Keywords: Neurological disorders - Psychiatric disorders
Abstract: Major Depressive Disorder (MDD) is a very serious mental illness that can affect the daily lives of patients. Accurate diagnosis of this disorder is necessary for planning individualized treatment. However, diagnosing MDD requires the clinicians to personally interview the subjects and rate the symptoms based on Diagnostic and Statistical Manual of Mental Disorders (DSM), which can be very time consuming. Discovering quantifiable signals and biomarkers associated with MDD using functional magnetic resonance imaging (fMRI) scans of patients have the potential to assist the clinicians in their assessment. This paper explores the use of resting-state functional connectivity and network features to classify MDD vs. healthy subjects. For each subject, mean time-series are extracted from 85 brain regions and they are decomposed to 4-frequency bands. Mean time-series for each of the frequency bands are utilized to compute the Pearson correlation and network characteristics. Features are selected separately from correlation and network characteristics using Minimum Redundancy Maximum Relevance (mRMR) to create the final classifier. The proposed scheme achieves 79% accuracy (65 out of 82 subjects classified correctly) with 86% sensitivity (42 out of 49 MDD subjects identified correctly) and 70% specificity (23 out of 33 controls identified correctly) using leave-one-out classification with in-fold feature selection. Pearson correlation had the highest discrimination in band 0.015-0.03 Hz and network based features had the highest discrimination in band 0.03-0.06 Hz for distinguishing MDD vs. healthy subjects.

Keywords: Neurological disorders - Stroke, Neurorehabilitation, Brain physiology and modeling - Sensory-motor
Abstract: Ischemic stroke is a major cause of disability among adults worldwide. Despite its prevalence, few effective treatment options exist to alleviate sensory and motor dysfunctions that result from stroke. In the past, rodent models of stroke have been the primary experimental models used to develop stroke therapies. However, positive results in these studies have failed to replicate in human clinical trials, highlighting the importance of nonhuman primate (NHP) models as a preclinical step. Although there are a few NHP models of stroke, the extent of tissue damage is highly variable and dependent on surgical skill. In this study, we employed the photothrombotic stroke model in NHPs to generate controlled, reproducible ischemic lesions. Originally developed in rodents, the photothrombotic technique consists of intravenous injection of a photosensitive dye such as Rose Bengal followed by illumination of an area of interest to induce endothelial damage resulting in the formation of thrombi in the illuminated vasculature. We developed a quantitative model to predict the extent of tissue damage based on the light scattering profile of light in the cortex of NHPs. We then employed this technique in the sensorimotor cortex of two adult male Rhesus Macaques. In vivo optical coherence tomography imaging of the cortical microvasculature and subsequent histology confirmed the formation of focal cortical infarcts and demonstrated its reproducibility and ability to control the sizes and locations of light-induced ischemic lesions in the cortex of NHPs. This model has the potential to enhance our understanding of perilesional neural dynamics and can be used to develop reliable neurorehabilitative therapeutic strategies to treat stroke.

Keywords: Neurological disorders, Brain functional imaging - fMRI, Brain physiology and modeling - Neural dynamics and computation
Abstract: Understanding the neuronal network dynamics underlying the third most common movement disorder, cervical dystonia, can be achieved using dynamic causal modelling. Current literature establishes structures of the midbrain network for covert attentional orienting as dysfunctional in patients with cervical dystonia. One of these structures is the superior colliculus, for which it is hypothesised that deficient GABAergic activity therein causes cervical dystonia. To understand the role that this node plays in cervical dystonia, various connectivity models of the midbrain network were compared under the influence of a loom-recede visual stimulus fMRI paradigm. These models included the thalamus and striatum, crucial nodes in the direct/indirect pathways for motor movement and inhibition. The parametric empirical Bayes approach was used to quantify the difference in connection strengths across the winning models between patients and controls. Our findings demonstrated greater modulation by a looming stimulus event on the strength of connection from the striatum to the superior colliculus in patients. These results offer new means to understanding the pathophysiology of cervical dystonia.

Keywords: Neurological disorders, Human performance - Speech, Neurological disorders - Diagnostic and evaluation techniques
Abstract: This study has investigated the use of inter-personnel mutual information computed from the phonetic sound recordings to differentiate between Parkinson’s disease (PD) and control subjects. The normalized mutual information (NMI) denotes the amount of information shared between the voice recordings of people within the same group: PD and Control. The hypothesis of this study was that within-group NMI will be significantly different when compared with intergroup NMI. For each phonetic sound, the NMI was computed for every pairing of recordings for both the PD and control groups. Pearson correlation coefficient analysis was used to determine the association of NMI with clinical parameters including Unified Parkinson’s Disease Rating Scale (UPDRS), Montreal cognitive assessment (MoCA) and disease duration. ANOVA test for the three phonetic sounds of control and PD subjects showed that there is a significant difference between the intra-group mean NMI for the two groups (p < 0.003) and also showed significant association with the UPDRS motor examination score, MoCA and disease duration.

Keywords: Neurological disorders - Diagnostic and evaluation techniques, Neural signals - Nonlinear analysis, Neuromuscular systems - Postural and balance
Abstract: Balance maintenance is commonly analyzed by evaluating the center of pressure (COP) displacement, which presents an acknowledged non-stationary behavior. The latter led to an evaluation of COP regularity through complexity measures such as the approximate (AppEn) and sample entropy (SampEn). These indexes quantify the regularity of time-series in terms of inner pattern recurrence; however, they are highly dependent on the input parameters used for their computation. Thus, this study aimed to evaluate the use of the AppEn, SampEn and a recently proposed entropy measure, the fuzzy entropy (FuzzyEn) for the analysis of COP time-series in type-2 diabetic subjects with and without neuropathy during quiet standing trials in eyes open condition. Results highlighted consistency of entropy measures for different values of input parameters, showing significant differences between the two populations in terms of COP regularity for both anteriorposterior and medial-lateral directions. Findings of this study outline low complexity in postural control of neuropathic subjects, also in the medial-lateral direction, which could indicate a limited capacity of producing adaptable responses, relying on fixed balance control patterns. Further, they support the use of complexity measures for the analysis of patients with diabetic neurological impairment.

Keywords: Neurological disorders - Diagnostic and evaluation techniques, Neurological disorders - Psychiatric disorders
Abstract: Idiopathic Parkinson’s disease and atypical parkinsonian syndromes have similar symptoms at early disease stages, which makes the early differential diagnosis difficult. Positron emission tomography with ¹⁸F-FDG shows the ability to assess early neuronal dysfunction of neurodegenerative diseases and is well established for clinical use. In the past decades, machine learning methods have been widely used for the differential diagnosis of parkinsonism based on metabolic patterns. Unlike these conventional machine learning methods relying on hand-crafted features, the deep convolutional neural networks, which have achieved significant success in medical applications recently, have the advantage of learning salient feature representations automatically and effectively. This advantage may offer more appropriate invisible features extracted from data for the enhancement of the diagnosis accuracy. Therefore, This paper develops a 3D deep convolutional neural network on ¹⁸F-FDG PET images for the automated early diagnosis. Furthermore, we depicted in saliency maps the decision mechanism of the deep learning method to assist the physiological interpretation of deep learning performance. The proposed method was evaluated on a dataset with 920 patients. In addition to improving the accuracy in differential diagnosis of parkinsonism compared to state-of-the-art approaches, the deep learning methods also discovered saliency features in a number of critical regions (e.g., midbrain), which are widely accepted as characteristic pathological regions for movement disorders but was ignored in the conventional analysis of FDG PET images.

Keywords: Neural networks and support vector machines in biosignal processing and classification, Data mining and processing - Pattern recognition, Signal pattern classification
Abstract: Parkinson’s Disease (PD) is the second most common neurodegenerative disorder worldwide, causing both motor and non-motor symptoms. In the early stages, symptoms are mild and patients may ignore their existence. As a result, they do not undergo any related clinical examination; hence delaying their PD diagnosis. In an effort to remedy such delay, analysis of data passively captured from user’s interaction with consumer technologies has been recently explored towards remote screening of early PD motor signs. In the current study, a smartphone-based method analyzing subjects’ finger interaction with the smartphone screen is developed for the quantification of fine-motor skills decline in early PD using Convolutional Neural Networks. Experimental results from the analysis of keystroke typing in-the-clinic data from 18 early PD patients and 15 healthy controls have shown a classification performance of 0.89 Area Under the Curve (AUC) with 0.79/0.79 sensitivity/specificity, respectively. Evaluation of the generalization ability of the proposed approach was made by its application on typing data arising from a separate self-reported cohort of 27 PD patients’ and 84 healthy controls’ daily usage with their personal smartphones (data in-the-wild), achieving 0.79 AUC with 0.74/0.78 sensitivity/specificity, respectively. The results show the potentiality of the proposed approach to process keystroke dynamics arising from users’ natural typing activity to detect PD, which contributes to the development of digital tools for remote pathological symptom screening.

