Cardiovascular, respiratory, and sleep devices - Smart systems, Cardiovascular, respiratory, and sleep devices - Nearables
Abstract: This document presents an algorithm for a non-obtrusive recognition of Sleep/Wake states using signals derived from ECG, respiration, and body movement captured while lying in a bed. As a core mathematical base of system data analytics, multinomial logistic regression techniques were chosen. Derived parameters of the three signals are used as the input for the proposed method. The overall achieved accuracy rate is 84% for Wake/Sleep stages, with Cohen’s kappa value 0.46. The presented algorithm should support experts in analyzing sleep quality in more detail. The results confirm the potential of this method and disclose several ways for its improvement.

Keywords: Sleep - Obstructive sleep apnea, Sleep - Periodic breathing & central apnea
Abstract: Sleep is a very important part of life. Lack of sleep or sleep disorder can cause a negative impact on day to day life and can have long term serious consequences. In this work, we propose an end-to-end trainable neural network for automated sleep arousal scoring. The network consists of two main parts. Firstly, a trend statistics network computes the moving average of the filtered signals at different scales. Secondly, we propose a channel invariant EEG network to detect the arousals in any Electroencephalography (EEG) channel. Finally, we combine the features from various channels through a convolution network and a bi-directional long short-term memory to predict the probability of arousal. Further, we propose an objective function that uses only respiratory effort related arousal (RERA) and non-arousal regions to optimize the network. We also propose a method to estimate the respiratory disturbance index (RDI) from the probability predicted by the network. Evaluation on Physionet Challenge 2018 database shows that the proposed method detects RERA with mean area under the precision-recall curve (AUPRC) of 0.50 in a 10-fold cross validation setup. The mean absolute error of RDI prediction is 6.11, while a two-class RDI severity prediction yields a specificity of 75% and a sensitivity of 83%.

Keywords: Sleep - Periodic breathing & central apnea, Cardiac mechanics, structure & function - Heart failure, Respiratory transport, mechanics and control - Periodic breathing
Abstract: Sleep disordered breathing is one of the most common comorbidities in chronic heart failure (HF). Central sleep apnea (CSA) with Cheyne-Stokes-respiration pattern is supposed to emerge with deterioration of heart failure. CSA is known as negative predictor for mortality in HF. Actually, guidelines on heart failure recommend optimization of HF-treatment. We report a case of a 65-year old male patient of caucasian ethnicity with dilative cardiomyopathy with left bundle branch block which was first diagnosed in 2003. After optimization of HF-therapy with ARNi, we could not prove a resolution of CSR-CSA in our patient. However, we could show relevant phenotypic changes of CSA-CSR. Thus, cycle length decreased, possibly reflecting an improvement of cardiac function as proved in a slightly increased LV-EF. This goes along with an improvement of oxygen pulse (VO2/Hf) in CPET which is considered a surrogate marker for cardiac stroke volume. While cycle and apnea length decreased, we could see an increase of AHI. In fact, we do not think, that this increase is reflecting a worsening of the conditions, as the patient reported of an improvement of his symptoms and sleep quality. Under treatment with ARNi basal oxygen saturation (SaO2) improved with reduction of time below 90% of oxygen saturation (T90) down to 0minutes. T90 was shown to be a prognostic factor in HF-patients with SDB. Under ARNi also circulatory delay decreased, reflecting an improved cardiocirculatory function. Furthermore, we could prove a lowered loop gain suggesting a reduced likelihood for evolution of CSA-CSR. This could explain a total resolution at least in some HF-patients with CSA-CSR, especially in those with a more preserved cardiovascular function. Interestingly, with ARNi not only cardiocirculatory parameters improved. A lowered loop gain and a reduced breathing frequency could be a sign for changes of central chemosensitivity.

Keywords: Sleep - Cardiovascular & Metabolic consequences of sleep disorders, Sleep - Sleep apnea therapy, Sleep - Obstructive sleep apnea
Abstract: Abstract— monitoring apnea-induced cerebral blood flow oscillations is of importance for assessing apnea patient brain health. Using an autoregressive moving average model, peak and trough values of cerebral blood flow were estimated from a concurrently recorded forehead photoplethysmography signal. Preliminary testing of the method in 7 subjects (4 F, 32±4 yrs., BMI 24.57±3.87 kg/m2) using a breath hold paradigm for simulating apnea shows that maximum mean and standard deviation of the prediction error was -1.10±8.49 cm/s and the maximum root mean squared of the error was 8.92 cm/s

Keywords: Sleep - Obstructive sleep apnea, Sleep - Cardiovascular & Metabolic consequences of sleep disorders, Cardiovascular and respiratory system modeling - Sleep-cardiorespiratory Interactions
Abstract: Obstructive Sleep Apnea severity is commonly determined after a sleep polysomnographic study by the Apnea-Hypopnea Index (AHI). This index does not contain information about the duration of events, and weights apneas and hypopneas alike. Significant differences in disease severity have been reported in patients with the same AHI. The aim of this work was to study the effect of obstructive event type and duration on the subsequent oxygen desaturation (SaO2) by mixed-effects models. These models allow continuous and categorical independent variables and can model within-subject variability through random effects. The desaturation depth dSaO2, desaturation duration dtSaO2 and desaturation area SaO2A were analyzed in the 2022 apneas and hypopneas of eight severe patients. A mixed-effects model was defined to account for the influence of event duration (AD), event type, and their interaction on SaO2 parameters. A two-step backward model reduction process was applied for random and fixed effects optimization. The optimum model obtained for dtSaO2 suggests an almost subject-independent proportional increase with AD, which did not significantly change in apneas as compared to hypopneas. The optimum model for dSaO2 reveals a significantly higher increase as a function of AD in apneas than hypopneas. Dependence of dSaO2 on event type and duration was different in every subject, and a subject-specific model could be obtained. The optimum model for SaO2 A combines the effects of the other two. In conclusion, the proposed mixed-effects models for SaO2 parameters allow to study the effect of respiratory event duration and type, and to include repeated events within each subject. This simple model can be easily extended to include the contribution of other important factors such as patient severity, sleep stage, sleeping position, or the presence of arousals.

Keywords: Image segmentation
Abstract: Segmentation for tracking surgical instruments plays an important role in robot-assisted surgery. Segmentation of surgical instruments contributes to capturing accurate spatial information for tracking. In this paper, a novel network, Refined Attention Segmentation Network, is proposed to simultaneously segment surgical instruments and identify their categories. The U-shape network which is popular in segmentation is used. Different from previous work, an attention module is adopted to help the network focus on key regions, which can improve the segmentation accuracy. To solve the class imbalance problem, the weighted sum of the cross entropy loss and the logarithm of the Jaccard index is used as loss function. Furthermore, transfer learning is adopted in our network. The encoder is pre-trained on ImageNet. The dataset from the MICCAI EndoVis Challenge 2017 is used to evaluate our network. Based on this dataset, our network achieves state-of-the-art performance 94.65% mean Dice and 90.33% mean IOU.

Keywords: Ultrasound imaging - Interventional
Abstract: Over the past decade, Freehand 3D Ultrasound(US) reconstruction using only image information has become a widely researched topic because it eliminates the need for an external tracking system and provides real-time volumetric information. But most of the state-of-art methods are inhibited by their inability to find a simple and robust similarity metric that could learn and estimate the spatial transformation between two US slices in a US sweep. In this work, we propose a novel similarity metric (TexSimAR), which computes the similarity value between two consecutive US images by correlating the parametric representation of the image-texture instead of the image itself. The purpose of this approach is to capture and compare the dynamics in the texture characteristics of two US images. We modelled these dynamics using a parametrical auto-regressive (AR) model. Experiments were performed on forearm datasets of three subjects. For every pair of consecutive US slices, we computed our TexSimAR similarity value and out-of-plane transformation from the ground truth to train a SVM based regression model, which was then used to predict the out-of-plane transformation with the similarity value as input. The proposed method shows promising results with predictions better than state-of-the-art methods even with one eighth of training data compared to other methods in the literature.

Keywords: Ultrasound imaging - Other organs, Ultrasound imaging - High-frequency technology, Image segmentation
Abstract: A device and technique to acquire and construct 3D ultrasound volumes of the skin of the hand and arm were developed. The Repeated Skin Thickness Measurement (RSTM) Device moves a high frequency ultrasound probe linearly in 3 axes in a water tank and images a submerged arm. These images are combined into an ultrasound volume, the skin layer segmented, and the thickness extracted. One particular application is measuring progression of scleroderma, a skin thickening disease. The current ultrasound-based scleroderma diagnostic processes assess skin thickness based on a single ultrasound image taken by a clinician holding the ultrasound probe, resulting in low measurement repeatability. The imagery that results from the instrumentation and analysis presented here can be used to create quantitative maps of skin thickness, to monitor the progression of skin-thickening diseases, and to observe the structures of tendons, ligaments, and the other soft tissue of the hand.

Keywords: Magnetic resonance imaging - Dynamic contrast-enhanced MRI
Abstract: We investigated if blood flow changes induced through the presence of a stent could be detected using in vitro dynamically scaled 4D Phase-Contrast Magnetic Resonance Imaging (PC-MRI). Using idealized and patient-specific left main coronary artery bifurcations, we 3D-printed the dynamically large scaled geometries and incorporated them into a flow circuit for non-invasive acquisition with a higher effective spatial resolution. We tested the effects of using non-Newtonian and Newtonian fluids for the experiment. We also numerically simulated the same geometries in true scale for comparison using computational fluid dynamics (CFD). We found that the experimental setup increased the effective spatial resolution enough to reveal stent induced blood flow changes close to the vessel wall. Non-Newtonian fluid replicated all of the flow field well with a strong agreement with the computed flow field (R2 > 0.9). Fine flow structures were not as prominent for the Newtonian compared to non-Newtonian fluid consideration. In the patient-specific geometry, arterial non-planarity increased the difficulty to capture the near wall slow velocity changes. Findings demonstrate the potential to dynamically scale in vitro 4D MRI flow acquisition for micro blood flow considerations.

Keywords: Magnetic resonance imaging - Interventional MRI, Image visualization
Abstract: Conventional MR techniques have difficulty to accurately localize the stent position and access the stent restenosis because of the effects of susceptibility and radiofrequency (RF) shielding artifacts caused by stent mesh. Previous studies have demonstrated that a susceptibility-based positive contrast MR method exhibits excellent efficacy for visualizing MR compatible metal devices by taking advantage of their high magnetic susceptibility. However, the method is not evaluated in the visualization of stents. Therefore, the purpose of this study is to prospectively assess whether the susceptibility-based positive contrast method can be used to visualize the nitinol stents, with the comparison of two typical MR positive contrast techniques, i.e., susceptibility gradient mapping using the original resolution (SUMO) and the gradient echo acquisition for super-paramagnetic particles with positive contrast (GRASP). Experiment results showed that the susceptibility-based method provided better visualization and more precise localization of the stent than SUMO and GRASP.

Keywords: X-ray - Interventional radiology, X-ray imaging applications, Image reconstruction and enhancement - Filtering
Abstract: — The aim of this study is to propose a new semi-automated method for three-dimensional (3D) reconstruction of coronary bifurcations arteries using X-ray Coronary Angiographies (CA). Considering two monoplane angiographic views as the input data, the method is based on a four-step approach. In the first step, the image pre-processing and the vessel segmentation is performed. In the second step the 3D centerline is reconstructed by implementing the back-projection algorithm. In the third step, the lumen borders are reconstructed around the centerline to result to the fourth step, during which the 3D point cloud of the side branch is adjusted to the main branch, to produce the final 3D model of the coronary bifurcation artery. Imaging data from seven patients (pre and post-stenting) were reconstructed in the 3D space. The validation of the proposed methodology was based on the comparison of the 3D model with the 2D CA. Blood flow simulations were performed both for the vessels before and after the angioplasty procedure. Decreased Endothelial Shear Stress (ESS) values were observed for the vessels after the Percutaneous Transluminal Coronary Intervention (PTCI).

Keywords: Portable miniaturized systems, Sensor systems and Instrumentation, Bio-electric sensors - Sensor systems
Abstract: This paper presents a new multi-modality readout system for potentiometric electrochemical sensors. The design adopts a pulse modulation at the gate and drain of the Bio-FET sensors to reduce the effects of charge accumulation between the surface of the electrodes and the ion analytes. The adjustable duration and amplitude of stimuli signals provide flexibility for different biosensing applications and a wide range of detectable concentration. Also, an oscillator-based architecture is proposed for digitization and integration. The counting time can be adjusted to enhance the resolution of the readout system. The proposed potentiometric sensing system was tested with 0.1-10 mM Potassium Ferricyanide (K3[Fe(CN)6]), and the results are interpreted in the micro-LCD on the board. The design offers the opportunity for a handheld medical device with fast and real-time monitoring of biomarkers and ion analytes.

Keywords: Wearable body-compliant, flexible and printed electronics, Physiological monitoring - Novel methods, Wearable sensor systems - User centered design and applications
Abstract: Inkjet-printing is a well-known technology that has been recently revalued for the production of flexible sensors and biosensors, thank to the use of engineered nanostructured inks. In a previous work, we developed a general-purpose biosensors printing platform that made use of a simple and low-cost consumer printer and allowed to produce customized flexible electrodes with an ultra-short concept-to-prototype time, without requiring any sintering step. In this study we show the preliminary results about the use of such a newly easily-accessible, low-cost inkjet-based platform to produce flexible and fully customizable electrodes for reliable surface electromyographic (sEMG) recordings.

Keywords: New sensing techniques, Physiological monitoring - Novel methods
Abstract: In patients with life-threatening illnesses, secretion and excretion of lactate is impaired which leads to a build-up of lactate levels in the body. In critical care units, the changes in lactate levels are measured invasively using intermittent blood gas analysers. Continuous monitoring of these changes can, however, be used for early prognosis and to guide therapy. Currently, there is no means to continuously measure lactate levels, particularly non-invasively. The motivation of this paper is to understand the interaction of lactate with light in the Near Infra-Red (NIR) region of the electromagnetic spectrum. This will enable an opportunity to explore the possibility of finding a non-invasive sensing technology to continuously monitor lactate. In-vitro studies were performed in solution samples with varying concentration levels of sodium lactate in an isotonic Phosphate Buffer Solution (PBS) of a constant pH (7.4). These samples were prepared using solution stoichiometry and spectra of each sample were taken using a state-of-the-art spectrometer in the NIR region. The spectra was then analysed qualitatively by 2D correlation analysis which prompted the regions of interest. Further analysis in these regions by linear regression in four randomly selected wavelengths shows bathochromic shifts, which indicates systematic variations in the changes of the concentrations of lactate.

Keywords: Physiological monitoring - Instrumentation
Abstract: The use of multiple surface EMG electrodes (High-Density surface EMG – HD-sEMG) allows the extraction of anatomical and physiological information either at the muscle or at the motor unit level with applications in several fields ranging from clinical neurophysiology to the control of prosthetic devices. These applications need to acquire monopolar sEMG signals free from power line interference arising from the capacitive coupling between the subject, the acquisition system and the power line. The aim of this work is to provide a common mode analysis of the detection system used to collect monopolar sEMG signals, characterizing different configuration of the reference electrodes leading to different behaviors in terms of immunity to the power line interference. Based on the experimental results, a new impedance-controlled HD-sEMG signal amplifier is proposed and discussed.

Keywords: Physiological monitoring - Novel methods, Portable miniaturized systems, New sensing techniques
Abstract: Out-of-clinic, continuous monitoring of vital signs is envisaged to become the backbone of future e-health. The emerging wrist worn devices have already proven to be a success in the measurement of pulse, however, a susceptibility to artefacts and missing data caused by regular motion in everyday activities, and the inability to continuously acquire the electrocardiogram call into question the utility of this technology in future e-Health.With this in mind, the head, and in particular the ear canals,have been investigated as possible locations for wearable devices.The ears offer a stable position relative to the vital signs during everyday activities, such as sitting, walking, running and sleeping,as well as being a practical and widely accepted base for wearable accessories. This all suggests that the ear canals are a most natural location for physiological sensing in the community. This work addresses the feasibility of recording the ECG from the ear canals,from a one-fits-all, user-friendly device. For rigour and clarity, we quantitatively compare the timings of the identified P-, QRS-, and T-waves within Ear-ECG and standard arm-ECG. Finally, to depict a future e-Health scenario for the Ear-ECG technology, a case study with an abnormal heart condition, the ventricular bigeminy,is presented. A comprehensive study over ten subjects demonstrates conclusively the possibility of in-ear cardiac monitoring in normal daily life.

