Keywords: Neural stimulation, Brain functional imaging - Source localization, Neural signal processing
Abstract: Individually targeted multi-channel transcranial Electric Stimulation (tES) has been suggested as a promising approach for manipulation of brain networks. Our somatosensory study investigates the effect of individualizing the targeting using combined Electro- and Magneto- EncephaloGraphy (EMEG) source analysis, so that the stimulation montage could be subsequently optimized for multi- electrode tES. We focus on the P20/N20 component of combined somatosensory evoked potential (SEP) and field (SEF) data and use calibrated realistic finite element method (FEM) head volume conductor models for targeting and optimization. Individual source analysis results, differing especially in the source orientation components and the resulting differences in optimized tES montages are presented.

Keywords: Neural stimulation, Brain physiology and modeling
Abstract: Individualized volume conductor models of the human head anatomy have become increasingly important in non-invasive brain stimulation (NIBS) for accurate simulation of the electric field distribution in the head. The first step in generating such head models is the segmentation of a structural magnetic resonance imaging (MRI) scan into different tissue classes. Here, we describe an atlas-based segmentation framework, which works robustly with different MRI sequences and achieves high segmentation accuracies.

Keywords: Neural stimulation, Brain physiology and modeling, Brain functional imaging - Segmentation
Abstract: Controlling dose parameters in transcranial current stimulation (tCS) still yields substantial intersubject variability in the results of stimulation. One cause for this variability is related to individual differences in head geometry, which lead to different electric field distributions for the same dose parameters. In this work we discuss a pipeline to control for these differences across subjects, by employing montage optimization combined with personalized head models obtained from subject-specific MRIs.

Keywords: Neuromuscular systems - Computational modeling, Neural stimulation - Deep brain, Neuromuscular systems - Central mechanisms
Abstract: First results on modeling transcranial electric stimulation via boundary element fast multipole method are reported including convergence rate. Some technical questions such as imprinting electrodes are detailed.

Keywords: Neural stimulation
Abstract: We review modelling studies of transcutaneous spinal cord direct current stimulation, report on our own findings and discuss the need for individualized dosing.

Keywords: Time-frequency and time-scale analysis - Empirical mode decomposition in biosignal analysis
Abstract: Photoplethsmography (PPG) is as a non-invasive photometric technique utilised extensively in pulse oximetry for the measurement of arterial blood oxygen saturation (SpO2). The focus of this paper will be the utilization of PPG at different organs and tissues.

Keywords: Physiological systems modeling - Multivariate signal processing
Abstract: This study aimed to quantitatively compare the volume distensibility difference between older and younger subjects using an external cuff equivalent to the arm length (around 50 cm) around the arm to induce external pressure in combination with a simple non-invasive pulse method for assessing the healthiness of the arteries. The evaluation was compared with the traditional technique without external pressure.

Keywords: Time-frequency and time-scale analysis - Time-frequency analysis
Abstract: This study investigated the accuracy of respiratory frequency (RF) derived from photoplethysmographic (PPG) signals measured from different body sites under normal breathing pattern. The results concluded that the relatively better site for accurate RF estimation was the forehead for normal breathing pattern.

Keywords: Physiological systems modeling - Signal processing in physiological systems, Data mining and processing in biosignals, Signal pattern classification
Abstract: There has been a rapid growth in R&D activity for the vascular optics technique photoplethysmography (PPG), particularly in device development, physiological measurements, advanced signal processing, and cardiovascular assessment. PPG can be low-cost and simple-to-do; it can be miniaturized and employ sophisticated signal processing to extract diagnostic information - including using modern machine learning. This paper (from my mini-symposium: ‘PPG measurements, advanced signal processing and cardiovascular applications’) summarizes a range of published work undertaken by the Newcastle group on innovative multi-site photoplethysmography (MPPG) measurement and analysis.

Keywords: Physiological systems modeling - Signal processing in physiological systems, Physiological systems modeling - Signals and systems
Abstract: Photoplethysmography (PPG) carries important information on cardiovascular diseases. Many useful features have been extracted from PPG signal for applications of vital physiological signal measurement and mobile healthcare. This work analyzed the effects of aging on five commonly-used PPG features, including pulse peak time, dicrotic notch time, reflection index, and two areas of the two PPG segments separated by the dicrotic notch. The PPG data were recorded from the right earlobe and fingertip of participants in five age groups (ages ranging from 20 to over 60 years). Results showed the effects of aging on the PPG waveform characteristics, and that the impact of the tidal wave in PPG waveform to pulse peak time is much smaller than to the other features.

Keywords: Data mining and processing in biosignals
Abstract: Arterial pulse wave analysis has been attempted to monitor the maternal physiological changes of circulatory system during pregnancy. This study aimed to quantify the difference of Gaussian modelling characteristics derived from radial pulses measured from the three trimesters of healthy pregnant women. Radial pulses were recorded from seventy pregnant women between gestational week 11-13, week 20-22 and then week 37-39. They were then normalized and decomposed into three independent Gaussian waves for deriving four key modelling characteristic parameters: including the peak time interval (T) and peak amplitude ratio (R) between the first and second Gaussian waves (T1,2 and R1,2), and their corresponding values between the first and third Gaussian waves (T1,3 and R1,3). Post-hoc multiple comparisons after analysis of variance was then applied to study the within-subject differences in Gaussian modelling characteristics between the three trimesters. The key results were that T1,2 and T1,3 increased significantly (T1,2: 12.8±1.3 vs 13.2±1.3, p<0.05; T1,3: 39.5±4.3 vs 45.4±5.1, p<0.001), and R1,3 decreased significantly from the first to second trimester (0.60±0.15 vs 0.53±0.11, p<0.001). From the second to third trimester, T1,2 decreased significantly (13.2±1.3 vs 12.8±1.2, p<0.01), and T1,3 and R1,3 decreased slightly but non-significantly. Since larger T1,2 and T1,3 and smaller R1,3 are associated with more compliant peripheral arteries, our results indicated that peripheral arteries become more compliant from the first to second trimester and then have a tendency of returning to baseline during normal pregnancy. In conclusion, this study has quantitatively demonstrated significant changes of Gaussian modelling characteristics derived from radial pulses at the three trimesters of normal pregnant women, suggesting that these modelling characteristics could be used as parameters in monitoring maternal physiological changes during normal pregnancy.

Keywords: Novel imaging modalities, CT imaging, Multimodal imaging
Abstract: Magnetic Particle Imaging (MPI) is a recent imaging modality depicting the distribution of superparamagnetic iron oxide nanoparticles. Since it does not show anatomic background it needs to be combined with other techniques in order to unfold its full potential for diagnostics. In this paper the first combined scanner with computed tomography (CT) is presented, which allows simultaneous data acquisition for both modalities.

Keywords: Novel imaging modalities, Image reconstruction and enhancement - Tomographic reconstruction
Abstract: In magnetic particle imaging, the selection field is important for confining the particle response to a defined region to enable the reconstruction of their spatial distribution. A system design featuring a high gradient Halbach permanent magnet field-free line is presented.

Keywords: Image reconstruction - Performance evaluation
Abstract: Experimental studies in the fields of magnetic particle spectroscopy (MPS) and magnetic particle imaging (MPI) are presented. Using these results, we examine the quantitative nature of MPS and MPI measurement signals, and the extent to which this translates into quantitative mapping of tracer distributions in reconstructed images from MPI. Using specially developed phantoms, we study the effect of various parameters which may adversely affect the capability for quantitative imaging in MPI, and examine the temporal stability of an MPI scanner over an extended time period. We discuss the outlook for the future in terms of developing and validating reliable quantitative MPI technology, and the potential applications thereof.

Keywords: Image enhancement, Image reconstruction and enhancement - Tomographic reconstruction
Abstract: The tomographic imaging method magnetic particle imaging is capable of imaging magnetic nanoparticles with high sensitivity. During in-vivo experiments it happens frequently that high particle concentrations shadow smaller ones since the dynamic range of MPI is limited to less than two orders of magnitude. Within this work we will introduce an algorithmic approach for increasing the dynamic range by using a two-step method that reconstructs the high and the low concentration parts in two individual steps.

Keywords: Magnetic sensors and systems, Sensor systems and Instrumentation, Portable miniaturized systems
Abstract: We present results of our conformal pediatric MEG system. It consists of 128 microfabricated OPMs, arranged on a flexible helmet. The OPMs are connected to a small control rack outside the shielded room.

Keywords: Magnetic sensors and systems, Physiological monitoring - Instrumentation
Abstract: We present the first proof of concept of magnetoencephalographic (MEG) signals recorded with 4He Optically Pumped Magnetometers (OPMs). The main advantage of this kind of OPM is the possibility to provide a tri-axis vector measurement of the magnetic field at room-temperature (the 4He gas is neither cooled nor heated).

Keywords: Magnetic sensors and systems, New sensing techniques
Abstract: Magnetoencephalography is a well establish technique with numerous SQUID (superconducting quantum interference device) systems worldwide. Over the past decade optical magnetometry has seen a rapid progress, offering an alternative for measuring biomagnetic fields: optically-pumped magnetometers (OPM). To make the most of optical magnetometry we followed an individual anatomy-derived approach and designed helm shaped sensor holder based on magnetic resonance images.

Keywords: Magnetic sensors and systems
Abstract: Magnetoencephalography (MEG) is used by neuroscience researchers and clinicians to understand and treat the human brain in health and disease. While MEG is unique in terms of functional neuroimaging spatial and temporal resolutions, the sensors utilized today pose many limitations because of their extreme cryogenic operating temperatures. A transition to so-called high-Tc SQUID sensors can increase signal levels and enable unique flexibility for improved neuroimaging on a broader spectrum of the population. To those ends, we develop high-Tc SQUID-based MEG systems wherein all sensors are in close proximity to the scalp surface. Herein, we present comparisons of such systems with a commercial MEG one, with focus on our newly-developed seven-channel on-scalp MEG system.

Keywords: Magnetic sensors and systems, Integrated sensor systems, Sensor systems and Instrumentation
Abstract: We fabricated a multichannel optically pumped magnetometer and carried out magnetic field distribution measurements with it. The results demonstrated that the magnetometer has sufficient sensitivity for magnetoencephalography and high spatial resolution.

Keywords: Signal pattern classification - Markov models, Physiological systems modeling - Signal processing in physiological systems, Data mining and processing in biosignals
Abstract: The advent of portable cardiac monitoring devices has enabled real-time analysis of cardiac signals. These devices can be used to develop algorithms for real-time detection of dangerous heart rhythms such as ventricular arrhythmias. This paper presents a Markov model based algorithm for real-time detection of ventricular tachycardia, ventricular flutter, and ventricular fibrillation episodes. The algorithm does not rely on any pre-processing or peak annotation of the original signal. When evaluated using ECG signals from three publicly available databases, the model resulted in an AUC of 0.96 and F1-score of 0.91 for 5-second long signals and an AUC of 0.97 and F1-score of 0.93 for 2-second long signals.

Keywords: Signal pattern classification, Neural networks and support vector machines in biosignal processing and classification, Nonlinear dynamic analysis - Biomedical signals
Abstract: Objective: This study concentrates on subtle electrocardiogram (ECG) spatiotemporal characteristics in the repolarization phase, and describes a deterministic learning-based methodology for the detection of abnormal cardiac dynamics induced by ischemia. Methods: ST-T complex of the surface 12-lead ECG signals are identified and extracted. Cardiac dynamics underlying ST-T complex signals is captured using deterministic learning algorithm. This kind of dynamics information represents the beat-to-beat temporal change of electrophysiological modifications in ventricular repolarization, which is shown to be sensitive to the variance during myocardial ischemia. Cardiodynamicsgram (CDG) is proposed as the three-dimensional graphic representation of cardiac dynamics information. Results: Encouraging evaluation results are achieved on electrocardiograms from public PTB database and hospital patients. Significant correlations are found between the CDG morphology and ischemia. Conclusion: Anormal dynamics of cardiac repolarization during ischemia can be detected using a deterministic learning-based methodology. The extracted cardiac dynamics information within routine ECG is expected to provide early detection for latent ischemia before obvious pathological changes are present in ECG. Significance: The proposed techniques can be considered as a complementary tool to the generally accepted ECG method for detection of abnormal dynamics in cardiac repolarization, which are important for identifying patients at risk of myocardial ischemia.

Keywords: Data mining and processing in biosignals
Abstract: Electrocardiography (ECG) signal analysis is an effective method for diagnosis of heart disease. However, the quality of ECG, corrupted by artifacts, limits the automatic ECG classification. In order to extract good quality ECG, we proposed a new ECG enhancement method based on tunable Q-factor wavelet transform (TQWT). In the proposed method, the original ECG signal was decomposed into high Q-factor component and low Q-factor component with dual-Q TQWT. According to the morphological of P, QRS, T waves in ECG, low Q-factor component was chosen for the representation of ECG. The proposed method was tested on 52 healthy volunteers and 52 myocardial infarction patients from the open dataset of PTB diagnostic ECG. A total of 288 features, covering time, frequency, nonlinear, and entropy domains, were extracted from R-R interval and ECG (in the window of 5s) across 12 leads. The features were selected by Relief method, and 22 discriminate features were fed into six different classifiers. The classification accuracy for dual-Q TQWT was 86.3%, which was 4.7% higher than the filtered data based on k-nearest neighbors (KNN) algorithm. The comparison results verified that the proposed dual-Q TQWT method provides good feasibility for ECG de-noising.

Keywords: Signal pattern classification, Neural networks and support vector machines in biosignal processing and classification, Data mining and processing in biosignals
Abstract: The classification of the heartbeat type is an essential function in the automatical electrocardiogram (ECG) analysis algorithm. The guideline of the ANSI/AAMI EC57 defined five types of heartbeat: non-ectopic or paced beat (N), supraventricular ectopic beat (S), ventricular ectopic beat (V), fusion of a ventricular and normal beat (F), pace beat or fusion of a paced and a normal or beat that cannot be classified (Q). In the work, a deep neural network based method was proposed to classify these five types of heartbeat. After removing the noise from ECG signals by a low-pass filter, the two-lead heartbeat segments with 2-s length were generated on the filtered signals, and classified by an adaptive ResNet model. The proposed method was evaluated on the MIT-BIH Arrhythmia Database with the patient-specific pattern. The overall accuracy was 98.6% and sensitivity of N, S, V, F were 99.4%, 85.4%, 96.6%, 90.6% respectively. Experimental results show that the proposed method achieved a good performance, and would be useful in the clinic practice.

Keywords: Signal pattern classification, Time-frequency and time-scale analysis - Wavelets, Data mining and processing - Pattern recognition
Abstract: Rhythm annotation of out-of-hospital cardiac episodes (OHCA) is key for a better understanding of the interplay between resuscitation therapy and OHCA patient outcome. OHCA rhythms are classified in five categories, asystole (AS), pulseless electrical activity (PEA), pulsed rhythms (PR), ventricular fibrillation (VF) and ventricular tachycardia (VT). Manual OHCA annotation by expert clinicians is onerous and time consuming, so there is a need for accurate and automatic OHCA rhythm annotation methods. For this study 852 OHCA episodes of patients treated with Automated External Defibrillators (AED) by the Emergency Medical Services of the Basque Country were analyzed. Six expert clinicians reviewed the electrocardiogram (ECG) of 4214 AED rhythm analyses and annotated the rhythm. Their consensus decision was used as ground truth. There were a total of 2418 AS, 294 PR, 1008 PEA, 472 VF and 22 VT. The ECG analysis intervals were extracted and used to develop an automatic rhythm annotator. Data was partitioned patient-wise into training (70%) and test (30%). Performance was evaluated in terms of per class sensitivity (Se) and F-score (F1). The unweighted mean of sensitivity (UMS) and F-score were used as global performance metrics. The classification method is composed of a feature extraction and denoising stage based on the stationary wavelet transform of the ECG, and on a random forest classifier. The best model presented a per rhythm Se/F1 of 95.8/95.7, 43.3/52.2, 85.3/81.3, 94.2/96.1, 81.9/72.2 for AS, PR, PEA, VF and VT, respectively. The UMS for the test set was 80.2%, 2-points above that of previous solutions. This method could be used to retrospectively annotate large OHCA datasets and ameliorate the workload of manual OHCA rhythm annotation.

Keywords: Signal pattern classification, Physiological systems modeling - Signal processing in physiological systems
Abstract: Biomedical signal analysis often depends on methods to detect and distinguish abnormal or high noise/artifact signal from normal signal. A novel unsupervised clustering method suitable for resource constrained embedded computing contexts, classifies arterial blood pressure (ABP) beat cycles as normal or abnormal. A cycle detection algorithm delineates beat cycles, so that each cycle can be modeled by a continuous time Fourier series decomposition. The Fourier series parameters are a discrete vector representation for the cycle along with the cycle period. The sequence of cycle parameter vectors is a non-uniform discrete time representation for the ABP signal that provides feature input for a clustering algorithm. Clustering uses a weighted distance function of normalized cycle parameters to ignore cycle differences due to natural and expected physiological modulations, such as respiratory modulation, while accounting for differences due to other causes, such as patient movement artifact. Challenging cardiac surgery patient signal examples indicate effectiveness.

Keywords: Motor learning, neural control, and neuromuscular systems, Neuromuscular systems - Postural and balance
Abstract: We developed a new technique to measure the contributions of rapid visuomotor feedback responses to the stabilization of a simulated inverted pendulum. Human participants balanced an inverted pendulum simulated on a robotic manipulandum. At a random time during the balancing task, the visual representation of the tip of the pendulum was shifted by a small displacement to the left or right while the motor response was measured. This response was either the exerted force against a fixation position, or the motion to re-stabilize the pendulum in the free condition. Our results demonstrate that rapid involuntary visuomotor feedback responses contribute to the stabilization of the pendulum.

Keywords: Motor learning, neural control, and neuromuscular systems, Neuromuscular systems - Postural and balance, Neuromuscular systems - Learning and adaption
Abstract: We recently developed a simulated inverted pendulum in order to examine human sensorimotor control strategies for stabilization. This simulated system allows us to manipulate the visual and haptic feedback independently from the physical dynamics of the task. Here we examine the effect of sensory delay in a balancing task. Human participants attempted to balance an inverted pendulum (simulated on a robotic manipulandum) with three different added delays (25, 50, and 75 ms), where the same delay was added to both the visual and haptic feedback. Increasing sensory delays decreased the ability of the participants to stabilize the pendulum. Investigation into the online control of the pendulum showed that with longer delays participants reduced their movement frequency but increased the amplitudes of their corrections.

Keywords: Motor learning, neural control, and neuromuscular systems, Neurological disorders - Diagnostic and evaluation techniques, Neurological disorders - Stroke
Abstract: Pediatric hemiplegia, caused by a unilateral brain injury during childhood, can lead to motor deficits such as weakness and abnormal joint torque coupling patterns which may result in a loss of independent joint control. It is hypothesized that these motor impairments are present in the paretic lower extremity, especially at the hip joint where extension may be abnormally coupled with adduction. Previous studies investigating lower extremity isometric joint torques in children with spastic cerebral palsy used tools that limited data collection to one degree of freedom, making it impossible to quantify these coupling patterns. We describe the adaptation of a multi-joint lower extremity isometric torque measurement device to allow for quantification of weakness and abnormal joint torque coupling patterns at the hip in the pediatric population. We also present preliminary data in three children without hemiplegia to highlight how the presence of atypical femoral bony geometry, often observed in childhood hemiplegia, can be accounted for in the Jacobian transformations and affect joint torque measurements at the hip.

Keywords: Motor learning, neural control, and neuromuscular systems, Neuromuscular systems - EMG processing and applications, Neuromuscular systems - Locomotion
Abstract: The muscle synergy theory was widely used in literature to assess the modular organization of the central nervous system (CNS) during human locomotion. The extraction of muscle synergies may be strongly influenced by the pre-processing techniques applied to surface electromyographic (sEMG) signals. The aim of this contribution is to assess the robustness improvement in muscle synergy extraction obtained using an innovative pre-processing technique with respect to the standard procedure. The new pre-processing technique that we propose is based on the extraction of principal muscle activation intervals (necessary to accomplish a specific biomechanical task during gait) from the original sEMG signals, discarding the secondary muscle activation intervals (activations that occur only in some strides with auxiliary functions). Results suggest that the extraction of the principal activation intervals from sEMG provide a more consistent and stable description of the modular organization of the CNS with respect to the standard pre-processing procedure.

Keywords: Motor learning, neural control, and neuromuscular systems, Neurorehabilitation, Human performance - Sensory-motor
Abstract: Visual amplification of kinematic errors has successfully been applied to improve performance for upper limb movements. In this study, we investigated whether visual error augmentation can promote faster adaptation during a full-body balance task. Healthy volunteers controlled a cursor by shifting their weight on the THERA-Trainer coro platform. Two experimental groups and one control group were asked to reach visual targets. For the experimental groups, the cursor’s deviation from the ideal straight line trajectory was augmented by a gain of 1.5 and 2, respectively, while the control group did not experience visual error amplification (gain of 1). Error augmentation with a gain of 1.5 enhanced the speed and the amount of motor adaptation, while the highest gain might have decreased the stability of adaptation. As visual feedback is commonly used in balance training, our preliminary data suggest that integrating visual error augmentation in postural exercises may facilitate balance control.