Keywords: Neural networks and support vector machines in biosignal processing and classification, Physiological systems modeling - Signal processing in physiological systems, Signal pattern classification
Abstract: Modeling transcranial magnetic stimulation (TMS) evoked potentials (TEP) begins with classification of stereotypical single-pulse TMS responses in order to select validation targets for generative dynamical models. Several dimensionality reduction techniques are commonly in use to extract statistically independent features from experimental data for regression against model parameters. Here, we first designed a 3-dimensional feature space based on commonly described event-related potentials (ERP) from the literature. We then compared classification schemes which take as inputs either the 3D projection space or the original full rank input space. Their ability to discriminate TEP recorded from different brain regions given a stimulus site were evaluated. We show that a deep learning architecture, employing Convolutional Neural Network (CNN) and Multi-Layer Perceptron (MLP), yields better accuracy than the 3D projection and raw TEP input combined with Support Vector Machines. Such supervised feature extraction models may therefore be useful for scoring neural circuit simulations based on their ability to reproduce the underlying dynamical processes responsible for differential TEP responses.

Keywords: Neural networks and support vector machines in biosignal processing and classification, Time-frequency and time-scale analysis - Nonstationary processing, Signal pattern classification
Abstract: Hybrid, passive brain-computer (h/pBCI) interfaces have received much attention in regards to measuring various mental states. A high classification rate of operator workload state is necessary in order to be able to enhance operator performance. Physiological measures have been used with machine learning algorithms to classify workload state, however, these measures are hypothesized to suffer from inherent nonstationarity. To attain a more generalizable classifier, a prior solution has been to use a multi-day learning paradigm to train classifier models. In earlier work, we have shown that increasing the number of unique data sessions used to form a learning set can improve the accuracy of classifying mental workload where improved generalizability is partly attributable to the multi-day paradigm. To further investigate methods that produce more generalizable classifiers, we look to ensemble learning. Here we implement ensemble learning to increase accuracies, reduce variance, and leverage theoretical performance of the ensemble as compared to observed to make inference about generalization. An adaptive boosting method (AdaBoost) is used to train a “base learning algorithm” multiple times, adaptively adjusting to errors and forming a vote out of the resulting hypotheses using three different base learning algorithms: an artificial neural network (ANN), a support vector machine (SVM), and linear discriminant analysis (LDA). We observed that the ensemble converged on theoretical performance with respect to error and variance only when the training sets were formed using the multi-day paradigm. These results indicate that ensemble learning approaches can be used in examples of pBCI such as those designed here, but they are also affected by theorized nonstationarity in physiological response. The observation of ensemble convergence on theoretical performance may be used to make inference about generalizability when simple accuracy of detection can be misleading.

Keywords: Neural networks and support vector machines in biosignal processing and classification
Abstract: Brain-Machine Interfaces (BMIs) aim to help disabled people brain control the external devices to finish a variety of movement tasks. The neural signals are decoded into the execution commands of the apparatus. However, most of the existing decoding algorithms in BMI are only trained for a single task. When facing a new task, even if it is similar to the previous one, the decoder needs to be re-trained from scratch, which is not efficient. Among the different types of decoders, reinforcement learning (RL) based algorithm has the advantage of adaptive training through trial-and-error over the recalibration used in supervised learning. But most of the RL algorithms in BMI do not actively leverage the acquired knowledge in the old task. In this paper, we propose a kernel RL algorithm with a weight transfer mechanism for new task learning. The existing neural patterns are clustered according to their similarities. A new pattern will be assigned with the weights that are transferred from the closest cluster. In this way, the most similar experiences from the previous task could be re-utilized in the new task to fasten the learning speed. The proposed algorithm is tested on the synthetic neural data. Compared with the policy of re-training from scratch, the proposed weight transfer mechanism could maintain a significantly higher performance and achieve a faster learning speed on the new task.

Keywords: Neural networks and support vector machines in biosignal processing and classification
Abstract: Extreme Learning Machine (ELM) with Radial Basis Function (RBF) Kernel has demonstrated strong capability in pattern recognition and classification problems. NS1 is a biomarker for flavivirus related diseases, where current detection methods are serum based and hence invasive. Our previous work has captured NS1 molecular fingerprint in saliva using Surface Enhanced Raman Spectroscopy (SERS) that could amount to non-invasive detection method. SERS is an improved Raman spectroscopic technique, which can amplify spectral intensity by 103 to l07 times, to yield usable spectra of low concentration NS1 in saliva. The spectra produced contain 1801 features for each of the 284 samples collected. Principal Component Analysis (PCA) transforms a high dimensional data to a lower dimension principal components (PCs), at no sacrifice of important information of the original data. Both termination criteria of PCA and kernel parameters of ELM have effect on performance of the classifier models. This paper aims to unravel an optimal ELM-RBF classifier model for classification of NS1 salivary SERS spectra. Performance of a total of 864 classifier models are examined and compared in terms of [accuracy, kappa, precision, sensitivity and specificity]. Results show that CPV- and EOC-ELM-RBF classifier models are on par and outperform the Scree-ELM-RBF classifier models.

Keywords: Signal pattern classification, Data mining and processing in biosignals, Data mining and processing - Pattern recognition
Abstract: Pediatric sleep apnea-hypopnea syndrome (SAHS) is a highly prevalent breathing disorder that is related to many negative consequences for the children’s health and quality of life when it remains untreated. The gold standard for pediatric SAHS diagnosis (overnight polysomnography) has several limitations, which has led to the search for alternative tests. In this sense, automated analysis of overnight oximetry has emerged as a simplified technique. Previous studies have focused on the extraction of ad-hoc features from the blood oxygen saturation (SpO2) signal, which may miss useful information related to apnea and hypopnea (AH) events. In order to overcome this limitation of traditional approaches, we propose the use of convolutional neural networks (CNN), a deep learning technique, to automatically detect AH events from the SpO2 raw data. CHAT-baseline dataset, composed of 453 SpO2 recordings, was used for this purpose. A CNN model was trained using 60-s segments from the SpO2 signal using a training set (50% of subjects). Optimum hyperparameters of the CNN architecture were obtained using a validation set (25% of subjects). This model was applied to a third test set (25% of subjects), reaching 93.6% accuracy to detect AH events. These results suggest that the application of CNN may be useful to detect changes produced in the oximetry signal by AH events in pediatric SAHS patients.

Keywords: Optical imaging and microscopy - Diffuse optical tomography, Optical imaging and microscopy - Near infra-red spectroscopy, Optical imaging and microscopy - Neuroimaging
Abstract: A non-contact scanning system was developed based on a picosecond supercontinuum laser, a fast-gated single-photon avalanche diode, a galvanometer scanner and time-correlated single photon counting. We demonstrated the feasibility of dynamic non-contact imaging of brain activation covering an area of e.g. 4×4 cm2 with 32×32 pixels at a frame rate of 1/s.

Keywords: Infra-red imaging
Abstract: Infrared Thermography (IRT) is a passive imaging modality which uses electromagnetic radiation emitted from a black body as a function of temperature. Over the last few decades, it has been demonstrated that the temperature signatures carry information on generalized infections (like in e.g. fever screening during worldwide epidemics) and on local (superficial) metabolic activities. It has also been shown that this method can be applied to vital sign monitoring. This paper and the corresponding talk introduce the state of the art in this area

Keywords: Optical imaging and microscopy - Near infra-red spectroscopy
Abstract: Medical devices intended for the diagnostic of neurodegenerative diseases are promising alternatives to study neural activities underlying cognitive functions and pathologies, and eventually to recover lost neural vital functions. This talk concerns a non-invasive functional near infrared spectroscopy (fNIRS) based platform mainly intended for epileptic foci localization and seizures onset detection. This platform is composed of an array of optodes based on fast detectors which are intended to exploit time-domain photon-counting approach to quantify changes in the number of photons scattering back to nearly where they came from. The source and detector of an optode are placed in null/small source-detector distance (ns-SDD) configuration. We describe the multidimensional design challenges to achieve the low-power small area system-in-package optical emitter and receiver. Application-specific microsystem architectures will be discussed, and experimental results will be demonstrated.

Keywords: Novel imaging modalities
Abstract: Pulse oximetry (POX) estimates arterial oxygen saturation (SpO2) from the relative amplitudes of PPG (photo-plethysmography) signals at two or more distinct wavelengths, typically red and near infra-red (nIR). Implicit assumptions in POX are: 1) only arterial blood is pulsatile and 2) the penetration depths of the used wavelengths are the same. This paper addresses the second assumption in the context of differences between contact- (e.g. finger probes) and remote POX, referred herein as conPOX and remPOX, respectively. Spekcle imaging as ‘colorblind PPG’ is proposed to probe relative penetration depths and correct for them to enhance POX accuracy

Keywords: Image analysis and classification - Machine learning / Deep learning approaches, Image feature extraction, Functional image analysis
Abstract: Patient vital sign monitoring is essential in many medical domains, from its traditional application to critical care, to recent uses in digital therapeutics and mental health. Vital sign variability over time contains a vast amount of clinical information. However, population- based metrics and human observation can easily lose this information in the sea of inter-patient noise. This challenges the ability (of algorithms and doctors alike) to identify patients’ deterioration with sufficient time to intervene. The loss of patient-specificity, in turn, hurts clinical performance via high false-alarm rates, missed early warning signs, and emergency readmissions. Intelli- gent computational methods are promising in their ability to both (i) handle large volumes of data and (ii) implement models sufficiently complex to examine the implications of intra-patient variability. Building smart algorithms for real-world vital sign monitoring scenarios is a challenge that begins with the measurement device and ends with clinical interpretation of an algorithm’s output. We will focus on three types of algorithms that are key to advancing the field: 1) Algorithms that secure reliable data (for analysis by subsequent algorithms) 2) Algorithms that parameterize complex machine learning models 3) Algorithms that are robust and interpretable by clinical staff

Keywords: Physiological monitoring - Modeling and analysis, Novel methods, Wearable sensor systems - User centered design and applications
Abstract: The gold standard for tidal volume measurement is spirometry. Based on retrospective data, this study evaluates different geometric lung models in their ability to deliver accurate tidal volumes from changes in thoracic and abdominal circumference. The geometric lung models showed good coefficients of determination (adjusted R2 >0.97) compared to the tidal volumes measured by a body plethysmograph. Tidal volumes obtained by circumference changes might be used in surveillance systems to analyze respiration without a face mask.