Keywords: Wearable wireless sensors, motes and systems, Acoustic sensors and systems, Sensor systems and Instrumentation
Abstract: Detecting critical events in postoperative care and improving comfort, costs and availability in sleep assessment are two of many areas in which wearable biosignal acquisition can be a viable tool. Modern sensors as well as patch and textile integration facilitate unobtrusive biosignal acquisition, yet placing sensors at different locations across the body is still prevailing. Actigraphy and the electrocardiogram (ECG) are commonly integrated modalities. The stethoscope however, despite its wide range of applications, has been neglected from these developments. The introduction of digital stethoscopes, recently led to an objectification and increased interest in the field. We present the prototype of a wearable, Bluetooth 5.0 LE enabled multimodal sensor patch combining five modalities: MEMS stethoscope, ambient noise sensing, ECG, impedance pneumography (IP) and 9-axial actigraphy. The system alleviates the need for sensors at different body positions and enables long-term auscultation. Using high sampling rates and online synchronization, multimodal sensor fusion becomes feasible. The patch measures 70 mm x 60 mm and is attached using three 24 mm Ag/AgCl electrodes. High quality cardiac and pulmonary auscultation as well as ECG and IP acquisition are demonstrated. We derived respiration surrogates with linear correlations to a reference exceeding 0.91 and conclude that the system can be utilized in fields requiring unobtrusive yet high quality signal acquisition. Future research will include the integration of additional sensors and further size reduction.

Keywords: Time-frequency and time-scale analysis - Nonstationary processing, Physiological systems modeling - Signal processing in physiological systems
Abstract: The main objective of this study was to characterize EEG resting-state activity in 55 Alzheimer's disease (AD) patients and 29 healthy controls by means of TREND, a measure based on recurrence quantification analysis. TREND was computed from 60-second recordings of consecutive EEG activity, divided into non-overlapping windows of length 1, 2, 3, 5, 10, 15, 20 and 60 seconds. This measure was computed in the conventional EEG frequency bands (delta, theta, alpha, beta-1, beta-2 and gamma). The parameters delay (tau) and embedding dimension (m) were first optimized for every window size and frequency band under study. These embedding parameters proved to be frequency-dependent. Furthermore, 10 s epochs were set as the minimum length required to avoid spurious results. Statistically significant differences between both groups were found (p < 0.05, Mann-Whitney U-test). The groups showed differences in TREND in the theta (4-8 Hz), beta-1 (13-19 Hz) and beta-2 (19-30 Hz) frequency bands. Our results using TREND suggest that AD disrupts resting-state neural dynamics. Furthermore, these findings indicate that AD induces a frequency-dependent pattern of alterations in the non-stationarity levels of resting-state neural activity.

Keywords: Time-frequency and time-scale analysis - Nonstationary processing, Nonlinear dynamic analysis - Deterministic chaos, Signal pattern classification
Abstract: Neuromodulations as observed in the extracellular electrical potential recordings obtained from Electroencephalograms (EEG) manifest as organized, transient patterns that differ statistically from their featureless noisy background. Leveraging on this statistical dissimilarity, we propose a non- iterative robust classification algorithm to isolate, in time, these neuromodulations from the temporally disorganized but structured background activity while simultaneously incorporating temporal sparsity of the events. Specifically, we exploit the ability of correntropy to asses higher - order moments as well as imply the degree of similarity between two random variables in the joint space regulated by the kernel bandwidth. We test our algorithm on DREAMS Sleep Spindle Database and further elaborate on the hyperparameters introduced. Finally, we compare the performance of the algorithm with two algorithms designed on similar ideas; one of which is a quick, simple norm based technique while the other parallels the state-of-the-art Robust Principal Component Analysis (RPCA) to achieve classification. The algorithm is able to match the performance of the state-of-the-art techniques while saving tremendously on computation time and complexity.

Keywords: Time-frequency and time-scale analysis - Nonstationary processing, Time-frequency and time-scale analysis - Time-frequency analysis, Nonlinear dynamic analysis - Biomedical signals
Abstract: Characterizing coordinated brain dynamics present in high-density neural recordings is critical for understanding the neurophysiology of healthy and pathological brain states and to develop principled strategies for therapeutic interventions. In this research, we propose a new modeling framework called State Space Global Coherence (SSGC), which allows us to estimate neural synchrony across distributed brain activity with fine temporal resolution. In this modeling framework, the cross-spectral matrix of neural activity at a specific frequency is defined as a function of a dynamical state variable representing a measure of Global Coherence (GC); we then combine filter-smoother and Expectation-Maximization (EM) algorithms to estimate GC and the model parameters. We demonstrate a SSGC analysis in a 64-channel EEG recording of a human subject under general anesthesia and compare the modeling result with empirical measures of GC. We show that SSGC not only attains a finer time resolution but also provides more accurate estimation of GC.

Keywords: Time-frequency and time-scale analysis - Time-frequency analysis, Neural networks and support vector machines in biosignal processing and classification, Signal pattern classification
Abstract: Deep neural networks enable learning directly on the data without the domain knowledge needed to construct a feature set. This approach has been extremely successful in almost all machine learning applications. We propose a new framework that also learns directly from the data, without extracting a feature set. We apply this framework to detecting bursts in the EEG of premature infants. The EEG is recorded within days of birth in a cohort of infants without significant brain injury and born <30 weeks of gestation. The method first transforms the time-domain signal to the time–frequency domain and then trains a machine learning method, a gradient boosting machine, on each time-slice of the time–frequency distribution. We control for oversampling the time–frequency distribution with a significant reduction (<1%) in memory and computational complexity. The proposed method achieves similar accuracy to an existing multi-feature approach: area under the characteristic curve of 0.98 (with 95% confidence interval of 0.96 to 0.99), with a median sensitivity of 95% and median specificity of 94%. The proposed framework presents an accurate, simple, and computational efficient implementation as an alternative to both the deep learning approach and to the manual generation of a feature set.

Keywords: Time-frequency and time-scale analysis - Time-frequency analysis, Physiological systems modeling - Signal processing in simulation, Time-frequency and time-scale analysis - Nonstationary processing
Abstract: Electroencephalographam (EEG) monitoring of neural activity is widely used for identifying underlying brain states. For inference of brain states, researchers have often used Hidden Markov Models (HMM) with a fixed number of hidden states and an observation model linking the temporal dynamics embedded in EEG to the hidden states. The use of fixed states may be limiting, in that 1) pre-defined states might not capture the heterogeneous neural dynamics across individuals and 2) the oscillatory dynamics of the neural activity are not directly modeled. To this end, we use a Hierarchical Dirichlet Process Hidden Markov Model (HDP-HMM), which discovers the set of hidden states that best describes the EEG data, without a-priori specification of state number. In addition, we introduce an observation model based on classical asymptotic results of frequency domain properties of stationary timeseries, along with the description of the conditional distributions for Gibbs sampler inference. We then combine this with multitaper spectral estimation to reduce the variance of the spectral estimates. By applying our method to simulated data inspired by sleep EEG, we arrive at two main results: 1) the algorithm faithfully recovers the spectral characteristics of the true states, as well as the right number of states. 2) the incorporation of the multitaper framework produces a more stable estimate than traditional periodogram spectral estimates.

Keywords: Time-frequency and time-scale analysis - Time-frequency analysis, Signal pattern classification, Data mining and processing in biosignals
Abstract: Maintaining anesthetic states using automated brain-state prediction systems is expected to reduce drug overdosage and associated side-effects. However, commercially available brain-state monitoring systems perform poorly on drug-class combinations. We assume that current automated brain-state prediction systems perform poorly because they do not account for brain-state dynamics that are unique to drug-class combinations. In this work, we develop a k-nearest neighbors model to test whether improvements to automated brain-state prediction of drug-class combinations are feasible. We utilize electroencephalogram data collected from human subjects who received general anesthesia with sevoflurane and general anesthesia with the drug-class combination of sevoflurane-plus-ketamine. We demonstrate improved performance predicting anesthesia-induced brain-states using drug-specific models.

Keywords: Ultrasound imaging - Cardiac, Image segmentation, Cardiac imaging and image analysis
Abstract: The assessment of the severity of arterial stenoses is of utmost importance in clinical practice. Several image modalities invasive and non-invasive are nowadays available and can be utilized for the 3-dimensional (3D) reconstruction of the arterial geometry. Following our previous study, the present study was conducted to further strengthen the evaluation of three reconstruction methodologies, namely: (i) the Quantitative Coronary Analysis (QCA), (ii) the Virtual Histology Intravascular Ultrasound VH-IVUS-Angiography hybrid method and (iii) the Coronary Computed Tomography Angiography (CCTA). Data from 13 patients were employed to perform a quantitative analysis using specific metrics, such as, the Mean wall shear stress (mWSS), the Minimum lumen diameter (MLD), the Reference vessel diameter (RVD), the Degree of stenosis (DS%), and the Lesion length (LL). A high correlation was observed for the mWSS factor between the three reconstruction methods, especially between the QCA and CCTA (r=0.974, P<0.001).

Keywords: Ultrasound imaging - Cardiac, Ultrasound imaging - Doppler, Image reconstruction and enhancement - Parametric image reconstruction
Abstract: Echodynamography (EDG) is a computational method to deduce two-dimensional (2D) blood flow vector from conventional color Doppler ultrasound image by considering that the blood flow is divided into vortex and base flow components. Left ventricular (LV) vortices indicate cardiac flow status influenced by LV wall motion. Thus, quantitative assessment of LV vortices may become new and sensitive parameters for cardiac function. In the present study, quantitative parameters of LV vortices such as vortex index, vortex size, and Reynolds number were calculated and relation between each parameter was assessed. Six healthy volunteers and three patients with myocardial infarction (MI) who underwent color Doppler echocardiography (CDE) were involved in the study. Serial CDE images in apical three-chamber view were recorded and 2D blood flow vector was superimposed on the CDE image. Vortex index, vortex size, and Reynolds number were compared between the normal volunteers and the MI patients. The results showed that vortex index (3.09±2.06 vs. 3.34±2.33, p<0.05), vortex size (1.76±0.69 vs. 2.01±0.68, p<0.05), Reynolds number (1020±603 vs. 1312±1046, p<0.05) were significantly greater in the MI patients than in the healthy volunteers. Vortex equivalent diameter in LV showed significant positive correlation with Reynolds number (R²=0.799, y=0.001x+0.7098, p<0.05) in healthy volunteers and (R²=0.6404, y=0.0005x+1.3185, p<0.05) in MI patients. Vortex index showed positive correlation with Reynolds number (R²=0.9351, y=0.002x+0.1397, p <0.05) in healthy volunteers and (R²=0.758, y=0.0019x+0.7957, p<0.05) in MI patients. In conclusion, EDG provides information on LV hemodynamics by quantitative LV vortices parameters both in healthy volunteers and MI patients.

Keywords: Ultrasound imaging - Doppler, Ultrasound imaging - Vascular imaging, Optical imaging
Abstract: An experimental evaluation and characterization of two transparent phantoms of arms, to be used as non-invasive experimental models to measure the blood pressure continuously, was performed in an in vitro study. The first phantom was made of polyvinyl alcohol (PVA) gel and the second phantom is based on gel wax. For the first time, a PVA and gel wax phantom for the arm, respectively, were developed, including tissue, radial artery, and ulnar and radius bones. The optoacoustic parameters of various samples for both phantoms were characterized and systematically compared, obtaining a maximum acoustic transmission of T=(78.3±1.9)% at 921 nm in the PVA hydrogel, due to its high transparency and homogeneity. The gel wax phantom possesses similar optical properties as the PVA hydrogel and presents acoustic characteristics similar to those of the soft tissue. Thus, both phantoms are well-suited as in vitro models for the development of new methods in optical-acoustic medical diagnosis.

Keywords: Ultrasound imaging - Vascular imaging, Ultrasound imaging - Doppler, Ultrasound imaging - Interventional
Abstract: We have previously proposed the use of acoustic radiation force in blood vessels for therapeutic application of ultrasound. For this purpose, we have developed a blood vessel network reconstruction algorithm to fuse between B-mode and Doppler-mode volumes. However, a size of ultrasound volume was insufficient to recognize the network for treatment. Therefore, using multiple ultrasound volumes, we propose a method to extend a network with neighbor networks. First, an ultrasound volume was analyzed to extract tree-structure including the nodes and the edges. Then we configured an extension method between two tree-structures, which performs the insertion of nodes in a source tree to a target tree. Next, similarity between two networks were evaluated by introducing edge lengths in the network and the edit distance. By analyzing in vivo blood vessel of porcine liver, we confirmed that the construction of the network was reliable according to the extension with the similarity of 60% compared with CT data. We confirmed that the proposed method is effective for reconstructing blood vessel network.

Keywords: Image registration, segmentation, compression and visualization - Machine learning / Deep learning approaches, Image segmentation, Ultrasound imaging - Vascular imaging
Abstract: Ultrasound (US) is an affordable and important imaging modality in medical imaging without potential hazards for patients and medical practitioners as compared to computed tomography which uses X-rays, magnetic resonance imaging which uses a magnetic field and radio waves that could heat up the patient's body during long examinations, nuclear imaging, etc. Texture classification of anatomical structures in US images is an essential step for disease diagnosis and monitoring. In this work, we employed a convolutional neural network to segment thyroid gland in US images. This is particularly important for thyroid diseases diagnosis as they involve changes in the shape and size of the thyroid over time. The training of the Convolutional Neural Network (CNN) was not done directly on the acquired US images but on texture database that is created by dividing the thyroid US images of size 760 x 500 pixels into smaller texture patches of size 20 x 20 pixels. We obtained a Dice coefficient (DC) of 0.876 and Hausdorff Distance (HD) of 7.3 using the trained CNN that classifies the thyroid tissues as thyroid or non-thyroid. This approach was compared to the classic image processing approaches like active contours with edges (ACWE), graph cut (GC) and pixel-based classifier (PBC) which obtained a DC of 0.805, 0.745 and 0.666 respectively and Volumetric and Mass-Spring Models which obtained a HD of 11.1 and 9.8 respectively.

Keywords: Ultrasound imaging - Vascular imaging, Image feature extraction, Image segmentation
Abstract: Ultrasound (US) imaging is one of the most cost-effective imaging modality that utilizes sound waves for generating medical images of anatomical structure. However, the presence of speckle noise and low contrast in the US images makes it difficult to use for proper classification of anatomical structures in clinical scenarios. Hence, it is important to devise a method that is robust and accurate even in the presence of speckle noise and is not affected by the low image contrast. In this work, a novel approach for thyroid texture characterization based on extracting features utilizing higher order spectral analysis (HOSA) was used. A Support Vector Machine (SVM) was applied on the extracted features to classify the thyroid texture. Since HOSA is a well suited technique for processing non-Gaussian data involving non-linear dynamics, good classification of thyroid texture can be obtained in US images as they also contain non-Gaussian Speckle noise and non-linear characteristics. A final accuracy of 93.27%, sensitivity of 0.92 and specificity of 0.62 were obtained using the proposed approach.

Keywords: Design and development of robots for human-robot interaction, Surgical robotics, Computer-assisted surgery
Abstract: Laparoendoscopic single-site surgery (LESS) has been shown to reduce the invasiveness of surgery by requiring only one incision to access the abdominal cavity. However, single-site surgery integrating with a compact robotic surgical platform remains as a unique challenge. To address this challenge, we have designed a comprehensive robotic surgery platform that consists of three 6-DOF manipulators and a laparoscope camera can all be inserted into the operation field through a single 18 mm cannula holding by one 4 degrees of freedom light-weight supporting frame. Each dexterous manipulator is 5+1 degree-of-freedom (DOF), serially inserted and removable, and remotely driven by 12 actuation tendons and is composed of rigid links joined by hybrid flexure hinges. This paper introduces the compact multiple-tool single port laparoscopic robot platform for the first time. Details of the mechanical design of the trocar and manipulator including joint design and tendon routing are presented. The forward and inverse kinematics of the manipulator are also discussed along with an analysis and simulation of the cooperative workspace of two manipulators. A preliminary dynamic model of the manipulator was also constructed to study the effect of tendon-sheath friction forces at various joint configurations. Future work will illustrate the existing supporting frame mechanism for posing tools and trocar.