Keywords: Motor learning, neural control, and neuromuscular systems, Human performance - Sensory-motor, Neuromuscular systems - Computational modeling
Abstract: Motor adaptation studies can provide insight into how the brain handles ascending and descending neural signals during motor tasks, revealing how neural pathologies affect the capacity to learn and adapt to movement errors. Such studies often involve reaches towards prompted target locations, with adaptation and learning quantified according to Euclidean distance between reach endpoint and target location. This paper describes methods to calculate steady-state error using knowledge of the distribution of univariate, bivariate, and multivariate steady-state reaches. Additionally, in cases where steady-state error is known or estimated, it does not fully describe underlying reach distributions that could be observed at steady-state. Thus, this paper also investigates methods to describe univariate, bivariate, and multivariate steady-state reaching behavior using knowledge of the estimated steady-state error. These methods may yield a clearer understanding of adaptation and steady-state reaching behavior, allowing greater opportunities for inter-study comparison and modeling.

Keywords: Sensor Informatics - Physiological monitoring, Sensor Informatics - Body sensor networks, Health Informatics - Telehealth
Abstract: This preliminary study investigates the potential of predicting on-field recovery rates of athletes to inform performance management. A finite-state machine (FSM) combined with a exponential model is presented to predict recovery times. The results show that this approach yields positive initial results based on well-established physiological measurements and supports further exploration of other respiratory variables for on-field performance prediction. The aim of in-game recovery time prediction provides a new method to manage athletes in real-time.

Keywords: Health Informatics - Personal/consumer health informatics, Health Informatics - Mobile health, Health Informatics - Outcome research
Abstract: ECG contains a wealth of physiological information but requires multiple adhesive electrodes and precision placement. This reduces patients' compliance (due to discomfort) and signal quality (due to poor electrode positioning). Wearable PPG, however, can work on a single site with no adhesives. This reduces the time and expertise needed for application and minimizes patient irritation. For detection of heart rate and arrhythmias (such as atrial fibrillation), the application of ECGs are surplus to the requirements, while PPG-based wearables offer a practical solution for real-world applications.

A key factor in denoising and interpreting waveform data is the personalized and physiological context that generated the data. Quantifying these contexts creates simple building blocks from which the waveform analysis can be improved and thereby more readily adapted to a variety of patient monitoring environments, from the bedside monitoring to ambulatory clinics and homecare. We focus on the benefits of multivariate and multichannel adaptive algorithms based on signal quality indices to reduce false readings and reconstruct correct readings from noisy signals.

Keywords: General and theoretical informatics - Algorithms, General and theoretical informatics - Artificial Intelligence, General and theoretical informatics - Computational phenotyping
Abstract: The continuous monitoring of patients' vital signs is an ideal setting to showcase the success of personalized medicine. A patient's vital signs are invaluable in settings from acute care to homecare. It is widely accepted that population-based approaches to monitoring vital signs are hindered by high inter-patient variability. However, data collection is even more fundamental, and requires patient-specific analysis in its own right. Patient-specific approaches to real-time monitoring of (i) data quality and (ii) a patient's adherence to protocol is far from trivial. Fortunately, these same methods feed directly into patient-specific machine learning algorithms such as those used to identify patient vital sign phenotypes or forewarn physiological deterioration deterioration.

Keywords: Computational modeling - Biological networks, Model building - Network modeling, Models of medical devices
Abstract: We have entered the age of the neural connectome — connectivity maps between neural groups in the brain, spinal cord, and peripheral nervous system. What are the goals and requirements of connectomes? We enumerate components of connectome models that will shed light on the operation of the nervous system and open the door to pharmaceutical and electroceutical methods to modulate central and peripheral nervous system activity at will and treat disease and disorder with innumerable new targets. We give examples of connectomes for neuromodulation of the human spinal cord for neuropathic pain, vagus nerve stimulation for epilepsy. and depression for modeling early drug development.

Keywords: Computational modeling - Biological networks, Model building - Network modeling
Abstract: Local field potentials (LFPs) of the subthalamic nucleus (STN) are being studied in Parkinson’s patients as a potential tool to optimize the localization of abnormal neural network pathways in order to apply neuromodulation therapies that can alleviate the symptoms of Parkinson’s disease. Using a computation based model of key neural circuits thought to be involved in Parkinson’s disease we were able to demonstrate changes in power in the beta band that correlates to clinical data obtained from humans.

Keywords: Systems biology and systems medicine - Modeling of biomolecular system dynamics
Abstract: Computational modeling has become more efficient and widespread in its use in creating hypotheses for testing in vivo and in vitro and in helping to optimize current neural interface devices, such as brain and peripheral stimulators. While the scale of the biological system being examined by modeling varies from ion channel protein sub-component dynamics to large-scale rudimentary neural network nodes with only simplified anchoring to the underlying biophysics of actual neurons, there remain concerns that neural circuit dynamics can be modeled to reproduce almost any phenomenon if parameter adjustments are made appropriately. Since many parameters of the actual anatomy and physiology involved in neural circuits are unknown or only imperfectly known, a large amount of latitude is thought to be afforded to a model in what it purports to show about the underlying circuitry being examined. This analysis and study discusses the roots of complex systems homeostasis and stability, reviewing the concepts of robustness and Lyapunov functions, while also showing results from basic densely interconnected circuitry models wherein single and multiple parameter changes result in marked stability in the overall dynamics of the neurons in the model.

Keywords: Modeling of cell, tissue, and regenerative medicine - Ionic modeling, Models of medical devices, Organs and medical devices - Multiscale modeling and the physiome
Abstract: Improvements in the efficacy of retinal neuroprostheses can be achieved through more sophisticated neural stimulation strategies that enable selective activation of specific retinal ganglion cells (RGCs). Computational models are particularly well suited for such tasks. Stimulating electric fields can be accurately reconstructed, and target neurons can be ‘probed’ at resolutions well beyond those achievable by today’s state-of-the-art biological techniques. In this study, we used a population-based computational model to explore the ability of subretinal electrical stimulation to differentially activate ON and OFF RGC subtypes.

Keywords: Model building - Algorithms and techniques for systems modeling, Computational modeling - Biological networks, Models of medical devices
Abstract: Optimization of stimulus waveforms for medical therapeutics today are limited to rectangular biphasic stimulus waveforms, where clinicians and researchers alike can only adjust the pulse duration, amplitude and frequency of the stimulus. Studies have shown that non-rectangular pulse waveforms are often more energy efficient, and may potentially open the door to greater insights into neuromodulatory approaches. Here, we examine the use of intelligent evolutionary algorithms to find more optimal stimulus waveforms beyond the standard rectangular biphasic shapes.

Keywords: Models of organs and medical devices - Inverse problems in biology, Data-driven modeling, Systems biology and systems medicine - Modeling of signaling networks
Abstract: Computational models of spinal cord stimulation have been used to develop an understanding of the mechanism of action of spinal cord stimulation. This has occurred without any electrophysiological evidence for comparison. Recently measurements of the electrically evoked compound action potentials (ECAP) of dorsal column axons have been compared with single fibre action potentials calculated in by computational models. ECAP recordings are amplitude modulated with propagation distance. The modulation is due to small diameter fibers leaving the population of fibers contributing to the ECAP potential. This observation is consistent with fibers are terminating as they ascend and descend the dorsal columns. The axon model predicts the shape of the single fiber action potential remarkably well.

Keywords: Diagnostic devices - Physiological monitoring
Abstract: The effective scattering and absorption coefficients of tissue phantoms were determined using an integrating sphere setup and solutions of the radiative transfer equation in the visible and near-infrared wavelength range with errors smaller than of a few percent.

Keywords: FNIR (functional near infra-red) spectroscopy and near-infrared scanning and assessment, Diagnostic devices - Physiological monitoring
Abstract: Multilaboratory initiatives for performance assessment of diffuse optics instruments are presented, together with the overarching methodology and future actions.

Keywords: FNIR (functional near infra-red) spectroscopy and near-infrared scanning and assessment
Abstract: Time-domain methods play an important role in the measurement of optical and physiological parameters of macroscopic tissue regions in vivo by near-infrared spectroscopy. Using systematic studies on tissue equivalent phantoms and investigations on humans in vivo we show that the accuracy of this method can be improved when the commonly used diffusion model is replaced by Monte Carlo simulations of light transport. Furthermore, the accuracy of optical properties can be assessed by evaluating the noise level of the reduced scattering coefficients at different NIR wavelengths with respect to the theoretical spectral dependency according to the scatter power law. Accurate knowledge of phantom optical properties is of high importance for the calibration of other optical techniques such as spatially resolved diffuse reflectance. We discuss this method as a tool for the characterization of superficial tissue regions which cannot be assessed with time domain measurements of picosecond time resolution. Both phantom and in vivo studies demonstrate that the method of calibrated spatially resolved diffuse reflectance yields reliable results. In particular, in vivo test interventions on the exposed kidney of rats such as arterial and venous occlusion or inspiration gas induced hypoxia, hyperoxia and hypercapnia yield hemoglobin concentrations and oxygen saturations, that are in accordance to the dynamics of tissue partial oxygen tension and laser Doppler flux measured at the renal cortex by an invasive probe.

Keywords: FNIR (functional near infra-red) spectroscopy and near-infrared scanning and assessment
Abstract: Tissue Oximetry is currently expanding its scope of application into more and more clinical fields. But the group of currently available instruments is heterogeneous and standardization is urgently needed. Blood-lipid phantoms are an important tool to test and compare oximeter performance in a reproducible way.

Keywords: Computer modeling for treatment planning
Abstract: While usefulness of modeling of light in skin is undisputed, a false sense of accuracy can arise when implicit limitations are overlooked. Similarly, ill-posedness of inverse problems in specific biomedical optics methods is sometimes overlooked. Lastly, arguments for precise terminology are given.

Keywords: Pulmonary and critical care - Ventilatory Assist Devices, Cardiovascular and respiratory system modeling - Gas exchange models, Respiratory transport, mechanics and control - Pulmonary mechanics in disease
Abstract: Mechanical ventilation (MV) is a primary therapy in the intensive care unit (ICU). Sub-optimal ventilator settings can cause lung damage, but optimal selection is confounded by significant inter- and intra- patient variability in response to MV. Titrating PEEP (positive end expiratory pressure) to minimum elastance is a proven approach. However, in clinical practice finding this value is difficult. A predictive elastance model, or virtual patient, would directly assess a current PEEP level should be changed based on whether a nearby PEEP had lower elastance, as well as enable safe PEEP titration. A predictive, virtual patient model for MV in the ICU is presented.

Keywords: Pulmonary and critical care - Bioengineering applications in Intensive care, Pulmonary and critical care - Ventilatory Assist Devices
Abstract: Mechanical ventilation (MV) is the primary life support for critically ill respiratory failure patients. However, patients’ condition and their response to treatment are heterogeneous and variable. Thus, a more individualised approach could improve care and outcomes compared to a generalised, one size fits all treatment. Model-based treatment offers the opportunity to optimize MV to patient-specific needs at any given interval. It uses real-time estimated patient-specific respiratory system model parameters to personalise and optimise care. This talk focuses on the journey of bringing conceptual model-based ideas, to clinical proof of concept, and finally to a large randomised controlled trial.

Keywords: Pulmonary and critical care – Multimodality monitoring in intensive care
Abstract: Electrical impedance tomography (EIT) is a radiation-free imaging method capable of continuously assessing regional lung ventilation and aeration at the bedside. EIT parameters acquired during ongoing mechanical ventilation or during specific ventilator manoeuvres (e.g. decremental positive end-expiratory pressure trial or low-flow inflation and deflation pressure-volume manoeuvres) allow the identification of deleterious effects like regional overdistension, alveolar cycling or the presence of atelectasis. This information on regional lung function, which is otherwise not accessible at the bedside in the clinical setting, can be utilized for personalized finding of optimum ventilator settings and minimizing ventilator-associated lung injury. EIT also allows the identification of other adverse events encountered during mechanical ventilation like endotracheal tube malposition or pneumothorax.

Keywords: Pulmonary and critical care - Ventilatory Assist Devices
Abstract: Historically, EIT difference images were explored before absolute images since difference images are more stable in the presence of modeling errors. Absolute images require more detailed models of contact impedance, stray capacitance, and a properly refined finite element mesh where the electric potential gradient is high, and the use of priors has improved resolution. Recently the computation time has diminished with advent of the real time D-Bar method, the non linear Kalman Filters (KF), and the acceleration of the Simulated Annealing (SA). Anatomy and physiology-based priors have improved organ localization, volume estimation and resistivity accuracy. Clinical use of absolute EIT images may become a reality as the image robustness is increasing and computation time is decreasing. We describe here several medical concepts computed from absolute EIT images with potential utility for ventilation support, functional residual capacity, pneumothorax detection and cellularity on pleural effusions.

Keywords: Pulmonary and critical care - Bioengineering applications in Intensive care, Cardiovascular and respiratory system modeling - Gas exchange models, Cardiovascular and respiratory system modeling - Pulmonary Circulation
Abstract: Models of the respiratory system for monitoring or understanding the impact of invasive mechanical ventilation (MV) range from those that only require several equations to model and predict the lung’s mechanical response to a change in pressure, to those that use many hundreds-of-thousands of equations to simulate full air, blood, and gas exchange function in a structure-based model. Both type of model has some disadvantages that limit their application to optimizing treatment in the individual patient. However, both have distinct advantages that can be exploited by their integration. Here we present a modelling framework that exploits the advantages of each ‘level’ of model to understand the interaction between patient and MV device.

Keywords: Pulmonary and critical care – Multimodality monitoring in intensive care, Pulmonary and critical care - Pulmonary disease
Abstract: Electrical Impedance Tomography (EIT), that is currently used on intensive care units to visualize ventilation distribution in mechanically ventilated patients, is being developed into an examination tool for spontaneously breathing patients. Extended reconstruction methods efficiently calculate 3D information and reveal relevant regional features. Thus, lung function information derived from EIT and appropriately displayed to the clinician may benefit diagnosis and treatment of lung diseases like COPD (chronic obstructive pulmonary disease) or CF (cystic fibrosis).

Keywords: EEG imaging, Brain imaging and image analysis, Electrical source brain imaging
Abstract: Due to the major role of balance in our everyday lives and the unsatisfying understanding of the role of neural mechanism on balance control, the focus of this study was to explore the role of the cerebral cortex and its effects on stability. We investigated the effects of non-visual (eyes closed) and dual-task balance tasks on balance performance and cortical theta response in a small, convenient sample. The dual-tasks were N-back and Sustained Attention Response Task (SART). Cortical theta activity showed strong correlations with balance performance metrics. Particularly, central regions showed an increase in theta power in more cognitively challenging tasks but not statistically significant. Parietal theta power had a statistically significant increase in tasks that led to a heavier reliance on proprioception and vestibular information. This study shows the efficacy of EEG recording during balance control tasks. Future studies on neurodegenerative diseases that affect neuromotor control could investigate these outcomes.

Keywords: EEG imaging, Electrical source brain imaging, Functional image analysis
Abstract: 30-60% of traumatic brain injury (TBI) patients suffer from long-term balance deficit. Even though motor preparation and execution are altered and slowed in TBI, their relative contribution and importance to posture instability remain poorly understood. This study investigates the impaired cortical dynamics and neuromuscular response in TBI in response to balance perturbation and its relation to balance deficit. 12 TBI and 6 healthy control (HC) participants took the Berg Balance Scale (BBS) test and participated in a balance perturbation task where they were subjected to random anterior/posterior translation, while brain (EEG), muscle (EMG) activities, and center of pressure (COP) were continuously recorded. Using independent component analysis (ICA), the component most responsible for the N1 component of the perturbation evoked potential (PEP) was selected and its amplitude and latency were extracted. Balance task performance was measured by computing the COP displacement during the task. TBI had a significantly lower BBS, larger COP displacement and lower N1 amplitude compared to the HC group. No group differences was found for N1 latency and muscle activity onset delay to the perturbation. BBS was correlated with the COP displacement and N1 amplitude, and COP displacement was correlated with N1 latency. TBI balance deficit may be associated with more impaired than delayed cortical response to balance perturbation.

Keywords: Electrical impedance imaging
Abstract: Electrical impedance tomography is an accepted and validated tool to analyze and support mechanical ventilation at the bedside. In the future it could furthermore clinically provide information of the pulmonary perfusion and other blood volume changes within the thorax by exploiting a cardiosynchronous EIT component. In the presented study, the spatial forward sensitivity against different background lung tissue distributions was analyzed. Spheres with a 10 % change of the background conductivity were introduced in the lungs and in the heart. The cranio-caudal distribution of sensitivity had a bell shape and was similar between all simulated scenarios, varying only in magnitude. If the background tissue conductivity within the lungs was chosen to be the one of deflated tissue, the overall sensitivity was 46 % smaller compared to the overall sensitivity against inflated lung tissue conductivity. Within the heart region, the sensitivity was increased for fully deflated lung tissue conductivity (23 % relative to the sensitivity in the lungs) compared to a homogeneous distribution of inflated lung tissue conductivity (10 % relative to the sensitivity in the lungs).

Keywords: Electrical impedance imaging, Image reconstruction and enhancement - Machine learning / Deep learning approaches, Image reconstruction and enhancement - Tomographic reconstruction
Abstract: Electrical impedance tomography (EIT) is a non-invasive and non-radiative medical imaging technique based on detecting the inhomogeneous electrical properties of the tissue. The inverse problem of EIT is a highly nonlinear ill-posed problem, which is the main reason that affects image quality. Our goal is to solve the EIT inverse problem using the nonlinear mapping properties of artificial neural networks (ANNs) and convolutional neural networks (CNNs). In this paper, the adaptive moment estimation (ADAM) optimization method and mean-square-error (MSE) function are used to train an ANN to solve the inverse problem and a CNN to process the ANN image. The networks are trained on datasets of simulated data, and tested on datasets of simulated data and experimental data. Results for time-difference EIT (td-EIT) images are presented using simulated EIT data from EIDORS and experimental EIT data from our EIT systems. The results are used to compare the proposed method with the one-step Gauss–Newton linear method and RBFNN method. The proposed method offers improved resolution (RES), low position error (PE) and excellent artefact removal compared to the existing methods. The experimental results show that our method can improve the RES by 50 to 70 percent and reduce the PE by 60 to 70 percent. The improvements in RES and processing speed are essential for clinical EIT measurement of dynamic physiological processes.

Keywords: Electrical source brain imaging, MEG imaging, Magnetic resonance imaging - MR neuroimaging
Abstract: Introduction: Patients with medically refractory epilepsy (MRE) need surgical resection of the epileptogenic zone (EZ) to gain seizure-freedom. High-frequency oscillations (HFOs, >80 Hz) are promising biomarkers of the EZ that are typically localized using intracranial electroencephalography (icEEG). The goal of this study was to localize the cortical generators of HFOs non-invasively using high-density (HD) EEG and magnetoencephalography (MEG) and validate the localization against the gold-standard given by the icEEG-defined HFO-zone. Methods: We analyzed simultaneous HD-EEG and MEG data from six children with MRE who underwent icEEG and surgery. We detected interictal HFOs (80-160 Hz) on HD-EEG and MEG separately, using an in-house automatic detector followed by visual human review, and distinguished between HFOs with and without spikes. We localized the cortical generators of each HFO on HD-EEG or MEG using the wavelet Maximum Entropy on the Mean (wMEM). For the HFOs localized in the brain area covered by icEEG, we estimated the localization error (Eloc) with respect to the gold-standard, and classified them as either concordant (Eloc≤15mm) or not. Results: We found that: (i) HD-EEG presented a higher rate of HFOs than MEG (with spikes: 0.44 vs 0.21, p=0.031 HFOs/min; without spikes: 0.36 vs 0.08, p=0.063); (ii) HFOs without spikes were more likely to be localized outside any brain region of interest (i.e. covered by icEEG) than HFOs with spikes; and (iii) HD-EEG and MEG showed high precision to the gold-standard (93%; 100%). Conclusion: We reported quantitative evidence that HD-EEG and MEG can localize the HFO cortical generators with high precision to the icEEG gold-standard in children with MRE, suggesting that they may possibly limit the need for icEEG prior to surgery. We also showed that HFOs with spikes on HD-EEG/MEG are more likely to be epileptogenic than those independent from spikes, which may represent physiological events from normal brain.

Keywords: Electrical source imaging
Abstract: ECG imaging estimates the cardiac electrical activity from body surface potentials. As this involves solving a severly ill-posed problem, additional information is required to get a unique and stable solution. Recent progress is based on introducing more problem-specific information by exploiting the structure of cardiac excitation. However, added information must be either certain or general enough to not impair the solution. We have recently developed a method that uses a spatio-temporal basis to restrict the solution space. In the present work, we analyzed this method with respect to one of the most fundamental assumptions made during basis creation: cardiac (an)isotropy. We tested the reconstruction using simulations of ventricular pacings and then applied it to clinical data. In simulations, the overall median localization error was smallest with a basis including fiber orientation. For the clinical data, however, the overall error was smallest with an isotropic basis. This observation suggests that modeling priors should be introduced with care, whereby further work is needed.