Keywords: Physiological monitoring - Novel methods, Sensor systems and Instrumentation, Portable miniaturized systems
Abstract: Cigarette smoking has severe health impacts on those who smoke and the people around them. Several wearable sensing modalities have recently been investigated to collect objective data on daily smoking, including detection of smoking episodes from breathing patterns, hand to mouth behavior, and characteristic hand gestures or cigarette lighting events. In order to provide new insight into ongoing research on the objective collection of smoking-related events, this paper proposes a novel method to identify smoking events from the associated changes in heart rate parameters specific to smoking. The proposed method also accounts for the breathing rate and body motion of the person who is smoking to better distinguish these changes from intense physical activities. In this research, a human study was first performed on 20 daily cigarette smokers to record heart rate, breathing rate, and body acceleration collected from a wearable chest sensor consisting of an ECG and bioimpedance measurement sensor and a 3D inertial sensor. Each participant spent ~2 hours in a laboratory environment (mimicking daily activities that included smoking 4 cigarettes) and ~24 hours under unconstrained free-living conditions. A support vector machine-based classifier was developed to automatically detect smoking episodes from the captured sensor signals using fifteen features selected by a forward-feature selection method. In a leave one subject out cross-validation, the proposed approach detected smoking events (187 out of total 232) with the sensitivity and F-score accuracy of 0.87 and 0.79, respectively, in the laboratory setting (known activities) and 0.77 and 0.61, respectively, under free-living conditions. These results validate the proof-of-concept that, although further research is necessary for performance improvement, characteristic changes in heart rate parameters could be a useful indicator of cigarette smoking even under free-living conditions.

Keywords: Physiological monitoring - Instrumentation, Sensor systems and Instrumentation, Portable miniaturized systems
Abstract: In this study, we show the measurement of respiratory rate on exercise using a nanoparticle-based humidity sensor. A portable respiratory rate sensor is comprised of a colloidal silica nanoparticle-based humidity sensor chip. The impedance of the silica nanoparticle film is dependent on humidity and it is used for the detection of humid exhaled air. The respiratory rate sensor can be attached on an oxygen mask and the sensor signal is remotely monitored via Bluetooth. We show that the sensor follows a respiratory rate up to 60 bpm. We compare the sensor signal with that of a conventional respiratory measurement unit, which monitors a respiratory volume. The nanoparticle-based sensor can monitor a respiratory rate of an exercising person on a treadmill. The sensor operates stably for almost one year.

Keywords: Physiological monitoring - Novel methods, Integrated sensor systems, Mechanical sensors and systems
Abstract: We present a new algorithm for peak detection in a ballistocardiography (BCG) signal and use it for heart rate estimation. Ejection waves of the BCG signal are enhanced and coarse heart beat locations estimated. Ejection waves I, J and K are detected simultaneously around coarse locations, only using detection of local maxima and weighted summation of peak heights. Due to a lack of reference BCG annotations, the algorithm’s performance is gaged by using the detected peaks for heart rate estimation. On two datasets acquired with a pneumatic BCG system, we evaluate the heart rate estimation performance and compare it against other methods found in literature. The first dataset features high-quality signals obtained from 11 volunteers. The second dataset is gathered from 42 patients in a clinical environment and provides lower quality taken from a more realistic scenario. With a mean absolute percentage error of 2.58 % at 65 % coverage on the clinical dataset, the presented method is on par with the best-performing state-of-the-art algorithms investigated. Limits of agreement (5th/95th percentiles) in a comparison with ECG-based heart rate measurements lie within P5=-3.63 and P95=5.78 beat/min.

Keywords: Sensor systems and Instrumentation, Integrated sensor systems, Physiological monitoring - Modeling and analysis
Abstract: Many emotion recognition schemes have been proposed in the state-of-the-art. They generally differ in terms of the emotion elicitation methods, target emotional states to recognize, data sources or modalities, and classification techniques. In this work several biosignals are explored for emotion assessment during immersive video visualization, collecting multimodal data from Electrocardiography (ECG), Electrodermal Activity (EDA), Blood Volume Pulse (BVP) and Respiration sensors. Participants reported their emotional state of the day (baseline), and provided self-assessment of the emotion experienced in each video through the Self-Assessment Manikin (SAM), in the valence-arousal space. Multiple physiological and statistical features extracted from the biosignals were used as inputs to an emotion recognition workflow, targeting user-independent classification with two classes per dimension. Support Vector Machines (SVM) were used, as it is considered one of the most promising classifiers in the field. The proposed approach lead to accuracies of 69.13% for arousal and 67.75% for valence, which are encouraging for further research with a larger training dataset and population.

Keywords: Sensor systems and Instrumentation, New sensing techniques, Wearable sensor systems - User centered design and applications
Abstract: A wearable system for monitoring non-invasively signals invaluable when examining person suffering from vasovagal syncope is presented in the paper. Following signals are continuously recorded: electrocardiogram, photopletysmogram, impedance cardiogram and electrodermal resistance

Keywords: Physiological systems modeling - Signal processing in physiological systems, Physiological systems modeling - Multivariate signal processing, Causality
Abstract: Epilepsy, a prevalent neurological disorder characterized by recurrent seizures, presents a unique opportunity to study and understand the transition into and out of crises of arguably the most complicated biological system, the human brain. Success in this endeavor will contribute to more accurate diagnosis and better treatment of this disorder, much to the benefit of the epilepsy patient. We will present fundamental questions about ictogenesis that remain unresolved despite decades of research, as well as examples and insights from the brain network perspective that could advance the field.

Keywords: Nonlinear dynamic analysis - Phase locking estimation, Time-frequency and time-scale analysis - Wavelets, Data mining and processing in biosignals
Abstract: The canonical unpredictability of epileptic seizures belies the existence in many patients of rhythms in seizure timing. Here, by analyzing chronic (years-long) EEG recordings from patients with epilepsy, we identify distinct types of rhythms that may be regulated differentially. Some highly regular rhythms appear to be externally-cued, such as by linkage to days of the week, while other more variable rhythms may be regulated by endogenous factors. In contrast to previous reports, we find that when flexible, spectral-domain analytical methods are employed, these rhythms are apparent in most patients. Leveraging rhythms of brain activity to anticipate seizures may be a broadly applicable approach to seizure forecasting.

Keywords: Causality, Kalman filtering, Signal pattern classification
Abstract: Despite recent technological developments in EEG signal processing, EEG is mainly analyzed visually in clinical practice by the treating epileptologist to assess the type of epilepsy or to localize the epileptogenic focus. In this study we present recent advances in EEG processing and how these can be used in clinical practice. We show that automated ESI and functional brain connectivity help to localize the epileptogenic focus and that resting state functional connectomes can be used to diagnose and lateralize temporal lobe epilepsy.

Keywords: Connectivity measurements, Independent component analysis, Data mining and processing in biosignals
Abstract: The possibility of recording together non-invasive electrophysiology (EEG, MEG) and intracerebral signals acquired in patients in presurgical evaluation of epilepsy opens new venues in mapping epileptic networks. Firstly, it permits to assess sensitivity of surface signals and to validate source reconstruction techniques. Secondly, it gives the opportunity of building a fused modality, combining the advantages of intracerebral signals (which provides a local view of brain activity), and surface signals (which gives a global view). We review here the advantages given by simultaneous MEG-SEEG recordings in the characterization of epileptic networks.

Keywords: Neuromuscular systems - Postural and balance, Motor learning, neural control, and neuromuscular systems, Neuromuscular systems - Central mechanisms
Abstract: Temporal pattern of postural sway during human quiet stance is an indicator that can characterize postural control strategies used by the central nervous system. The stiffness control model with a ceaseless active feedback controller and the intermittent feedback control model that switches the active feedback control off and on intermittently are two representative hypotheses for human postural control, among others. Here, we examined a methodology to quantify the performance of those two hypothetical models for simulating postural sway as “similar” as possible to experimentally observed actual sway data. To this end, we utilized a data assimilation technique with the approximate Bayesian inference.

Keywords: Neuromuscular systems - Postural and balance, Neuromuscular systems - Learning and adaption, Sensory neuroprostheses - Somatosensory
Abstract: We propose a novel method to investigate the adaptation ability of human postural control system. Our preliminary experiment indicates that the postural control system modifies the control parameters of position and velocity gains according to type of artificially imposed novel dynamics.

Keywords: Human performance - Cognition, Motor learning, neural control, and neuromuscular systems, Brain physiology and modeling - Cognition, memory, perception
Abstract: Humans can intuitively improvise and use objects in our environment as tools to assist interaction tasks. Here, starting from findings in our Neuroscientific studies on tool use and embodiment, we are developing an algorithm that enables similar human like tool cognition by robots. Our algorithm enables a robot without any tool experience, to automatically recognize an object (seen for the first time) as a potential tool for an otherwise unattainable task, and use the tool to perform the task thereafter.