Keywords: Computer-assisted surgery, Human machine interfaces and robotics applications, Surgical robotics
Abstract: Surgical Skill Assessment has increased interest through which the training and objective feedback to surgeons can be given based on the task performance. In this paper, motor control features which are a part of psychomotor learning, are developed based on the camera plane coordinates of the tip of the tools from the videos of surgeons performing the Urethro-Vesicle Anastomosis (UVA) surgical task. Classification into Novices (N) and Experts (E), when compared to the manual encoding of subject expertise based on the Dreyfus model, resulted in high accuracy. Additionally, this study could form a basis for closed loop surgical training, specifically for the novitiate surgeons.

Keywords: Haptics in robotic surgery, Haptic interfaces, Clinical robots
Abstract: Remote manipulation in robotized percutaneous procedures can offer increased safety to radiologists as well as patients. Providing feedback to the radiologist on needle-tissue interactions is however mandatory in addition to medical images. A tactile feedback strategy is developed in this paper. Two types of information are considered: tissue puncture and nature of tissues. A haptic device is developed for that purpose, using a tactile display to provide information. Adequate signals are identified experimentally, with analysis of reaction times and the ability to discriminate one information from the other.

Keywords: Haptics in robotic surgery, Surgical robotics
Abstract: Transoral oral robotic surgery (TORS) presents unique challenges due to difficulty manipulating surgical instruments within the tight confines of the oral cavity. Collisions between the end effectors and anatomical structures can be visualized through the endoscope; however, instrument shaft collisions are outside of the field-of-view. Acquiring the requisite skill set to minimize these collisions is challenging due to the lack of an appropriate training platform. In this paper, we present a TORS training platform with an integrated collision sensing system and real-time haptic feedback. Preliminary testing involved the recruitment of 10 Otolaryngology residents assigned to ‘feedback’ (N=5) and ‘no feedback’ (N=5) groups. Each trainee performed three mock surgical procedures involving the resection of a tumor from the base of the tongue. Superior surgical performance was observed in the feedback group suggesting that haptic feedback will enhance the acquisition of surgical skills.

Keywords: Image guided surgery, Robot-aided surgery - Remote surgery systems / telesurgery, Computer-assisted surgery
Abstract: This paper presents an accurate and robust tracking vision algorithm for Fetoscopic Laser Photo-coagulation (FLP) surgery for Twin-Twin Transfusion Syndrome (TTTS). The aim of the proposed method is to assist surgeons during anastomosis localization, coagulation and review using a teleoperated robotic system. The algorithm computes the relative position of the fetoscope tool tip with respect to the placenta, via local vascular structure registration.

The algorithm uses image features (local superficial vascular structures of the placenta's surface) to automatically match consecutive fetoscopic images. It is composed of three sequential steps: image filtering, binarization and vascular structures segmentation; relevant POIs seletion that are used on the final step: image registration between consecutive images.

The algorithm has to deal with the low quality of fetoscopic images: the liquid and dirty environment inside the placenta jointly with the thin diameter of the fetoscope optics and low amount of environment light reduces the image quality. The obtained images are blurred, noisy and with very poor color components.

The tracking system has been tested using real video sequences of Fetoscopic Laser Photo-coagulation (FLP) surgery for Twin-Twin Transfusion Syndrome (TTTS). The computational performance enables real time tracking, locally guiding the robot over the placenta's surface with enough accuracy.

Keywords: Surgical robotics, Planning and execution in surgical robotics
Abstract: Inserting a fine needle presents a trade-off problem between safety and accuracy. As one of the serious complications due to tissue damages during needle insertion, severe bleeding often occurs owing to blood vessel puncture. However, there are few researches to evaluate the safety quantitatively regarding bleeding during the fine needle insertion. Therefore, the purpose of this study was the quantitative evaluation of the amount of bleeding due the artery and vein puncture depending on the needle size. We developed a blood circulation system for measuring the amount of bleeding due to blood vessel puncture. Using the system, the amount of bleeding due to different needle sizes was evaluated. The results suggested that the amount of bleeding per unit time increased depending on the needle radius. According to ordinal safety standards, the 22-gauge needle is appropriate for insertion into the lower abdomen.

Keywords: Wearable sensor systems - User centered design and applications, Wearable wireless sensors, motes and systems, Implantable systems
Abstract: Measuring the severity of hyperkinetic movement disorders like tremor and myoclonus is challenging. Although many accelerometers are available to quantify movements, the vast majority lacks real-time analysis and an interface that makes it possible to real-time adjust therapy like deep brain stimulation (DBS). Here, we developed a smartwatch / smartphone application that is capable of real-time analysing movement disorder severity. Movement analysis was realised by integrating acceleration values, to velocity and subsequently to distance. Measured distances were compared with a validated accelerometer already applied for quantifying movement disorders. Further validation was done by quantitative assessment of simulated movement disorders in 10 healthy volunteers. Finally, the approach was tested in two patients treated with DBS to quantify the effect of different DBS settings on myoclonus and tremor severity, respectively. The distance measured with the application had a 96% accuracy. This was non-inferior (p = 0.76) compared to accelerometers already clinically applied. Furthermore, (simulated) movement disorder severity could be classified correctly in 93% of the cases. Finally, the method was capable of distinguishing effective from non-effective DBS parameters in two patients. In summary, with our approach we realised an instantaneous and reliable estimation of the severity of movement disorders which can assist in real time titrating therapy like DBS.

Keywords: Wearable sensor systems - User centered design and applications, Modeling and analysis
Abstract: To measure the life log of humans and enjoy virtual or augmented reality video games, several wearable devices have been developed that allow users to intuitively input commands. However, monitoring and estimating threedimensional human motions for extended periods using the wearable devices is difficult. Therefore, this study aims to develop a method that estimates the joint angles of the upper human body using a wearable suit implanted with strain sensors with a nonlinear characteristic. We used a convolutional neural network (CNN) to estimate the joint angles. We established a CNN estimator based on the training data of two adult males and confirmed that this estimator could estimate the joint angles of other adult males. To monitor the caretakers in a care facility, we measure the care-working motion, such as motions that care workers transform the elder persons, estimate each joint ngle, and visualize the motions on Unity.

Keywords: Wearable sensor systems - User centered design and applications, Integrated sensor systems, Portable miniaturized systems
Abstract: In this paper we propose a sensor embedded knee brace to monitor knee flexion and extension and other lower limb joint kinematics after anterior cruciate ligament (ACL) injury. The system can be easily attached to a standard post-surgical or post-injury brace and uses a novel sensor fusion algorithm that does not require calibration and is robust to magnetic interference. The sensor fusion algorithm integrates information from accelerometers, gyroscopes, and knee brace kinematic restrictions to calculate sensor orientation with low computational requirements. The wearable system and the sensor fusion algorithm were validated for various physical therapy exercises against a previously validated optical motion capture system. Root mean squared error (RMSE) and correlation coefficients (CCC and ICC) from the novel sensor fusion algorithm were compared with a benchmark Kalman filtering algorithm. The proposed sensor fusion algorithm demonstrated excellent correlation coefficients and significantly lower RMSE than the benchmark algorithm. The demonstrated error for most exercises was lower than other devices in the literature. The quantitative measures obtained by this system can be used to obtain longitudinal range-of-motion and functional biomarkers. These biomarkers can be used to improve patient outcomes through the early detection of at-risk patients, tracking patient function outside of the clinic, and the identification of relationships between patient presentation, intervention, and outcomes.

Keywords: Wearable sensor systems - User centered design and applications
Abstract: In this study, we propose a technique to increase the reliability of muscle signals in case of weak muscle signal measurements using pressure sensors in a prosthesis wearer for measuring the degree of contraction of the muscle. This technique is applied when determining the intention of the wearer of an intelligent prosthesis. Through this technique, it is possible to determine the intention of a prosthesis wearer, which was previously not possible.

Keywords: Wearable sensor systems - User centered design and applications, Wearable body sensor networks and telemetric systems, Textile-electronic integration
Abstract: We propose a new area for wearable technology and interaction by acquiring gastrointestinal signals non-invasively from the abdomen. The mind-gut connection has flourished as a research area in the past two decades, elucidating the gut’s key role in stress, affect, and memory. Meanwhile, engineering advancements have shown potential in accuracy of non-invasive gastric recordings. Here, we investigate the design and specification of a wearable system for the recording of gut-brain activity non-invasively. We also present results from a preliminary pilot test of a wearable gut-brain computer interface (GBCI).

Keywords: Wearable sensor systems - User centered design and applications
Abstract: Parkinson’s Disease (PD) can lead to impaired/slowed movement, gait impairments and increased risk of falling. Wearable technology-based gait analysis is emerging as a powerful tool to detect early disease and monitor progression. Here we present a novel approach to producing an objective, compact and personalised overview of a patient’s gait pattern. Phase plots were constructed in 41 people with PD and 38 controls (CL) from accelerometry data collected during straight intermittent walks with a single triaxial accelerometer placed on the lower back. Phase plots were analysed using bivariate Gaussian mixture models and classified based on several apparent features derived from the parameters of said model. Significant differences in phase plot form were found between and PD and CL subjects; with a very high within-subject consistency (reproducibility) (p < 0.0001). PD and CL subjects differ in the types of phase plots produced (p < 0.001). Strong connections between spatio-temporal (ST) gait characteristics and phase plot types were found. The presented novel methodology not only showed to be sensitive to pathology (PD vs CL), but can quickly produce a unique fingerprint of a person’s gait. This work presents encouraging results for clinical application of an objective, personalised gait feature for disease monitoring and clinical applications.

Keywords: Neural stimulation, Brain functional imaging - Source localization, Brain functional imaging - Evoked potentials
Abstract: Transcranial direct current stimulation (tDCS) is a noninvasive method that delivers current through the scalp to enhance or suppress brain activity. The standard way of applying tDCS is by the use of two large rectangular sponge electrodes on the scalp. The resulting currents often stimulate a broad region of the brain distributed over brain networks. In order to address this issue, recently, multi-electrode transcranial direct current stimulation with optimized montages has been used to stimulate brain regions of interest (ROI) with improved trade-off between focality and intensity of the electrical current at the target brain region. However, in many cases only the location of target region is considered and not the orientation. Here we emphasize the importance of calculating the individualized target location and orientation by combined electroencephalography and magnetoencephalography (EMEG) source analysis in individualized skull-conductivity calibrated finite element method (FEM) head models and stimulate the target region by four different tDCS montages. We have chosen the generator of the P20/N20 component, located at Brodmann area 3b and oriented mainly from posterior to anterior directions as our target for stimulation because it can be modeled as a single dipole source with a fixed position and orientation. The simulations will deliver optimized excitatory and inhibitory electrode montages that are in future investigations compared to standard and sham tDCS in a somatosensory experiment. We also present a new constrained maximum intensity (CMI) optimization approach that better distributes the currents over multiple electrodes, therefore should lead to less tingling and burning sensations at the skin and thus allows an easier realization of the sham condition, without significantly reducing the current intensity parallel to the target.

Keywords: Neural stimulation
Abstract: Experimental studies on transcutaneous spinal cord direct current and magnetic stimulation (tsDCS and tsMS) show promising results in the neuromodulation of spinal sensory and motor pathways, with possible application in spinal functional rehabilitation. Modelling studies on the electric field (EF) distribution during tsDCS and tsMS are powerful tools to understand the underlying biophysics and to select and optimize stimulation protocols for a specific clinical target. The study presented here compares the EF during cervical tsDCS and tsMS. The EF predictions show the same spatial profiles along the cervical spinal cord using both types of stimulation. tsMS presents higher average magnitudes per spinal segment, with a maximum value of 14.61 V/m, whereas tsDCS is approximately 30 times lower, reaching 0.44 V/m. According to previous studies, tsDCS and tsMS induce EF values which are sufficient for spinal neuromodulation.

Keywords: Neural stimulation, Brain functional imaging - EEG, Motor learning, neural control, and neuromuscular systems
Abstract: Transcranial alternating current stimulation (tACS) is becoming an important method in the field of motor rehabilitation because of its ability to non-invasively influence ongoing brain oscillations at arbitrary frequencies. However, substantial variations in its effect across individuals are reported, making tACS a currently unreliable treatment tool. One reason for this variability is the lack of knowledge about the exact way tACS entrains and interacts with ongoing brain oscillations. The present crossover stimulation study on 20 healthy subjects contributes to the understanding of cross-frequency effects of gamma (70 Hz) tACS over the contralateral motor cortex by providing empirical evidence which is consistent with a role of low- (12 -20 Hz) and high- (20- 30 Hz) beta power as a mediator of gamma-tACS on motor performance.

Keywords: Neural stimulation
Abstract: Neural activation by infrared nerve stimulation (INS) gains growing interest as a potential alternative to conventional electric nerve stimulation, since unambiguous advantages like contact-free operation, enhanced spatial selectivity and lack of (electrical) stimulation artifacts are promising for both future electrophysiological research and clinical application. For the systematic investigation of laser nerve activation, we recently introduced a novel experimental approach. Comprising a defined focused beam profile, it enables remote controlled, contact-free pulsed laser stimulation of the rat sciatic nerve, simultaneous to high-speed temperature measurement in vivo. Up to now, successful neural activation with single laser pulses (2 – 6 mJ) was observed in all performed experiments, however, it strongly depended on the particular nerve location. Hence, we depict the investigation of spatial dependency of the nerve response and identify ‘regions of excitability’ on the nerve surface, that are highly susceptible to INS. By means of thermal imaging, we simultaneously monitored the nerve surface temperature, where we observed progressing temperature build-up during single pulse stimulation with repetition rates above 4 Hz. In this work, we present current results of our ongoing research.

Keywords: Neural stimulation, Brain physiology and modeling
Abstract: Realistic electric field (E-field) models of the brain have cast doubt on classical targeting approaches used in transcranial direct current stimulation (tDCS). In apparent contradiction with physiological results, modeling studies predict similar or even higher E-field values in regions between the electrodes distant to the presumed targeted areas. As an explanation, not only the magnitude, but the direction of the E-field over specific cortical structures, have been shown to be determinant for the stimulation outcome. This work examines the magnitude and distribution of tangential and normal E-field components over different cortical areas in a representative brain atlas for various electrode montages commonly used in clinical applications. We have confirmed a general trend in the distribution of tangential and normal E-fields on gyri and sulci areas, respectively, partially independent of electrode configuration. The differences found between the various montages are also discussed.

Keywords: Neural stimulation, Brain functional imaging - EEG, Brain functional imaging - Mapping
Abstract: Repetitive Transcranial magnetic stimulation (rTMS) is a noninvasive brain stimulation technique that able to influence cortical excitability. Low-frequency rTMS (stimulation frequency ≤1Hz) induces long-lasting inhibitory effects on cortical excitability. In order to study the effects of 1Hz rTMS of the motor cortex on neuronal activity, 20 healthy subjects were recruited to receive rTMS, and electroencephalography (EEG) in resting condition with eye open were recorded before rTMS, at 0min, 20min, 40min, 60min after rTMS. In multiple frequency bands, power values on each channel were calculated, and functional connectivity between two channels was assessed using phase synchronization. We found an increase in power of theta-band oscillations in the frontal and the central brain areas immediately after rTMS. And alpha resting power in the central-parietal brain area did not change immediately after rTMS, but increased at 20min after rTMS. Moreover, there is a widespread increase in functional connectivity after rTMS in the theta band, whereas widespread decreases in the functional connectivity were found in the alpha band after rTMS. At the same time, there was no significant recovery on power and functional connectivity at 60min after rTMS. These results provide an evidence for a transient reorganization of neuronal activity after 1Hz rTMS over the motor cortex. In addition, low-frequency rTMS produces widespread long-lasting alterations in cortical functional connectivity.