Keywords: Chemo/bio-sensing - Biological sensors and systems, Chemo/bio-sensing - Chemical sensors and systems, Portable miniaturized systems
Abstract: The relevance of pH assessment in clinical analysis, environmental and industrial control, has raised the demand for the development of portable, low cost and easy-to-use monitoring systems. This paper proposes a pH sensor printed on a paper support passivated with a solid-ink coating. The sensor exploits the pH sensitivity of a reduced graphene oxide functionalized with 3-(4-aminophenil)propionic acid. The sensor responded in the pH range [4, 10] and had a sensitivity of 46 mV/pH. Tests on human plasma and seawater proved this pH sensor to have similar performances than those of a commercial pH-meter with an uncertainty of 0.1 and 0.2 pH unit in plasma and seawater, respectively.

Keywords: Chemo/bio-sensing - Biological sensors and systems, Physiological monitoring - Instrumentation
Abstract: Here we present a cost-effective point-of-use wireless platform for the electrochemical detection of low concentrations of glucose and hydrogen peroxide (H2O2), simultaneously. The electrochemical system utilizes a dual sensor integrated with a portable bipotentiostat. The bipotentiostat hardware implements a basic designed that reduces the cost of construction and increase the affordability of the instrument, while providing similar functionality as the more expensive bench-top potentiostats. The bipotentiostat utilizes inexpensive components and common Ag/AgCl reference and platinum counter electrodes and two working electrodes, and it is designed to detect currents within the range of 20 uA – 7 mA. Additionally, the bipotentiostat is integrate with wireless module ESP8266 that interfaces with a smartphone to enable real-time monitoring and visualization of the analyte concentration levels. The results show that the self-designed bipotentiostat is capable of performing chronoamperometry and demonstrate an electrochemical detection system that is a portable alternative system for laboratory and point-of-use testing.

Keywords: Chemo/bio-sensing - Micrototal analysis and lab-on-chip systems, Chemo/bio-sensing - Biological sensors and systems
Abstract: Protein testing in blood is important for clinical analysis. Traditional blood tests are performed in centralized laboratories and are slow to provide results. In contrast, point-of-care devices deliver rapid results in non-laboratory settings, allowing timely analysis, which can in turn reduce healthcare costs. Successful point-of-care platforms require seamless integration of chemical assays, fluid management and signal readout. In this regard, we present an integrated, compact and automated electronic system for point-of-care sensing of protein biomarkers. The electronic system comprises of a microfluidic-based electrochemical biosensor, amperometry circuitry and automated microfluidic fluid handling circuitry. This platform utilizes magnetic microbeads to expedite an electronic enzyme-linked immunosorbent assay and microfluidics to manage small volumes and automate operations. A commercial single-chip potentiostat is utilized for amperometry measurements and microfluidics control. Using this electronic system, we demonstrate an integrated and automated assay for human interleukin-6.

Keywords: Chemo/bio-sensing - Micrototal analysis and lab-on-chip systems, Bio-electric sensors - Sensing methods, Chemo/bio-sensing - Biological sensors and systems
Abstract: We present a novel method for label-free detection of cytokines in serum after ten minutes of stabilization at nanomolar concentrations. Detection of proteins in blood using label-free impedance-based techniques is difficult due to high salt concentration of the matrix, which results in screening of the charge of the target proteins. The microwell array design provides enhanced electrochemical sensitivity by electric field focusing in the wells. We describe an on-chip multiwell plate sensing configuration where sensitivity benefits from the high salt concentration of the matrix and demonstrate robust performance through testing in rat serum.

Keywords: Novel methods, Mechanical sensors and systems, Integrated sensor systems
Abstract: We have designed and fabricated a low-cost modular electrical and fluidic integration platform for microelectromechanical systems (MEMS) based biosensors, paving the way for a disposable, low-cost Lab-on-a-Chip. We demonstrate seamless integration using an additive manufacturing enabled “plug-and-play” platform that does not require permanent electronic or fluidic integration. This paper describes the fabrication steps and assembly of the method and highlights its advantages over the more traditional methods, such as ‘wire bonding’ and ‘flip chip’. We also provide design guidelines for improved biosensing, taking transport and binding kinetics into consideration in the context of prostate cancer diagnosis. Our novel approach combined with the design guidelines, opens up new opportunities for low-cost disposable high-density MEMS-based lab-on-a-chips for biosensing applications.

Keywords: Chemo/bio-sensing - Chemical sensors and systems, Chemo/bio-sensing - Techniques
Abstract: Lung cancer high mortality rate is mainly related to late-stage tumor diagnosis. Survival rates and treatments could be greatly improved with an effective early diagnosis. Volatile organic compounds (VOCs) in exhaled breath have been known for long to be linked to the presence of a disease. Exhaled breath analysis for early diagnosis of lung cancer represents a non-invasive, low-cost and user-friendly approach. In this paper we present the design and development of an electronic nose based on a metal oxide sensors array for the early diagnosis of lung cancer. Breath samples collected from healthy controls (n=10) and lung cancer subjects (n=6) were analyzed by the electronic nose, and classification was performed using an artificial neural network (ANN). A sensitivity of 85.7%, specificity of 100%, and accuracy of 93.8% were reached with leave one out cross validation (LOOCV). The presented device demonstrates that a simple, cost-effective, and non-invasive approach based on exhaled breath analysis has the potential to be of great help in decreasing lung cancer mortality.

Keywords: Connectivity measurements, Physiological systems modeling - Signal processing in physiological systems, Physiological systems modeling - Signals and systems
Abstract: The High Frequency (HF) band of the power spectrum of the Heart Rate Variability (HRV) is widely accepted to contain information related to the respiration. However, it is known that this often results in misleading estimations of the strength of the Respiratory Sinus Arrhythmia (RSA). In this paper, different approaches to characterize the change of the RSA with age, combining HRV and respiratory signals, are studied. These approaches are the bandwidths in the power spectral density estimations, bivariate phase rectified signal averaging, information dynamics, a time-frequency representation, and a heart rate decomposition based on subspace projections. They were applied to a dataset of sleep apnea patients, specifically to periods without apneas and during NREM sleep. Each estimate reflected a different relationship between RSA and age, suggesting that they all capture the cardiorespiratory information in a different way. The comparison of the estimates indicates that the approaches based on the extraction of respiratory information from HRV provide a better characterization of the age-dependent degradation of the RSA.

Keywords: Data mining and processing - Pattern recognition, Signal pattern classification, Time-frequency and time-scale analysis - Time-frequency analysis
Abstract: In this study, we investigated if audio signals may carry information related to the site of obstruction of the upper airway. The information regarding the site of collapse could improve obstructive sleep apnoea (OSA) treatment by allowing more individualized or structure-specific therapy. In this preliminary study, we developed an algorithm for automatically determining the site of collapse in 13 OSA patients through snoring analysis. Audio was recorded with a ceiling mounted microphone, simultaneously with full-night polysomnography during sleep. The surrogate measure of the site of airway collapse was identified by manual analysis of the nasal pressure signal. We extracted various time and frequency features of audio signal to classify the signal into “lateral wall”, “palate” and “tongue base” related collapse. The classification was carried out with a Gaussian mixture model classifier. Performance of the proposed model showed that it can achieve an overall accuracy of 78.9±0.96% with specificity and PPV of (89.3±0.81%, 78±1.5%) (73.2±1.3% , 83.2±1.8%) (61.3±2.5%, 71.8±1.3%) for lateral wall, palate and tongue base related collapse respectively. Our preliminary results suggest that the audio signal may be helpful in identifying the site of obstruction and therefore maybe a useful tool for deciding appropriate therapy.

Keywords: Data mining and processing - Pattern recognition, Neural networks and support vector machines in biosignal processing and classification, Signal pattern classification
Abstract: Polysomnography (PSG) is the gold standard for diagnosing sleep apnea severity. It allows monitoring of obstructive apneas and hypopnea events throughout the night. The detection of these events is usually done by trained sleep experts. However, this task is tedious, highly time-consuming and subject to important inter-scorer variability. In this study, we adapt our state-of-the-art deep learning method for sleep event detection, DOSED, to the detection of sleep apnea-hypopnea events in PSG for the diagnosis of sleep apnea severity. We used a dataset of 52 PSG recordings with apnea-hypopnea scoring from 5 trained sleep experts. We assessed the performance of the automatic approach and compared it to the inter-scorer performance for both the diagnosis of sleep apnea severity and the detection of single apnea-hypopnea events. We observed that human sleep experts reached an average accuracy of 75% while the automatic approach reached 81% for sleep apnea severity diagnosis. F1 Score for individual event detection was 0.55 in average for experts and 0.57 for the automatic approach. These results demonstrate that the automatic approach can perform at a sleep expert level when diagnosing sleep apnea severity.

Keywords: Data mining and processing - Pattern recognition, Signal pattern classification
Abstract: Sleep apnea is a common chronic respiratory disorder which occurs due to the repetitive complete or partial cessations of breathing during sleep. The gold standard assessment of sleep apnea requires full night polysomnography in a sleep laboratory which is expensive, time consuming, and inconvenient. Hence, there is an urgent need for a convenient, robust and wearable monitoring device for screening of sleep apnea. A simple and convenient accelerometer-based portable system is presented to estimate the severity of sleep apnea by analyzing tracheal movements. Respiratory related movements were recorded over the suprasternal notch using a 3D accelerometer. Twenty-one physiological features (7 features, 3 accelerometer channels) were extracted. Performance of three different deep learning models – convolutional neural network, recurrent neural network, and their combination – were evaluated for estimating the apnea hypopnea index (AHI). The estimated AHI is compared to the gold standard polysomnography. In 3-fold cross-validation experiments with 20 participants (9 female, age=47.8±18.0 years, BMI=30.8±4.8, AHI=22.2±21.8 events/hr), we achieved a correlation coefficient between gold standard and estimated values (r-value = 0.84). The proposed system is an accurate, convenient, and portable device suitable for home sleep apnea screening.

Keywords: Time-frequency and time-scale analysis - Time-frequency analysis, Physiological systems modeling - Signals and systems, Signal pattern classification
Abstract: Background and Rational: Obstructive Sleep Apnea (OSA) is a common disorder, affecting almost 10% of adults, but very underdiagnosed. This is largely due to limited access to overnight sleep testing using polysomnography (PSG). Our goal was to distinguish OSA from healthy individual using a simple maneuver during wakefulness in combination with machine learning.

Methods: Participants had undergone an overnight PSG to determine their ground truth OSA severity. Separately, they were asked to breathe through a nasal mask or a mouth piece through which negative pressure (NP) was applied, during wakefulness. Airflow waveforms were acquired and several features were extracted and used to train several classifiers to predict OSA.

Results and Discussion: The performance of each classifier and experimental setup was calculated. The best results were obtained using random forest classifier for distinguishing OSA from healthy individuals with a very good area under the curve of 0.80. To our knowledge, this is the first study to deploy machine learning and NP with promising path to diagnose OSA during wakefulness.

Keywords: Physiological systems modeling - Signal processing in physiological systems, Data mining and processing - Pattern recognition, Signal pattern classification
Abstract: In this paper, we used ECG signals and repiratory inductance plethysmography (RIP) or respiratory bands. We evaluated the performance of the signals individually as well as different combinations of features and signals for sleep apnoea detection. We implemented two methods (QRS area, and fast principal component analysis (PCA) methods) for estimating the ECG derived respiratory (EDR) signal and the cardiopulmonary coupling (CPC) spectrum. We then extracted features from the time and frequency representations of the ECG and RIP signals. Finally, we applied different features sets to a linear discriminant analysis (LDA) for classification. The results were examined on the MIT PhysioNet Apnea-ECG database. Apnoea classification was carried out using leave-one-record-out cross-validation approach. The highest performance of our algorithm was achieved using the RIP and RR-interval features as well as using the RIP and PCA CPC features with an accuracy of 90% and AUC of 0.97. The highest performance results of using only RIP or ECG features achieved an accuracy of 87% and AUC of 0.95. We conclude that although ECG sensors are more convenient for patients in sleep studies, using both RIP and ECG sensors enhances the performance results for automated diagnosis of sleep apnoea.

Keywords: Image analysis and classification - Machine learning / Deep learning approaches, Image classification, Optical imaging and microscopy - Microscopy
Abstract: Melanoma is one of the most deadly skin lesion, which often uses the skin dermoscopy to detect it. However, the low interclass variations between melanoma images make manual dermoscopic detection time-consuming and laborious. Therefore, an automatic recognition algorithm of skin image is highly desirable. However, the traditional methods still have the limitations (e.g., weak robustness and generalization ability). To meet the challenge, we propose an effective architecture based on residual – squeeze – and - excitation -Inception-v4 network (MelanomaNet) to detect melanoma. Specifically, Inception-v4 structure is utilized to get the rich spatial features and increase feature diversity. We also consider the relationship between feature channels by adding residual-squeeze-and-excitation (RSE) blocks in Inception- v4 network using the feature re-calibration strategies. Finally, we use the support vector machine (SVM) as the classifier for the skin lesion classification. We evaluate our proposed method on the public available ISIC skin lesion challenge datasets in 2018 for training and evaluation. The experimental results show that the proposed method has achieved better performance over the state-of-the-arts methods.

Keywords: Image analysis and classification - Machine learning / Deep learning approaches, Image segmentation
Abstract: Based on an image dataset of 88 in-vivo dental images taken with an intra-oral camera, we show that a Deep Learning model (Mask R-CNN) can detect and classify dental caries on occlusal surfaces across the whole 7-class ICDAS (International Caries Detection and Assessment System) scale. This is accomplished without any image pre-processing method and by utilizing superpixels segmentation for the experts' annotations and the evaluation of the classifier. In the proposed methodology, transfer learning and data augmentation are employed during the training of the model. The paper discusses technical details, provides initial results and denotes points for further improvement by fine-tuning the classifier along with an extended dataset.

Keywords: Optical imaging, Image analysis and classification - Machine learning / Deep learning approaches, Image feature extraction
Abstract: Face anti-spoofing is a crucial part of face recognition system to protect subject’s privacy and life safety. Most current face anti-spoofing algorithms are based on feature extraction and machine learning. The performance of machine learning based approaches depends on the quantity and quality of the training data. In this paper, we propose an unsupervised face anti-spoofing method based on feature extraction and matching of a dual camera setup, which does not require offline training. The principle of our method is simple, intuitive, and generally applicable. The core idea of our method is exploiting the fact that a 3D face has different feature representations in images from two cameras with different view angles, as compared to that of a 2D spoofing face (either printed in a paper or showing on a screen). The proposed method has been benchmarked on a dataset created by our dual camera setup and shows an accuracy of 94.2%.

Keywords: Image analysis and classification - Machine learning / Deep learning approaches, Image feature extraction, Novel imaging modalities
Abstract: Epilepsy monitoring involves the study of videos to assess clinical signs (semiology) to assist with the diagnosis of seizures. Recent advances in the application of vision-based approaches to epilepsy analysis have demonstrated significant potential to automate this assessment. Nevertheless, current proposed computer vision based techniques are unable to accurately quantify specific facial modifications, e.g. mouth motions, which are examined by neurologists to distinguish between seizure types. 2D approaches that analyse facial landmarks have been proposed to quantify mouth motions, however, they are unable to fully represent motions in the mouth and cheeks (ictal pouting) due to a lack of landmarks in the the cheek regions. Additionally, 2D region-based techniques based on the detection of the mouth have limitations when dealing with large pose variations, and thus make a fair comparison between samples difficult due to the variety of poses present. 3D approaches, on the other hand, retain rich information about the shape and appearance of faces, simplifying alignment for comparison between sequences. In this paper, we propose a novel network method based on a 3D reconstruction of the face and deep learning to detect and quantify mouth semiology in our video dataset of 20 seizures, recorded from patients with mesial temporal and extra-temporal lobe epilepsy. The proposed network is capable of distinguishing between seizures of both types of epilepsy. An average classification accuracy of 89% demonstrates the benefits of computer vision and deep learning for clinical applications of non-contact systems to identify semiology commonly encountered in a natural clinical setting.

Keywords: Image analysis and classification - Machine learning / Deep learning approaches, Image registration, segmentation, compression and visualization - Machine learning / Deep learning approaches
Abstract: Solving the classification problem, unbalanced number of dataset among the classes often causes performance degradation. Especially when some classes dominate the other classes with its large number of datasets, trained model shows low performance in identifying the dominated classes. This is common case when it comes to medical dataset. Because the case with a serious degree is not quite usual, there are imbalance in number of dataset between severe case and normal cases of diseases. Also, there is difficulty in precisely identifying grade of medical data because of vagueness between them. To solve these problems, we propose new architecture of convolutional neural network named {bf Tournament based Ranking CNN} which shows remarkable performance gain in identifying dominated classes while trading off very small accuracy loss in dominating classes. Our Approach complemented problems that occur when method of Ranking CNN that aggregates outputs of multiple binary neural network models is applied to medical data. By having tournament structure in aggregating method and using very deep pretrained binary models, our proposed model recorded 68.36% of exact match accuracy, while Ranking CNN recorded 53.40%, pretrained Resnet recorded 56.12% and CNN with linear regression recorded 57.48%. As a result, our proposed method is applied efficiently to cataract grading which have ordinal labels with imbalanced number of data among classes, also can be applied further to medical problems which have similar features to cataract and similar dataset configuration.

Keywords: Image analysis and classification - Machine learning / Deep learning approaches
Abstract: In urology endoscopic procedures, the Ureteral Orifice (UO) finding is crucial but may be challenging for inexperienced doctors. Generally, it is difficult to identify UOs intraoperatively due to the presence of a large median lobe, obstructing tumor, previous surgery, etc. To automatically identify various types of UOs in the video, we propose a real- time deep learning system in UO identification, localization and tracking in urinary endoscopy videos, and it can be applied to different types of urinary endoscopes. Our UO detection system is mainly based on Single Shot MultiBox Detector (SSD), which is one of the state-of-the-art deep-learning based detection networks in natural image domain. For the preprocessing, we apply both general and specific data augmentation strategies which have significantly improved all evaluation metrics. For the training steps, we only utilize rescetoscopy images which have more complex background information, and then, we use ureteroscopy images for testing. Simultaneously, we demonstrate that the model trained with rescetoscopy images can be successfully applied in the other type of urinary endoscopy images with four evaluation metrics (precision, recall, F1 and F2 scores) greater than 0.8. We further evaluate our model based on four independent video datasets which comprise both rescectoscopy videos and ureteroscopy videos. Extensive experiments on the four video datasets demonstrate that our deep-learning based UO detection system can identify and locate UOs of two different urinary endoscopes in real time with average processing time equal to 25 ms per frame and simultaneously achieve satisfactory recall and specificity.

Keywords: Design and development of robots for human-robot interaction, Assistive and cognitive robotics in rehabilitation, Humanoid robotics
Abstract: Socially assistive robots have shown potential benefits in therapy of child and elderly patients with social and cognitive deficits. In particular, for autistic children, humanoid robots could enhance engagement and attention, thanks to their simplified toy-like appearance and the reduced set of possible movements and expressions. The recent focus on autism-related motor impairments has increased the interest on developing new robotic tools aimed at improving not only the social capabilities but also the motor skills of autistic children. To this purpose, we have designed two embodied mirroring setups using the NAO humanoid robot. Two different tracking systems were used and compared: Inertial Measurement Units and the Microsoft Kinect, a marker-less vision based system. Both platforms were able to mirror upper limb basic movements of two healthy subjects, an adult and a child. However, despite the lower accuracy, the Kinect-based setup was chosen as the best candidate for embodied mirroring in autism treatment, thanks to the lower intrusiveness and reduced setup time. A prototype of an interactive mirroring game was developed and successfully tested with the Kinect-based platform, paving the way to the development of a versatile and powerful tool for clinical use with autistic children.

Keywords: Design and development of robots for human-robot interaction
Abstract: Hemiplegics have difficulty conducting various daily tasks because they must perform such tasks using only their unaffected arm. If a robot arm that can replace their missing upper-limb function is made available, it will greatly contribute to an improvement in their lives. To realize such a robot arm, cooperative actions of the robot arm and the unaffected arm are important. The purpose of this research is to develop a “cybernic robot arm” that can provide work support through cooperation with the unaffected arm for realization of upper-limb tasks in the daily life of hemiplegics. The developed system can detect the external force applied to a gripped object using a tactile force sensor embedded in the robot fingers. The system estimates the motion of the unaffected arm and the user’s intention regarding its operation based on the external force, and provides a support action in cooperation with the unaffected arm based on the estimated information. In addition, to confirm the applicability of the developed system, we developed a support function for opening tasks that are difficult for hemiplegics to carry out on their own, and conducted an experiment confirming the applicability of the system. The results confirm that the proposed system provides proper support by cooperating with the user’s arm, allowing an opening task to be performed. The developed system will support various upper-limb tasks that are difficult for hemiplegics.