Keywords: Motor learning, neural control, and neuromuscular systems, Neuromuscular systems - Learning and adaption, Neuromuscular systems - Postural and balance
Abstract: Skilful movement arises through continuous adaptation of our sensorimotor control system to the external environment. While most previous research focused on the adaptation of predictive force or impedance control, recent research has started to highlight the importance of feedback adaptation. I review my recent research in these areas and highlight some new directions that might broaden our understanding of these complementary processes.

Keywords: Neuromuscular systems - Postural and balance, Motor learning, neural control, and neuromuscular systems, Neuromuscular systems - Learning and adaption
Abstract: Sustained regulation of human balance is believed to be maintained by automated, reflexive feedback control. This control is typically modelled as continuous, linear predictive feedback. Our alternative hypothesis is that sustained balance is a sequential process of discretely executed motor programs. We note that balance is characterized by a sustained 0.5-2 Hz oscillatory component which is not related linearly to an input disturbance. We are investigating the source of this oscillation in ankle moment. Our question is whether the mechanism of the feedback control loop can cause this peak at 0.5-2 Hz? We test two deterministic models, continuous linear-predictive control (cc) and intermittent predictive control (ic), each without (cc, ic) and with (ccn, icn) addition of observation and motor noise.

Thirteen healthy participants, strapped to a one degree of freedom robot with dynamics of upright standing, used visual and haptic feedback and myoelectric control signals from the anterior and posterior lower leg to maintain balance for 250s. An input disturbance of discrete steps in external force was applied. A linear time invariant model (ARMA) was used to extract the component of the control signal linearly related to the disturbance. Subtraction of the linear component from the experimental control signal gave the non-linear component. Using optimization of parameters and white noise amplitude, we fitted four models (cc, ccn, ic, icn) to replicate concurrently (i) the linear component of the time series in the time domain and (ii) the magnitude-frequency spectrum of the non-linear component.

All models replicated the linear component to comparable, high fit. Only the intermittent control models (ic, icn) replicated the non-linear 0.5-2 Hz component.

Our tentative, working conclusion is that the non-oscillatory component arises from a combination of intermittent sampling and an intermittent predictive controller mistuned to the controlled system.

Keywords: Motor learning, neural control, and neuromuscular systems, Neuromuscular systems - Learning and adaption, Neuromuscular systems - Postural and balance
Abstract: Activities of daily living are characterized by complex tasks, often unstable. These tasks, most of the times, have more than one goal to be achieved and permit to choose more than one control strategy to master them. In this work we critically analyze how it is possible to design experiments in a multi strategies/multi goals unstable scenario. In the first experiment results revealed how changing the intrinsic dynamics of the task permitted to force the subjects to adopt specific strategies. In the second experiment results showed how two simultaneous unstable goals can be achieved composing different motor control mechanisms.

Keywords: Cellular force transduction - Cell mechanics, Micro- and nano-technology
Abstract: Mechanical force provides a novel physical approach for cell fate regulation. Magnetic field with spatiotemporal control and excellent tissue permeability holds the promise for magneto-mechanical actuation via magnetic materials. Magnetic nanoparticles (MNPs) functionalized with targeting moieties can recognize specific cell components and induce mechanical actuation under magnetic field. Their size is adequate for reaching tumors and targeting cancer cells. However, due to the nanometric size, the force generated by MNPs is smaller than the force required for largely disrupting key components of cells. Here, we show the magnetic assembly process of the nanoparticles inside the cells, to form elongated aggregates with the size required to produce elevated mechanical forces. We synthesized iron oxide nanoparticles doped with zinc, to obtain high magnetization, and functionalized for targeting cancer cells. Under a low frequency rotating magnetic field at 15 Hz and 40 mT, the internalized MNPs formed elongated aggregates and generated pN to initiate programmed cell death and necrosis. Our work provides a novel strategy of designing magnetic nanomedicines for mechanical control the cell fate.

Keywords: Micro- and nano-technology, Electric fields - Tissue regeneration, Biomaterial-cell interactions - Stimuli-sensitive biomaterials
Abstract: Highly integrated micromachines featuring remote electrical stimulation for cell differentiation are developed for targeted cell therapy. The micromachines consist of piezoelectric and magnetic materials. The magnetic nanoparticles serve as the component for the actuation of the device. The piezoelectric component acts as acoustically/magnetically responsive cell electrostimulation platform. The microrobots can swim by actuation of rotating magnetic fields in different liquid environments that mimic the human body fluids.

Keywords: Micro- and nano-technology, Cellular force transduction - Mechanical stimuli and mechanotransduction, Biomaterial-cell interactions - Functional biomaterials
Abstract: As a promising nanomaterial, magnetic nanoparticles (MNPs) with remote control and manipulation properties have been explored extensively for magnetic hyperthermia and magneto-mechanical destruction of cancer cells, respectively. However, the combination of these two treatment modalities has not yet been explored. Herein, we design a 60 nm integrin-targeted zinc-doped iron oxide nanocube as the magneto-responder to perform magneto-mechanical destruction and hyperthermia (MMDH) therapy to fight against glioma modulated with dual frequencies of magnetic fields. The hyperthermia effect sensitized by magnetic forces was explored in vitro. By means of the 15 Hz rotating magnetic field (RMF) and 375 kHz alternating magnetic field (AMF) in order, we can manipulate the nanocubes to produce a localized mechanical torque to impair lysosome structure, improve reactive oxygen species (ROS) level and induce mitochondria depolarization sensitizing cancer cells; thus, followed by a mild magnetic hyperthermia treatment, we can initiate cells apoptosis and achieve an effective therapeutic effect. This MMDH approach is a powerful tool to treat cancer cells in a spatiotemporal, effective and noninvasive fashion based on one simple MNP and adjusted magnetic fields.

Keywords: Biomaterial-cell interactions - Stimuli-sensitive biomaterials, Gene and drug delivery - Cell specific delivery
Abstract: Sophisticated formulations responsive to light offer an excellent opportunity for the controlled release of biomolecules, enabling the control of timing, duration, location and their dosage. During my presentation, I will cover some of the experimental results obtained my group in the design of advanced drug release systems based on light-triggerable materials.

Keywords: Biomaterial-cell interactions - Stimuli-sensitive biomaterials, Biomaterial-cell interactions - Functional biomaterials, Biomaterial-cell interactions - Biologics
Abstract: This paper and the related talk aim at describing the potential of ultrasonic stimulation, alone or in combination with piezoelectric nanomaterials, to induce regenerative phenomena in cells and tissues.

Keywords: Health Informatics - Behavioral health informatics, Sensor Informatics - Data inference, mining, and trend analysis, General and theoretical informatics - Data intelligence
Abstract: People Living with Dementia (PLwD) often exhibit behavioral and psychological symptoms of dementia; with agitation being one of the most prevalent symptoms. Agitated behaviour in PLwD indicates distress and confusion and increases the risk to injury to both the patients and the caregivers. In this paper, we present the use of wearable devices to detect agitation in PLwD. We hypothesize that combining multi-modal sensor data can help in building better classifiers to identify agitation in PLwD in comparison to a single sensor. We present a unique study to collect motion and physiological data from PLwD. This multi-modal sensor data is subsequently used to build predictive models to detect agitation in PLwD. The results on Random Forest for 28 days of data from PLwD show a strong evidence to support our hypothesis and highlight the importance of using multi-modal sensor data for detecting agitation events amongst them.

Keywords: Health Informatics - internet of things, Health Informatics - Behavioral health informatics, Sensor Informatics - Wearable systems and sensors
Abstract: A less-invasive method for the diagnosis of the major depressive disorder can be useful for both the psychiatrists and the patients. We propose a machine learning framework for automatically discriminating patients suffering from the major depressive disorder (n=14) and healthy subjects (n=17). To this end, spontaneous physical activity data were recorded via a watch-type computer device equipped by the participants in their daily lives. Two machine learning models are investigated and compared, i.e., support vector machines, and deep recurrent neural networks. Experimental results show that, both of the two methods, i.e., the static model fed with human hand-crafted features, and the sequential model fed with raw data can reach a promising performance with an unweighted average recall at 76.0% and 56.3%, respectively.

Keywords: Public Health Informatics - Public health management solutions, Health Informatics - Behavioral health informatics, Public Health Informatics - Non-medical data analytics in public health
Abstract: Obesity is a preventable disease that affects the health of a significant population percentage, reduces the life expectancy and encumbers the health care systems. The obesity epidemic is not caused by isolated factors, but it is the result of multiple behavioural patterns and complex interactions with the living environment. Therefore, in-depth understanding of the population behaviour is essential in order to create successful policies against obesity prevalence. To this end, the BigO system facilitates the collection, processing and modelling of behavioural data at population level to provide evidence for effective policy and interventions design. In this paper, we introduce the behaviour profiles mechanism of BigO that produces comprehensive models for the behavioural patterns of individuals, while maintaining high levels of privacy protection. We give examples for the proposed mechanism from real world data and we discuss usages for supporting various types of evidence-based policy design.