Keywords: Image segmentation, X-ray - Fluoroscopy, Cardiac imaging and image analysis
Abstract: X-ray angiograms are currently the gold-standard in percutaneous guidance during cardiovascular interventions. However, due to lack of contrast, to overlapping artifacts and to the rapid dilution of the contrast agent, they remain difficult to analyze either by cardiologists, or automatically by computers. Providing, a general yet accurate multi-arteries segmentation method along with the uncertainty linked to those segmentations would not only ease the analysis of medical imaging by cardiologists, but also provide a required pre-processing of the data for tasks ranging from 3D reconstruction to motion tracking of arteries. The proposed method has been validated on clinical data providing an average accuracy of 94.9%. Additionally, results show good transposition of learning from one type of artery to another. Epistemic uncertainty maps provide areas where the segmentation should be validated by an expert before being used, and could provide identification of regions of interest for data augmentation purposes.

Keywords: Image segmentation, Image registration, segmentation, compression and visualization - Machine learning / Deep learning approaches, Magnetic resonance imaging - MR neuroimaging
Abstract: In this paper, we proposed and validated a novel and fully automatic pipeline for simultaneous tissue classification and lateral ventricle segmentation via a 2D U-net. The 2D U-net was driven by a 3D fully convolutional neural network (FCN). Multiple T1-weighted atlases which had been pre-segmented into six whole-brain regions including the gray matter (GM), white matter (WM), cerebrospinal fluid (CSF), lateral ventricles (LVs), skull, and the background of the entire image were used. In the proposed pipeline, probability maps of the six whole-brain regions of interest (ROIs) were obtained after a pre-segmentation through a trained 3D patch-based FCN. To further capture the global context of the entire image, the to-be-segmented image and the corresponding six probability maps were input to a trained 2D U-net in a 2D slice fashion to obtain the final segmentation map. Experiments were performed on a dataset consisting of 18 T1-weighted images. Compared to the 3D patch-based FCN on segmenting five ROIs (GM, WM, CSF, LVs, skull) and another two classical methods (SPM and FSL) on segmenting GM and WM, the proposed pipeline showed a superior segmentation performance. The proposed segmentation architecture can also be extended to other medical image segmentation tasks.

Keywords: Image segmentation, Image analysis and classification - Machine learning / Deep learning approaches
Abstract: Nematodes are plant parasites that cause damage to crops. In order to quantify nematode infestation based on soil samples, we propose an instance segmentation method that will serve as the basis of automatic quantitative analysis. We consider light microscopic images of cluttered object collections as they occur in realistic soil samples. We introduce an algorithm, LMBI (Local Maximum of Boundary Intensity) to propose instance segmentation candidates. In a second step, a SVM classifier separates the nematode cysts among the candidates from soil particles. On a data set of soil sample images, the LMBI detector achieves near-optimal recall with a limited number of candidate segmentations, and the combined detector/classifier achieves recall and precision of about 0.7. The pipeline only requires simple dot annotations and moderately sized training data, which enables quick annotating and training in laboratory applications.

Keywords: Image segmentation
Abstract: Automatic pancreas segmentation with high precision in Computed Tomography (CT) images is a fundamental issue in both medical image analysis and computer-aided diagnosis (CAD). However, pancreas segmentation is challenging because of the high variability in location and anatomy of the organs, while occupying only a very small part of the entire abdominal CT scans. Due to the rapid development of the CAD system and the urgent need for clinical treatment, the pancreas image segmentation with high precision is demanded. In this paper, we propose a new approach for automatic pancreas segmentation of CT images using inter-/intra-slice contextual information with a cascade neural network. Fully convolutional neural networks (FCN) are used to extract intra-slice contextual information for pancreas segmentation. Recurrent neural networks (RNNs) is introduced to extract inter-slice contextual information. With the setting bounding boxes, the proposed method outperforms the state-of-the-arts with an average Dice Similarity Coefficient (DSC) of 87.72 for NIH dataset with 4-fold cross-validation.

Keywords: Image segmentation, Image registration, segmentation, compression and visualization - Machine learning / Deep learning approaches, Image feature extraction
Abstract: Left atrial voltage maps are routinely acquired during electroanatomic mapping in patients undergoing catheter ablation for atrial fibrillation (AF). For patients, who have prior catheter ablation when they are in sinus rhythm (SR), the voltage map can be used to identify low voltage areas (LVAs) using a threshold of 0.2 - 0.45 mV. However, such a voltage threshold for maps acquired during AF has not been well established. A prerequisite for defining a voltage threshold is to maximize the topologically matched LVAs between the electroanatomic mapping acquired during AF and SR. This paper demonstrates a new technique to improve the sensitivity and specificity of the matched LVA. This is achieved by computing omni-directional bipolar voltages and applying Gaussian Process Regression based interpolation to derive the AF map. The proposed method is evaluated on a test cohort of 7 male patients, and a total of 46,589 data points were included in analysis. The LVAs in the posterior left atrium and pulmonary vein junction are determined using the standard method and the proposed method. Overall, the proposed method showed patient-specific sensitivity and specificity in matching LVAs of 75.70% and 65.55% for a geometric mean of 70.69%. On average, there was an improvement of 3.00% in the geometric mean, 7.88% improvement in sensitivity, 0.30% improvement in specificity compared to the standard method. The results show that the proposed method is an improvement in matching LVA. This may help develop the voltage threshold to better identify LVA in the left atrium for patients in AF.

Keywords: Image segmentation, Image classification, Image feature extraction
Abstract: Early screening in Colorectal Cancer consists in finding and removing small precancerous masses or neoplastic lesions developed from the mucosa, usually lesions smaller than 10 mm. Localization of small neoplastic lesions is a very challenging task since colon exploration is highly dependent on the expert training and colon preparation. Several strategies have attempted to locate neoplasias, but usually they are huge lesions that a trained gastroenterologist could hardly miss. This work presents a saliency-based strategy to localize polypoid and non-polypoid neoplastic lesions smaller than 10 mm in colonoscopy videos by combining spatio-temporal descriptors. For doing so, a per-frame-multi-scale representation is computed and edge, texture and motion features are extracted. Each of these features is used to construct a primary saliency map which are then combined to obtain a coarse saliency map. Finally, the neoplasia is localized as the bounding box of a circular region, approximated by the Hough transform, with the largest salience. The proposed approach was evaluated in 8 short colonoscopy videos obtaining an average of Annotated Area Covered of 0.75 and a precision of 0.82.

Keywords: Brain-computer/machine interface, Motor neuroprostheses - Neuromuscular stimulation, Neural signal processing
Abstract: The aim of the MoreGrasp project is to develop a non-invasive, multimodal user interface including a brain-computer interface (BCI) for control of a grasp neuroprostheses in individuals with high spinal cord injury (SCI). The first results of the ongoing MoreGrasp clinical feasibility study involving end users with SCI are presented. This includes BCI screening sessions, in which we investigate the electroencephalography (EEG) patterns associated with single, natural movements of the upper limb. These will later be used to control the neuroprosthesis. Additionally, the MoreGrasp grasp neuroprosthesis consisting of electrode arrays embedded in an individualized textile forearm sleeve is presented. The general feasibility of this electrode array in terms of corrections of misalignments during donning are shown together with the functional results in end users of the electrode forearm sleeve.

Keywords: Brain-computer/machine interface, Neural signal processing, Brain physiology and modeling - Sensory-motor
Abstract: The speed of visual brain-computer interfaces (BCIs) has been greatly improved in recent years. However, traditional visual BCI paradigm requires users to directly gaze at the intensive flickering items, which would cause severe problems in practical applications, such as visual fatigue and excessive visual resources consumption. A promising solution is to use small visual stimuli outside the central visual area to encode instructions, which had been demonstrated to be effective in our previous study. This study aims to further investigate the effects of stimulus position on the classification of miniature asymmetric visual evoked potentials (aVEPs). Small peripheral visual stimuli were designed with different eccentricities (1° and 2°) and directions (0°, 45°, 90°, 135°, 180°, -135°, -90°, and -45°) to induce different kinds of miniature aVEPs. Five subjects participated in this experiment. Discriminative canonical pattern matching (DCPM) was used to classify all possible pairs of miniature aVEPs. Study results showed that visual stimuli with less eccentricity could induce more distinct miniature aVEPs. The highest single-trial accuracy achieved was about 83% for the binary classifications of miniature aVEPs pairs corresponding to (1°, -135°) Vs (1°, 0°), (1°, -45°) Vs (1°, -135°) and (1°, -45°) Vs (1°, 180°). This finding is very important for the design and development of the miniature aVEPs-based BCIs.

Keywords: Brain-computer/machine interface, Human performance - Cognition, Brain functional imaging - EEG
Abstract: Electroencephalography (EEG) neurofeedback (NF) training has been shown to produce long-lasting effects on the improvement of cognitive function as well as the normalization of aberrant brain activity in disease. However, the impact of the sensory modality used as the NF reinforcement signal on training effectiveness has not been systematically investigated. In this work, an EEG-based NF-training system was developed targeting the individual upper-alpha (UA) band and using either a visual or an auditory reinforcement signal, so as to compare the effects of the two sensory modalities. Sixteen healthy volunteers were randomly assigned to the Visual or Auditory group, where a radius-varying sphere or a volume-varying sound, respectively, reflected the relative amplitude of UA measured at EEG electrode Cz. Each participant underwent a total of four NF sessions, of approximately 40 min each, on consecutive days. Both groups showed significant increases in UA at Cz within sessions, and also across sessions. Effects subsequent to NF training were also found beyond the target frequency UA| and scalp location Cz, namely in the lower-alpha and theta bands and in posterior brain regions, respectively. Only small differences were found on the EEG between the Visual and Auditory groups, suggesting that auditory reinforcement signals may be as effective as the more commonly used visual signals. The use of auditory NF may potentiate training protocols conducted under mobile conditions, which are now possible due to the increasing availability of wireless EEG systems.

Keywords: Brain-computer/machine interface, Human performance - Driving
Abstract: Existing wheelchair control interfaces, such as sip & puff or screen based gaze-controlled cursors, are challenging for the severely disabled to navigate safely and independently as users continuously need to interact with an interface during navigation. This puts a significant cognitive load on users and prevents them from interacting with the environment in other forms during navigation. We have combined eye tracking/gaze-contingent intention decoding with computer vision context-aware algorithms and autonomous navigation drawn from self-driving vehicles to allow paralysed users to drive by eye, simply by decoding natural gaze about where the user wants to go: A.Eye Drive. Our “Zero UI” driving platform allows users to look and interact visually with at an object or destination of interest in their visual scene, and the wheelchair autonomously takes the user to the intended destination, while continuously updating the computed path for static and dynamic obstacles. This intention decoding technology empowers the end-user by promising more independence through their own agency.

Keywords: Brain-computer/machine interface, Neural signal processing, Human performance - Modelling and prediction
Abstract: Motor Imagery (MI) is a typical paradigm for Brain-Computer Interface (BCI) system. In this paper, we propose a new framework by introducing a tensor-based feature representation of the data and also utilizing a convolutional neural network (CNN) architecture for performing classification of MI-EEG signal. The tensor-based representation that includes the structural information in multi-channel time-varying EEG spectrum is generated from tensor discriminant analysis (TDA), and CNN is designed and optimized accordingly for this representation. Compared with CSP+SVM (the conventional framework which is the most successful in MI-based BCI) in the applications to the BCI competition III-IVa dataset, the proposed framework has the following advantages: (1) the most discriminant patterns can be obtained by applying optimum selection of spatial-spectral-temporal subspace for each subject; (2) the corresponding CNN can take full advantage of tensor-based representation and identify discriminative characteristics robustly. The results demonstrate that our framework can further improve classification performance and has great potential for the practical application of BCI.

Keywords: Brain-computer/machine interface, Neural signal processing
Abstract: For the past few years, electroencephalogram (EEG)-based brain-computer interfaces (BCIs) have gotten tremendous progress and attracted increasing attention. To broaden the application of BCIs, researchers have focused on the increasement of the BCI instruction number in recent years. However, with a large number of instructions, the BCI calibration time will be too long to be accepted in practical usage. This study proposed a new coding method based on multifocal steady-state visual evoked potentials (mfSSVEPs), in which 16 targets were binary coded by 4 frequencies. Notably, the training data needed for calibration corresponded to only five out of the sixteen targets. Five volunteers were recruited to test this paradigm. Task-related component analysis combined with a probabilistic model were employed for target recognition. As a result, the accuracy could reach as high as 93.1% with 1s-length data. The highest information transfer rate (ITR) was 101.1 bits/min with an average of 73.9 bits/min. The results indicate that this new paradigm is promising to encode a large BCI instruction set with less trainings.

Keywords: Time-frequency and time-scale analysis - Time-frequency analysis, Time-frequency and time-scale analysis - Wavelets, Time-frequency and time-scale analysis - Nonstationary processing
Abstract: In high-income countries, fetal hypoxia affects 3 to 8 newborns per 1000 live births with subsequent moderate or severe Hypoxic-Ischemic Encephalopathy (HIE) in 0.5 to 1 per 1000 live births. Visual interpretation of FHR signal issued from a Doppler ultrasound cardiotocography is the gold standard to monitor fetal condition. Unfortunately, its analysis presents a high rate of inter-observer variability and a low specificity to predict poor neonatal outcomes. Under hypoxia, the fetus develops several adaptive mechanisms regulated by the autonomic nervous system inducing changes in the fetal heart rate variability. Though fetal heart rate variability methods demonstrated abilities to predict perinatal asphyxia, most of the Doppler ultrasound technologies used in clinical practice do not provide sufficiently accurate fetal heart rate signals for heart rate variability analysis. Indeed, Doppler ultrasound cardiotocography usually provides fetal heart rate values averaged over 2 or 3 beats which can constitute a limitation for spectral analysis. We developed a fetal heart rate variability analysis method: the Fetal Stress Index (FSI). The objective of this study was to investigate the influence of averaged fetal heart rate on this new index in order to check the feasibility of computing the FSI from the signal issued from Doppler ultrasound cardiotocography.

Keywords: Data mining and processing in biosignals, Time-frequency and time-scale analysis - Time-frequency analysis
Abstract: Fetal heart rate (FHR) is an important characteristic in fetal well-being follow-up. It is classically estimated using the cardiotocogram (CTG) non-invasive reference technique. However, this latter presents some significant drawbacks. An alternative non-invasive solution based on the fetal phonocardiogram (fetal PCG) can be used. But most of proposed methods based on the PCG signal need to detect and to label the fetal cardiac S1 and S2 sounds, which may be a difficult task in certain conditions. Therefore, in this paper, we propose a new methodology for FHR estimation from fetal PCG with one single cardio-microphone and without the distinction constraint of heart sounds. The method is based on the non-negative matrix factorization (NMF) applied on the spectrogram of fetal PCG considered as a source-filter model. The proposed method provides satisfactory results on a preliminary dataset of abdominal PCG signals. When compared to the reference CTG, correlation on FHR estimations between PCG and CTG is around 90%.

Keywords: Nonlinear dynamic analysis - Biomedical signals
Abstract: Monitoring uterine activity by electrohysterogram (EHG), associated with contractions both in pregnancy and labor, may contribute to the knowledge for evaluating possible risks to the binomial mother-fetus. In this context, the aim of the present study was to explore the complexity of EHG generated by women during the third trimester of pregnancy (group P) and at term labor (group L). The EHG was obtained by band-pass filtering in the range from 0.1 to 3 Hz the monopolar raw signal of the electrode number 1, of a 4-by-4 sensor array, which was located near to the tocodynamometer transducer. Multiscale entropy (MSE) analysis measures the entropy over multiple time scales to provide the complexity of the EHG time series. The results pointed out that such nonlinear technique has the potential to discriminate contractions from both groups using the area under the MSE curve (AUC) as index. The highest complexity was obtained for group P (N= 8) as AUC was 13.9233 + 0.2015 while the lowest complexity was for group L, with N=8 and AUC of 5.1675 + 0.0783 (p<0.0001). Consequently, the complexity of EHG by MSE could provide an index to discriminate between the electrical uterine activity generated during pregnancy or at labor.