Keywords: Human machine interfaces and robotics applications, Brain machine interfaces and robotics application in robot-aided living
Abstract: This paper analyzes the galvanic skin response (GSR) recorded from healthy and motor disabled people while steering a robotic wheelchair (RobChair ISR-UC prototype), to infer whether GSR can help in the recognition of stressful situations. Seven healthy individuals and six individuals with motor disabilities were asked to drive the RobChair by means of a brain-computer interface in indoor office environments, including complex scenarios such as passing narrow doors, avoiding obstacles, and with situations of unexpected trajectories of the wheelchair (controlled by an operator without users knowledge). All these driving situations can trigger emotional arousals such as anxiety and stress. A method called featurebased peak detection (FBPD) was proposed for automatic detection of skin conductance response (SCR) which proved to be very effective compared to the state-of-the-art methods. We found that SCR was elicited in 100% of the occurrences of collisions (lateral scrapings) and 94% of unexpected trajectories.

Keywords: New technologies and methodologies in medical robotics, Computer-assisted surgery, Surgical robotics
Abstract: Safe interactions between humans and robots require the robotic arms and/or tools to recognize and react to the surrounding environment via pressure sensing. With small-scale surgical interventions in mind, we have developed a flexible skin with tens of pressure sensing elements, designed to cover a 5 mm diameter tool. The prototype uses only biocompatible materials: soft silicones, carbon powder and metal wires. The material performance, sensing element, manufacturing technology, and the readout electronics are described. Our prototype demonstrates the feasibility of using this technology in various intervention scenarios, from endoscopic navigation to tissue manipulation. We conclude by identifying research directions that maximise the potential of the proposed technology.

Keywords: Assistive and cognitive robotics in aided living, Human machine interfaces and robotics applications, Neural control of movement and robotics applications
Abstract: Assistive technology is critical to improving daily life of those with muscular issues such as Cerebral Palsy and Duchenne Muscular Dystrophy by augmenting their activities of daily living (ADL). Robotic manipulators are one solution for helping with ADL; however, intuitive, accurate interfaces for higher degrees of freedom (DOF) robotic arms are still lacking. An intuitive control system based on artificial neural network (ANN) classification of real-time surface electromyography (sEMG) signals from the user’s forearm to detect nine hand gestures and control the movement of the 6 DOF robotic arm is proposed in this paper. The regular machine learning classifiers with the highest classification accuracies were ensemble-bagged trees at 90.3% and cubic SVM at 89.6%, with linear SVM being 84.8%. However, the classifier chosen was a scaled conjugate gradient backpropagation neural network model, with a classification accuracy of 85%, due to accuracy and usability in a Simulink model. The trained ANN model was incorporated into the control system for the robotic arm and tested in virtual environment. Preliminary testing of the robotic arm shows that the forward kinematic control system works well for most hand poses. Future improvements will include more processing of the sEMG signals and training on sEMG data from multiple subjects for a generalized ANN model.

Keywords: Assistive and cognitive robotics in aided living, Robot-aided mobility - Wheelchairs, canes, crutches, and mobility tools, New technologies and methodologies in human movement analysis
Abstract: In Japan, the number of people who have difficulty walking has been increasing with the rise in the aging population and that of people with physical disabilities. Individuals with athetosis-type cerebral palsy may use electric wheelchairs due to abnormal walking. However, since they have problems with fine motor control, including the occurrence of involuntary movements and difficulty maintaining posture, they have difficulty intentionally controlling their hand movements. Therefore, they cannot operate a joystick, even if they desire to use electric wheelchairs, and there are risks of accidents. In this study, by considering the arch structure of hand, we developed a new joystick grip that enables the suppression of involuntary movement. We evaluated our proposed grip by comparing running stability with a conventional grip, and demonstrated the effectiveness of proposed method.

Keywords: Drug delivery routes - Transdermal drug delivery, Drug delivery routes - Parenteral drug delivery, Micro and Nano formulation - Nanotechnology/Nanoparticles
Abstract: Vitamin B-12 (cobalamin) deficiency in humans is a worldwide problem emanating from varied causes. As oral supplementation is limited by its bioavailability due to the absorptive property of intrinsic factor, clinicians often prescribe parenteral forms of administration to replenish diminished levels rapidly. The gold standard in parenteral delivery of cobalamin is subcutaneous and/or intramuscular injections. The relatively large molecular size of cobalamin (1355.39 Da) makes passive transdermal patch-based delivery via the stratum corneum quite challenging. Hence, the primary goal of this study is to investigate the feasibility of intradermal (ID) delivery of Vitamin B-12 via an almost painless microneedle injection and subsequent comparison with standard subcutaneous (SC) delivery. This work reports on a custom-made microneedle device built from a commercial insulin needle and it’s use to perform ID delivery of Co-57 radiolabeled Vitamin B-12 in-vivo in rabbits. The pharmacokinetic profile and bioavailability were studied and compared with SC delivery. It is the first comprehensive study, to our best knowledge, that compares a micronutrient (eg. Vitamin B-12) delivery via ID and SC routes in-vivo. While the bioavailability for the SC route is found to be slightly higher compared to the ID route (99% vs. 96%), the Tmax for both are almost identical. Thus, ID delivery of Vitamin B-12 using a microneedle injection could be a viable and minimally invasive alternative to existing parenteral options.

Keywords: Drug release and solubility - Controlled/Sustained/Modified release, Micro and Nano formulation - Microparticles/Microspheres
Abstract: In this paper, we have presented a novel Drug Delivery Substrate (DDS) that that is responsive to external stimuli of high-frequency alternating magnetic fields. The DDS is constituted of chitosan crosslinked with PEGDMA (polyethylene glycol dimethacrylate), loaded with Fe3O4 magnetic nanoparticles and vancomycin. In another experiment, a 19-hour elution was observed where three magnetic stimuli of 25 mT, 109.9 kHz were given for 60 min to the test samples. The stimuli were separated by several hours. After excitation span, it was observed that the stimulated samples released a significantly higher amount of vancomycin by as much as 21% compared to non-stimulated samples. In another study, preliminary results showing the effect of different PEGDMA chain lengths have been discussed. These results show evidence of a smart, controllable DDS that allows modulation of its normal passive antibiotic elution by applying external stimuli per personalized needs.

Keywords: Drug release and solubility - Controlled/Sustained/Modified release, Drug delivery systems and carriers - Liposomes/Lipid-based drug delivery, Micro and Nano formulation - Nanotechnology/Nanoparticles
Abstract: The increasing interest towards biocompatible nanotechnologies in medicine, combined with electric fields stimulation, is leading to the development of electro-sensitive smart systems for drug delivery applications. Common examples of electro-sensitive materials include phospholipids that can be used to design nano-sized vesicles suitable for external electric actuation. To this regard, recently the use of pulsed electric fields to trigger release across phospholipid membranes has been numerically studied, for a deeper understanding of the phenomena at the molecular scale. Aim of this work is to give an experimental validation of the feasibility of controlling drug release from liposomes mediated by nanosecond pulsed electric fields.

Keywords: Drug delivery systems and carriers - Peptide and protein drug delivery, Clinical pharmacology - Computer simulation/modeling, Clinical pharmacology - Pharmacokinetics/Pharmacodynamics
Abstract: Insulin adsorption has been observed in ICU delivery lines, and is especially problematic in the delivery of insulin within the neonatal ICU. This paper presents a two state model with adsorptive loss described as a predator-prey term with parameters K_1 and B_eq. This model is discretized to N sub volumes along the length of an infusion set. The model was found to converge to a solution for N>~100-150. The model was fit to literature data, and it was found that the total adsorptive capacity of a material (B_eq, U/m2) was hyperbolically related to flow rate. If the average rate constant K_1 was used with the hyperbolic relationship, the model was able to describe adsorption dynamics at all 3 examined flow rates for a poly-ethylene line.

Keywords: Drug delivery routes - Transdermal drug delivery
Abstract: Most transdermal drug delivery systems are designed to inject drugs through the skin in a direction normal to the skin surface. However, in some applications, such as local anaesthesia, it is desirable to disperse the drug in a direction parallel to the surface of the skin. In this paper we present nozzles for needle-assisted jet injection that are designed to laterally disperse the fluid drug at a chosen depth in tissue. These nozzles were manufactured by laser machining holes in the walls of 0.57 mm (24 G) hypodermic needles, and sealing the ends of the needles. An existing controllable jet injection system was used to test the nozzles. High-speed video recordings were taken to examine the shape of the high-speed jets emitted from the orifices, and jet injections into post mortem porcine tissue were performed to evaluate the resulting dispersion pattern. These injections demonstrated the ability of these nozzles to achieve a widely spread dispersion at a depth of 3 mm to 4 mm in tissue. We observed that the widest dispersion occurred at the same depth as the orifices, and dispersion was greater in the direction of the jets. Further investigation, including an in vivo study, is now required to evaluate whether this technique can reduce the time, cost or pain associated with transdermal local anaesthetic delivery.

Keywords: Brain functional imaging - Source localization, Brain functional imaging - Mapping, Brain functional imaging - EEG
Abstract: sLORETA is one of the well-established EEG source localization methods that is popular for its satisfactory estimation, simplicity, and fast computation. However, the method has a low-resolution and requires manual post-processing thresholding to provide a sparser solution with acceptable resolution in source detection. Here we propose a subspace based thresholding that results in a higher resolution brain imaging based on minimizing a desired least square source detection error. Simulation results show the proposed method, denoted by HR-sLORETA, provides stable and high resolution solution in terms of Percentage of Undetected Sources (PUS) and Spatial Dispersion (SD) compared to the existing manual thresholding approaches as well as Otsu thresholding approach. It is shown that HR-sLORETA outperforms Otsu, which is the only other available automatic thresholding method, in scenarios with three or more sources.

Keywords: Brain functional imaging - Connectivity and information flow, Neurological disorders, Brain functional imaging - fMRI
Abstract: Cervical Dystonia (CD) is a neurological movement disorder characterized by intermittent muscle contractions in the head and neck. The pathophysiology and neural networks underpinning this condition are incompletely understood. There is increasing evidence that isolated focal dystonias are due to network-wide functional alterations. An abnormal temporal discrimination threshold (TDT) is believed to be a mediational endophenotype due to its prevalence in unaffected first-degree relatives as well as patients. However the neural correlates linking abnormal TDT and CD remain poorly understood. Probing changes in large-scale network topology via graph theory with resting state fMRI data from relatives and patients may provide further insight into the pathophysiology of CD. In this study, resting state fMRI data were acquired and analyzed from 16 CD patients with abnormal TDT, 32 unaffected first degree relatives (16 with normal TDT and 16 with abnormal TDT) and 16 healthy controls. Graph theory metrics demonstrating network topology were extracted. The results indicate large-scale functional reorganization of networks in relatives (with abnormal TDT) along with a manifestation of topological aberrations similar to patients.

Keywords: Brain functional imaging - Connectivity and information flow, Brain functional imaging - EEG, Neural signal processing
Abstract: Intrinsic connectivity networks (ICNs) have been widely studied using functional magnetic resonance imaging (fMRI) data and electrophysiological data (e.g., EEG or MEG). Two major methods, i.e., seed-based correlation analysis (SBCA) and independent component analysis (ICA), are widely used to extract ICNs. Among them, ICA usually involves a dual regression analysis in order to obtain final spatial definitions of ICNs. Recently, we proposed a framework that includes cortical source imaging, source-level ICA, and statistical correlation analysis, to extract cortical ICNs from resting-state EEG data. In the present study, we proposed an alternative framework that uses sensor-level ICA and regression analysis instead of source-level ICA and correlation analysis, considering the well-studied characteristics of sensor-level ICs in differentiating neural activities from artifacts and the benefit of regression in accommodating multivariate analysis over correlation. In the present study, we mainly investigated the performance of the proposed procedure in extracting cortical ICNs. Meanwhile, we also investigated different variants of the regressors sampled at different frequencies to formulate the regression model. The results demonstrated that cortical ICNs corresponding to major ICNs identified in literature could be obtained by the proposed framework. In general, spatial patterns of cortical ICNs obtained via both correlation and regression analyses show statistically significant similarity. However, the cortical ICNs reconstructed using the regression analysis exhibit more focal and more superficial spatial patterns, in general, that the cortical ICNs from the correlation analysis. The different variants of regressors at the same sampling frequency do not produce obvious impacts on spatial patterns of cortical ICNs, while the different sampling frequencies show large effects on extracted spatial patterns of cortical ICNs. In summary, it is suggested that the proposed

Keywords: Brain functional imaging - EEG, Human performance - Attention and vigilance, Neural interfaces - Body interfaces
Abstract: Everyday work is becoming increasingly complex and cognitively demanding. A person’s level of attention influences how effectively their brain prepares itself for action, and how much effort they apply to a task. However, the various distractions of the modern work environment often make it hard to pay and sustain attention. To address this issue, we present AttentivU - a system that uses wearable electroencephalography (EEG) to measure the attention of a person in real-time. When the user’s attention level is low, the system provides real-time, subtle feedback to nudge the person to become attentive again. Users can choose to turn the device on or off based on whether their current task requires focused attention. We tested the system on 12 adults in a real workplace setting. The preliminary results show that the biofeedback redirects the attention of the participants to the task at hand and improves their performance.

Keywords: Brain functional imaging - Connectivity and information flow
Abstract: This work aims at identifying characteristic features of EEG to demarcate a microsleep from preceding responsive states. The EEG signals, after reference electrode standardization technique (REST) re-referencing, were processed through a time-varying general linear Kalman filter (TVGLKF) to derive time-varying auto-regressive (TVAR) parameters. The time-varying effective connectivity measure of orthogonal partial directed coherence (OPDC) was obtained for every instant at 256 Hz. Effective connectivity matrices formed using these OPDC measures, with the scalp electrodes as nodes were processed further using graph theory. Community based measures were investigated and statistical significances compared. Non-parametric Wilcoxon signed rank test was used for significance analysis, with Cohen-type and Common Language effect size (CLES) as measures of effect sizes. The results showed a decrease in directional modularity from anterior to posterior, in theta, alpha, and beta bands in microsleeps. The alpha band showed the highest significance with a Cohentype effect size of 1.25 and a median percentage difference of 23% across subjects, with a range of 13-28%. Flexibility and integration also decreased with average percentage of 25% (17–35%) and 20% (16–32%), respectively, while recruitment increased on an average of 11% (3–16%), wherever significant across all bands. These community-based measures can help characterize and explain changes in brain mechanisms, and can also serve as potential biomarkers for microsleep detection.

Keywords: Brain functional imaging - MEG, Brain physiology and modeling - Cognition, memory, perception, Human performance
Abstract: Numerous studies have demonstrated that brain rhythms are modulated according to memory performance or memory processing. In sequential memory tasks, memory performance can be reduced by shortening the intervals between memory item presentations. To clarify the neurophysiological mechanism underlying this, we recorded magnetoencephalograms in 33 healthy volunteers performing two sequential memory tasks with either short or long intervals between memory items (hereafter, fast and slow conditions, respectively). Memory accuracy, and theta- and alpha-band activities originating from occipital and frontal brain areas were analyzed. Memory performance was significantly lower for the fast condition than the slow condition. Meanwhile, occipital and frontal theta activities were significantly lower for the fast condition than the slow condition. Increased occipital-alpha, a sign of active inhibition of task-irrelevant visual input, occurred regardless of condition. However, memory processing related to occipital- and frontal-theta activities had some temporal limitations. Namely, the shorter intervals of the fast condition attenuated theta activity, likely disrupting working memory processing, thereby leading to the observed decline in memory performance.

Keywords: General and theoretical informatics - Machine learning, General and theoretical informatics - Decision support systems, General and theoretical informatics - Statistical data analysis
Abstract: The acute respiratory distress syndrome (ARDS) is a fulminant inflammatory lung injury that develops in patients with critical illnesses including sepsis, pneumonia or trauma. However, many patients with ARDS are not recognized when they develop this syndrome nor given outcome-improving treatments. Because ARDS is a clinical syndrome, physicians may not have certainty about the diagnosis for some patients (label uncertainty). In addition, the diagnosis requires a chest x-ray, which may not be always be available in the clinical setting (privileged information). For this paper, we implemented the Learning Using Label Uncertainty and Partially Available Privileged Information (LULUPAPI) paradigm, built on classical SVM, to detect ARDS using Electronic Health Record (EHR) data and chest radiography. In comparison to SVM, this resulted in 3.55 percent improvement of test AUC.

Keywords: General and theoretical informatics - Machine learning, Health Informatics - Electronic health records, Health Informatics - Informatics for chronic disease management
Abstract: Chronicity is a problem that is affecting quality of life and increasing healthcare costs worldwide. Predictive tools can help mitigate these effects by encouraging the patients’ and healthcare system’s proactivity. This research work uses supervised learning techniques to build a predictive model of the healthcare status of a chronic patient, using Clinical Risk Groups (CRGs) as a measure of chronicity and prescription and diagnosis data as predictors. The model is addressed to the whole population in our healthcare system regardless of the disease, as data used are widely available in a consistent way for all patients. We explore different ways to encode data that are appropriate for machine learning. Results suggest that these data alone can be used to build accurate models, and show that, in our set, prescription information has a higher predictive value than diagnosis.

Keywords: General and theoretical informatics - Machine learning, General and theoretical informatics - Unsupervised learning method, Health Informatics - Personalized health/precision medicine
Abstract: Among the major challenges in training predictive models in wireless health, is adapting them to new individuals or groups of people. This is not trivial largely due to possible differences in the distribution of data between a new individual in a real-world deployment and the training data used for building the model. In this study, we aim to tackle this problem by employing recent advancements in deep Domain Adaptation which tries to transfer a model trained on a labeled dataset to a new unlabeled one that follows a different distribution as well. To show the benefits of our approach, we transfer an activity recognition model, trained on a popular adult dataset to children. We show that direct use of the adult model on children loses 25.2% in F1-score against a supervised baseline, while our proposed transfer approach reduces this to 9%.

Keywords: General and theoretical informatics - Machine learning, General and theoretical informatics - Supervised learning method, General and theoretical informatics - Pattern recognition
Abstract: Obesity is gaining increasing attention in modern society since it is associated with various health issues. The visceral adipose tissue (VAT) deposits around the abdominal organs and is considered an extremely important indicator of health risk. VAT can be assessed through magnetic resonance imaging (MRI) or computed tomography (CT) accurately, but the cost is prohibitive. Shape-based body composition prediction has become a promising topic thanks to the prevalence of commodity optical body scan systems, from which numerous anthropometries can be extracted automatically. In this paper, we propose an innovative shape-based hybrid VAT prediction model. The most appealing benefit of our method is to robustly handle the lack of knowledge about gender and demographics. First, we train a baseline VAT prediction model for each gender separately. Second, we train a classifier to predict the gender likelihood and a classifier to predict the shape likelihood of being overestimated in VAT baseline prediction. Third, we integrate the gender likelihood and shape likelihood into the baseline models to derive one hybrid VAT prediction model. We compare our prediction model with other state-of-the-art VAT prediction methods. The result shows that our method outperforms the comparison methods by 21.8% on average.

Keywords: General and theoretical informatics - Machine learning, General and theoretical informatics - Computational phenotyping, Health Informatics - Decision support methods and systems
Abstract: Identification of minimal residual disease (MRD) is important in assessing the prognosis of acute myeloid leukemia (AML) and myelodysplastic syndrome (MDS). The current best clinical practice relies heavily on Flow Cytometry (FC) examination. However, the current FC diagnostic examination requires trained physicians to perform lengthy manual interpretation on high-dimensional FC data measurements of each specimen. The difficulty in handling idiosyncrasy between interpreters along with the time-consuming diagnostic process has become one of the major bottlenecks in advancing the treatment of hematological diseases. In this work, we develop an automatic MRD classifications (AML, MDS, normal) algorithm based on learning a deep phenotype representation from a large cohort of retrospective clinical data with over 2000 real patients' FC samples. We propose to learn a cytometric deep embedding through cell-level autoencoder combined with specimen-level latent Fisher-scoring vectorization. Our method achieves an average AUC of 0.943 across four different hematological malignancies classification tasks, and our analysis further reveals that with only half of the FC markers would be sufficient in obtaining these high recognition accuracies.

Keywords: General and theoretical informatics - Machine learning, Sensor Informatics - Sensor-based mHealth applications, Health Informatics - Preventive health
Abstract: Foot ulcers are a common complication of diabetes and are the leading cause of amputation amongst those with diabetes. Research has shown that, an increase of two degrees Celsius in the skin temperature on the plantar surface of the foot can be an early indication of injury or inflammation. Early detection and treatment of a hotspot region may reduce the risk of an ulcer developing. This paper presents a thermography-based approach for detecting temperature hotspots on the foot. The system comprises a bespoke application and a thermal camera attachment which captures RGB images and a temperature matrix. Web-based services process the captured data and detect whether any regions of higher temperature are present on the foot, in comparison to the other foot. The accuracy of this system has been verified through a pilot study. Hotspots were simulated on the feet of 10 healthy participants. The results indicated that hotspots were correctly detected for 60% of the participants. We discuss some reasons why the results were inaccurate for the remaining four participants. Furthermore, we also suggest some potential enhancements to the system with the aim of increasing the precision of the results.