Keywords: Health Informatics - Behavioral health informatics, Health Informatics - Pervasive health, Health Informatics - Preventive health
Abstract: Despite the many mHealth solutions available, it remains unclear what their success factors are. Specifically, there has been controversy on the effectiveness of extrinsic rewards. This study evaluates two design elements of an mHealth solution -- i.e. social proof and tangible rewards -- and their impact on user engagement. During a four-week campaign, a sample of 143 university staff members engaged in a health promotion campaign. Participants were randomly distributed over one of three treatment groups. It was found that the introduction of a sufficiently meaningful, unexpected, and customized extrinsic reward can engage participants significantly more in a health promotion context.

Keywords: General and theoretical informatics - Algorithms, General and theoretical informatics - Machine learning, Health Informatics - Behavioral health informatics
Abstract: Mental health is a growing concern and its problems range from inability to cope with day-to-day stress to severe conditions like depression. Ability to detect these symptoms heavily relies on accurate measurements of emotion and its components, such as emotional valence comprising of positive, negative and neutral affect. Speech as a bio-signal to measure valence is interesting because of the ubiquity of smartphones that can easily record and process speech signals. Speech-based emotion detection uses a broad spectrum of features derived from audio samples including pitch, energy, Mel Frequency Cepstral Coefficients (MFCCs), Linear Predictive Cepstral Coefficients, Log frequency power coefficients, spectrograms and so on. Despite the array of features and classifiers, detecting valence from speech alone remains a challenge. Further, the algorithms for extracting some of these features are compute-intensive. This becomes a problem particularly in smartphone applications where the algorithms have to be executed on the device itself. We propose a novel time-domain feature that not only improves the valence detection accuracy, but also saves 10% of the computational cost of extraction as compared to that of MFCCs. A Random Forest Regressor operating on the proposed feature-set detects speaker-independent valence on a non-acted database with 70% accuracy. The algorithm also achieves 100% accuracy when tested with the acted speech database, Emo-DB.

Keywords: General and theoretical informatics - Algorithms, General and theoretical informatics - Artificial Intelligence, General and theoretical informatics - Machine learning
Abstract: The existing adaptive basal-bolus advisor (ABBA) was further developed to benefit patients under insulin therapy with multiple daily injections (MDI). Three different in silico experiments were conducted with the DMMS.R simulator to validate the approach of combined use of self-monitoring of blood glucose (SMBG) and insulin injection devices, e.g. insulin pen, as are used by the majority of type 1 diabetes patients under insulin therapy. The proposed approach outperforms the conventional method, as it increases the time spent within the target range and simultaneously reduces the risks of hyperglycaemic and hypoglycaemic events.

Keywords: Models of medical devices, Model building - Algorithms and techniques for systems modeling
Abstract: Abstract — Recently, skull-remodeling surgery was proposed as a means of enhancing the dose of tumor treating fields in glioblastoma treatment. This abstract describes the basic principles of the finite element methods used to plan the surgery and evaluate the treatment efficacy.

Keywords: Models of medical devices
Abstract: Cloud-based platforms have the potential to define a new collaborative framework in healthcare, engaging research centers to safely commercialize their IP (clinical data, virtual patients, model templates and simulation tools) and to facilitate the creation of innovative products during the pre-clinical and clinical trial phases of medical device development. This presentation will review the benefits of cloud computing for healthcare, using the InSilicoMRI application as an example. This application automates the set-up and solution of RF heating analysis in a manner that is intended to conform to guidelines and existing standards for in-vitro testing.

Keywords: Models of medical devices, Data-driven modeling, Translational biomedical informatics - Mining clinical data
Abstract: A simulation-based study investigating the connection between Tumor Treating Fields (TTFields) distribution in the brain and survival in a cohort of 340 newly diagnosed glioblastoma patients is presented. The analysis led to a definition for TTFields dose density with higher values of average dose density at the tumor bed associated with improved patient outcome. This study sets the foundations for a framework for TTFields dosimetry and treatment planning, ultimately expected to improve patient outcome.

Keywords: Data-driven modeling, Models of medical devices
Abstract: Among the potential hazards of MRI for patients with active implantable medical device is RF-induced unintended cardiac stimulation (UCS). RF energy incident on the device may be rectified by internal active components. In order to assess the risk to the patient, device manufacturers use computational human models to quantify the incident RF on the device and perform bench top testing to determine the likelihood of UCS.

Keywords: Systems modeling - Decision making, Model building - Algorithms and techniques for systems modeling
Abstract: Recently, a new numerical algorithm has been developed that couples the fast multipole method with the boundary element method. This new methodology has demonstrated a dramatic increase in simulation speeds and accuracies for computational electromagnetics problems. The increase in speed enables medical device developers and engineers the opportunity to simulate a significantly greater number of test cases, leading to better predictions of device safety and efficacy. This work demonstrates the method for these purposes, focusing on applications related to non-invasive brain stimulation. Relevant field and energy absorption results will be presented and compared with existing methods.

Keywords: Models of medical devices
Abstract: We evaluated the influence of 3T MRI whole-body birdcage coil modeling on the estimation of the electromagnetic exposure in a computational human body model. Our results show that simplifications of the coil geometry, realistic tuning, and capacitor losses had a significant impact on the estimated electromagnetic field distribution. Therefore, reliable MRI RF safety assessment requires to carefully consider variations in the generated electric field and diversity of the ratio of B1+ at the coil iso-center to the whole-body average SAR.

Keywords: Clinical engineering, Computer model-based assessments for regulatory submissions, Health technology management and assessment
Abstract: Developing techniques for automatic detections of ECG features can assist in accurate diagnosis of heart diseases. To this date, extensive research has been conducted on diagnosing heart diseases by investigating changes related to R peaks. Nevertheless, little research has been devoted to detecting end of T-wave which is associated with several heart disorders. This paper discusses a technique based on recurrence plot for the detection of end of T wave in fetal ECG signals.

Keywords: Cardiovascular assessment and diagnostic technologies
Abstract: Early diagnosis of the cardiac abnormalities during the pregnancy may reduce the risk of perinatal morbidity and mortality. DUS, which is commonly used for monitoring the fetal heart rate, can also be used for identifying the event timings of fetal cardiac valve motions. In this study, a cepstral domain signal analysis technique is proposed to analyze the fetal cardiac Doppler ultrasound signals for the fetal cardiac timing events estimation. The results show that the proposed cepstrum method can achieve a promising fetal cardiac timing events’ estimation accuracy and hence more robustness. Therefore, this technique would be useful for reliable screening of fetal wellbeing.

Keywords: Cardiovascular assessment and diagnostic technologies, Diagnostic devices - Physiological monitoring
Abstract: Abstract The aim of this review is to describe current techniques for cardiac monitoring of the fetus. Methods: We reviewed the literature for recent advances in evaluating the fetal heart during pregnancy. Results: Several methodologies exist for fetal cardiac monitoring, including cardiotocography, fetal ECG by blind source separation or adaptive interference cancellation and magnetocardiography. Each has distinct advantages and limitations. Conclusions: While no readily clinically available tools exist for fetal cardiac monitoring from 16-40 weeks gestation, there are several promising strategies for future research.

Keywords: Cardiac signal remote monitoring devices and technologies
Abstract: Fetal auscultation is a low-cost and noninvasive method as it captures the acoustic signals of fetal heart sounds from the mothers’ abdominal surface. However, fetal auscultation confronts many challenges because the signals are contaminated with noise from various sources making the research about fetal phonocardiogram signals a difficult study area demanding robust analysis techniques for results. In the last few years, Deep Learning (DL) analysis has been proven to present good results in the medical field by creating a new area of research. This study employs an existing DL library and tests its performance in the detection of fetal heart sounds in phonocardiogram signals leading the way for the creation of a tool able to provide the physician with information about the condition of the fetus.

Keywords: Sleep - Obstructive sleep apnea
Abstract: Tracheal sound (TS) analysis is a simple way to study the abnormalities of upper airway like airway obstruction. We present in this paper a summary of results from various clinical validation studies underlining the usefulness of TS in obstructive sleep apnea (OSA) diagnosis.

Keywords: Cardiovascular, respiratory, and sleep devices - Monitors, Cardiovascular, respiratory, and sleep devices - Smart systems, Cardiovascular, respiratory, and sleep devices - Wearables
Abstract: Sleepiness is a physiological condition that is characterized by an uncontrollable desire to sleep and by impairments of performance which can lead to poor quality of life and disastrous accidents. Therefore, detecting sleepiness is a critical issue of public health and safety. Since the use of parameters related to eyelids and pupil (i.e. ocular parameters) seems to be the most suitable technique for detecting sleepiness objectively, automatically, and in real-time, we have developed and validated a new sleepiness detection system based on the analysis of infrared (IR) eye images: the Drowsimeter. We conducted a study to show that there is a good correlation between the level of sleepiness determined automatically by the Drowsimeter and several references (i.e. ground truths).

Keywords: Cardiovascular, respiratory, and sleep devices - Diagnostics, Cardiovascular, respiratory, and sleep devices - Sensors, Cardiovascular, respiratory, and sleep devices - Smart systems
Abstract: Traditional techniques for sleep recording are limited for use in sleep labs with medical personnel assistance. We present Somnograph, a new self-administered sleep recording system for easy use in home environment.

Keywords: Sleep - Obstructive sleep apnea, Sleep - Sleep apnea therapy, Sleep - Periodic breathing & central apnea
Abstract: The aim of this prospective study was to investigate whether a contactless measurement by a 3-D-camera can detect sleep apnea and periodic leg movements. We also explored whether machine learning algorithm will be able to distinguish between waking and sleep time.