Keywords: Neural networks and support vector machines in biosignal processing and classification, Physiological systems modeling - Signal processing in physiological systems
Abstract: Neonatal EEG seizure detection algorithms (NSDAs) have an upper bound of performance related to the agreement between visual interpretation of human experts. No published algorithms have reported performance that has reached this upper bound. In this paper, we combined two recently developed NSDAs in order to improve detection performance. An outlier detection stage was also added to improve robustness in the presence of unseen data. A large database of EEG from 79 term infants labeled by three independent human experts was used to develop and test the sufficiency of the hybrid NSDA. The inter-observer agreement (IOA) between experts was high (K = 0.757, 95%CI: 0.665-0.836, n=79). The area under the receiver operator characteristic of the NSDA compared to the consensus annotation of the human experts was 0.952 (95%CI: 0.0927-0.971). The IOA of seizure detection between the NSDA and human experts was not significantly less than the IOA among human experts (dK = 0.022, 95%CI: -0.020 to 0.072) and was further improved by increasing the minimum seizure duration from 10s to 30s (dK = -0.002, 95%CI: -0.073 to 0.055). Automated methods of neonatal EEG seizure detection have sufficient accuracy to replace human interpretation, potentially, providing reliable interpretations for clinicians in the neonatal intensive care unit.

Keywords: Signal pattern classification, Neural networks and support vector machines in biosignal processing and classification
Abstract: Frequent pain and discomfort in premature infants can lead to long-term adverse neurodevelopmental outcomes. Video-based monitoring is considered to be a promising contactless method for identification of discomfort moments. In this study, we propose a video-based method for automated detection of infant discomfort. The method is based on analyzing facial and body motion. Therefore, motion trajectories are estimated from frame to frame using optical flow. For each video segment, we further calculate the motion acceleration rate and extract 18 time- and frequency-domain features characterizing motion patterns. A support vector machine (SVM) classifier is then applied to video sequences to recognize infant status of comfort or discomfort. The method is evaluated using 183 video segments for 11 infants from 17 heel prick events. Experimental results show an AUC of 0.94 for discomfort detection and the average accuracy of 0.86 when combining all proposed features, which is promising for clinical use.

Keywords: Data mining and processing in biosignals, Coupling and synchronization - Coherence in biomedical signal processing, Nonlinear dynamic analysis - Biomedical signals
Abstract: Early life stress in the neonatal intensive care unit (NICU) predisposes premature infants to adverse health outcomes. Although those patients experience frequent apneas and sleep-wake disturbances during their hospital stay, clinicians still rely on clinical scales to assess pain and stress burden. This study addresses the relationship between stress and apneic spells in NICU patients to implement an automatic stress detector. EEG, ECG and SpO2 were recorded from 40 patients for at least 3 hours and the stress burden was assessed using the Leuven Pain Scale. Different logistic regression models were designed to detect the presence or the absence of stress based on the signals reactivity to each apneic spell. The classification shows that stress can be detected with an area under the curve of 0.94 and a misclassification error of 19.23%. These results were obtained via SpO2 dips and EEG regularity. These findings suggest that stress deepens the physiological reaction to apneas, which could ultimately impact the neurological and behavioral development.

Keywords: Optical imaging and microscopy - Optical vascular imaging, Image registration, segmentation, compression and visualization - Volume rendering, Multimodal image fusion
Abstract: Diameter and volume are frequently used parameters for cardiovascular diagnosis, e.g., to identify a stenosis of the coronary arteries. Intra-vascular OCT imaging has a high spatial resolution and promises accurate estimates of the vessel diameter. However, the actual images are reconstructed from A-scans relative to the catheter tip and imaging is subject to rotational artifacts. We study the impact of different volume reconstruction approaches on the accuracy of the vessel shape estimate. Using X-ray angiography we obtain the 3D vessel centerline and the 3D catheter trajectory, and we propose to align the A-scans using both. For comparison we consider reconstruction along a straight line and along the centerline. All methods are evaluated based on an experimental setup using a clinical angiography system and a vessel phantom with known shape. Our results ilustrate potential pitfalls in the estimation of the vessel shape, particularly when the vessel is curved. We demonstrate that the conventional reconstruction approaches may result in an overestimate of the cross-section and that the proposed approach results in a good shape agreement in general and for curver segments, with DICE coefficients of approximately 0.96 and 0.98, respectively.

Keywords: Optical imaging and microscopy - Neuroimaging, Optical imaging, Ultrasound imaging - Photoacoustic/Optoacoustic/Thermoacoustic
Abstract: Graded-index (GRIN) fibers have been used as implantable optical waveguides to guide light and relay images through the depth of the tissue. We have recently shown that non-invasive ultrasound can generate refractive index gradients within the tissue that form virtual GRIN lenses for imaging and photostimulation deep into the tissue. Here we present the theory behind this idea by analyzing the coupled acoustic-photonic system that models the interaction of light with the ultrasonically modulated medium. We will discuss how changing the parameters of ultrasound will change the confinement and guiding of light within the modulated medium. We will also demonstrate that using a custom-designed cylindrical ultrasonic array, the pressure interference can be controlled to sculpt complex patterns of light in the medium, such as dipole and quadrupole shapes, suitable for multisite imaging. Finally, we will discuss experimental results corroborating the theoretical predictions to generate single and multisite in situ virtual lenses that can be used for fluorescent imaging of mouse brain tissue that expresses green fluorescent protein (GFP).

Keywords: Optical imaging and microscopy - Fluorescence microscopy, Image visualization, Image classification
Abstract: Due to the rapid globalization there is an increasing danger for pandemic outbreaks. The high death toll of fast spreading diseases like the Ebola infection demand the fast development of new medicines. Thus, the automation of pharmaceutical processes is an indispensable but challenging task. In cooperation with the Institute for Virology, Philipps-University, Marburg, Germany, recently, algorithms to detect and evaluate subviral particle tracks in live-cell fluorescence image sequences were developed. In steady interdisciplinary exchange between pharmacists and engineers it turned out that new measures to identify and classify subviral particle motion are required. This article focuses on the evaluation and optimization of a new method to classify subviral particle motion using fractal dimension estimation. The influence of global and local interpolation methods on fractal dimension estimation is investigated. The methods are tested on simulated data and applied to real image sequences. The results prospect a high benefit of using the presented methods for an effective classification of subviral particle behavior.

Keywords: Optical imaging and microscopy - Fluorescence microscopy
Abstract: Active molecular transport ensures a purposeful spatiotemporal distribution of cellular proteins and is therefore key to a wide range of processes such as morphogenesis, homeostasis or migration. However, redistributions of molecules in bulk are seldom quantified because the regions involved are too diffuse to be segmented consistently. To bridge this gap, we propose a Laplace-corrected Runge-Kutta advection that is based on mesh triangulation. Our framework can follow the movement and deformation of multiple parts of a diffuse region at once and offers a seamless combination with spatiotemporal line integration in Lagrangian coordinates. This allows the flexibility to taylor specific measures to the question at hand, e.g. mechanical work, bringing long-established physics concepts into biology grounds. We exemplify our approach by quantifying how the isotropy of intracellular protein distributions changes during cargo transport.

Keywords: Optical imaging - Confocal microscopy, Optical imaging and microscopy - Fluorescence microscopy, Image analysis and classification - Machine learning / Deep learning approaches
Abstract: Cellular imaging with confocal fluorescence laser microscopy gave rise to many new insights into the cellular machinery. One interesting observation suggests that morphology of cell nucleus plays a key role for neutrophilic function, which is an essential part of the innate immune system of most mammals. Due to the increasing availability of high resolution 3D images coming from the microscope, machine learning becomes a promising tool for automatically discovering underlying hidden structures. Here, the major difficulty consists of selecting an appropriate representation for characterizing the morphology of cell nucleus. In this work we tackle this problem and propose a fully unsupervised mechanism for finding structure in high-throughput 3D image data. The key component of our approach is based on Generic Fourier Transform (GFT) for 2D images, which for 3D involves spherical coordinate transformation prior to fast Discrete Fourier Transformation. On top on GFT we apply dimensionality reduction with Principal Component Analysis, followed by generative cluster analysis with a Gaussian Mixture Model. We validate our new approach first on a synthetic 3D-MNIST dataset with random rotations, where quantitative and qualitative results confirm the applicability of the proposed pipeline for exploring shape space in a purely unsupervised manner. Then we apply our proposed technique to a new collected dataset of high resolution 3D images of neutrophile nuclei suggesting a clustering model with six significant clusters of morphological cell nuclei prototypes. We visualize differences in the cell shape clusters by providing prototypical examples of neutrophilic cell nuclei.

Keywords: Optical imaging and microscopy - Diffuse optical tomography, Optical imaging and microscopy - Near infra-red spectroscopy, Image reconstruction and enhancement - Tomographic reconstruction
Abstract: Diffuse optical tomography (DOT), a subset of functional near-infrared spectroscopy (fNIRS), is a noninvasive functional imaging modality for studying the human brain in normal and diseased conditions. It measures changes in concentrations of oxygenated hemoglobin (HbO) and deoxygenated hemoglobin (Hb) in the blood vasculature of the brain. In contrast to functional magnetic resonance imaging (fMRI), the gold standard in human brain imaging, DOT offers the advantage of higher temporal resolution, portability, lower cost, multiple contrasts and usability for persons who cannot otherwise utilize MRI-based imaging modalities, including bedridden patients and infants, etc. The goal of the present study was to evaluate performance of a DOT method in studying dynamic patterns of brain activations involving motor control. CW-fNIRS data were acquired in four sessions from a healthy male participant when he performed a motor task in a block-design experiment. Results from experimental data showed pronounced activity in the primary motor cortex (M1), contralateral to the clenching hand. It was further observed that the M1 activity was consistent over four sessions. Furthermore, temporal dynamics of motor activity at each session further revealed well-sequenced activation patterns among M1, premotor cortex (PMC), and supplementary motor area (SMA). Timed ipsilateral motor activity suppression was also observed several seconds after the onset of contralateral M1 activity. More importantly, these temporal dynamics were similarly observed in all four sessions. These preliminary results suggest that the DOT method has the sensitivity, reliability, and spatio-temporal resolutions to study activities originated from the motor cortices. A full-scope evaluation and validation in more participants on the motor system can establish it as a promising neuroimaging tool to study, such as, infants at the risk of cerebral palsy or elders with Parkinson’s due to its portability and usability.

Keywords: Bio-electric sensors - Sensor systems, Bio-electric sensors - Sensing methods, Physiological monitoring - Instrumentation
Abstract: We present proof of concept experiment of a sensing method to detect skin hydration using a low-cost bio-impedance sensor. The sensing system is validated by testing its current output over frequencies between 1 kHz and 50 kHz and comparing measured values of impedance. A series of experiments with salt-water mixtures as well as a gelatin-based phantom were carried out to test the sensor’s ability to detect small changes in impedance due to changes in water content. We also compared impedance measurements from the phantom to human skin to confirm that the manufactured phantoms can mimic skin properties successfully. Our experimental results show that the sensor can detect small changes in salt concentration and can capture the correlation between the impedance and skin hydration in a reliable manner.

Keywords: Wearable sensor systems - User centered design and applications, Physiological monitoring - Novel methods, Physiological monitoring - Modeling and analysis
Abstract: A high-precision wearable bioimpedance sensor developed at Imec was extensively tested. Unlike known bioimpedance sensors on the market, the new device enables hydration shift measurement in a single person, with no need for averaging over a population. For reaching this target, a method for hydration monitoring in case of altered hydration is tested. An assessment of fluid shift with sensitivity of about 700 ml has been demonstrated, which is comparable with the capabilities of known methods because of the device accuracy, immunity to electrode-skin impedance variation, and due to establishing the impedance baseline prior to fluid shift.

Keywords: Bio-electric sensors - Sensor systems, Integrated sensor systems, Wearable low power, wireless sensing methods
Abstract: This paper presents a portable and modular wireless multichannel sensor system for high-density surface electromyography (HD-sEMG) signals acquisition. Featuring low-power and high-quality off-the-shelf components such as the Intan Technologies RHD2132 digital electrophysiology interface chip, the current iteration of the proposed sensor system allows the recording of 32 surface electromyography (sEMG) channels, each at a sampling rate of 1 kHz, and a sample resolution of 16 bits. It features the RHD2132’s typical input-referred noise of 2.4 μVrms, with <15% variation with amplifier bandwidth as specified by the manufacturer, and a total power consumption of 49.5 mW. Data is sent in real-time to a base station using a 2.4-GHz industrial, scientific and medical (ISM) wireless link. Along with the recording platform, the integrated sensor system includes a dry surface electrodes array prototype directly built on a printed circuit board. Intended for complex muscles activity patterns detection on the forearm, the flexible 32 surface electrodes array is designed to be placed flat or to fit a curved area like the forearm in a hand gestures recognition prosthetic system. In such applications, this device will offer improved prosthesis control scheme intuitiveness and ease-of-use. Among other core features of the system are its compact, light-weight and easy to install physical design. The complete system fits on a 2 by 6.5 cm2 printed circuit board mounted on a 7.6 by 11.8 cm2 electrodes array. HD-sEMG user forearm output data collected with the system is presented with a proposed frequency-time-space cross-domain preprocessing method for visualization of HD-EMG data and building training datasets.

Keywords: Bio-electric sensors - Sensor systems, Portable miniaturized systems, New sensing techniques
Abstract: In the presented work, we combine a commonplace commodity, tissue paper, with biosensing capabilities to provide on-the-go detection of biomarkers. We developed a simple light-weight sensor using single walled carbon nanotubes (SWCNT) on commercially available tissue paper. As a proof-of-concept, we show that these sensors can be used to quantify Interleukin-6 protein concentration. Detection is carried out by examining the modulation in electrical properties of the substrate induced by addition of protein on these sensors. Our results show that these sensors can provide the foundation for development of tissue paper based portable sensors for biomarkers relevant to on-the-go testing.

Keywords: Magnetic sensors and systems, Physiological monitoring - Instrumentation, Physiological monitoring - Novel methods
Abstract: In this paper, a contactless radio-frequency (RF) inductive probe is used to spatially localize and characterize a complex dielectric organic inclusion in a fluid. The effect of dielectric properties (DP) of this organic material is investigated experimentally and by numerical computations.

The used RF probe is a 135 MHz 3 cm diameter and 10 cm long, cylindrical bracelet resonator, placed close to a water tank filled with deionized water which includes a 1.5 cm diameter inclusion filled of air or NaCl solutions and placed in arbitrary positions. The water tank and the inclusion are used to model an organic material including a tumor. The RF probe is used as a transmit and receive sensor. It induces a magnetic field inside the water tank, which, by reciprocity, conveys information about the DP of the investigated material. The impedance changes at the end of the RF probe are directly related to the modifications of the magnetic field, and are measured by means of a network analyzer. A complex fit of the impedance frequency response around the resonance frequency gives access to two quantities proportional to the electrical conductivity and dielectric constant of the inclusion. The inclusion is moved into the water tank along the three axes by means of a robotic arm, so that two three dimensional maps of the equivalent dielectric quantities in function of the inclusion position are sensed by the probe. Then, the inclusion is filled with different conductive NaCl solutions from 0.1 to 1.1 S/m in order to test the ability of the probe to sense the modifications of the dielectric properties of the inclusion. Experimental as well as computation results obtained using the Distributed Point Source Method (DPSM) validate the ability of the proposed probe to localize the inclusion as deep as 1 cm into the water, and the ability of the probe to sense the dielectric property changes of the inclusion.