Keywords: Integrated sensor systems, Physiological monitoring - Modeling and analysis, Physiological monitoring - Instrumentation
Abstract: This paper describes a study performed in the frame of WEARABLES project and reports about preliminary analysis of the results on the activity, HR and breathing rate distribution. Objective of the study was the monitoring of employees’ well-being finalized at the investigation on the correlation between daily working activity and the observed physical parameters. The study has been performed by using sensing textiles, to collect objective work-correlated parameters during daily activity aiming at the acquisition of objective indicators for an improved management of people within teams. Scope of the project was to monitor a sample of 28 volunteers in environmental service delivery (at the Amey’s contract with Wolverhampton City Council), for a period of two non-consecutive weeks per volunteer, with a total of 275 data acquisition sessions.

Keywords: Smart textiles and clothings, Textile-electronic integration, Wearable sensor systems - User centered design and applications
Abstract: This research study investigates the impact of various insulating textile materials on the performance of smart textile pressure sensors made of conductive threads and piezo resistive material. We designed four sets of identical textile-based pressure sensors – each of them integrating a different insulating textile substrate material. Each of these sensors underwent a series of tests that linearly increased and decreased a uniform pressure perpendicular to the surface of the sensors. The controlled change of the integration layer altered the characteristics of the pressure sensors including both the sensitivity and pressure ranges. Our experiments highlighted that the manufacturing design technique of textile material has a significant impact on the sensor; with evidence of reproducibility values directly relating to fabric dimensional stability and elasticity.

Keywords: Smart textiles and clothings, Textile-electronic integration, Wearable sensor systems - User centered design and applications
Abstract: The aim of this work is to design and develop a sensorized sock in Electronic Textile (ET), SWEET-Sock. The device has been realized by three textile sensor placed in a specific points of plantar arch and an accelerometer unit, both embedded and connected by conductive thread. The sensors allows the acquisition of plantar pressure and acceleration signals deriving from the motion of the lower limbs. The detected biosignals have been condictionated by a voltage divider and then were acquired through a LilyPad Arduino microcontroller and transmitted using the Simblee BLE technology to a custom made mobile app. Data were afterwards uploaded through a smartphone on a dropbox cloud where a custom made MATLAB GUI platform has been developed for further digital signal processing of main biomechanical parameters of clinical interest in postural and gait analysis.

Keywords: Smart textiles and clothings, Bio-electric sensors - Sensing methods, Novel methods
Abstract: Phantom Motor Execution (PME) is a mechanism-based approach for the treatment of Phantom Limb Pain (PLP), which could potentially be self-administered in the home environment. However, the placement of electrodes aimed to acquire myoelectric signals from the residual stump muscles can be regarded as a difficult and time-consuming process by the patient. Thus, to increase patient compliance, the process must be made easier, faster, and cost effective. In this study, we developed and investigated a seamless integrated textrode-band for myoelectric recordings. The textrode-band can be easily donned/doffed, is reusable and washable. We demonstrated the viability of such concept by analyzing the signal-to-noise ratio (SNR), as well as offline and real time motion decoding performance, that in our experience are compatible with the PME treatment.

Keywords: Wearable sensor systems - User centered design and applications, Integrated sensor systems, Smart textiles and clothings
Abstract: This paper describes a study performed in the frame of Wearables project and reports preliminary results. Objective of the study was the implementation of an integrated service finalized to increase employees’ well-being through the investigation on the correlation between daily working activity and the observed physical parameters. The project monitored 28 volunteers employed in the field of waste collection (at the Amey’s contract with Wolverhampton City Council), for a total of 275 data acquisition sessions. The study has been performed using sensing textiles, to collect objective work-correlated parameters during daily activity, aiming at the acquisition of objective indicators for an improved wellbeing. Physical parameters like heart rate, energy expenditure and heart rate activity-zones distribution have been evaluated from data acquired during normal working activity. The service produced encouraging results both in terms of monitoring individual subjects and in identifying trends correlated to different roles or tasks covered by workers. Also in term of usability and acceptability the system showed interesting potentialities, proving how wearable technologies can trigger innovative approaches and open new prospective in the growing field of workplace wellness.

Keywords: Smart textiles and clothings, Wearable body-compliant, flexible and printed electronics, Physiological monitoring - Novel methods
Abstract: In this paper, a new improved mask printing method of liquid metal is developed, which realizes the fast fabrication of flexible electronics on fabrics. Here, polymethacrylates (PMA) glue is printed on fabrics to improve the adhesion effect of liquid metal (EGaIn) on fabrics. Combined with mask printing, liquid metal can be directly and rapidly printed on the fabrics with PMA glue to manufacture flexible electronics, such as LED array circuit, strain sensor and temperature monitoring circuit. With combined the advantages of favorable stretchability and rapid manufacture, the improved liquid metal mask printing method provides an approach with valuable prospects for individualized wearable health care devices. Besides, this method has extensive application prospect in mass production of smart electronic fabrics.

Keywords: Cardiovascular assessment and diagnostic technologies
Abstract: The performance of bioelectrodes in enzymatic glucose biofuel cell is not only dependent on the enzyme immobilization schemes but it is greatly influenced by the ability of the enzyme to exhibit favorable orientation for a direct electron transfer (DET) between the enzyme and the current collector. The electrochemical investigation of chitosan and nafion-chitosan coatings on multi-walled carbon nanotubes (MWCNTs) immobilized with pyrroloquinoline quinone dependent glucose dehydrogenase (PQQ-GDH) and bilirubin oxidase (BOD) at the bioanode and biocathode, respectively revealed interesting operational stability performance for the enzymatic biofuel cells. The bioelectrodes operated in DET mode and the chitosan coated biofuel cell system overall demonstrated higher power (156 W) output. The stability of PQQ-GDH bioanodes varied based on the enzyme concentrations, wherein a concentration of 2.5 mg/ml resulted in a significant enhancement in stability and the maximum power density of 1.6 mW/cm2 compared to enzyme concentrations of 5 mg/ml PQQ-GDH or higher.

Keywords: Cardiovascular assessment and diagnostic technologies, Defibrillators (implantable or external), Clinical engineering
Abstract: In most cases, the diagnosis of an electrical injury or electrocution is straightforward. However, there is a necessity for much closer analysis in many cases. There exist sophisticated electrical safety standards that predict outcomes for shocks of various currents applied to different parts of the body. Unfortunately, the actual current is almost never known in an accident investigation. A common source of errors is the assumption that the source (including the return) has zero im-pedance. Another surprisingly common problem is the erroneous assumption that the body current is equal to the source current capability. Methods: We used the following methodology for analyzing such cases: (1) Determine body pathway, (2) Estimate body pathway impedance, (3) Determine source voltage, (4) Determine source impedance, (5) Calculate delivered current using total pathway impedance, and (6) Ignore available cur-rent as it is largely confounding in most cases. Results: We analyzed 6 difficult cases using the above methodology. This includes 2 subtle situations involving pairs of matched case-control subjects where a subject was electrocuted while his work partner was not. Conclusions: Careful calculations of the amplitude and duration of the shock is required for understanding the limits and potential causation of such electrical injury. This requires the determination of both the source and body pathway impedance. Available current is usually irrelevant and over-emphasized.

Keywords: Cardiovascular assessment and diagnostic technologies, Diagnostic devices - Physiological monitoring
Abstract: Cerebrovascular diseases such as stenosis of the carotid artery are accountable for about 1 million death per year across Europe. Diagnostic tools like US, angiography or MRI require specific hardware and highly depend on the experience of the examining clinician. In contrast auscultation with a stethoscope can be used to screen for bruits - audible vascular sounds associated with turbulent blood flow. Dynamical changes in the flow due to pathological narrowing of the vessel can indicate the need for additional diagnostic investigations. A reliable auscultation setup is prerequisite to ensure high signal quality, adequate processing and the objective evaluation of a still subjectively assessed audible signal. We propose a computer assisted auscultation device for the characterisation of carotid bruits to facilitate the assessment of long-term changes in the vessel condition. Main goal of this work are design considerations regarding the mechanical interface of the proposed system to the skin. An experimental setup was used to compare the signal quality and morphology of different setups to a digital stethoscope. A combined system with two different interface configurations is proposed, current limitations of the system and potential improvements are discussed.

Keywords: Cardiovascular assessment and diagnostic technologies
Abstract: Atrial fibrillation (AF) and atrial flutter (AFL) represent atrial arrhythmias closely related to increasing risk for embolic stroke, and therefore being in the focus of cardiologists. While the reported methods for AF detection exhibit high performances, little attention has been given to distinguishing these two arrhythmias. In this study, we propose a deep neural network architecture, which combines convolutional and recurrent neural networks, for extracting features from sequence of RR intervals. The learned features were used to classify a long term ECG signals as AF, AFL or sinus rhythm (SR). A 10-fold cross-validation strategy was used for choosing an architecture design and tuning model hyperparameters. Accuracy of 88.28 %, with the sensitivities of 93.83%, 83.60% and 83.83% for SR, AF and AFL, respectively, was achieved. After choosing optimal network structure, the model was trained on the entire training set and finally evaluated on the blindfold test set which resulted in 89.67% accuracy, and 97.20%, 94.20%, and 77.78% sensitivity for SR, AF and AFL, respectively. Promising performances of the proposed model encourage continuing development of highly specific AF and AFL detection procedure based on deep learning. Distinction between these two arrhythmias can make therapy more efficient and decrease the recovery time to normal heart rhythm.

Keywords: Cardiovascular assessment and diagnostic technologies, Diagnostic devices - Physiological monitoring, Implantable devices for cardiac monitoring
Abstract: The aim of this preliminary study is to look how maternal-fetal heart rates and their coupling patterns are influenced by injection of beta blocker(propranolol) into pregnant mice. Total of 6 pregnant female mice were divided into two groups [control (N=3) and beta blockade (N=3)]. On 17.5-day mean heart rate of mothers and fetuses (MHR and FHR) were simultaneously measured for 20 minutes (10 minutes under normal condition and 10 minutes with saline (to control group) and propranolol (to the beta blockade group) solution by using an invasive maternal and fetal electrocardiogram techniques with needle electrodes. Results show that FHR decreased and maternal-fetal heart rate coupling (λ) patterns changed with propranolol infusion (no change with saline). Statistical test showed that changes (increase/decrease from pre to post values) in mean, rmssd and power spectral density (PSD) (2~4 Hz)) of MHR, short term variability of FHR, PSD (0.0~1.0 Hz) of FHR and λ were found to be significantly associated with treatment types (saline to propranolol). The presented results and protocol allow for assessment of beta adrenergic control of maternal and fetal heart, which will further enhance the value of the mouse as a model of heritable human pregnancy and hypertension.

Keywords: Cardiovascular assessment and diagnostic technologies, Diagnostic devices - Physiological monitoring, Defibrillators (implantable or external)
Abstract: Electrical safety limits for unidirectional pulses with short durations are increasingly important due to the proliferation of electric-car and factory energy storage systems with potentially dangerous voltages. Electrocution by a short-duration direct-current pulse is not understood as well as that by alternating current and the data are limited. The primary international guidance comes from IEC 60479-2 section 11. Methods: We have analyzed the dosimetry for short pulse safety limits based on a fuller understanding of the scientific principles involved and human data. Fortuitously many implantable defibrillators have been tested by intentionally externally delivering short-duration pulses to patients. Thus, today, we have the luxury of published human data which we analyze for this paper. Results: The present IEC current limit (60479-2:11) for short pulse durations is based on an exponent of -0.68 in the equation I = d-0.68, (d being pulse width), while the correct exponent should be -1.0 given the constant charge for the VF threshold of short pulses. We also propose a baseline charge value based on the human data. Conclusions: Charge-based VF thresholds give the correct dosimetry for short-duration pulses. Results from this paper should be considered in support of revising the IEC 60479-2 standard section 11.

Keywords: Sensory neuroprostheses - Visual, Neural stimulation, Neural interfaces - Tissue-electrode interface
Abstract: Retinal prostheses have the potential to restore vision to blind patients that have retinitis pigmentosa or similar hereditary degenerative disorders, by electrically stimulating surviving retinal neurons through implanted electrode arrays. Current retinal prostheses provide limited visual acuity and one challenge is to spatially control neural activation following electrical stimulation. Most of the retinal prostheses are either epi-retinal - in front of the retinal ganglion cell layer, or sub-retinal - behind photoreceptor layer. In this study, we performed calcium imaging of ganglion cells from whole mounted retinas and compared the spread of neural activation between epi-retinal stimulation with a fiber electrode and sub-retinal stimulation with a disk electrode. We investigated the effects of phase durations on the spatial resolution of biphasic stimulation. Our results suggest that with fiber electrode epi-retinal stimulation, the axon bundles activation can lead to significant spread of stimulation, and cannot be avoided simply by changing the phase durations. However, sub-retinal stimulation with very short pulses (phase duration 0.033ms) can effectively confine the activation of retinal ganglion cells.

Keywords: Sensory neuroprostheses, Sensory neuroprostheses - Auditory, Human performance - Speech
Abstract: Output Signal to Noise Ratio (OSNR) is the Signal to Noise ratio (SNR) at the output of a cochlear implant (CI) sound processor. Whereas other prediction metrics typically predict mean speech-in-noise test scores for a group of subjects, an OSNR-based model has been shown to accurately predict scores for individual CI recipients. The OSNR model was unable to predict scores for aggressive Ideal Binary Mask (IdBM) sound processing. This algorithm calculated Input Signal to Noise Ratio (ISNR), in each CI channel, and applied a gain function to suppress noise when a gain threshold was exceeded.

The current study investigated the effect of IdBM processing on the separate speech and noise signals to determine whether audibility was affecting intelligibility. A novel metric, “OSNR and Power” (OSNRP), which combined the effect of the reduction in output speech power with OSNR, was proposed.

It was found that the IdBM reduced the output speech level, likely causing audibility issues, at poor ISNRs. OSNRP accurately predicted individual speech-in-noise test scores for aggressive IdBM.

The novel OSNRP metric has potential as a tool for calculating optimum configurations for sound processor parameter settings for individual CI recipients. We propose using a prescribed set of reference test conditions, the results of which can be utilized to predict outcomes when using alternative sound processing parameters and techniques, and to tailor them to the individual needs of individual CI recipients.

Keywords: Sensory neuroprostheses - Somatosensory, Neural interfaces - Bioelectric sensors, Neural stimulation
Abstract: Providing high-quality somatosensory feedback from myoelectric prostheses to an upper-limb amputee user is a long-standing challenge. Various approaches have been investigated for tactile feedback, ranging from direct neural stimulation to noninvasive sensory substitution methods. However, only a few of studies evaluated the closed-loop performance, and real-time movement information of active prostheses still could not be transferred in the form of proprioceptive feedback so far. In the current study, an integrated closed-loop prosthesis system consisted of two types of sensors, programmable electrical stimulator and multichannel array electrodes was presented. The grasping angle and corresponding grasping force of a single-freedom myoelectric prosthesis were simultaneously coded with spatial and mixed (spatial and intensity of sensation) coding scheme and tested in 15 intact-bodied subjects. The experimental results demonstrated that the subjects were able to discriminate 4 types of object sizes, 3 kinds of different softness and 4 levels of grasping forces in relatively high correct identification rates (CIRs) (size: 87.5%, Softness: 94%, grasping force: 73.8%). The study outcomes and specific conclusions provide valuable guidance for the design of closed-loop myoelectric prostheses equipped with electrotactile feedback.

Keywords: Neural stimulation, Sensory neuroprostheses, Sensory neuroprostheses - Somatosensory
Abstract: Intracortical microstimulation has proven to be effective in a variety of sensory applications, such as returning touch percepts to paralyzed patients. The parameters of microstimulation play an important role in the perception quality of the stimulus. Eliciting naturalistic percepts is essential for the adaptability and functionality of this technology. Compared to the typical biphasic symmetric waveforms, asymmetric waveforms enhance activation selectivity by preferentially activating cell bodies. Behavioral studies have shown that asymmetric waveforms can elicit behavioral responses, but these require higher charges than typical symmetric waveforms. Here, we investigated the effects of phase duration and waveform asymmetry for somatosensory cortex intracortical microstimulation of freely-behaving rats. Detection thresholds were obtained using a conditioned avoidance behavioral paradigm. Our results indicate that phase duration has significant effects on threshold regardless of symmetry, polarity, and phase order of the waveform. Specifically, shorter phase durations tend to elicit lower behavioral thresholds. Analogous to studies in the auditory cortex, asymmetric waveforms in which the short pulse was cathodic were more effective than those with short-anodic pulses. With short phase durations, these short-cathodic waveforms are capable of evoking behavioral responses at low charges (<5 nC/Phase). Altogether, these results suggest the possibility of cell body selective microstimulation at safe thresholds, as well as the potential translatability of neuromodulation parameters across distinct sensory cortical areas.

Keywords: Sensory neuroprostheses - Visual, Neural stimulation
Abstract: Retinal implants are currently the only commercially available devices that can restore vision in patients suffering from a wide range of outer retinal degenerations. In order to improve the clinical outcome, i.e. the quality of elicited vision, a large number of in-vitro experiments probing the impact of electric stimulation on activation of the retina have been conducted. In these studies, however, retinas from many different species have been used which impedes comparisons between studies. Therefore, we measured the responses from four major ganglion cell types to light and electric stimulation in rabbit and mouse retina and compared their responses. We found strong similarities between the two species in transient cells whereas responses in sustained cell types typically did not match as well.

Keywords: Sensory neuroprostheses - Visual, Neural stimulation, Neural interfaces - Tissue-electrode interface
Abstract: Stimulation threshold is a key parameter to enable an efficient design for retinal implants. Stimulation parameters such as stimulus pulse duration, pulse amplitude, pulse repetition, pulse shape and polarity have been shown to be the governing factors that can influence the efficacy of retinal prosthetics. The effectiveness of these devices should best be evaluated both in the retina and in the visual cortex. Prior electrophysiological studies in the retina have shown that introducing an interphase gap make stimulation more efficient. Previous in vitro studies have also demonstrated the response properties of retinal ganglion cells are frequency dependent. However, the effect of these two stimulus parameters are not well explored at the cortical level where higher visual processing signals are processed. In this study, we examined the response properties of neurons in the primary visual cortex (V1) under stimulation of retinal ganglion cells in rat using a single-channel electrode of diameter 75 µm. We compared the response strength curves as a function of stimulus current amplitudes under different stimulus pulse duration, interphase gap and stimulus rate. Localized response to single channel epiretinal stimulation was robustly observed in V1 neurons. We found that V1 neurons were more sensitive to longer pulse and stimulus with an interphase gap, similar to previously reported results in the retina. We were also able to examine the effect of stimulus frequency on threshold in the visual cortex. Our results indicate that electrical activation of V1 neurons are more efficient at low frequency.

Keywords: Signal pattern classification, Neural networks and support vector machines in biosignal processing and classification
Abstract: This paper presents a novel method to estimate systolic blood pressure (SBP) and diastolic blood pressure (DBP) from time domain features extracted on auscultatory waveforms (AWs) using a long short term memory (LSTM) recurrent neural network (RNN). The proposed LSTM-RNN can effectively discover the latent structure in AW sequences and automatically learn such structures. The SBP and DBP points are then detected as the cuff pressures at which AW sequence changes its structure. Our LSTM-RNN is a powerful technique for sequence learning and can be used in blood pressure estimation as an alternative way for replacing traditional approaches.

Keywords: Signal pattern classification, Neural networks and support vector machines in biosignal processing and classification, Time-frequency and time-scale analysis - Wavelets
Abstract: We have recently demonstrated that micro-scale Sharp waves in the first few hours EEG of asphyxiated preterm fetal sheep models are the reliable prognostic biomarkers for Hypoxic-Ischemic Encephalopathy (HIE). Higher number of sharp waves within the first 2 hours from a hypoxic insult is shown to be significantly correlated to subcortical neuronal survival in caudate nucleus of striatum. Cerebral therapeutic hypothermia is also shown to be optimally neuroprotective only if initiated as soon as possible during a short window of opportunity within the first 2-3 hours of HI insult, called the latent phase. Therefore there is an urgent necessity for reliable automated algorithms to robustly identify such biomarkers to help early diagnosis of HIE, in real time at birth, before the optimal window of opportunity for treatment is missed. We have previously introduced successful automated signal processing strategies based on the fusion of wavelet and fuzzy techniques, for real-time identification and quantification of sharp waves along a profoundly suppressed EEG/ECoG background, post HI-insult, during the latent phase of sheep models. This work, in particular, for the first time represents a novel online fusion strategy based on the combination of a deep Convolutional Neural Network (CNN) in conjunction with Wavelet Scalogram (WS) for the real-time identification and classification of micro-scale sharp wave biomarkers within the 1024Hz high resolution ECoG recordings as well as the down-sampled 256Hz signals, from in utero preterm fetal sheep. The WS-CNN classifier highlights ability in the identification of HI sharp waves with remarkable high accuracies of 95.34% for 1024Hz and 94.62% for 256Hz data tested over one hour HI ECoG within the most important interval during the first 2 hours of the latent phase, where experiments have suggested hypothermia is optimally effective.