Keywords: Vascular mechanics and hemodynamics - Pulmonary Circulation, Vascular mechanics and hemodynamics - Vascular Hemodynamics, Cardiac mechanics, structure & function - Cardiac structure from imaging
Abstract: Pulmonary venous hypertension (PVH) is a life-threatening disease with a complex etiology characterized by elevated hemodynamic pressure in the pulmonary arteries (PA) of the right heart but as a result of left heart dysfunction. However, the process at which a left heart dysfunction alters PA hemodynamics leading to right heart dysfunction remains unclear. Here, we studied 16 PVH patients using 4D Flow MR images. We computed total 3D vorticity in the right heart’s left (LPA) and right pulmonary arteries (RPA) separately and assessed 1) the association of LPA and RPA volumetric vorticity to bi-ventricular (i.e., left (LV) and right ventricular (RV)) function.

Keywords: Cardiac mechanics, structure & function - Heart failure, Cardiac mechanics, structure & function - Ventricular mechanics
Abstract: Using 4D flow magnetic resonance imaging (MRI) and flow modelling using the Navier-Stokes equations, it is possible to compute the hemodynamic forces exchanged between the blood and myocardium. Hemodynamic forces show potential as a new measure of pathophysiology, e.g. in heart failure and congenital heart disease. This talk will describe 4D flow and modelling methods used and summarize the current knowledge from studies in healthy controls, elite athletes and patients.

Keywords: Vascular mechanics and hemodynamics - Vascular Hemodynamics, Vascular mechanics and hemodynamics - Vascular Disease
Abstract: Wall shear stress (WSS) is the shear force of blood flow on the endothelial cells and is associated with remodeling of the vessel wall. Here, we revisit the clinical utility of advanced WSS mapping in aortas with bicuspid aortic valves.

Keywords: Modeling and analysis, Sensor systems and Instrumentation, Wearable wireless sensors, motes and systems
Abstract: Gait is an extraordinary complex function of human body that involves the activation of entire visceral nervous system, making human gait definite to various functional abnormalities. Diagnosis and treatment of such disorders prior to their development can be achieved through integration of modern technologies with state-of-the-art developed methods. Modern machine learning techniques have outperformed and complemented the use of conventional statistical methods in bio-medical systems. In this research a wearable sensor system is presented, which combines plantar pressure measurement unit and Inertial Measurement Units (IMU’s) integrated with a stacked Long short-term memory (LSTM) model to detect human gait abnormalities that are prone to the risk of fall. The computed metrics and gait parameters show significant differences between normal and abnormal gait patterns. Three specific abnormalities involving Hemiplegic, Parkinsonian and Sensory-Ataxic gaits are simulated to validate the proposed model and show promising results. The proposed research aims to demonstrate how advanced technologies can be used in gait diagnosis and treatment assistant systems.

Keywords: Modeling and analysis, Sensor systems and Instrumentation, Bio-electric sensors - Sensor systems
Abstract: This study aims at estimating a virtual surface Electromyography (sEMG) channel through a Recurrent Neural Network (RNN) by using Long Short-Term Memory (LSTM) nodes. The virtual channel is used to classify hand postures from the publicly NinaPro database with a multi-class, one-against-all Support Vector Machine (SVM) using the Root Mean Square RMS of the sEMG signal as feature. The classification of the signals through the virtual channel was compared with uncontaminated data and data contaminated with noise saturation. The hit rate from the clean data has averaged 73.96% ± 3.02%. The classification from the contaminated data of one of the channels has improved from 9.29% ± 4.42% to 66.48% ± 6.11% with the virtual channel.

Keywords: Physiological monitoring - Modeling and analysis
Abstract: Human gait has been regarded as a useful behavioral biometric trait for personal identification and authentication. This study aimed to propose an effective approach for classifying gait, measured using wearable inertial sensors, based on neural networks. The 3-axis accelerometer and 3-axis gyroscope data were acquired at the posterior pelvis, both thighs, both shanks, and both feet while 29 semi-professional athletes, 19 participants with normal foot, and 21 patients with foot deformities walked on the 20-meter straight path. The classifier based on the gait parameters and fully connected neural network was developed by applying 4-fold cross-validation to 80% of the total samples. For the test set that consisted of the remaining 20% of the total samples, this classifier showed an accuracy of 93.02% in categorizing the athlete, normal foot, and deformed foot groups. Using the same model validation and evaluation method, up to 98.19% accuracy was achieved from the convolutional neural network-based classifier. This classifier was trained with the gait spectrograms obtained from the time-frequency domain analysis of the raw acceleration and angular velocity data. The classification based only on the pelvic spectrograms exhibited an accuracy of 94.25% even without requiring a time-consuming and resource-intensive process for feature engineering. The notable performance and practicality in gait classification achieved by this study suggest potential applicability of the proposed approaches in the field of biometrics.

Keywords: Sensor systems and Instrumentation
Abstract: To enable on-time and high-fidelity lower-limb exoskeleton control, it is effective to predict the future human motion from the observed status. In this research, we propose a novel method to predict future plantar force during the gait using IMU and plantar sensors. Deep neural networks (DNN) are used to learn the non-linear relationship between the measured sensor data and the future plantar force data. Using the trained network, we can predict the plantar force not only during walking but also at the start and end of walking. In the experiments, the performance of the proposed method is confirmed for different prediction time.

Keywords: Physiological monitoring - Modeling and analysis, Physiological monitoring - Instrumentation
Abstract: Heart rate (HR) estimation using wearable reflectance-type photoplethysmographic (PPG) signals is challenging due to low signal-to-noise ratio (SNR). Especially during intensive exercise, motion artifacts (MAs) overwhelm PPG signals in an unpredictable way. To overcome the issue, an acceleration signal as a reference signal has been adopted by simultaneously measuring with PPG signal. However, MAs are frequently uncorrelated with accelerometer signals under various activities. In this paper, we present a learning-based framework for HR estimation. The proposed framework is based on the deep neural network (DNN). For the feasibility study, we presented a simple network with two fully connected layers. We first extracted power spectra from the measured PPG signal and the acceleration signal. The two power spectra were then used for the input layer in the network. In addition, to inform the PPG signal quality, we added the acceleration signal intensity for the input layer. The proposed simple DNN network was trained for 10 epochs in IEEE Signal Processing Cup 2015 (ISPC) dataset (n=23). Subsequently, the trained network provided low mean absolute error (MAE) of 2.31 bpm in the ISPC dataset. We further tested the network on the new BAMI dataset (n=4), and found that it provided 4.72 bpm of MAE. On the other hand, the MAE without the learning frame was 15.73 bpm. In this study, we found that the simple DNN technique is effective. More training issues were also discussed.

Keywords: Wearable sensor systems - User centered design and applications, Wearable wireless sensors, motes and systems, Novel methods
Abstract: The Eyelid Drive System (EDS) is an assistive technology device intended to allow users to wirelessly control other devices, such as power wheelchairs and personal computers, using commands consisting only of blinking and winking. In this paper, four machine learning classifiers are trained on data taken from one subject and validated offline on the training subject plus two additional subjects. The classifiers are assessed for accuracy, computational and memory requirements, and transferability from the “training” subject to the other two subjects. A support vector machine (SVM) achieved the highest level of accuracy (97.5%) while using a potentially prohibitive level of computational and memory resources. A logistic regression classifier also achieved excellent accuracy (96.5%) while using two to three orders of magnitude fewer computational and memory resources than the SVM.

Keywords: Neural networks and support vector machines in biosignal processing and classification, Data mining and processing - Pattern recognition, Physiological systems modeling - Signal processing in physiological systems
Abstract: Automatic sleep staging provides a cheaper, faster and more accessible alternative for evaluating sleep patterns and quality compared with manual hypnogram scoring performed by a clinician. Traditionally, classification methods treat sleep stages independently of their temporal order, despite sleep patterns themselves being highly sequential. Such independent sleep stage classification can result in poor sensitivity and precision, in particular when attempting to classify the sleep stage N1, otherwise known as the transition stage of sleep which links periods of wakefulness to periods of deep sleep. To this end, we propose a novel transition sleep classification method which aims to improve classification accuracy. This is achieved by utilising both the temporal information of previous stages and treating the transitions between stages as classes in their own right. Simulations on publicly available polysomnography (PSG) data and a comprehensive performance comparison with standard classifiers demonstrate a marked improvement achieved by the proposed method in both N1 sensitivity and precision across all considered classifiers. This includes an increase in N1 precision from 0.01% to 36.75% in an MLP classifier, and an increase in both accuracy and Cohen’s kappa value in two of the three classifiers. Overall best mean performance is obtained by transition classification with a random forest classifier (RF) which achieved a kappa value of κ = 0.75 (substantial agreement), and an N1 stage precision of 58%.

Keywords: Time-frequency and time-scale analysis - Time-frequency analysis, Physiological systems modeling - Signal processing in physiological systems
Abstract: It is well known that the coordination among several subsystems in the newborns is effectively changing as a function of behavioral states. For this reason, sleep state characterization is an essential procedure in neonatal monitoring. Despite its importance, methodologies assessing sleep states are discrete in time and usually based on visual inspection. In this work, we validate a point process framework on a population of 113 full-term infants to the aim of providing a continuous in time sleep state characterization. After determining the best suitable probability density distribution to fit the neonatal RR series, we compare traditional heart rate variability (HRV) parameters with respective point process-extracted sets of time and frequency domain instantaneous measures in order to validate the proposed framework. Our results provide relevant insights into the point process ability to capture HRV dynamics within a high degree of reliability, thus providing evidence that our framework might be employed for an instantaneous estimate of behavioral states.