Keywords: Physiological monitoring - Instrumentation, Bio-electric sensors - Sensing methods, Sensor systems and Instrumentation
Abstract: In order to effectively record from electrically active cells through microelectrode arrays a low-noise amplification and data acquisition system is required. Although commercially available, these systems can be expensive and lack the freedom to customise hardware and software. In this work we present a low-cost (US21 for the first channel + US11 for each additional channel), low-noise amplifier coupled with an analog to digital converter from National Instruments. The amplifier was designed to (i) operate between 0 and 5 V utilising a DC battery power supply, (ii) operate within a bandwidth of 10 kHz, (iii) remove DC voltage created at the electrode/electrolyte interface with a high-pass cut-off frequency of 0.7 Hz and (iv) have a gain of 2000. Strategies to reduce environment electromagnetic interference at the amplifier front end were employed and involved a customised neural interface board connected between the microelectrode array and amplifier. The constructed amplifier achieved an intrinsic noise amplitude of 0.8 μVrms, which facilitated high quality recordings as exemplified by in vitro recordings from primary hippocampal neurons.

Keywords: Time-frequency and time-scale analysis - Empirical mode decomposition in biosignal analysis, Adaptive filtering, Physiological systems modeling - Signal processing in physiological systems
Abstract: In recent years, considerable progress has been made in the non-contact based detection of the respiration rate from video sequences. Common techniques either directly assess the movement of the chest due to breathing or are based on analyzing subtle color changes that occur as a result of hemodynamic properties of the skin tissue by means of remote photoplethysmography (rPPG). However, extracting hemodynamic parameters from rPPG is often difficult especially if the skin is not visible to the camera.. In contrast, extracting respiratory signals from chest movements turned out to be a robust method. However, the detectability of chest regions cannot be guaranteed in any application scenario, for instance if the camera setting is optimized to provide close-up images of the head. In such a case an alternative method for respiration detection is required.

It is reasonable to assume that the mechanical coupling between chest and head induces minor movements of the head which, like in rPPG, can be detected from subtle color changes as well. Although the strength of these movements is expected to be much smaller in scale, sensing these intensity variations could provide a reasonably suited respiration signal for subsequent respiratory rate analysis.

In order to investigate this coupling we conducted an experimental study with 12 subjects and applied motion- and rPPG-based methods to estimate the respiratory frequency from both head regions and chest. Our results show that it is possible to derive signals correlated to chest movement from facial regions. The method is a feasible alternative to rPPG-based respiratory rate estimations when rPPG-signals cannot be derived reliably and chest movement detection cannot be applied as well.

Keywords: Signal pattern classification, Neural networks and support vector machines in biosignal processing and classification
Abstract: Individuals with Autism Spectrum Disorder (ASD)have been shown to have atypical scanning patterns during face and emotion perception. While previous studies characterized ASD using eye-tracking data, this study examined whether the use of eye movements combined with task performance in facial emotion recognition could be helpful to identify individuals with ASD. We tested 23 subjects with ASD and 35 controls using a Dynamic Affect Recognition Evaluation (DARE) task that requires an individual to recognize one of six emotions (i.e., anger, disgust, fear, happiness, sadness, and surprise) while observing a slowly transitioning face video. We observed differences in response time and eye movements, but not in the recognition accuracy. Based on these observations, we proposed a machine learning method to distinguish between individuals with ASD and typically developing (TD) controls. The proposed method classifies eye fixations based on a comprehensive set of features that integrate task performance, gaze information, and face features extracted using a deep neural network. It achieved an 86% classification accuracy that is comparable with the standardized diagnostic scales, with advantages of efficiency and objectiveness. Feature visualization and interpretations were further carried out to reveal distinguishing features between the two subject groups and to understand the social and attentional deficits in ASD.

Keywords: Neural networks and support vector machines in biosignal processing and classification
Abstract: In recent years, more studies focus on the Ultra-Wide Band (UWB) radar to provide a noncontact vital sign monitoring service. To further improve the accuracy of vital sign monitoring, the UWB radar network composed by multiple radars is considered for providing information on different angles. The radar deployment is a key factor in the radar network to impact the monitoring results, since that different deployments bring diverse combinations of vital sign information. To the best of our knowledge, few studies attempt to optimize the radar deployment for the purpose of improving the vital sign monitoring accuracy. This paper provides an analysis on the number and angle in radar deployment for vital sign monitoring. To firstly validate the superiority of utilizing multiple radars rather than the single radar, the theoretical analysis is performed by combining the vital sign monitoring model and the Cramer-Rao Low Bound (CRLB) theory. Then experiments for discussing the effect of the radar number and the angle between radars are conducted in realistic environment. Considering the radar number from 2 to 4, signals are acquired from radars located symmetrically with the subject sitting at the center. Additionally, the deployments of two radars with angles of 0°, 60°, 120°, 180°, 240° and 300° are discussed, ensuring that at least one radar is directed to the chest of the subject. A Neural Network (NN) based data fusion method is performed to obtain the fused vital signal from the radars. The accuracy of the NN is discussed as the evaluating indicators for these deployments.

Keywords: Time-frequency and time-scale analysis - Time-frequency analysis
Abstract: Continuous HR (Heart Rate) monitoring enables the stress estimation in daily life. A Doppler sensor could be a key device to facilitate the non-contact HR estimation. As one of the Doppler sensor-based HR estimation methods, we have previously proposed a MUSIC (MUltiple SIgnal Classification)-based HR estimation method. MUSIC is the algorithm widely used as a tool to estimate DOA (Direction of Arrival). In our previous method, MUSIC spectrum is calculated in each sliding window, and then HR is estimated by the maximum peak detection over the MUSIC spectrum. However, when HR changes largely within the window, several peaks due to heartbeats appear over the MUSIC spectrum, which might cause the incorrect peak detection. Hence, an adaptive window is required so that only one peak appears. In this paper, we propose a MUSIC-based HR estimation method with an adaptive window size setting. When several peaks due to heartbeats appear over the MUSIC spectrum, our proposed method shortens the time window and re-calculates the MUSIC spectrum, which is repeated until only one peak appears. The experimental results showed that our method outperformed not only our previous one but also the other existing MUSIC-based HR estimation one in terms of the estimation accuracy of the HR, the stress indexes CVI (Cardiac Vagal Index) and CSI (Cardiac Sympathetic Index).

Keywords: Nonlinear dynamic analysis - Biomedical signals, Principal and independent component analysis - Blind source separation, Adaptive filtering
Abstract: Laboratory animal science plays a crucial role in medical and biological research. In the last decades, stricter regulations were enforced to safeguard laboratory animals. Following the “3Rs” guiding principles, animal trials should be replaced, reduced and refined, whenever possible.

A contactless modality capable of assessing the respiratory rate (RR) and additional breath related characteristics can potentially refine anesthetic interventions in rodents by continuously monitoring their anesthetic depth. This can reduce complications and thus the number of needed animals.

We were able to extract the instantaneous RR in rodents with a sum squared error (SSE) of 0.26 breaths/min from color video. A correlation of 0.9781 compared to an Electrocardiography (ECG) based reference was achieved. Furthermore, additional temporal and morphological characteristics were extracted, which are sensitive for changes in the anesthetic depth.

Keywords: Neural stimulation, Brain physiology and modeling - Neuron modeling and simulation, Neurological disorders - Treatment methodologies
Abstract: Transcranial direct current stimulation (tDCS) is a neuromodulation technique currently used in a variety of clinical applications. To optimize medical procedures and establish safety guidelines, this contribution investigates the influence of anatomical alterations of the brain on electric potential, current density and electric field distributions. A 2D simulation is run in COMSOL to generate the electric field distribution inside a human head model. By scaling the gray and white matter regions only, the electric field and current densities are substantially reduced compared to the reference state. Anatomical alterations should be considered for improving focality and targeting as the efficiency and effectiveness of tDCS might otherwise be compromised.

Keywords: Neural stimulation, Brain physiology and modeling
Abstract: MRI-based computational models have proved a useful tool to estimate the E-field distribution in the brain for transcranial electric stimulation (tES). This paper reviews the main current tES modeling approaches and the remaining controversial issues the modeling community faces.

Keywords: Neural stimulation, Brain physiology and modeling
Abstract: Personalization is becoming standard practice in medical diagnosis and treatment. Recent work has highlighted the importance of physical modeling of the electric field in brain stimulation research and clinical practice. We present two different approaches to personalize transcranial stimulation based on realistic biophysical and physiological models of the human brain. The first, already used with success in clinical pilots, relies on MRI-driven finite element modeling of the electric field produced by multichannel transcranial current stimulation, using a genetic algorithm to optimize the number, current intensity and location of the electrodes based on a specific target. The second, more advanced, is based on modeling the brain as a network of neural masses embedded in a realistic physical matrix, representing both measurable electrical brain activity and electric interactions.

Keywords: Neural stimulation
Abstract: MRI-derived computational models of transcranial electrical stimulation (tES) serve as a principle method for understanding the path and density of current flow in tES applications and how it might relate to clinical outcomes. These models require accurate foreknowledge regarding the electrode assembly used to deliver current to the scalp. While a variety of necessary factors can be estimated with relative ease, the positioning/placement of electrodes often assumes alignment with the 10-20 International measurement system or another form of head position metric. In clinical application, there is often significant variation in actual placement of electrodes, even when using a physical measurement-based electrode positioning approach. Prior research demonstrates that even 1 cm of inaccuracy in electrode positioning significantly alters computational modeling estimates as well as the impact of tES on functional outcomes. To optimize the accuracy of individualized computational models, variation in placement in clinical conditions must be accurately captured and integrated into modeling solutions. We present a method for enhancing the accuracy and application of individualized MRI derived computational models by using 3D capture of real-world electrode positioning with 3D scanning cameras, 3D capture software, and analytic approaches for integration of electrode position into individualized computational models.

Keywords: Brain physiology and modeling - Neuron modeling and simulation
Abstract: Transcranial direct current stimulation (tDCS) is a non-invasive, safe, inexpensive technique with rapidly expanding therapeutic application. Despite promising results there is variability in response. One potential source of variability is individual anatomy that influences the current flow in the cortex. We are interested in quantifying the scale of variability and customization of dosing to account for this. We summarize results of variation across pediatric, adults,aging and ethnic factors based on previous published work both including ours and others. We also consider variability due to gender differences based on our on-going study. We consistently observe that current flow pattern is influenced by individual factors. Subject-specific modeling provides a rational path towards standardizing tDCS dose.

Keywords: Micro- and nano-sensors, Cellular force transduction - Atomic force microscopy, Electric fields - Tissue regeneration
Abstract: A new digitally controlled field generator is integrated into a laser microscope. Magnetic particles powered non-invasively by the field generator treat cell populations inside a dish above the microscope lens, which can be observed in real time.

Keywords: Micro- and nano-sensors, Microfluidic applications, Translational issues in tissue engineering and biomaterials - Cell seeding and viability in scaffolds
Abstract: Microphysiometry showed to be a useful tool to monitor the energy metabolism of living cells and their interaction with other cells. In the past, the technique was mainly used for monitoring of 2D monolayers of cells. Recently, the cellasys group showed that it is possible to monitor the extracellular acidification rate (EAR) and transepithelial electrical resistance (TEER) of 3D skin constructs in an automated assay maintaining an air liquid interface (ALI) with the IMOLA-IVD technology.

Keywords: Stem cells - Tissue morphogenesis, Translational issues in tissue engineering and biomaterials, Scaffolds in tissue engineering - Self-assembled
Abstract: Interaction between immune cells and bone metabolism plays a crucial role in the pathogenesis of several disorders affecting the musculoskeletal system. However, current engineering approaches exclude the immune system due to its complexity. Here, we aim to present the importance of modelling the crosstalk between immune cells and bone as exemplified by mimicking the initial phase of fracture healing in vitro.

Keywords: Microfluidic techniques, methods and systems, Scaffolds in tissue engineering - Biofabrication, Translational issues in tissue engineering and biomaterials - Bioreactors
Abstract: Organ-on-chip systems are used to recreate the physiology of a certain tissue or organ of interest, thereby enabling more detailed investigations than conventional cell culture allows for. In this project a bone-on-a-chip system is presented that allows for control of oxygen levels and mechanical loading. Culturing osteoblasts in the bioreactor under mechanical load results in an increase in their capacity for matrix mineralization and therefore in their bone forming capability.

Keywords: Translational issues in tissue engineering and biomaterials - Cell seeding and viability in scaffolds, Scaffolds in tissue engineering - Self-assembled, Microfluidic techniques, methods and systems
Abstract: Multicellular tumor spheroids are 3D models generated from various cell types, including heterogeneous tissues of solid human tumors. They have proven utility for predictive testing of patient samples to identify the most promising pharmaceutical anti-cancer treatment.

Keywords: General and theoretical informatics - Natural language processing, General and theoretical informatics - Machine learning, Health Informatics - Electronic health records
Abstract: In this work, we present FREGEX a method for automatically extracting features from biomedical texts based on regular expressions. Using Smith-Waterman and Needleman-Wunsch sequence alignment algorithms, tokens were extracted from biomedical texts and represented by common patterns. Three manually annotated datasets with information on obesity, obesity types, and smoking habits were used to evaluate the effectiveness of the proposed method. Features extracted using consecutive sequences of tokens (n-grams) were used for comparison, and both types of features were mathematically represented using the TF-IDF vector model. Support Vector Machine and Naïve Bayes classifiers were trained, and their performances were ultimately used to assess the ability of the feature extraction methods. Results indicate that features based on regular expressions not only improved the performance of both classifiers in all datasets but also use fewer features than n-grams, especially in those datasets containing information related to anthropometric measures (obesity and obesity types).

Keywords: General and theoretical informatics - Natural language processing, General and theoretical informatics - Machine learning, Health Informatics - Electronic health records
Abstract: Diagnosis codes are used as a billing mechanism in the Electronic Health Record and have the capability to benefit decision support systems, which aim to assist coders by suggesting a relevant subset of potential codes to choose from. Due to the large set of possible labels and length of patient records, automatic ICD code assignment is considered to be a challenging task within the field of multi-label classification. This paper introduces a baseline for automatic ICD code assignment using Support Vector Machines (SVM) and FastText with Unified Medical Language System (UMLS) metathesaurus mappings into word embedding models. Training data is obtained from the Medical Information Mart for Intensive Care (MIMIC-III) database and extended with ’is-a’ relationships from ICD-9 hierarchy. FastText is evaluated with different label count estimations, of which an approach based on label cardinality yields a F1-Score of 62:2%. FastText achieves high recall results and mentionable performance improvements over previous models. Reported values are obtained through 10-fold cross-validation.

Keywords: General and theoretical informatics - Natural language processing, General and theoretical informatics - Data mining, Health Informatics - e-communities, social networks and social media
Abstract: Continuous monitoring the safe use of medication is an important task in pharmacovigilance. The first-hand experiences of medication effects come from the consumers of the pharmaceuticals. Social media have been considered as a possible alternative data source for gathering consumer-generated information of their experience with medications. Identifying personal experience in social media data is a challenging task in natural language processing. In this study, we investigated a method of predicating personal experience tweets using Google’s Bidirectional Encoder Representations from Transformers (BERT) and neural networks, in which BERT models contextually represented the tweet text. Both pre-trained BERT models and our BERT model trained with 3.2 million unlabeled tweets were examined. Our results show that our trained BERT model performs better than Google’s pre-trained models (p < 0.01). This suggests that domain-specific data may contribute to the BERT model yielding better classification performance in predicting personal experience tweets of medication use.

Keywords: General and theoretical informatics - Computational disease profiling, Health Informatics - Patient tracking, General and theoretical informatics - Natural language processing
Abstract: Amyotrophic lateral sclerosis (ALS) is a degenerative disease which causes death of neurons controlling voluntary muscles. It is currently assessed with subjective clinical measurements, but it would benefit from alternative surrogate biomarkers that can better estimate disease progression. This work analyzes speech and fine motor coordination of subjects recruited by the Answer ALS foundation using data from a mobile app. In addition, clinical variables such as speech, writing and total ALSFRS-R scores are also acquired along with forced and slow vital capacity. Cross-sectional and longitudinal analyses were performed using speech and fine motor features. Results show that both types of features are useful to infer clinical variables especially for males (R2=0.79 for ALSFRS-R total score), but their initial values are not helpful to predict speech and motor decline. However, we found that longitudinal progression for bulbar and spinal ALS onset are different and they can be identified with high accuracy by the extracted features.