Keywords: Neural networks and support vector machines in biosignal processing and classification, Signal pattern classification
Abstract: Sleep staging, a process of identifying the sleep stages associated with polysomnography (PSG) epochs, plays an important role in sleep monitoring and diagnosing sleep disorders. We present in this work a model fusion approach to automate this task. The fusion model is composed of two base sleep-stage classifiers, SeqSleepNet and DeepSleepNet, both of which are state-of-the-art end-to-end deep learning models complying to the sequence-to-sequence sleep staging scheme. In addition, in the light of ensemble methods, we reason and demonstrate that these two networks form a good ensemble of models due to their high diversity. Experiments show that the fusion approach is able to preserve the strength of the base networks in the fusion model, leading to consistent performance gains over two base networks. The fusion model obtain the best modelling results we have observed so far on the Montreal Archive of Sleep Studies (MASS) dataset with 200 subjects, achieving an overall accuracy of 88.0%, a macro F1-score of 84.3%, and a Cohen's kappa of 0.828.

Keywords: Neural networks and support vector machines in biosignal processing and classification, Physiological systems modeling - Signal processing in physiological systems, Physiological systems modeling - Signals and systems
Abstract: Wearable technology for the automatic detection of gait events has recently gained growing interest, enabling advanced analyses that were previously limited to specialist centres and equipment (e.g., instrumented walkway). In this study, we present a novel method based on dilated convolutions for an accurate detection of gait events (initial and final foot contacts) from wearable inertial sensors. A rich dataset has been used to validate the method, featuring 71 people with Parkinson’s disease (PD) and 67 healthy control subjects. Multiple sensors have been considered, one located on the fifth lumbar vertebrae and two on the ankles. The aims of this study were: (i) to apply deep learning (DL) techniques on wearable sensor data for gait segmentation and quantification in older adults and in people with PD; (ii) to validate the proposed technique for measuring gait against traditional gold standard laboratory reference and a widely used algorithm based on wavelet transforms (WT); (iii) to assess the performance of DL methods in assessing high-level gait characteristics, with focus on stride, stance and swing related features. The results showed a high reliability of the proposed approach, which achieves temporal errors considerably smaller than WT, in particular for the detection of final contacts, with an inter-quartile range below 70 ms in the worst case. This study showes encouraging results, and paves the road for further research, addressing the effectiveness and the generalization of data-driven learning systems for accurate event detection in challenging conditions.

Keywords: Signal pattern classification, Data mining and processing - Pattern recognition
Abstract: Dysarthria speakers suffer from poor communication, and voice conversion (VC) technology is a potential approach for improving their speech quality. This study presents a joint feature learning approach to improve a sub-band deep neural network-based VC system, termed J_SBDNN. In this study, a listening test of speech intelligibility is used to confirm the benefits of the proposed J_SBDNN VC system, with several well-known VC approaches being used for comparison. The results showed that the J_SBDNN VC system provided a higher speech intelligibility performance than other VC approaches in most test conditions. It implies that the J_SBDNN VC system could potentially be used as one of the electronic assistive technologies to improve the speech quality for a dysarthric speaker.

Keywords: Signal pattern classification, Neural networks and support vector machines in biosignal processing and classification, Data mining and processing - Pattern recognition
Abstract: In recent years, machine learning algorithms have become increasingly popular for analyzing biomedical signals. This includes the detection of cyclic alternating pattern (CAP) in electroencephalography recordings. Here, we investigate the performance gain of a recurrent neural network (RNN) for CAP scoring in comparison to standard classification methods. We analyzed 15 recordings (n1-n15) from the publicly available CAP Sleep Database on Physionet to evaluate each machine learning method. A long short-term memory (LSTM) network increases the accuracy and F1-score by 0.5-3.5% and 3.5-8%, respectively, compared to commonly used classification algorithms such as linear discriminant analysis, k-nearest neighbour or feed-forward neural network. Our results show that by using a LSTM classifier the quantity of correctly detected CAP events can be increased and the number of wrongly classified periods reduced. RNNs significantly improve the precision in CAP scoring by taking advantage of available information from the past for deciding current classification.

Keywords: Novel imaging modalities, Image reconstruction - Fast algorithms, Image reconstruction and enhancement - Compressed sensing / Sampling
Abstract: Microwave ablation (MWA) is one of the most promising treatment tool to achieve a minimal invasion for human body. For safety and effective ablation for cancerous tissue, the accurate and real-time imaging for a temporal evolution of the ablation zone is highly demanded. This paper assumes the microwave based MWA monitoring, and introduces the novel boundary reconstruction algorithm, which has been demonstrated to achieve a real-time, accurate and noise-robust characteristic. This paper also newly introduces the S11 based complex permittivity estimator, which is necessary for estimating the ablation boundary. The both finite difference time domain (FDTD) based 3-D numerical test and the experimental investigation demonstrate that the proposed method provides accurate and high-speed 3-D imaging for the ablation zone.

Keywords: Novel imaging modalities, Optical imaging, Infra-red imaging
Abstract: Thermal ablation is a minimally invasive technique used to induce a controlled necrosis of malignant cells by increasing the temperature in localized areas. This procedure needs an accurate and real-time monitoring of thermal effects to evaluate and control treatment outcome. In this work, a hyperspectral imaging (HSI) technique is proposed as a new and non-invasive method to monitor ablative therapy. HSI provides images of the target object in several spectral bands, hence the reflectance/absorbance spectrum for each pixel. This paper presents a preliminary and original HSI-based analysis of the thermal state in the in vivo porcine liver undergoing laser ablation. In order to compare the spectral response between treated and untreated areas of the organ, proper Regions of Interest (ROIs) were chosen on the hyperspectral images; for each ROI, the absorbance variation for the selected wavelengths (i.e., 630, 760, and 960nm, for deoxyhemoglobin, methemoglobin, and water respectively) was assessed. Results obtained during and after laser ablation show that the absorbance of the methemoglobin peaks increases up to 40% in the burned region with respect to the non-ablated one. Conversely, the relative change of deoxyhemoglobin and water peaks is less marked. Based on these results, absorbance threshold values were retrieved and used to visualize the ablation zone on the images. This preliminary analysis suggests that a combination of the absorbance information is essential to achieve a more accurate identification of the ablation region. The results encourage further studies on the correlation between thermal effects and the spectral response of biological tissues undergoing thermal ablation, for final clinical use.

Keywords: Novel imaging modalities, MEG imaging, Ultrasound imaging - Prenatal
Abstract: The levator ani muscles (LAM) are integral to pelvic floor support and injury to this muscle complex has been associated with pelvic floor disorders, but our ability to evaluate their neuromuscular integrity is limited. During pregnancy, gravidas undergo systemic functional and anatomic modifications, including pelvic floor muscular adaptations. Magnetomyography (MMG) is a novel and non-invasive tool to passively measure the magnetic fields generated by depolarization activity of muscles and offers a unique method to evaluate the LAM. We collected serial MMG data in a pregnant woman with singleton gestation. Pregnant woman performed LAM contractions (Kegels) with intervening rest periods. Kegel signals were isolated by using the frequency dependent subtraction (SUBTR) and independent component analysis (ICA) methods. Concurrent body-surface electromyography (EMG) was used to evaluate for accessory-muscle recruitment by placing bipolar electrodes on the perineum, abdomen, and thigh. Amplitude and spectral-related indicators were computed across moderate intensity MMG Kegel epochs: root-mean square (RMS) amplitude, power spectrum density (PSD) and relative PSD (rPSD) in three frequency bands. Indicators were extracted from two pregnancy recordings and one postpartum. Parameters were represented in terms of gestation and postpartum weeks. We observed that postpartum RMS Kegel amplitudes had lower values than seen in pregnancy. Changes in spectral indicators were observed between pregnancy and postpartum.

Keywords: Novel imaging modalities
Abstract: This study presents a breast microwave radar imaging system designed for bistatic operation in air. The system operates in the range of 1-8 GHz and employs a double-rigged horn antenna array with four degrees of freedom. A hemispherical breast phantom with a tumor inclusion of 1.5 cm diameter was used to validate the system. Reconstructed datasets resulted in artifact-free images where the tumor presence was detected. Signal to clutter ratios greater than 30 dB and tumor to clutter ratios of 3 dB were measured. The encouraging images obtained with the system validate its potential as a clinical tool for breast cancer detection.

Keywords: Novel imaging modalities, Image reconstruction and enhancement - Tomographic reconstruction, Iterative image reconstruction
Abstract: Microwave mammography is one of the most promising alternative for the X-ray based breast cancer detection technique, where a malignant tumor has a certain level of contrast of dielectric property compared with those in normal tissues. However, an inverse problem of reconstructing complex permittivity is a non-linear and ill-posed problem, and the appropriate selection of such algorithms is the key for success of microwave mammography. The contrast source inversion (CSI) method is most promising solution of the above problem, where the iterative procedure does not require a computationally expensive forward solver like finite difference time domain (FDTD) method. However, the conventional CSI method assumes the non-dispersive dielectric model, while the breast or other human tissues have a non-negligible dispersive property. To address with this problem, this paper introduces the extended CSI method, being suitable for dispersive medium, where the multi-frequency integration is introduced to enhance the reconstruction accuracy. The FDTD numerical test, using realistic breast phantom by MRI, demonstrates that our proposed method efficiently enhance the reconstruction accuracy even in dispersive medium.

Keywords: Novel imaging modalities, Image reconstruction and enhancement - Filtering, Image reconstruction - Performance evaluation
Abstract: This paper focuses on the data preprocessing scheme, as well as on the frequency selection and spatial filtering modules integrated with a Time-Reversal Multiple Signal Classification (TR-MUSIC) algorithm, for microwave breast imaging. This algorithm is part of the data processing chain of the Wavelia Microwave Breast Imaging (MBI) system prototype, which has been recently installed at the University Hospital of Galway, Ireland, for a first-in-human clinical trial. Indicative results from application of the algorithm on an experimental phantom dataset, and on a first patient dataset, are presented in this paper. Good correspondence between the two datasets is demonstrated, confirming the validity of the experimental setup used so far for the on-site acceptance of the Wavelia system, after installation at the hospital for clinical testing.

Keywords: Modeling of cell, tissue, and regenerative medicine - Ionic modeling, Models of organ physiology, Modeling of cell, tissue, and regenerative medicine - Cells
Abstract: Changes of serum and extracellular ion concentrations occur regularly in patients with chronic kidney disease (CKD). Recently, hypocalcemia, i.e. a decrease of the extracellular calcium concentration [Ca²⁺]_o, has been suggested as potential pathomechanism contributing to the unexplained high rate of sudden cardiac death (SCD) in CKD patients. In particular, there is a hypothesis that hypocalcaemia could slow down natural pacemaking in the human sinus node to fatal degrees. Here, we address the question whether there are inter-species differences in the response of cellular sinus node pacemaking to changes of [Ca²⁺]_o. Towards this end, we employ computational models of mouse, rabbit and human sinus node cells. The Fabbri et al. human model was updated to consider changes of intracellular ion concentrations. We identified crucial inter-species differences in the response of cellular pacemaking in the sinus node to changes of [Ca²⁺]_o with little changes of cycle length in mouse and rabbit models (<83ms) in contrast to a pronounced bradycardic effect in the human model (up to 705ms). Our results suggest that experiments with human sinus node cells are required to investigate the potential mechanism of hypocalcaemia-induced bradycardic SCD in CKD patients and small animals models are not well suited.

Keywords: Data-driven modeling, Model building - Algorithms and techniques for systems modeling, Models of organ physiology
Abstract: Analysis of beat-to-beat spontaneous cerebral hemodynamic data has yielded predictive dynamic models of cerebral hemodynamics and has shown previously that patients with Mild Cognitive Impairment (MCI) exhibit significantly reduced cerebral vasomotor reactivity to CO2 relative to cognitively normal control subjects [1]. The present work examines the heart-rate reflex (HRR) dynamics of 46 MCI patients compared to 20 control subjects, using closed-loop modeling of HRR under resting conditions of spontaneous variations of arterial blood pressure (baroreflex) and end-tidal CO2 (chemoreflex). These subject-specific predictive dynamic models are obtained via the methodology of Principal Dynamic Modes [2] and allow the computation of model-based markers of baroreflex and chemoreflex function. We found that the chemoreflex gain is significantly weakened in MCI patients relative to controls (p=0.0086), while the baroreflex is not significantly affected. These findings offer another tool for diagnosis and monitoring of MCI (via model-based markers), when used in conjunction with current methods.

Keywords: Models of organs and medical devices - Inverse problems in biology, Computational modeling - Biological networks
Abstract: Non-invasive fetal electrocardiography (NI-FECG) is an emerging technology that demonstrates potential for providing novel physiological information compared to traditional ultrasound-based cardiotocography (CTG). However, few studies have investigated the reliability of signal features derived via this technique for diagnostic use. One feature of NI-FECG recordings proposed for the purpose of identifying fetal distress is the T/QRS ratio, which has been indicated to change in response to fetal hypoxia. As the T/QRS ratio measures characteristics of the heart’s electrical activity in 3D space (represented as the vectorcardiogram), it is critical to understand how changes in the vectorcardiogram orientation may influence the reliability of this feature.

To study this influence, this work simulates NI-FECG recordings using eight finite element models of the maternal-fetal anatomy and calculates the T/QRS ratio for a range of vectorcardiogram orientations and sensor positions. To quantify the potential for T/QRS ratio estimation error in real world data, these results are compared to those observed in a homogeneous volume conductor model, as assumed by many existing signal processing techniques.

Our results demonstrate that the fetal vectorcardiogram orientation has a significant influence on the reliability of the T/QRS ratio obtained via NI-FECG. Varying the vectorcardiogram orientation through a range of -30 to +30 degrees along each coordinate axis results in the potential for the T/QRS ratio to be underestimated by up to 94% and overestimated by up to 240% if a homogeneous volume conductor model is assumed. Furthermore, we find that the sensor positioning on the maternal abdomen strongly affects the range of the T/QRS ratio estimation error. These results confirm that further study must be undertaken to determine the relationship between the physiological and signal processing domains before utilizing the T/QRS ratio obtained via NI-FECG for diagnostic purposes.

Keywords: Models of medical devices, Organ modeling
Abstract: In treating recalcitrant low back pain, extreme lateral lumbar interbody fusion (XLIF) with a large cage is reported to have better stability compared to approach of transforaminal lumbar interbody fusion (TLIF) using a small cage. In addition, bilateral pedicle screw fixation (PSF) in comparison with unilateral fixation achieved no inferior fusion rate, but with a significant reduction in operation time and blood loss. The aim of the study was to understand the mechanism underpinning the stability of lumbar interbody fusion using different cage sizes with unilateral or bilateral PSF. A computer model of human lumbar vertebrae L4 and L5 with implants was reconstructed based on CT scans and simulated in Ansys Workbench. Simulation results demonstrated that for either XLIF or TLIF cages, the maximum values of rod stress were comparable with bilateral and unilateral PSF. However, the stability was considerably reduced with unilateral PSF for TLIF due to significantly increased facet joint strain for TLIF; whereas for XLIF with left unilateral PSF, the max facet joint strain was comparable to bilateral PSF, possibly due to facet tropism of this specific subject.

Keywords: Systems modeling - Clinical applications of biological networks, Model building - Parameter estimation
Abstract: In the present work, we introduce a novel technique to identify the infarct time from time-series meaurements of the cardiac troponin T (cTnT) into plasma. Although this information is extremely valuable from a clinical standpoint, it is not always possible to establish with certainty the exact infarct time. Here, we show how the infarct time can be reliably estimated from the cTnT release data in the first few hours after AMI, by using an optimization-based procedure and a model-based approach. To validate the present approach, we have used a clinical dataset of patients in whom the infarct has been induced and, therefore, the infarct time is certain.

Keywords: Models of medical devices
Abstract: Radiofrequency (RF) induced heating of tissues near an active implantable medical device can occur in patients undergoing magnetic resonance imaging. In vivo temperature measurement is a complex task, thus a reliable assessment that avoids in vivo experimentation is needed. A modeling workflow for obtaining the lead electromagnetic model (LEM) and validating the LEM was successfully developed. The proposed workflow was applied to investigate the RF responses, i.e., the net dissipated electrode power and net temperature increase, above background, at the electrodes of realistic coax leads with helical wires.

Keywords: Neural networks and support vector machines in biosignal processing and classification, Time-frequency and time-scale analysis - Time-frequency analysis, Physiological systems modeling - Signal processing in physiological systems
Abstract: Photoplethysmogram (PPG) is increasingly used to provide monitoring of the cardiovascular system under ambulatory conditions. Wearable devices like smartwatches use PPG to allow long-term unobtrusive monitoring of heart rate in free-living conditions. PPG based heart rate measurement is unfortunately highly susceptible to motion artifacts, particularly when measured from the wrist. Traditional machine learning and deep learning approaches rely on tri-axial accelerometer data along with PPG to perform heart rate estimation. The conventional learning based approaches have not addressed the need for device-specific modeling due to differences in hardware design among PPG devices. In this paper, we propose a novel end-to-end deep learning model to perform heart rate estimation using 8-second length input PPG signal. We evaluate the proposed model on the IEEE SPC 2015 dataset, achieving a mean absolute error of 3.36+/-4.1BPM for HR estimation on 12 subjects without requiring patient-specific training. We also studied the feasibility of applying transfer learning along with sparse retraining from a comprehensive in-house PPG dataset for heart rate estimation across PPG devices with different hardware design.

Keywords: Neural networks and support vector machines in biosignal processing and classification, Time-frequency and time-scale analysis - Wavelets, Adaptive filtering
Abstract: Chest compressions delivered during cardiopulmonary resuscitation (CPR) induce artifacts in the ECG that may make the shock advice algorithms (SAA) of defibrillators inaccurate. There is evidence that methods consisting of adaptive filters that remove the CPR artifact followed by machine learning (ML) based algorithms are able to make reliable shock/no-shock decisions during compressions. However, there is room for improvement in the performance of these methods. The objective was to design a robust ML framework for a reliable shock/no-shock decision during CPR. The study dataset contained 596 shockable and 1697 nonshockable ECG segments obtained from 273 cases of out-of-hospital cardiac arrest. Shock/no-shock labels were adjudicated by expert reviewers using ECG intervals without artifacts. First, CPR artifacts were removed from the ECG using a Least Mean Squares (LMS) filter. Then, 38 shock/no-shock decision features based on the Stationary Wavelet Transform (SWT) were extracted from the filtered ECG. A wapper-based feature selection method was applied to select the 6 best features for classification. Finally, 4 state-of-the-art ML classifiers were tested to make the shock/no-shock decision. These diagnoses were compared with the rhythm annotations to compute the Sensitivity (Se) and Specificity (Sp). All classifiers achieved an Se above 94.5%, Sp above 95.5% and an accuracy around 96.0%. They all exceeded the 90% Se and 95% Sp minimum values recommended by the American Heart Association.

Keywords: Neural networks and support vector machines in biosignal processing and classification, Time-frequency and time-scale analysis - Wavelets
Abstract: Atrial fibrillation (AF) is one of the most common arrhythmias. The automatic AF detection is of great clinical significance but at the same time it remains a big problem to researchers. In this study, a novel deep learning method to detect AF was proposed. For a 10s length single lead electrocardiogram (ECG) signal, the continuous wavelet transform (CWT) was used to obtain the wavelet coefficient matrix, and then a convolutional neural network (CNN) with a specific architecture was trained to discriminate the rhythm of the signal. The ECG data in multiple databases were divided into 4 classes according to the rhythm annotation: normal sinus rhythm (NSR), atrial fibrillation (AF), other types of arrhythmia except AF (OTHER), and noise signal (NOISE). The method was evaluated using three different wavelet bases. The experiment showed promising results when using a Morlet wavelet, with an overall accuracy of 97.56%, an average sensitivity of 97.56%, an average specificity of 99.19%. Besides, the area under curve (AUC) value is 0.9983, which showed that the proposed method was effective for detecting AF.

Keywords: Neural networks and support vector machines in biosignal processing and classification, Signal pattern classification
Abstract: Electrocardiogram (ECG) delineation is a process to detect multiple characteristic points, which contain critical diagnostic information about cardiac diseases. We treat the ECG delineation task as an one-dimensional segmentation problem, and propose a novel end-to-end deep learning method to segment sections of ECG signal. Our neural network consists of two parts: a segmentation network composed of multiple 1D Convolutional Neural Networks (CNN) and a postprocessing network composed of a sequential Conditional Random Field (CRF). Our method is trained and validated on QT database. The experimental results show that our method yields competitive overall performance compared with other state-of-the-art works and outperform them on onset of the P wave and offset of the T wave.