Keywords: Neural networks and support vector machines in biosignal processing and classification, Time-frequency and time-scale analysis - Wavelets, Signal pattern classification
Abstract: Elderly and patients with neurodegenerative diseases (NDD) often complain about sleep problems and show altered sleep structure. Automated algorithms for efficient and specific sleep staging are needed. We propose a new algorithm using only one electroencephalographic and two electrooculographic channels to score wakefulness, rapid eye movement (REM) sleep and non-REM sleep in a cohort of elderly healthy controls (HC), patients with Parkinson’s disease (PD), isolated REM sleep behavior disorder (iRBD), considered the prodromal stage of PD, and patients with PD and RBD (PD+RBD). The proposed method scores both standard 30-s epochs (macro-staging) and 5-s mini-epochs (micro-staging), whose evaluation may help to better understand sleep micro-structure. Moreover, the algorithm is interactive, as it labels the classified sleep epochs as either certain or uncertain, so that experts can manually review the uncertain ones. The algorithm performances were evaluated for macro-sleep staging, where it achieved overall accuracies of 0.87±0.05 in 41 HC, 0.86±0.10 in 57 PD, 0.76±0.10 in 31 iRBD and 0.77±0.10 in 30 PD+RBD patients when all 30-s epochs were considered. The accuracies increased to 0.91±0.05, 0.90±0.08, 0.85±0.09, 0.88±0.08 respectively when considering only the certain ones. The epochs labeled as uncertain were 9.95±4.15%, 11.13±7.86%, 18.39±7.38% and 18.90±8.00% in HC, PD, iRBD and PD+RBD respectively. The proposed interactive micro and macro sleep staging algorithm can be used in clinics to reduce the burden of manual sleep staging in elderly and patients with NDD.

Keywords: Neural networks and support vector machines in biosignal processing and classification, Signal pattern classification
Abstract: OBJECTIVE The distinction of snoring and loud breathing is often subjective and lies in the ear of the beholder. The aim of this study is to identify and assess acoustic features with a high suitability to distinguish these two classes of sound, in order to facilitate an objective definition of snoring based on acoustic parameters. METHODS A corpus of snore and breath sounds from 23 subjects has been used that were classified by 25 human raters. Using the openSMILE feature extractor, 6 373 acoustic features have been evaluated for their selectivity comparing SVM classification, logistic regression, and the recall of each single feature. RESULTS Most selective single features were several statistical functionals of the first and second mel frequency spectrumgenerated perceptual linear predictive (PLP) cepstral coefficient with an unweighted average recall (UAR) of up to 93.8%. The best performing feature sets were low level descriptors (LLDs), derivatives and statistical functionals based on fast Fourier transformation (FFT), with a UAR of 93.0%, and on the summed mel frequency spectrum-generated PLP cepstral coefficients, with a UAR of 92.2% using SVM classification. Compared to SVM classification, logistic regression did not show considerable differences in classification performance. CONCLUSION It could be shown that snoring and loud breathing can be distinguished by robust acoustic features. The findings might serve as a guidance to find a consensus for an objective definition of snoring compared to loud breathing.

Keywords: Physiological systems modeling - Signal processing in physiological systems, Data mining and processing in biosignals, Time-frequency and time-scale analysis - Wavelets
Abstract: Sleep arousal is generally defined as an abrupt shift in EEG frequency with a duration of 3-16 seconds. Arousal from sleep expected to cause sudden changes in the cardiovascular system that can manifest as cardiac responses. In this paper, our objective was to investigate how cardiac characteristics change due to arousal. We focused on the QT interval fluctuations in ECG during the occurrence of arousals. We analysed 7373 sleep arousals collected from 50 males that were older than 65 years. We analysed the ECG signal 5 seconds prior to and 10 seconds after each arousal onset (Pre and Post-Onset). Q and T waves were detected for all Pre and Post-Onset windows to estimate their time intervals. To find out whether the QT interval, a marker of ventricular activation, is modulated by arousal onset, we have applied graphical and statistical analysis. Our observations indicate that in 47 out of 50 subjects (94%), the average QT interval of all arousal significantly shortened at arousal onset. We observed similar outcomes for different types of arousals, indicating that the shortening in average QT interval is independent of the type of arousal. We also studied the relative QT interval change during arousal. The distribution of relative QT interval changes demonstrates that around 60% of arousals increase or decrease QT interval by a maximum of 20%. The probability of QT time interval shortening was twice that of QT interval lengthening.

Keywords: Signal pattern classification, Time-frequency and time-scale analysis - Time-frequency analysis
Abstract: This paper describes an automatic classification algorithm for nocturnal hypoxemia in patients receiving home oxygen therapy (HOT). Nocturnal hypoxemia is a well-known complication in patients with chronic respiratory disease, and the number of patients receiving HOT has increased in recent years. Many studies have reported that 40% of patients receiving HOT have sleep-related oxygen desaturation. To deal with this situation, a nocturnal pulse oximetry is used to measure oxygen saturation (SpO2) and control the flow rate of highly concentrated oxygen. However, in some cases, the flow rate is not controlled properly and the same flow rate is adopted both during the day and night. There are several types of nocturnal hypoxemia, and it is difficult to classify these types only according to a subjective assessment of a medical doctor. Furthermore, it is difficult to continuously monitor the measurement results of pulse oximetry, although a flexible treatment depending on the state of hypoxemia is desired. To overcome these difficulties, an automatic classification method for SpO2 measured by the nocturnal pulse oximetry is proposed in this paper. The proposed method uses the time domain waveform and the frequency characteristics of SpO2. The classification performance of the method is evaluated by using 48 measured SpO2 values from patients receiving the HOT. The classification results are validated with decisions of ten chest physicians.

Keywords: Image analysis and classification - Machine learning / Deep learning approaches, Multiscale image analysis, Functional image analysis
Abstract: Cleft lip and palate is the most common congenital malformation in oral and maxillofacial region. As a kind of facial plastic surgery, the most important factor for the success of cleft lip and palate repair surgery is the design of surgical markers and incisions. However, general hospitals especially in rural areas lack dependable medical resources, which makes the effect of the surgery hard to guarantee. To solve this problem, we propose a novel robotic surgery assistant technology based on deep learning to help reduce the technical threshold and improve the overall effect of cleft lip and palate repair surgery. For the first time, a robust dataset of cleft lip and palate cases is established, which can be used to train the model to locate surgical markers and incisions. Secondly, we build a strong baseline on this dataset by using state-of-the-art Hourglass architecture and residual learning, with two neoteric block designs, one of which enables stronger capability of generalization, while the other greatly reduces the complexity of the model, thus making efficient application possible. Finally, by comparing with other facial feature extraction methods, our models achieve the best results on multiple metrics, showing their strong superiority and adaptability on this task.

Keywords: Image analysis and classification - Machine learning / Deep learning approaches, Image feature extraction
Abstract: Analysis of laboratory animal behavior allows assessment of animal wellbeing. We present a method for the classification of different activities of laboratory mice by analyzing video clips using three deep learning methods. Animals placed in observation cages are filmed and short video clips are labelled as belonging to one of five defined behaviors. Subsequently, three different methods based on convolutional neural networks (CNNS) are applied to classify the clips. The best performing method - a two-stream network that analyzes individual frames as well as the video's optical flow - achieves an accuracy of 86.4%, including detection of important behavioral patterns such as self-grooming. These results show that the presented analysis protocol allows automated assessment of animal behavior by algorithmic analysis of videos of mice on observation boxes.

Keywords: Image analysis and classification - Machine learning / Deep learning approaches, Functional image analysis, Image feature extraction
Abstract: We present the use of an error correcting autoencoder stage to a convolutional neural network model as a means of improving image based automatic Positive Airway Pressure (PAP) mask sizing accuracy. A single convolutional layer neural network was pre-trained using MUCT dataset and transfer learning was applied to mitigate against the relatively small custom dataset. The base model was then augmented with an additional error correcting autoencoder and trained against the custom dataset. The presented model increased PAP sizing accuracy against the baseline by 15.3% while reducing overfitting.

Keywords: Image analysis and classification - Machine learning / Deep learning approaches, Image classification, Image feature extraction
Abstract: In this paper, we propose a clustering method with temporal ordering information for endoscopic image sequences. It is difficult to collect a sufficient amount of endoscopic image datasets to train machine learning techniques by manual labeling. The clustering of endoscopic images leads to group-based labeling, which is useful for reducing the cost of dataset construction. Therefore, in this paper, we propose a clustering method where the property of endoscopic image sequences is fully utilized. For the proposed method, a deep neural network was used to extract features from endoscopic images, and clustering with temporal ordering information was solved by dynamic programming. In the experiments, we clustered the esophagogastroduodenoscopy images. From the results, we confirmed that the performance was improved by using the sequential property.