Keywords: General and theoretical informatics - Predictive analytics, Health Informatics - Computer-aided decision making, General and theoretical informatics - Natural language processing
Abstract: Sepsis and septic shock are major concerns in public health as the leading contributors to hospital mortality and cost of treatment in the United States. Early treatment is instrumental for improving patient outcome; to this end, algorithmic methods for early prediction of septic shock have been developed using electronic health record data, with the goal of decreasing treatment delay. We extend a previously-developed method, using a gradient boosting algorithm (XGBoost) to compute a time-evolving risk of impending transition into septic shock, by combining physiological data from the electronic health record with features obtained from natural language processing of clinical note data. We compare two different methods for generating natural language processing features, with the best method obtaining improved performance of 0.92 AUC, 84% sensitivity, 82% specificity, 49% positive predictive value, and a median early warning time of 7.0 hours. This degree of early warning is sufficient to enable intervention many hours in advance of septic shock onset, with the improved prediction performance of this method resulting in fewer false alarms and thus more actionable predictions.

Keywords: Health Informatics - Mobile health, Health Informatics - eHealth, Health Informatics - e-communities, social networks and social media
Abstract: Suicide is a leading cause of death globally, and certain locations experience clusters of increased frequencies of suicidal behaviours. Prevention efforts are warranted in these locations to prevent contagion effects, and there is increasing interest in technology-supported suicide prevention interventions. Crisis support services are also implementing online and mobile health support. This study investigated the relationship between geospatial suicide clusters in the US and service use, and emergency responses initiated by, a text message-based crisis support service. 103,570 conversations involving 64,391 unique users over a two-year period were de-identified, analysed, and mapped to the state and county level. Moderate correlations were observed between service user rate and suicide mortality (rho=0.53), and active rescues and suicide mortality (rho=0.46). Suicide clusters were associated with a non-significant increase in service use (p=0.06) and active rescues (p=0.48). While it was not possible to observe significant cluster effects within this dataset, future analysis involving data collected through mobile health platforms may lead to better identification of risk at an individual level.

Keywords: Organ modeling, Models of organs and medical devices - Inverse problems in biology, Organs and medical devices - Multiscale modeling and the physiome
Abstract: This paper presents the methods and the materials towards characterizing frontal sinus anatomy and developing representative anatomical models which reflect the variance of the anatomy with three different sizes: small, medium and large. Anatomical characterization was performed using computer tomography data of up to 50 anonymous patients. Dimensional and volumetric measurements were conducted using the .stl files generated by segmentation and 3-D reconstruction. Three representative data sets were chosen to be realized in the form of models with frontal sinuses of small, medium and large sizes. The models include bone, mucosa and skin structures, whereas bone structures were manufactured by selective laser sintering of polyamide and the soft tissues by casting of gelatin and silicone. To ensure realistic optical and mechanical properties of the mucosa, verification tests were performed and the results were integrated into the manufacturing process.

Keywords: Models of organs and medical devices - Inverse problems in biology, Models of medical devices
Abstract: Detailed computational anatomical models of the entire head are needed for accurate in silico modeling in a variety of transcranial stimulation applications. Models from different subjects help to understand and account for population variability. To this end, we have developed a new library of head models of 20 individuals, segmented from co-aligned multi-modal medical image data. The acquired image modalities allow to accurately model tissues with different material properties, such as electrical conductivity or spatially varying acoustic properties. The usefulness of the models is illustrated for two example applications.

Keywords: Models of organs and medical devices - Inverse problems in biology, Computational modeling - Biological networks, Systems modeling - Decision making
Abstract: Damaged cardiac muscle tissue caused by ischemia leads to compromised cardiac function. While conventional imaging can view the ischemic tissue, currently there is no clinical way to quantitatively predict improved heart function after revascularization treatment. This increases the decision difficulty of treatment planning as there is no guarantee the heart function will improve enough to justify the cost of revascularization treatment. The complement of biomechanical modelling with conventional imaging offers an alternative method to determine the amount of ischemic tissue which can then be used as a potential predictor to estimate the range of functional improvement. A novel shape optimization technique is presented to predict the contractility of ischemic tissue in an in-silico left ventricle model that has suffered acute myocardial infarction. Preliminary results show that the proposed technique can reconstruct the damage caused by ischemic tissue within 18%. A range of minimum to maximum predicted cardiac improvement can then be given based on this error to help decide if the cost of revascularization treatment is justified.

Keywords: Models of organ physiology, Systems modeling - Decision making, Models of medical devices
Abstract: Modeling and simulation of human eye movement have a wide range of applications in many domains. Various attempts have been made to model and simulate eye movements in a physically accurate manner. All the existing models show limitations and problems in simulating secondary and tertiary eye movements. Recent investigation of pulley models (passive and active hypotheses) in representing human eye motion has recognized mathematical complexity in modeling eye behavior. Sophisticated techniques of modeling are required to investigate eye movements. This paper presents a procedure for eye movement simulation through geometrical modeling (an OpenSim script with its recent MATLAB binding) for binocular vision. First order neural dynamics with Millard’s muscle model are used to actuate six Extra Ocular Muscles (EOMs). Pulse-step inputs are used to generate the muscle forces around the eye globe. The implemented model is successful in simulating horizontal and vertical movements of the human eye with respect to the prescribed activation. The developed technique is evaluated using responses from lumped parameter models and EOG recordings.

Keywords: Models of organ physiology, Systems modeling - Decision making
Abstract: Facial palsies due to stroke, accidental and sportive injuries or sometimes without etiology, affect the professional and personal lives of involved patients. These disorders are not only a functional handicap but also a social integration impairment. The recovery of facial mimics with a normal and symmetrical facial expression allows involved patients to improve their living conditions and social identity. To monitoring facial mimics and head movements, a computer-aided animation system needs to be developed. Numerous systems have been proposed using single camera, stereo camera, 3-D scanner, and Kinect approaches. In particular, Kinect contactless sensor has been proven to be very suitable for 3-D facial simulation applications. However, little studies have employed the Kinect sensor for real-time head animation applications. Consequently, this study aimed to develop a real-time head and facial mimic animation system using the contactless Kinect sensor and system of systems approach. To evaluate the accuracy of the subject-specific Kinect-based geometrical models, magnetic resonance imaging (MRI) data were used. As results, the mean distance deviation between generated Kinect-based and reconstructed MRI-based geometrical head models are approximately 1 mm for two tested subjects. The generation times are 9.7 s ± 0.3 and 0.046 s ± 0.005 by using the full facial landmark and MPEG-4 facial landmark respectively. Real-time head and facial mimic animations were illustrated. Particularly, the system could be executed at a very high framerate (60 fps). Further developments relate to the integration of texture information and internal structures such as skulls and muscle network to develop a full subject specific head and facial mimic animation system for facial mimic rehabilitation.

Keywords: Data-driven modeling, Computational modeling - Structural bioinformatics, Computational modeling - Biological networks
Abstract: We develop a system-level approach to modelling optogenetic-neurons firing behaviors in in-vivo condition. This approach contains three sub-modules:1) a Mie/Rayleigh scattering mode of light penetration in tissue; 2) a classic likelihood Poisson spiking train model; 3) A 4-state model of the Channelrhodopsin-2(ChR2) channel with a CA3 based Hodgkin-Huxley model. We first investigate opto-neurons light-to-spike mechanisms in in-vivo: the background noises (synaptic currents) play a dominant role in generating spikes rather than light intensities which at in-vitro condition (Typically the required light intensity is less than 0.3 〖mW/mm〗^2at in-vivo). Then the spiking fidelity is quantitative analyzed at different background noise levels. Next, by combining light penetration profiles, we show that how neuron firing rates decay as tissue distance increases in a 2D dimension. This preliminary data clearly demonstrate that at given light stimulation protocol, the maximum effected distance in-vivo is 250 µm with small frequency decay rate, while at in-vitro is 50µm with considerable frequency decay rate. Therefore, the developed model can be used for designing sensible light stimulation strategies in-vivo and opto-electronics systems.

Keywords: Neural stimulation (including deep brain stimulation), Neuromodulation devices, Wireless technologies for interrogation of implantable therapeutic devices
Abstract: A novel minimally invasive wirelessly powered medical device, a magnetic induction extra-cranial brain stimulation (MI-ECBS) system is implemented for treating neurological disorders, Alzheimer’s disease (AD) and Epilepsy. The proposed system provides 2 different types of clinically significant stimulation waveforms for the therapy. For high frequency stimulation (HFS), we used 1mA, 10Hz, rectangular, charge balanced (0.5msec pulse width) pulses for 3sec with 21sec rest (total 600 pulses). Subsequently, under same configuration, a low frequency stimulation (LFS; 1Hz, 600 pulses) protocol was applied to canine-animal models. As a result, complementary neuro-modulation, facilitation and an inhibition are successfully demonstrated with an EEG power spectrum monitoring and the stimulation delivery efficacy is enhanced to 39.57x comparing to conventional transcutaneous direct current stimulation (tDCS).

Keywords: Diagnostic devices - Physiological monitoring, Muscle stimulation, Neural stimulation (including deep brain stimulation)
Abstract: This paper is concerned with the automatic evaluation of selected tasks performed by people with intellectual Disabilities. According to the International Classification of Functioning, Disability and Health (ICF) system, subjects must be divided into two groups: group with no difficulty (N) and group with difficulty (D) being this classification based on performances obtained in a conventional table (CT) soccer. Three tasks, with different levels of difficulty, were proposed for performance evaluation. Experimental results were obtained on the basis of the task execution in both a CT and a robotics table (RT) soccer. All participants were able to perform tasks with the joystick on the RT soccer and the automatic evaluation system identified differences in reaction times with and without red color flag in the participants, on RT soccer. One of the tasks was completely performed by all the participants by using the RT soccer.

Keywords: Muscle stimulation, Neural stimulation (including deep brain stimulation), Cardiovascular assessment and diagnostic technologies
Abstract: Abstract — The G.L.O.V.E (Generated Low Output Voltage Emitter) is a tool for law-enforcement professionals operating in a wider band of the force continuum than other intermediate weapons. The goal of this paper was to analyze the G.L.O.V.E electrical output and to compare it to safety and efficacy requirements of relevant international standards. Methods — Four G.L.O.V.Es were tested. Measurements on fresh, skinned animal tissue established a G.L.O.V.E operational impedance range of 140 – 300 ohm. Their voltage, current, charge, pulse duration and rate were measured when applied to loads from 140 to 330 ohm. Measurements were also taken with two G.L.O.V.E devices simultaneously applied to a resistive network which simulated a hand-hand, or shoulder-shoulder, current-flow path. Such path may involve cardiac risk. The measurement results were compared to relevant safety and efficacy requirements of following standards: UL 69, IEC 60335-7-26, IEC 60479-1 and -2, and ANSI/CPLSO-17. Results — Within its operational load range, the G.L.O.V.E device delivers maximum voltages in the range of 210 – 320 V, maximum currents of 0.9 – 1.5 A, charge levels of 84 – 125 µC with pulse durations between 105 – 115 µs, with repetition rates of 29.7 – 30.8 pps and duty cycles of 0.32 – 0.35%. These parameters were all within relevant ranges required by UL 69, IEC 60335-7-26, IEC 60479-1 and -2, and ANSI/CPLSO-17. Conclusion — Based on our measurement data, the G.L.O.V.E device is in compliance with relevant requirements for safety and efficacy stated by standards such as UL 69, IEC 60335-7-26, IEC 60479-1 and -2, and ANSI/CPLSO-17.

Keywords: Cardiac, Fibrillation, Safety, Standards, G.L.O.V.E, CEW.

Keywords: Defibrillators (implantable or external)
Abstract: Sudden cardiac death (SCD) is caused by a change in heart rhythm. Ventricular fibrillation (VF) is the most common cause of SCD, and electrical defibrillation is the most effective method of terminating VF. Miniaturization of defibrillators has extraordinary significance. In this study, we proposed a miniaturized design for the defibrillator charge/discharge circuit. The core circuit, composed by low-operating-voltage and surface-mount elements, achieved a size of 4.9 cm x 4.9 cm. We specially designed a voltage management system to power the system by a low-voltage battery. The proposed circuit could quickly charge the storage capacitor to 800 V within 8 s and release a biphasic exponential truncated waveform whose pulse width and delivered energy could be adjusted flexibly. With further optimization and suitable batteries and capacitors, this defibrillator charge/discharge circuit can be used for miniaturized wearable cardiac defibrillators (WCDs) and implantable cardioverter defibrillators (ICDs).

Keywords: Neural stimulation (including deep brain stimulation), Computer modeling for treatment planning, Neuromodulation devices
Abstract: In this paper we present a simulation framework for automated optimization of deep brain stimulation (DBS) parameters based on the hand kinematics signal as the feedback signal, in patients with essential tremor. We used Gaussian Process regression (GPR) models to develop patient-specific models for predicting the effect of DBS on the hand kinematics using the clinical data that was recorded during DBS programming. In this framework, we characterized the performance of a Bayesian Optimization method to identify the optimal DBS parameters that maximized the clinical efficacy. Our results demonstrate the feasibility of using black-box optimization methods for automated identification of optimal DBS parameters in clinical settings.

Keywords: Diagnostic devices - Physiological monitoring
Abstract: Contemporary cardiovascular device manufacturing companies still employ manual processes to assemble active or passive catheters. Further automation is limited due to the small size and fragility of the components. To overcome this obstacle we propose two novel methods to facilitate a highly cost efficient and time saving assembly process that emphasizes the use of thin-film, flexible printed circuit boards (FPCBs) made of liquid crystal polymer (LCP). The device that resulted from both processes was a esophageal diagnostic catheter for the detection of paroxysmal arrhythmias with an array of 14 cylindrically shaped electrodes placed in equidistant steps along the catheter tube. Small adhesion promoting holes were laser cut in to the LCP to create a mechanically durable interlocking structure with the solidified catheter tube material. The first presented process comprises a lamination procedure that thermo-mechanically embeds the FPCB in to the catheter tube. The second proposed process is laser welding the FPCB on to the catheter tube resulting in small welding seams that provide the catheter with increased adhesion properties. Peel strength tests of first catheter prototypes, manufactured with both processes, have shown that the resulting bond is strong and durable in accordance with accepted standards for intravascular catheters. Overall peel strength values of up to 9N with a circumferential coverage of 90 % have been observed. The catheter retained the required mechanical flexibility and strechabilty demanded from contemporary catheter designs intended for the introduction in to the human body. The presented methods could lead to a paradigm change in the manufacturing processes used for the production of invasive cardiovascular catheters.

Keywords: Pulmonary and critical care - Bioengineering applications in Intensive care, Cardiovascular and respiratory system modeling - Gas exchange models, Pulmonary and critical care - Pulmonary disease
Abstract: Primary blast lung injury (PBLI) is the most important component of a multisystem syndrome of injury resulting from exposure to an explosive shockwave. The majority of such casualties require ventilation in an intensive care unit. Using a novel primary blast lung injury simulator, we evaluate the potential use of continuous positive airway pressure (CPAP) in 6 in-silico casualties over 24 hours after injury. This study suggests that acute lung injury in PBLI can be effectively managed with continuous positive airway pressure. In an austere environment or in circumstances where medical resources are overwhelmed, CPAP using ambient air may be of considerable benefit.

Keywords: Cardiovascular and respiratory system modeling - Cardiovascular control models, Cardiovascular and respiratory signal processing - Time-frequency, time-scale analysis of cardiorespiratory variability
Abstract: The neural control systems that generate the drive to breathe and the autonomic modulation of heart function continuously integrate sensory feedback with current physiological needs, imparting a combination of random and temporally scaled characteristics into the variability of breath-to-breath interval (BBI) and heartbeat-to-heartbeat interval (RRI). Although there are analytical techniques that quantify the temporal or fractal-like scaling in BBI and RRI, methods to computationally reproduce or predict the behavior of these rhythms are unavailable. The purpose of this effort is to demonstrate the design and implementation of a stochastic and mathematically integrative model (SIM) of cardiorespiratory function that replicates the intrinsic combination of random and temporally-scaled characteristics of human breathing or heartrate. The first step is to extract BBI and RRI sequences from previously recorded heart and breathing rate measurements. Next, the inter-breath and inter-beat memory are estimated with an autocorrelation function. A discrete probability density functions (dPDF) is created for each sequence by fitting a polynomial curve to each normalized histogram. The SIM builds each artificial BBI or RRI sequence by randomly selecting interval values from a dPDF, and then integrating the interval series with a function that is optimized with the memory parameters from the autocorrelation analysis. Temporal scaling of the original and SIM-generated sequences is quantified with detrended fluctuation analysis. Previous studies using this technique have successfully reproduced the innate mix of stochastic and temporally scaled characteristics in BBI and RRI sequences, and most recently, a gender-specific difference in BBI patterns. In conclusion, the SIM is a new analytical technique that can reproduce the natural balance of stochastic and fractal scaling characteristics in human cardiorespiratory patterns.