Keywords: Neural networks and support vector machines in biosignal processing and classification, Data mining and processing in biosignals
Abstract: Cuff-less blood pressure (BP) is a potential method for BP monitoring because it's undisturbed and continuous monitoring. Existing cuff-less estimation methods are easily influenced by signal noise and unideal and untypical signal morphology. In this study we propose a novel well-designed Convolutional Neural Network (CNN) model named Deep-BP for BP estimation task. The deep structure of Deep-BP can help to capture more underlying data factors associated with BP than handcraft features, thus increase the robustness and estimation accuracy. We carry out experiments with and without calibration procedure in training stage to evaluate the performance of new method in different application scenarios. The experiment results show that the Deep-BP model achieves high accuracy and outperform existing methods both in the experiments with and without calibration.

Keywords: Neural networks and support vector machines in biosignal processing and classification, Signal pattern classification, Adaptive filtering
Abstract: Pulse detection during out-of-hospital cardiac arrest remains challenging for both novel and expert rescuers because current methods are inaccurate and time-consuming. There is still a need to develop automatic methods for pulse detection, where the most challenging scenario is the discrimination between pulsed rhythms (PR, pulse) and pulseless electrical activity (PEA, no pulse). Thoracic impedance (TI) acquired through defibrillation pads has been proven useful for detecting pulse as it shows small fluctuations with every heart beat. In this study we analyse the use of deep learning techniques to detect pulse using only the TI signal. The proposed neural network, composed by convolutional and recurrent layers, outperformed state of the art methods, and achieved a balanced accuracy of 90% for segments as short as 3 s.

Keywords: Neuromuscular systems - EMG processing and applications, Neurorehabilitation, Human performance - Modelling and prediction
Abstract: The aim of the proposed work is to utilize the heterogeneity information, in input signal extraction, to improve the joint force estimation from high-density surface electromyography (HD-sEMG). For this purpose, joint force and HD-sEMG signals from biceps brachii and brachialis were collected synchronously during isometric elbow flexion. The input signal of the force model was obtained after the following procedures: first, HD-sEMG signals were decomposed by principal component analysis into principal components and weight vectors; second, the first several weight maps were segmented to obtain heterogeneity information by the Otsu and Moore-Neighbor tracing methods, and the principal component covering the most activated areas (maximum heterogeneity) was selected; and last, the selected principal component was low-pass filtered to obtain the input signal. The force model was built using a polynomial fitting technique. The conventional power-based input signal was compared with our obtained input signal. According to the obtained results, the proposed heterogeneity-based input signal can reduce the force estimation error significantly than the power-based input signal. The proposed heterogeneity-based input signal extraction methods had more neuromuscular control information and will be tested in more muscles and force tasks in future works.

Keywords: Neuromuscular systems - EMG processing and applications, Neural interfaces - Body interfaces, Neurological disorders
Abstract: We are developing a wearable neural interface based on high-density surface electromyography (HDEMG) for detecting and decoding signals from spared motor units in the forearms of people with tetraplegia after spinal cord injury (SCI). A lightweight, form-fitting garment containing 150 disc electrodes and covering the entire forearm was used to map the myoelectric activity of forearm muscles during a wide range of voluntary tasks of a person with chronic tetraplegia after SCI (C5 motor and C6 sensory American Spinal Injury Association Impairment Scale B spinal cord injury). Despite exhibiting no overt finger motion, myoelectric signals were detectable for attempted movements of individual digits and were highly discriminable. Motor unit decomposition was used to identify the activity of >30 motor neurons, active specifically during rotation, pronation of the wrist (4 units), and flexion of the elbow joint (7 units), and during attempted movements of individual hand digits (1-5 units). In addition, we performed a neural connectivity analysis based on the power of the common oscillations of the identified motor neurons in the delta (∼5Hz), alpha (∼6-12 Hz), and beta bands (∼15-30 Hz). This analysis showed clear common synaptic inputs to the identified motor neurons in all the analyzed frequency bands. This neural interface offers a new potential for the control of assistive technologies, whereby the motor neurons spared after SCI may serve as a direct readout of motor intent that allows proportional control over several distinct degrees of freedom. Moreover, this framework can be used to study the reorganization and recovery of spinal networks after injury and rehabilitation.

Keywords: Neuromuscular systems - Central mechanisms, Motor neuroprostheses - Epidural stimulation, Neurorehabilitation
Abstract: Fall risk is a serious problem especially for the elderly. Fall accident causes fracture, and it leads to bedridden. Early detection of balance inability, and rehabilitation training is important to decrease the fall risk. Quiet standing test is one of the physical tests to assess the balance inability of human. When the human is standing, he/she controls the plantar force to keep the balance. Center of pressure (COP) is the representative parameter to explain the plantar force movement. During the quiet standing test, COP fluctuates unconsciously. There are many researches that analyze the relationship between the COP fluctuation and balance inability. Many statistical indicators have been developed to assess the fluctuation specification. In contrast, several researchers have been tried to reveal the COP fluctuation mechanism by introducing stochastic mathematical model. Ornstein-Uhlenbeck process and diffusion equation are often introduced to the model. The mathematical models have been developed, and analyzed how visual input is related to the model parameters. The subjects stand quietly for 60 seconds with and without visual input in the past researches. If the mathematical model can explain the COP fluctuation dynamics completely, it becomes easier to assess the balance ability by model parameters. However, statistical indicators are still used in the clinical cases, thus there is room to discuss which statistical indicators should be used for assessment. The purpose of this research is to explore the relationship between statistical indicators and mathematical model by evaluating whether the fitted mathematical model reproduces the same values with real data. The reproduced indicators and original indicators are compared, and the magnitude of the errors are evaluated in the view point of two error definitions.

Keywords: Neuromuscular systems - Computational modeling, Motor learning, neural control, and neuromuscular systems, Neuromuscular systems - Postural and balance
Abstract: Successful manipulation of objects requires forming internal representations of the object dynamics. To do so, the sensorimotor system uses visual feedback of the object movement allowing us to estimate the object state and build the representation. One way to investigate this mechanism is by introducing a discrepancy between the visual feedback about the object’s movement and the actual movement. This causes a decline in the ability to accurately control the object, shedding light about possible factors influencing the performance. In this study, we show that an optimal feedback control framework can account for the performance and kinematic characteristics of balancing an inverted pendulum when visual feedback of pendulum tip did not represent the actual pendulum tip. Our model suggests a possible mechanism for the role of visual feedback on forming internal representation of objects’ dynamics.

Keywords: Neuromuscular systems - EMG processing and applications, Neural signals - Blind source separation (PCA, ICA, etc.), Neuromuscular systems - Peripheral mechanisms
Abstract: Mechanical vibration applied directly to the muscle belly or tendon has been reported to elicit a specific reflex loop named tonic vibration reflex (TVR), which involves motor unit (MU) activation synchronized and un-synchronized within the vibration cycle. Indirect application of vibration to the muscle by vibration exercise (VE) has also been suggested to evoke TVR, as evidenced by the spectral peaks observed at the vibration frequency in the surface electromyography (sEMG). However, other studies interpreted these spectral peaks as the result of motion artifacts (MAs). The aim of the present study is, therefore, to investigate MU activation patterns during VE in order to clarify the nature of those spectral peaks. To this end, low-intensity isometric contractions were executed with and without VE, and high-density sEMG measurements were performed during the contraction tasks. MU action potential (MUAP) trains were extracted by decomposing the recorded high-density sEMG signals. The spectra of the MUAP trains were then calculated and compared between vibration and no-vibration conditions. Clear MU synchronization was observed during VE, confirming the spectral peaks at the vibration frequency to be mainly due to the reflex loop rather than MAs.

Keywords: Neuromuscular systems - EMG processing and applications, Human performance - Fatigue, Neuromuscular systems - EMG models
Abstract: This paper presents the estimation and analysis of surface electromyogram (EMG) signals during fatiguing contractions based on a stochastic EMG model. In the model, the probability distribution of EMG signals is assumed to be a mixture of Gaussians with the same mean but different variances, facilitating the representation of the variance distribution of EMG signals. The paper proposes a continuous estimation method for variance distribution parameters using a sliding window, enabling the evaluation of the time-varying stochastic properties of EMG signals. Estimation experiments were conducted on six healthy young adults to analyze changes in EMG variance distribution with the progression of muscle fatigue. The results reveal the linear and nonlinear relationships between muscle fatigue and variance distribution parameters.

Keywords: Health Informatics - Behavioral health informatics, Health Informatics - eHealth, Health Informatics - Informatics for chronic disease management
Abstract: There is a rapid growth in the emergence of data-driven e-Health to support healthcare professionals and empower patients. In this session of the mini-symposium, we will explore how human factors, such as usability, can play a major role in the reliability and scalability of such solutions. This will be done using as example experiences in mobile and wearable solutions currently under development in Qatar to support childhood obesity.

Keywords: Health Informatics - eHealth, Health Informatics - Outcome research, Health Informatics - Informatics for chronic disease management
Abstract: Despite the developments of recombinant growth hormone (GH), many children with growth hormone deficiency (GHD) fail to achieve their target adult height; this has largely been attributed to suboptimal adherence to treatment. Clinic visits are undertaken routinely, but reported between-visit adherence data is frequently under or over-estimated. Automatic transmission of injection data provides a more accurate insight into adherence patterns. Most importantly, it connects clinician, nurse, patient and caregiver and this can elicit a prompt intervention to improve low adherence if necessary. This talk will assess the long term clinical and pharmacoeconomic outcomes of the eHealth ecosystem which supports patients, who are receiving Saizen® for growth disorders.

Keywords: Health Informatics - eHealth, Health Informatics - Informatics for chronic disease management, Health Informatics - Information technologies for the management of patient safety and clinical outcomes
Abstract: Prof. Dimitri is a Senior Consultant in Paediatric Endocrinology and Director of Research & Innovation at Sheffield Children’s NHS Foundation Trust in the UK. He is also the NIHR National Children's Speciality Lead in the UK. Based upon his experience in starting the National Technology Innovation Transforming Child Health (TITCH) Network and the National Institute of Health Research (NIHR) Children and Young People MedTech Cooperative he has experience in supporting technology development from proof of concept to product development and scaling up opportunities in child health technology.

Keywords: Health Informatics - Disease profiling and personalized treatment, Health Informatics - Mobile health, Health Informatics - eHealth
Abstract: Clinical data is not enough for a proper management of patient diseases. There is a movement that empowers the patient to get a more personalized care based on Patient Reported Outcomes (PRO), collecting data from the patient and his context, where mobile technology is playing a key role. In this session, we will introduce the MyCyFAPP success case, focused on children with cystic fibrosis and their relatives promoting and maintaining adequate nutritional behaviors and involving them in an active way. The platform used to collect such data could be easily transferred to other use cases for the collection of PROs to enhance research and patient support.

Keywords: Health Informatics - eHealth, Health Informatics - Information technologies for healthcare delivery and management
Abstract: Although the use of digital health in pediatrics started over two decades ago, it is not until recently massive adoption of mobile solutions has started. This emerging adoption of digital health includes connected devices (e.g. connected injectable devices), mobile applications and wearable devices. In this talk, we will provide an overview of how technology and data can play a role in the clinical work and research of pediatric endocrinologists. Prof. Sävendahl will provide an overview on the day-to-day work of pediatric endocrinologists with a special focus on the importance of using data to support the care of patients with growth disorders. This talk will by completed by more technical presentations, which are part of the mini-symposium.

Keywords: Model building - Parameter estimation, Model building - Algorithms and techniques for systems modeling, Data-driven modeling
Abstract: In mathematical modeling of cell physiological processes, measurements required for parameter determination are often available only as aggregated data in the literature. Physiological measurements contain relatively large observation errors due to intrinsic variations in physiological processes, and the errors cause uncertainties in parameter values. This paper reports analyses of the uncertainty in parameter estimates of a simple mathematical model of an ion channel from a set of published experimental data. A conventional approach for estimating model parameters from aggregated data is applying the method of least squares to a series of the mean values of measurements. The parameter estimates by the conventional method significantly differed from those by a statistical approach, maximum likelihood estimation considering the standard errors of the means. Exhaustive analyses on the likelihood of parameter values show high parameter uncertainties and wide distribution of parameter values with no significant differences in the likelihood. These results imply the importance of considering variances of observations and uncertainties in parameter estimates.

Keywords: Computational modeling - Biological networks, Systems modeling - Clinical applications of biological networks, Systems biology and systems medicine - Modeling of signaling networks
Abstract: The molecular mechanisms underlying Alzheimer’s disease (AD) have been and are still under heavy scrutiny to better understand what leads to the onset and progression of the disease, and to design and develop efficacious therapeutic strategies. These decade-long studies have taught us a lot regarding the various molecular pathways involved in the pathology, but a complete dynamic picture of the underlying pathological mechanisms is still missing. We propose to provide a technological answer to fill this gap by developing and using a computational approach that integrates AD-related experimental findings and their effects on multiple aspects of neuronal function. The present study focuses on implementing one known pathogenic process: the binding of amyloid beta, the hallmark of AD, on NMDA receptors, receptors present in the main type of excitatory synapses in the brain, thereby affecting synaptic transmission and downstream pathways. We describe model implementation and calibration; we then quantify the downstream effects of this disruption both in terms of electrical activity (changes in short-term spiking activity of the postsynaptic neuron), and biochemical pathways activation through changes in calcium dynamics (an important trigger to longer-term changes). The computational approach outlined constitutes an insightful instrument to examine the downstream consequences of multiple pathogenic dysfunctions on higher level observables and sets the path for in-silico discovery and testing of therapeutic agents.

Keywords: Systems biology and systems medicine - Modeling of signaling networks, Computational modeling - Biological networks, Data-driven modeling
Abstract: There has been a recent surge of dynamical systems models (DSMs) constructed from brain activity to investigate how neural firing patterns evolve over time, and how such patterns in turn generate measured behavior. An advantage with DSMs are their ability to accurately reconstruct neural patterns and behavior simultaneously, capturing variability in data due to different tasks. In this paper, we demonstrate how to use a general DSM beyond reconstruction. In particular, we show that the general DSM can also be used to (i) quantify which states, i.e. which neural regions drive the dynamics of the observed neural activity throughout behavior, and (ii) whether interactions between populations are primarily within the same brain region (intra) or across brain regions (inter). We first begin with an intuitive dynamical system example - a coupled two-mass spring system, and then perform the same analyses for a DSM estimated from neural data collected from a nonhuman primate executing a reach-to-grasp motor task. These examples show that DSM is a modeling approach that not only can accurately reconstruct observed neural and behavioral activities, but can predict important dynamical neural drivers as well as interactions between different neural populations that cannot be gleaned from data alone.

Keywords: Data-driven modeling, Computational modeling - Biological networks, Model building - Parameter estimation
Abstract: Network formation from neural activity is a foundational problem in systems neuroscience. Functional networks, after downstream analysis, can provide key insights into the nature of neurobiological structure and computation. The validity of such insights hinges on accurate selection and estimation of the edges connecting nodes. However, commonly used statistical inference procedures generally fail to identify the correct features, and further introduce consequential bias in the estimates. To address these issues, we developed Union of Intersections (UoI), a flexible, modular, and scalable framework for enhanced statistical feature selection and estimation. Methods based on UoI perform feature selection and feature estimation through intersection and union operations, respectively. In the context of linear regression (specifically UoI-Lasso), we summarize extensive numerical investigation on synthetic data to demonstrate tight control of false-positives and false-negatives in feature selection with low-bias and low-variance estimates of selected parameters, while maintaining high-quality prediction accuracy. We demonstrate, with UoI-Lasso, the extraction of sparse, predictive, and interpretable functional networks from human electrocorticography recordings during speech production and the inference of parsimonious coupling models from nonhuman primate single-unit recordings during reaching tasks. Our results establish that UoI-Lasso generates interpretable and predictive functional connectivity networks.

Keywords: Systems biology and systems medicine - Modeling of gene/epigene regulatory networks, Systems biology and systems medicine - Modeling of biomolecular system pathways, Systems biology and systems medicine - Modeling of signaling networks
Abstract: Gene regulatory networks depict the interactions among genes, proteins, and other components of the cell. These interactions are stochastic when large differences in reaction rates and small copy number of molecules are involved. Discrete Chemical Master Equation (dCME) provides a general framework for understanding the stochastic nature of these networks. Here we used the Accurate Chemical Master Equation method to directly compute the exact steady state probability landscape of the feed-forward loop motif (FFL). FFL is one of the most abundant gene regulatory networks motifs where the regulation is carried out from the top nodes to the bottom ones. We examine the behavior of stochastic FFLs under different conditions of various regulation intensities. Under the conditions with slow promoter binding, we show how FFL can exhibit different multistabilities in their landscapes. We also study the sensitivities of regulations of FFLs and introduce a new definition of stochastic sensitivity to characterize how FFLs respond in their probability distributions at the steady state to perturbations of system parameters. We show how change in gene expression under FFL regulations are sensitive to system parameters, including the state of multistability in FFLs

Keywords: Synthetic Biology - Control systems and circuits, Systems modeling - Decision making
Abstract: Recent progress in the development of new methods for cancer treatment has shown advantages of multiple therapies over mono-therapy. In particular, direct drug targeting (DDT) combined with mixed immunotherapy and chemotherapy has the potential to mitigate the undesired side-effects allied with conventional therapies, where nanorobots in DDT carry therapeutic agents through the blood vessel channel in order to localize and target diseased tissue with a safe drug interaction. This process can be modeled by a “touchable” (i.e., externally controllable and trackable) molecular communication (MC) system. However, in such a complex process overcoming unavoidable vascular channel uncertainties remains a great challenge. In this paper a multiple model predictive controller (MMPC) is proposed, which is robust against random channel uncertainties. The efficacy of the proposed method is illustrated through identification of globally optimized drug administration schedules. Furthermore, we introduce upper and lower bounds on the inputs and outputs which lead to clinically realistic constraints on the system. Simulation results demonstrate the promising performance of proposed MMPC to control tumor growth in presence of vascular channel uncertainties in MC-inspired DDT for cancer treatment.

Keywords: Image-guided devices - MRI-compatible instrumentation and device management, Image-guided drug delivery
Abstract: This paper introduces our compact and lightweight patient-mounted MRI-compatible 4 degree-of-freedom (DOF) robot with an improved transmission system for MRI-guided arthrography procedures. This robot could make the traditional two-stage arthrography procedure (fluoroscopy-guided needle insertion followed by a diagnostic MRI scan) simpler by converting it to a one-stage procedure but more accurate with an optimized system. The new transmission system is proposed, using different mechanical components, to result in higher accuracy of needle insertion. The results of a recent accuracy study are reported. Experimental results show that the new system has an error of 1.7 mm in positioning the needle tip at a depth of 50 mm, which indicates high accuracy.

Keywords: Clinical engineering - Device alarm, alert, and communication systems, Image-guided devices - Biopsy, Clinical engineering
Abstract: During stereotactic brain tumor biopsies, the detection of protoporphyrin IX (PpIX) fluorescence and microvascular perfusion using laser Doppler flowmetry (LDF) with a handheld fiber optic probe allows the identification of tumor tissue while decreasing the risk of intracranial hemorrhage. Neurosurgeons performing this procedure usually view the measurement values on a screen. When their visual focus is directed at the surgical site, they require an assistant to verbally relay the values. An auditory and visual user interface (UI), which displays measurement values accurately and allows fast and intuitive signal recognition, can improve this procedure. This paper experimentally evaluates an auditory and visual UI for providing real-time measurement feedback during stereotactic brain tumor biopsies. In a user study (n = 15), the accuracy of auditory and visual response was determined using function response tests, and user acceptance was evaluated. The auditory signals proved to be intuitive and easy to recognize and remember. The visual display of measurement values was easy to understand and facilitated the user's decision-making process. Moreover, the UI exhibited high user acceptance.

Keywords: Image-guided devices - MRI-compatible instrumentation and device management, Image-guided drug delivery
Abstract: This paper presents a 2 degrees-of-freedom (DOF) remotely actuated needle driving device for Magnetic Resonance Imaging (MRI) guided pain injections. The device is evaluated in phantom studies under real-time MRI guidance. The force and torque asserted by the device on the 4-DOF base robot are measured. The needle driving device consists of a needle driver, a 1.2-meter long beaded chain transmission, an actuation box, a robot controller and a Graphical User Interface (GUI). The needle driver can fit within a typical MRI scanner bore and is remotely actuated at the end of the MRI table through a novel beaded chain transmission. The remote actuation mechanism significantly reduces the weight and size of the needle driver at the patient end as well as the artifacts introduced by the motors. The clinician can manually steer the needle by rotating the knobs on the actuation box or remotely through a software interface in the MRI console room. The force and torque resulting from the needle driver in various configurations both in static and dynamic status were measured and reported. An accuracy experiment in the MRI environment under real-time image feedback demonstrates a small mean targeting error (<1.5 mm) in a phantom study.

Keywords: Image-guided devices - MRI-compatible instrumentation and device management, Image-guided drug delivery, Computer modeling for treatment planning
Abstract: To assist in percutaneous interventions in the lower back under magnetic resonance imaging guidance, a 4 degree-of-freedom body-mounted parallel robot is developed. The robot structure is improved comparatively to a previously developed robot, to increase the stability, enhance accuracy, and streamline the assembly and calibration process. The optimized assembly and calibration workflows are carried out, and the system accuracy is evaluated. The results demonstrate that the system positioning and angular accuracy are 2.28±1.1 mm and 1.94±1.01 degree respectively. The results show that the new system has a promising and consistent behavior.

Keywords: Image-guided devices - Biopsy, Image-guided devices - MRI-compatible instrumentation and device management
Abstract: This work describes an initial design concept for a spinal needle using new materials to optimize their visualization in magnetic resonance imaging. Common MRI needles made of Nickel-Titanium alloys still show poor visibility in imaging because they generate susceptibility artifacts due to materials interactions with the magnetic environment. The use of nonmetallic materials can reduce these artifacts. However, so far no non-metallic needle design has made it to clinical routine due to sharpness and cost issues. We propose a design of a coaxial needle with a fiber enforced inner core and an outer hollow sheet. The concept has been evaluated in the MRI environment. Additionally, mechanical tests were performed to examine and quantify the variation between a conventional spinal needle and our proposed design.

Keywords: Image-guided devices - Biopsy, Clinical engineering
Abstract: In radiology practices, the ultrasound-guided breast biopsy is among the most commonly performed minimally invasive procedures. However, many radiology residents in their graduate residencies are found with little or no hands-on experience with ultrasound-guided breast procedures. To enhance safety, the problem can be solved by the use of anthropomorphic training phantoms which can provide the resident with realistic ultrasound imaging and needle insertion haptic feedback. Stiffness and acoustic properties of breast tissues vary between different people. The training breast phantom should be able to possess different acoustic and mechanical properties which conform the inconsistencies found in real tissues among people. Therefore, this paper investigates the tunability of acoustic and mechanical behaviors in breast tissue mimicking materials (TMMs). Experiments of central composite design (CCD) with a center point, four corner points, and an additional four axis points were used to fit the non-linear regression model of the speed of sound. The same design of experiment approach was then used to fit the second-order response surface of the attenuation coefficient. Suitable series of tissue mimicking materials for the glandular tissue and malignant lesion were suggested. Latin hypercube design method was conducted to evaluate the main factors that affected the mechanical property (Young’s modulus) of tissue mimicking materials. The results showed that the recipe of tissue mimicking materials could be customized to possess different acoustic and mechanical properties which conform the inconsistencies found in real breast tissues.

Keywords: Cardiovascular regulation - Baroreflex, Cardiovascular regulation - Autonomic nervous system, Cardiovascular and respiratory signal processing - Heart Rate and Blood Pressure Variability
Abstract: The coupling and latency between heart period (HP) and systolic arterial pressure (SAP) variability can be investigated along the two arms of the HP-SAP closed loop, namely along the baroreflex feedback from SAP to HP, and along the feedforward pathway from HP to SAP. This study investigates the HP-SAP closed loop variability interactions through cross-correlation function (CCF). Coupling strength and delay between HP and SAP variability series were monitored in 13 subjects prone to develop orthostatic syncope (SYNC, 28±9 yrs, 5 males) and in 13 subjects with no history of postural syncope (noSYNC, age: 27±8 yrs, 5 males). Analysis was carried out at rest in supine position (REST) and during head-up tilt (TILT) at 60 degrees. The null hypothesis of HP-SAP uncoupling was tested through a surrogate analysis associating the HP series of a subject with a SAP sequence of a different one. Results showed that during TILT the coupling strength increased along the baroreflex and latency augmented along the mechanical feedforward pathway exclusively in noSYNC subjects. Finally, closed loop HP-SAP interactions were detected in about one third of subjects and the situation of full uncoupling was rarely found. CCF analysis was found to be a straightforward and easily applicable method to investigate HP-SAP control deserving a direct comparison with more sophisticated signal processing tools assessing causality.

Keywords: Cardiovascular and respiratory signal processing - Cardiovascular signal processing, Cardiovascular and respiratory signal processing - Non-linear cardiovascular or cardiorespiratory relations, Cardiovascular regulation - Baroreflex
Abstract: Cardiovascular diseases are one of the most common causes of death in elderly patients. The etiology of cardiomyopathies is difficult to discern clinically. The objective of this study was to classify cardiomyopathy patients using coupling analysis, through their cardiovascular behavior and the baroreflex response. A total of thirty-eight cardiomyopathy patients (CMP) classified as ischemic (ICM, 25 patients) and dilated (DCM, 13 patients) were analyzed. Thirty elderly control subjects (CON) were used as reference. Their electrocardiographic (ECG) and blood pressure (BP) signals were studied. To characterize the cardiovascular activity, the following temporal series were extracted: beat-to-beat intervals (from the ECG signal), and end– systolic and diastolic blood pressure amplitudes (from the BP signal). Non-linear characterization techniques like high resolution joint symbolic dynamics, segmented Poincaré plot analysis, normalized short-time partial directed coherence, and dual sequence method were used to characterize these times series. The best indices were used to build support vector machine models for classification. The optimal model for ICM versus DCM patients achieved 84.2% accuracy, 76.9% sensitivity, and 88% specificity. When CMP patients and CON subjects were compared, the best model achieved 95.5% accuracy, 97.3% sensitivity, and 93.3% specificity. These results suggest a disfunction in the baroreflex mechanism in cardiomyopathies patients.

Keywords: Cardiovascular regulation - Baroreflex, Cardiovascular regulation - Autonomic nervous system, Cardiovascular and respiratory signal processing - Non-linear cardiovascular or cardiorespiratory relations
Abstract: Short-term cardiovascular control, comprising cardiac baroreflex and mechanisms governing cardiac contractility and vascular properties, links heart period (HP) and systolic arterial pressure (SAP) fluctuations. It is activated during postural challenge and this activation is traditionally quantified via linear tools such as HP-SAP squared coherence function. In this study the ability of a nonlinear bivariate tool based on joint symbolic analysis (JSA) approach was tested against HP-SAP coherence function during orthostatic challenge in recreational athletes. We studied 9 men healthy cycling practitioners (age: 20-40 yrs) at rest in supine condition (REST) and during active standing (STAND). The JSA method is based on the definition of symbolic HP and SAP patterns and on the evaluation of the rate of their simultaneous occurrence in both HP and SAP series. HP-SAP squared coherence was computed in the low frequency band (LF, from 0.04 to 0.15 Hz). We found the expected response to the postural stimulus, namely the increase of sympathetic modulation and the contemporaneous vagal withdrawal. However, only JSA was able to detect the expected increase of association between HP and SAP variability series over slow time scales. This result suggests that recreational athletes have more relevant nonlinear interactions between HP and SAP that might be missed by traditional linear tools and might require nonlinear tools to be efficiently described.

Keywords: Cardiovascular and respiratory signal processing - Complexity in cardiovascular or respiratory signals, Cardiovascular and respiratory signal processing - Heart Rate and Blood Pressure Variability, Cardiovascular regulation - Autonomic nervous system
Abstract: Uncovering the physiological correlates of dreams is one of the most ambitious aim of multidisciplinary neuroscientific research. Here we investigated Autonomic Nervous System (ANS) dynamics associated with a dream recall, with a particular focus on the complexity assessment on cardiovascular control. We recorded electrocardiogram and arterial blood pressure signals from eight healthy subjects during rapid-eye-movement sleep before awakenings. Recordings were then split into two groups: the ones with a dream experience, and the ones without recall of dream experiences. The randomness of cardiovascular variability series was assessed through Sample Entropy metrics, which did not show any statistical difference between groups. On the other hand, a multiscale complexity analysis based on Distribution Entropy and Fuzzy Entropy revealed that a higher cardiovascular complexity is associated with a dreaming experience.

Keywords: Cardiovascular and respiratory signal processing - Cardiovascular signal processing, Cardiovascular, respiratory, and sleep devices - Monitors, Cardiovascular and respiratory signal processing - Heart Rate and Blood Pressure Variability
Abstract: Electroencephalogram (EEG)-based prediction systems are used to target anesthetic-states in patients undergoing procedures with general anesthesia. These systems are not widely employed in resource-limited settings because they are cost-prohibitive. Although anesthetic-drugs induce highly-structured, oscillatory neural dynamics that make EEG-based systems a principled approach for anesthetic-state monitoring, anesthetic-drugs also significantly modulate the autonomic nervous system (ANS). Because ANS dynamics can be inferred from electrocardiogram (ECG) features such as heart rate variability, it may be possible to develop an ECG-based system to infer anesthetic-states as a low-cost and practical alternative to EEG-based anesthetic-state prediction systems. In this work, we demonstrate that an ECG-based system using ANS features can be used to discriminate between non-general anesthetic and general anesthetic-states. With further refinement, ECG-based systems could be developed as a fully automated system for anesthetic-state monitoring in resource-limited settings.

Keywords: Cardiovascular and respiratory signal processing - Time-frequency, time-scale analysis of cardiorespiratory variability, Cardiovascular and respiratory signal processing - Heart Rate and Blood Pressure Variability
Abstract: Multiscale and multifractal (MF) analyses have been proven an effective tool for the characterisation of heartbeat dynamics in physiological and pathological conditions. However, pre-processing methods for the unevenly sampled heartbeat interval series are known to affect the estimation of MF properties. In this study, we employ a recently proposed method based on wavelet p-leaders MF spectra to estimate MF properties from cardiovascular variability series, which are also pre-processed through an inhomogeneous point-process modelling. Particularly, we exploit a non-Gaussian multiscale expansion to study changes in heartbeat dynamics as a response to a sympathetic elicitation given by the cold-pressor test. By comparing MF estimates from raw heartbeat series and the point-process model, results suggest that the proposed modelling provides features statistically discerning between stress and resting condition at different time scales. These findings contribute to a comprehensive characterization of autonomic nervous system activity on cardiovascular control during cold-pressor elicitation.

Keywords: Image registration, segmentation, compression and visualization - Machine learning / Deep learning approaches, Ophthalmic imaging and analysis, Optical imaging - Coherence tomography
Abstract: Intra-retinal cysts (IRCs) are significant in detecting several ocular and retinal pathologies. Segmentation and quantification of IRCs from optical coherence tomography (OCT) scans is a challenging task due to present of speckle noise and scan intensity variations across the vendors. This work proposes a convolutional neural network (CNN) model with an encoder-decoder pair architecture for IRC segmentation across different cross-vendor OCT scans. Since deep CNN models have high computational complexity due to a large number of parameters, the proposed method of depthwise separable convolutional filters aids model generalizability and prevents model over-fitting. Also, the swish activation function is employed to prevent the vanishing gradient problem. The optima cyst segmentation challenge (OCSC) dataset with four different vendor scans is used to evaluate the proposed model. Our model achieves a mean Dice score of 0.74 and mean recall/precision rate of 0.72/0.82 across different imaging vendors and it outperforms existing algorithms on the OCSC dataset.

Keywords: Ophthalmic imaging and analysis, Image analysis and classification - Machine learning / Deep learning approaches
Abstract: Deep learning has achieved great success in image classification task when given sufficient labeled training images. However, in fundus image based glaucoma diagnosis, we often have very limited training data due to expensive cost in data labeling. Moreover, when facing a new application environment, it is difficult to train a network with limited labeled training images. In this case, some images from some auxiliary domains (i.e., source domain) could be exploited to improve the performance. Unfortunately, direct using the source domain data may not achieve promising performance for the domain of interest (i.e., target domain) due to reasons like distribution discrepancy between two domains. In this paper, focusing on glaucoma diagnosis, we propose a deep adversarial transfer learning method conditioned on label information to match the distributions of source and target domains, so that the labeled source images can be leveraged to improve the classification performance in the target domain. Different from the most existing adversarial transfer learning methods which consider marginal distribution matching only, we seek to match the label conditional distributions by handling images with different labels separately. We conduct experiments on three glaucoma datasets and adopt multiple evaluation metrics to verify the effectiveness of our proposed method.

Keywords: Ophthalmic imaging and analysis, Image analysis and classification - Machine learning / Deep learning approaches, Image classification
Abstract: We describe and assess convolutional neural network (CNN) models for detection of glaucoma based upon optical coherence tomography (OCT) retinal nerve fiber layer (RNFL) probability maps. CNNs pretrained on natural images performed comparably to CNNs trained solely on OCT data, and all models showed high accuracy in detecting glaucoma, with receiver operating characteristic area under the curve (AUC) scores ranging from 0.930 to 0.989. Attention-based heat maps of CNN regions of interest suggest that these models could be improved by incorporation of blood vessel location information. Such CNN models have the potential to work in tandem with human experts to maintain overall eye health and expedite detection of blindness-causing eye disease.

Keywords: Ophthalmic imaging and analysis, Image visualization, Optical imaging - Coherence tomography
Abstract: Various eye diseases, including polypoidal choroidal vasculopathy (PCV) and age-related macular degeneration (AMD), affect choroidal vasculature early, but possibly minutely. However, due to the complex networked structure of the vasculature, it becomes hard to visualize, analyze and detect such changes in 2D OCT B-scan images. In contrast, algorithmic evaluation of cross-section facilitates clinicians in tracing minute variations in the vessel network, and quantifying those correlated with pathologies, potentially leading to early diagnosis. In this context, we proposed a novel method of estimating vessel cross-sections in choroidal Haller's layer. Accuracy of our method was evaluated on synthetic as well as clinical data by trained optometrists, and earned a confidence score of 90%, marking about 60% improvement over estimates based on a well-known tree-based method.

Keywords: Ophthalmic imaging and analysis, Image classification, Image analysis and classification - Machine learning / Deep learning approaches
Abstract: Diabetic retinopathy (DR) is one kind of eye disease that is caused by overtime diabetes. Lots of patients around the world suffered from DR which may bring about blindness. Early detection of DR is a rigid quest which can remind the DR patients to seek corresponding treatments in time. This paper presents an automatic image-level DR detection system using multiple well-trained deep learning models. Besides, several deep learning models are integrated using the Adaboost algorithm in order to reduce the bias of each single model. To explain the results of DR detection, this paper provides weighted class activation maps (CAMs) that can illustrate the suspected position of lesions. In the pre-processing stage, eight image transformation ways are also introduced to help augment the diversity of fundus images. Experiments demonstrate that the method proposed by this paper has stronger robustness and acquires more excellent performance than that of individual deep learning model.

Keywords: Image analysis and classification - Machine learning / Deep learning approaches, Optical imaging - Coherence tomography, Ophthalmic imaging and analysis
Abstract: The objective of this study was to build deep learning models with optical coherence tomography (OCT) images to classify normal and age related macular degeneration (AMD), AMD with fluid, and AMD without any fluid. In this study, 185 normal OCT images from 49 normal subjects, 535 OCT images of AMD with fluid, and 514 OCT mages of AMD without fluid from 120 AMD eyes as training data, while 49 normal images from 25 normal eyes, 188 AMD OCT images with fluid and 154 AMD images without any fluid from 77 AMD eyes as test data, were enrolled. Data augmentation was applied to increase the number of images to build deep learning models. Totally, two classification models were built in two steps. In the first step, a VGG16 model pre-trained on ImageNet dataset was transfer learned to classify normal and AMD, including AMD with fluid and/or without any fluid. Then, in the second step, the fine-tuned model in the first step was transfer learned again to distinguish the images of AMD with fluid from the ones without any fluid. With the first model, normal and AMD OCT images were classified with 0.999 area under receiver operating characteristic curve (AUC), and 99.2% accuracy. With the second model, AMD with the presence of any fluid, and AMD without fluid were classified with 0.992 AUC, and 95.1% accuracy. Compared with a transfer learned VGG16 model pre-trained on ImageNet dataset, to classify the three categories directly, higher classification performance was achieved with our notable approach. Conclusively, two classification models for AMD clinical practice were built with high classification performance, and these models should help improve the early diagnosis and treatment for AMD.

Keywords: Wearable sensor systems - User centered design and applications, Novel methods, Physiological monitoring - Instrumentation
Abstract: Magneto-Inertial technology is a well- established alternative to optical motion capture for human motion analysis applications, since it can allow long time monitoring in free-living conditions. Magneto and Inertial Measurement Unit (MIMU) integrates three-axial accelerometer, gyroscope and magnetometer in a single and lightweight device. The body segment orientation on which the MIMU is attached can be obtained by combining its sensor readings within a sensor fusion framework. Despite several sensor fusion implementations have been proposed, no well-established conclusion about the accuracy level achievable with MIMUs has been reached yet. The aim of this study was to perform a direct comparison among four popular sensor fusion algorithms at three different rotation rates, using the orientation provided by a stereophotogrammetric system as reference. To enable a meaningful comparison among the algorithms, a procedure for suboptimal determination of the values of the filter parameters was also proposed. The findings of this preliminary study highlighted that all the filters exhibited reasonable accuracies (rms errors < 6.4°). Moreover, in accordance with previous studies, the orientation accuracy worsened when increasing the rotation rate for all the algorithms. At fast speed, the algorithm from Sabatini (2011) showed the best performances with errors smaller than 4.1° rms.

Keywords: Physiological monitoring - Instrumentation, Physiological monitoring - Novel methods, Physiological monitoring - Modeling and analysis
Abstract: The quantification of postural control (PC) provides the opportunity to understand the function and integration of the sensorimotor subsystems. The increased availability of portable sensing technology, such as Wii Balance Boards (WBB), has afforded the capacity to capture data pertaining to motor function, outside of the laboratory and clinical setting. However, prior to its use in long-term monitoring, it is crucial to understand natural daily PC variation. Twenty-four young adults conducted repeated static PC assessments over 20 consecutive weekdays, using WBBs. 16/24 participants (eyes open) and 11/24 participants (eyes closed) exhibited statistically significant differences (p <0.05) between their initial ‘once-off’ measure and their daily measures of PC. This study showed that variations in PC exist in a healthy population, a once-off measure may not be representative of true performance and this inherent variation should be considered when implementing long-term monitoring protocols.

Keywords: Physiological monitoring - Instrumentation, Physiological monitoring - Novel methods, Wearable sensor systems - User centered design and applications
Abstract: Concussion is one of the most common injuries reported across a myriad of sports. Recent evidence suggests that individuals may possess sensorimotor deficits beyond clinical recovery, predisposing them to further injury. This preliminary prospective case series aimed to determine if an inertial sensor instrumented Y balance test can capture changes in dynamic balance, regardless of apparent ‘clinical recovery’, in six concussed elite rugby union players. The findings from this case series demonstrate that the inertial sensor-based measures can detect clinically meaningful changes in dynamic balance performance, not captured by the traditional clinical scoring methods, 48-hours post-injury and at the point of ‘clinical recovery’ (return to play). Further research should investigate the role such instrumented dynamic balance assessments may play in the management of sports-related concussion.

Keywords: Physiological monitoring - Instrumentation, Wearable wireless sensors, motes and systems, Mechanical sensors and systems
Abstract: Although concussion continues to be a major source of acute and chronic injury in automotive, athletic and military arenas, concussion injury mechanisms and risk functions are ill-defined. This lack of definition has hindered efforts to develop standardized concussion monitoring, safety testing and protective countermeasures. Recent research has provided evidence of the role of repetitive head impact exposure as a predisposing factor for the onset of concussion using developed instrumented helmets and mouthguards. To overcome this knowledge gap, we have developed, tested and deployed a head impact monitoring mouthguard (IMM) system. In this study, we deployed the IMM system to gather high quality estimates of athlete head impacts in situ. And with enough longer-term data collection, potential concussive events or mild traumatic brain injuries (mTBIs) will be gathered and ideally will provide actionable risk-based threshold.

Keywords: Physiological monitoring - Novel methods, Modeling and analysis, Wearable sensor systems - User centered design and applications
Abstract: This paper presents an architecture for generalized predictive control for an active prosthetic socket system, based on a cost function performance index measure for minimization of residual limb tissue injury. Finite element analysis of a transtibial residuum model donned with a total surface bearing socket was used to provide controller training data and biomechanical rationale for deep tissue injury risk assessment, by estimating the internal deformation state of the soft tissues and the residuum-socket interface loading under a range of prosthetic loading instances. The results demonstrate the concept of this approach for interface actuation modelled as translational spring and damper systems.

Keywords: Sensor systems and Instrumentation, Modeling and analysis, Novel methods
Abstract: Ambulatory sensing of gait kinematics using inertial measurement units (IMUs) usually uses sensor fusion filters. These algorithms require measurement updates to reduce drift between segments. A full body IMU suit can use biomechanical relations between body segments to solve this. However, when minimising the sensor set, we lose a lot of this information. In this study, we explore the assumptions of zero moment point (ZMP) as a possible source of measurement updates for the sensor fusion filters. ZMP is otherwise utilised for humanoid gait in robots. In this study, first, the relation between the ZMP and centre of pressure (CoP) is studied using a GRAIL system, consisting of opto-kinetic measurements. We find that the mean distance over the gait cycle between ZMP and CoP is 10.5(1.2) % of the foot length. Following this, we show how these results could be used to improve measurements in a minimal IMU based sensing setup.