Keywords: Pulmonary and critical care - Ventilatory Assist Devices, Respiratory transport, mechanics and control - Work of breathing, Cardiovascular and respiratory system modeling - Respiratory Control models
Abstract: Mechanically ventilated patients are often ventilated in support mode ventilation where the ventilator supports the patient’s spontaneous breathing effort. Support mode ventilation requires a balancing between over- and under-supporting the patient. This paper reviews modeling and evaluation work on a model-based decision support system for support mode mechanical ventilation.

Keywords: Pulmonary and critical care - Pulmonary function testing & instrumentation, Pulmonary and critical care - Pulmonary disease, Cardiovascular and respiratory signal processing - Complexity in cardiovascular or respiratory signals
Abstract: Lung diseases are one of the main causes of death worldwide. Identifying their phenotypes and staging their severity can be obtained through computed tomography (CT) images of the pathological site. However, CT can lead to a huge amount of data (many 2D-scans for one region are studied to represent its volume) that can be difficult to analyze and interpret. Therefore, we herein propose a new entropy-based measure, termed pseudo-three dimensional multiscale fuzzy entropy, to quantify the irregularity of high resolution CT scans (HRCT). Data from three groups of subjects are processed: one normal group and two other groups with chronic obstructive pulmonary diseases characterized by a progressive and permanent decline in lung function. The results are interesting for HRCT scan slices, allowing us to further extend this study to a larger number of patients in the future.

Keywords: Cardiac mechanics, structure & function - Cardiac muscle mechanics, Cardiovascular, respiratory, and sleep devices - Sensors, Cardiovascular and respiratory system modeling - Cardiovascular-Respiratory Interactions
Abstract: Cardio Respiratory Fitness (CRF) is a strong predictor of mortality. Because of this the American Heart Association recommends that ideally all patients should have measured CRF with a maximal effort test on a regular basis. However, this is not practically possible. Seismocardiography (SCG) is a measure of the vibration produced by the beating heart, measured on the sternum. This study investigates the correlation between amplitudes in the SCG and VO₂-max from 17 females and 9 males. All amplitudes investigated correlated significantly with VO₂-max.

Keywords: Implantable technologies, Implantable systems
Abstract: A preliminary study of capsule orientation estimation has been performed using a computational human body model [4]. It is shown that any directionality (or equivalently null) in the antenna radiation pattern can be exploited to sense changes in the orientation of the capsule through observation of the S21 vector at a set of on-body receivers. Two methodologies i.e. complex cross-correlation and Minkowski distance were used to assess capsule orientation estimation with respect to a reference position. The study was performed within the Ultra-Wide Band frequency range as this technology is considered to be an attractive candidate for the next generation of wireless capsule endoscopy.

Keywords: Wearable antennas and in-body communications, Modeling and analysis
Abstract: Wireless hearables have become increasingly popular in both consumer electronics and hearing aid markets. Due to its close proximity to the ear canal, a hearable device’s antenna design needs to take into account the bio-loading effect. The relative geometric relationship between antenna and bio-tissues around the ear contributes significantly to the antenna’s behavior. This study presents how the bio-loading impacts antennas’ performance and the sensitivity of bio-loading.

Keywords: Wearable body sensor networks and telemetric systems, Implantable systems, Modeling and analysis
Abstract: Accurate and numerically efficient human phantoms for computation electromagnetics studies represent a significant enabling technology in the areas of safety estimation, wearable device development and antenna design. When coupled with fast numerical methods, these phantoms accelerate the design process by replacing expensive and time consuming testing of potential solutions. In this way, engineers may examine and evaluate a design in ‘virtual space’ and determine if it is worth prototyping. Human phantoms also permit evaluation of important safety parameters related to electromagnetic energy absorption and corresponding heating near emitting devices that would otherwise not be obtainable. This work describes the use of several state-of-the-art human phantoms for numerical modeling of in- and on-body antenna designs engineered for communications applications. Relevant metrics are presented.

Keywords: Signal pattern classification, Connectivity measurements, Data mining and processing in biosignals
Abstract: A novel decoding scheme for motor imagery (MI) brain computer interfaces (BCI’s) is introduced based on the GFT concept. It considers the recorded EEG activity as a signal defined over (the graph of) the sensor array. A graph encapsulating the functional covariations emerging during the execution of a specific imagined movement is first defined, from a small training set of relevant trials. The ensemble of graphs signals corresponding to a multi-trial training dataset is then analyzed using a graph-guided decomposition and, based on discriminative Lasso (dLasso), an information-rich GFT subspace is defined. After training, only simple matrix operations are required for transforming the multichannel signal into features to be fed into a classifier that decides whether brain activity conforms with the graph structure associated with the targeted movement. The proposed decoding scheme is evaluated based on two different datasets and found to compare favorably against popular alternatives in the field.

Keywords: Signal pattern classification, Time-frequency and time-scale analysis - Time-frequency analysis, Data mining and processing - Pattern recognition
Abstract: Classification of various cognitive and motor tasks using electroencephalogram (EEG) signals is necessary for building Brain Computer Interfaces (BCI) that are noninvasive. However, achieving high classification accuracy in a multi-subject multitask scenario is a challenge. A noticeable reduction in accuracy is observed when the subjects between train and test are mismatched. Drawing a similarity from speaker adaptation approaches in speech, we propose a method to perform subject-wise adaptation of EEG in order to improve the task classification performance. A Common Spatial Pattern (CSP) approach is employed for feature extraction. Gaussian Mixture Model (GMM) based subject-specific models are built for each of the tasks. Maximum a-posterior (MAP) adaptation is performed, and an absolute improvement of 1.22-7.26% is observed in the average accuracy.

Keywords: Physiological systems modeling - Signal processing in physiological systems, Physiological systems modeling - Multivariate signal processing, Connectivity measurements
Abstract: In this work, we analyze brain-heart interactions during different mental states computing mutual information (MI) between the dynamic activity of different physiological systems. In 18 healthy subjects monitored in a relaxed resting state and during a mental arithmetic and a serious game task, multichannel EEG, one lead ECG, respiration and blood volume pulse were collected via wireless non-invasive biosensors. From these signals, synchronous 300-second time series were extracted measuring brain activity via the δ, θ, α, and β EEG power, and activity of the body district via the ECG R-R interval η, the respiratory amplitude ϱ and the pulse arrival time π. MI was computed using a linear estimator: (i) between {η,ϱ,π} and {δ,θ,α,β}, to measure overall brain-body interactions; (ii) between each time series and the others of the same district, to measure information shared within a district; and (iii) between each time series of a district and all series of the other district, to evaluate individual contributions to the information shared between brain and body. Results document the existence of statistically significant brain-body interactions, with high MI values involving mainly the η body dynamics and the δ and β brain dynamics. State-dependent variations were mostly relevant to the MI of the brain system involving δ, θ, α during mental arithmetic, and α and β during serious game. Thus, MI can be useful to detect correlated activity within and between brain and body systems monitored simultaneously during different mental states.

Keywords: Signal pattern classification, Physiological systems modeling - Signal processing in physiological systems
Abstract: Motor imagery (MI) based brain-computer interface systems (BCIs) are highly in demand for many real-time applications such as hands and touch-free text entry, prosthetic arms, virtual reality, movement of wheel chair, etc. Traditional sparse representation based classification (SRC) is a thriving technique in recent years and has been a successful approach for classifying MI EEG signals. To further improve the capability of SRC, in this paper, a weighted SRC (WSRC) has been proposed for classifying MI signals. WSRC constructs a weighted dictionary according to the dissimilarity information between the test data and the training samples. Then for the given test data the sparse coefficients are computed over the weighted dictionary using l_0-minimization problem. The sparse solution obtained using WSRC gives discriminative information than SRC and as a consequence, WSRC proves to be superior for MI EEG classification. The experimental results substantiate that WSRC is more efficient and accurate than SRC.

Keywords: Time-frequency and time-scale analysis - Time-frequency analysis, Signal pattern classification, Physiological systems modeling - Closed loop systems
Abstract: This study designed a brain-computer interface (BCI) with high frequency steady-state visual evoked potentials (SSVEPs) paradigm based on dual frequency modulation. The traditional low-frequency SSVEP-BCI communication speed is fast, but is very irritating to the human eye for long-term use. High-frequency stimulation can greatly improve the comfort of the system, but the communication rate of high-frequency systems is poor for the EEG response of high-frequency stimulation is weak. This study introduces a dual-frequency modulation method to improve the recognition accuracy of high-frequency BCI. Each target in the paradigm is composed of sinusoidal brightness modulated flicker light of the same initial phase with different stimulation frequencies in a space composition of the checkerboard. Using the above method, a relatively high-frequency SSVEP-BCI paradigm with a relatively complex code is proposed. Due to the complexity of the coding, only the training-based identification algorithm is used. With a data length of 0.5s, the average recognition accuracy is 91.02±7.77%, and the information transfer rates (ITR) is 267.85±39.36bits/min. The performance is higher than the existing high frequency SSVEP-based BCI paradigms.

Keywords: Neural networks and support vector machines in biosignal processing and classification
Abstract: Parkinson’s Disease (PD) is a neurodegenerative disorder that manifests through slowly progressing symptoms, such as tremor, voice degradation and bradykinesia. Automated detection of such symptoms has recently received much attention by the research community, owing to the clinical benefits associated with the early diagnosis of the disease. Unfortunately, most of the approaches proposed so far, operate under a strictly laboratory setting, thus limiting their potential applicability in real world conditions. In this work, we present a method for automatically detecting tremorous episodes related to PD, based on acceleration signals. We propose to address the problem at hand, as a case of Multiple-Instance Learning, wherein a subject is represented as an unordered bag of signal segments and a single, expert-provided, ground-truth. We employ a deep learning approach that combines feature learning and a learnable pooling stage and is trainable end-to-end. Results on a newly introduced dataset of accelerometer signals collected in-the-wild confirm the validity of the proposed approach.

Keywords: Ultrasound imaging - Cardiac, Cardiac imaging and image analysis
Abstract: Arrhythmia localization prior to catheter ablation is critical for clinical decision making and treatment planning. The current standard lies in 12-lead electrocardiogram (ECG) interpretation, but this method is non-specific and anatomically limited. Accurate localization requires intracardiac catheter mapping prior to ablation. Electromechanical Wave Imaging (EWI) is a high frame-rate ultrasound modality capable of non- invasively mapping the electromechanical activation in all cardiac chambers in vivo. In this study, we evaluate 3D- rendered EWI as a technique for consistently localizing the accessory pathway (AP) in Wolff-Parkinson-White (WPW) pediatric patients. A 2000 Hz EWI diverging sequence was used to transthoracically image 13 patients with evidence of ECG pre-excitation, immediately prior to catheter ablation and after successful ablation whenever possible. 3D-rendered activation maps were generated by co-registering and interpolating the 4 resulting multi-2D isochrones. A blinded electrophysiologist predicted the AP location on 12-lead ECG prior to ablation. Double-blinded EWI isochrones and clinician assessments were compared to the successful ablation site as confirmed by intracardiac mapping using a segmented template of the heart with 19 ventricular regions. 3D-rendered EWI was shown capable of consistently localizing AP in all the WPW cases. Clinical ECG interpretation correctly predicted the origin with an accuracy of 53.8%, respectively 84.6% when considering predictions in immediately adjacent segments correct. Our method was also capable of assessing the difference in activation pattern from before to after successful ablation on the same patient. These findings indicate that EWI could inform current diagnosis and expedite treatment planning of WPW ablation procedures.

Keywords: Ultrasound imaging - Breast, Image analysis and classification - Machine learning / Deep learning approaches, Image feature extraction
Abstract: Breast calcifications indicate the high possibility of malignancy in the radiological assessment of breast lesions. However, it is difficult to detect them from traditional B-mode ultrasound images due to the resolution limit and speckle noise. In this paper, we proposed a novel automatic calcification detection method based on ultrasound radio frequency (RF) signals by quantitative multi-parameter fusion. The proposed method consists of four steps: selecting the region of interest (ROI), extracting multiple features on sliding windows that traverse the entire ROI, classifying the window with or without calcifications using the Adaptive Boosting classifier, and obtaining the detection result by a threshold filter. Experiments were conducted on a database of 130 experienced doctor-proven breast tumors with calcifications. Compared to manual annotation, the proposed method achieved an average accuracy of 88%. The experiments demonstrated that our computerized RF signals feature system was capable of helping radiologists detect tumor calcifications more accurately and provided more guidance for the final decision.

Keywords: Ultrasound imaging - Doppler, Ultrasound imaging - Elastography, Ultrasound imaging - Vascular imaging
Abstract: Methods used in clinical practice to diagnose and monitor atherosclerosis present limitations. Imaging the mechanical properties of the arterial wall has demonstrated the potential evaluate plaque vulnerability and assess the risk for stroke. Adaptive PWI is a non-invasive ultrasound imaging technique, which automatically detects points of spatial mechanical inhomogeneity along the imaged artery and provides piecewise stiffness characterization. The aims of the present study are to: 1) demonstrate the initial feasibility of adaptive PWI to image the mechanical properties of an atherosclerotic plaque 2) demonstrate the feasibility to combine adaptive PWI with vector Doppler in a single imaging modality in order to simultaneously obtain information plaque mechanical properties and plaque hemodynamics. The common carotid arteries of 1 healthy subject and 2 carotid artery disease patients were scanned in vivo. One of the patients underwent carotid endarterectomy and a plaque sample was retrieved. In this patient, a higher compliance value of the stenotic segment was estimated by Adaptive PWI as compared with the adjacent arterial wall, and the healthy carotid artery. This was corroborated by histological staining of the plaque sample, which revealed the presence of a large necrotic core and a thrombus, characteristics associated with reduced plaque stiffness. Moreover, the same sequence demonstrated the feasibility to obtain both stiffness maps and vector flow information, showing promise in atherosclerosis diagnosis and patient care.

Keywords: Ultrasound imaging - Elastography, Image analysis and classification - Machine learning / Deep learning approaches
Abstract: Time delay estimation (TDE) is a critical and challenging step in all ultrasound elastography methods. A growing number of TDE techniques require an approximate but robust and fast method to initialize solving for TDE. Herein, we present a fast method for calculating an approximate TDE between two radio frequency (RF) frames of ultrasound. Although this approximate TDE can be useful for several algorithms, we focus on GLobal Ultrasound Elastography (GLUE), which currently relies on Dynamic Programming (DP) to provide this approximate TDE. We exploit Principal Component Analysis (PCA) to find the general modes of deformation in quasi-static elastography, and therefore call our method PCA-GLUE. PCA-GLUE is a data-driven approach that learns a set of TDE principal components from a training database in real experiments. In the test phase, TDE is approximated as a weighted sum of these principal components. Our algorithm robustly estimates the weights from sparse feature matches, then passes the resulting displacement field to GLUE as initial estimates to perform a more accurate displacement estimation. PCA-GLUE is more than ten times faster than DP in estimation of the initial displacement field and yields similar results.

Keywords: Ultrasound imaging - Elastography
Abstract: An instrumented ultrasound probe system is designed to estimate the distribution of pressure applied across the probe face during image acquisition. The pressure distribution is used to investigate the effects of varying, and non-uniform, pressure distributions on shear wave elastography measurements in phantoms and ex vivo samples. Pressure distribution has a notable effect on shear wave elastography of ex vivo samples. The gradient in applied pressure across the probe face is mirrored in the gradient of elasticity measurements across the ultrasound elastogram image.