Keywords: Image analysis and classification - Machine learning / Deep learning approaches
Abstract: Soft-tissue Sarcomas (STS) are a heterogeneous group of malignant neoplasms with a relatively high mortality rate from distant metastases. Early prediction or quantitative evaluation of distant metastases risk for patients with STS is an important step which can provide better personalized treatments and thereby improve survival rates. Positron emission tomography – computed tomography (PET-CT) image is considered as the imaging modality of choice for the evaluation, staging and assessment of STS. Radiomics, which refers to the extraction and analysis of the quantitative of high-dimensional mineable data from medical images, is foreseen as an important prognostic tool for cancer risk assessment. However, conventional radiomics methods that depend heavily on hand-crafted features (e.g. shape and texture) and prior knowledge (e.g. tuning of many parameters) and therefore cannot fully represent the semantic information of the image. In addition, convolutional neural networks (CNN) based radiomics methods present capabilities to improve, but currently they are mainly designed for single modality e.g., CT or a particular body region e.g., lung structure. In this work, we propose a deep multi-modality collaborative learning to iteratively derive optimal ensembled deep and conventional features from PET-CT images. In addition, we introduce an end-to-end volumetric deep learning architecture to learn complementary PET-CT features optimised for image radiomics. Our experimental results using public PET-CT dataset of STS patients demonstrate that our method has better performance when compared with the state-of-the-art methods.

Keywords: Image analysis and classification - Machine learning / Deep learning approaches, X-ray radiography
Abstract: Respiratory diseases account for a significant proportion of deaths and disabilities across the world. Chest X-ray (CXR) analysis remains a common diagnostic imaging modality for confirming intra-thoracic cardiopulmonary abnormalities. However, there remains an acute shortage of expert radiologists, particularly in under-resourced settings, resulting in severe interpretation delays. These issues can be mitigated by a computer-aided diagnostic (CADx) system to supplement decision-making and improve throughput while preserving and possibly improving the standard-of-care. Systems reported in the literature or popular media use handcrafted features and/or data-driven algorithms like deep learning (DL) to learn underlying data distributions. The remarkable success of convolutional neural networks (CNN) toward image recognition tasks has made them a promising choice for automated medical image analyses. However, CNNs suffer from high variance and may overfit due to their sensitivity to training data fluctuations. Ensemble learning helps to reduce this variance by combining predictions of multiple learning algorithms to construct complex, non-linear functions and improve robustness and generalization. This study aims to construct and assess the performance of an ensemble of machine learning (ML) models applied to the challenge of classifying normal and abnormal CXRs and significantly reducing the diagnostic load of radiologists and primary-care physicians.

Keywords: Micro-and nano-biorobotics, Biologically inspired robotics and micro-biorobotics - Modeling, Neural interfaces for robotic prosthetics
Abstract: Conducting polymer (CP) actuators are promising devices for biomedical applications such as artificial muscles and drug delivery systems. Here, we report a tri-layer actuator based on poly(pyrrole) (PPy) microtubes (PPy MTs) doped with poly(sodium-p-styrenesulfonate) (PSS) and constructed on a passive layer of gold-coated poly-propylene (PP) film. The PPy MTs were fabricated using electrochemical deposition of PPy around poly(lactic-co-glycolic acid) (PLGA) fiber templates, followed by template removal. The PPy MTs were subjected to a redox process using cyclic voltammetry in 0.1 M NaPSS electrolyte solution as the potential was swept between –0.8 V and +0.4 V for 5 cycles at the scan rates of 10, 50, 100, and 200 mV/s. The bending behavior of the PPy MTs actuator was investigated by measuring the deflection of actuator tip resulting from the expansion/contraction strain of PPy MTs. The PPy MTs actuator showed a reversible bending movement during each potential cycle. The maximum deflection of actuator decreased by increasing the scan rate that was confirmed by calculating the actuation strain generated during each cycle at various scan rates.

Keywords: Clinical robots, New technologies and methodologies in medical robotics, Mechanics of locomotion and balance
Abstract: Wireless capsule endoscopes provide a painless and non-invasive alternative to the flexible endoscope in various applications of the gastrointestinal tract diagnosis. Operating a wireless capsule endoscope in the colon may benefit from an active position control as the large colon diameter can lead to uncontrollable and unpredictable capsule trajectory. Robot assisted magnetic steering is an attractive technique that is being explored by researchers worldwide. This paper presents the implications of a novel capsule geometry to markedly improve capsule stabilization and locomotion compared to the cylinder-based capsule geometry that is commonly used.

Keywords: Biologically inspired robotics and micro-biorobotics - Biologically inspired locomotion, Biomimetic robotics, New technologies and methodologies in biomechanics
Abstract: This paper introduces a new approach to characterize the locomotion of self-assembling modular systems. By establishing the necessary components for a locomotive modules and a hierarchy of locomotive systems, further research can be done into each category. The systems are broken into three primary groups: limbed, limbless, and rolling. The majority of the variations exist within the legged domain, and hence the largest area of study rests there. This study contributes to the literature inasmuch as creating a hierarchical framework that can be used for the creation of algorithms for self-assembling micro and nano robots for drug delivery and surgery.

Keywords: New technologies and methodologies in medical robotics, Micro-and nano-biorobotics
Abstract: In this paper, a bi-directional linear capsule robot (capsulbot) for potential applications in Gastrointestinal (GI) tract inside human body is studied. Compared with the conventional endoscope limited by its poor locomotion and steering capabilities, active locomotion actuator will play an important role in the diagnosis of narrow organ tract of the human body in the future. This paper studies a new simple-structured actuator that can realize bi-directional linear motion by properly controlling the supplied current profile. It is demonstrated that the linear motion of the new capsule is affected by three main factors: current waveform, current duty ratio, and current amplitude. The optimized current profile can maximize the capsulbot displacement and the performance is verified experimentally on a prototype capsulbot.

Keywords: Micro-and nano-biorobotics, New technologies and methodologies in medical robotics, New technologies and methodologies in Milli, micro and nanorobots
Abstract: This study introduces a miniaturized capsule endoscope equipped with a marking module for intestinal tumor or lesion localization. The design concept is based on an active wireless capsule endoscope platform that is manipulated by an external electromagnetic actuation (EMA) system. The magnetic response of a permanent magnet inside the capsule is designed to have flexible movement in viscous environment of bowel. This magnet is also utilized to activate tattooing process by triggering a gas-generated chemical reaction. Once approaching to a target region, gradient magnetic field from EMA system is induced to push magnet down, releasing water to dry chemical powder mixture. Then the gas pressure increases and pushes the piston move to inject ink into target point. During traveling in digestive organs, injection needle is stowed inside the capsule to avoid damage to the organs. The whole procedure is manipulated by EMA system, the injection consumes no internal battery and is observable through capsule’s camera which provides clinician vision. Basic tests were conducted to evaluate the performance of proposed robotic capsule. The success of creating a black visible bled from serosa of intestine proves the feasibility and potential of the design. This study could be an alternative for traditional tattooing endoscopy and motivate other research groups for further development of functional wireless capsule endoscope.

Keywords: New technologies and methodologies in surgical planning, New technologies and methodologies in medical robotics, New technologies and methodologies in biomechanics
Abstract: Organ models are important for medical training and surgical planning. With the fast development of additive fabrication technologies, including 3D printing, the fabrication of 3D organ phantoms with precise anatomical features becomes possible. Here, we develop the first high-resolution kidney phantom based on soft material assembly, by combining 3D printing and polymer molding techniques. The phantom exhibits both the detailed anatomy of a human kidney and the elasticity of soft tissues. The phantom assembly can be separated into two parts on the coronal plane, thus large renal calculi are readily placed at any desired location of the calyx. With our sealing method, the assembled phantom withstands a hydraulic pressure that is four times the normal intrarenal pressure, thus it allows the simulation of medical procedures under realistic pressure conditions. The medical diagnostics of the renal calculi is performed by multiple imaging modalities, including X-ray, ultrasound imaging and endoscopy. The endoscopic lithotripsy is also successfully performed on the phantom. The use of a multifunctional soft phantom assembly thus shows great promise for the simulation of minimally invasive medical procedures under realistic conditions.

Keywords: Point of care - Respiratory monitoring, Point of care - Heart rate monitoring, Point of care - Home-based applications
Abstract: Non-contact vital sign monitoring of multiple people can be realized using a radar sensor that is able to provide a high resolution two-dimensional image of the monitored area. This is particularly user-friendly, since no electrodes have to be attached to the body, which is important especially in view of applications like monitoring neonates or burn victims. There are many further applications of remote cardiopulmonary monitoring in the area of home health care, driver monitoring, sleep monitoring, or even monitoring of arrested people in a prison cell. The radar is able to measure the tiny movements of the chest, caused by respiration and heartbeat. Due to the superposition of both movements a sophisticated processing of the measured data is necessary to extract the heartbeat and respiration signal from the measured overall signal. This paper presents radar measurements at 24 and 77 GHz, where vital signs of multiple people have been simultaneously monitored and proposes a signal processing method to separate heartbeat and respiration in the measured data.

Keywords: Point of care - Detection and monitoring, Point of care - Evaluation and validation, Point of care - Clinical use and acceptance
Abstract: This work aims to present a quantitative metric to assess the impact of feeding teats on the nutritive sucking of newborns. Two different teat models are compared: a classical model (model C), and a model provided with two opposite recesses to match the anatomical characteristics of the mouth of a newborn (model I). This latter feeding teat model has been specifically designed to promote the attachment of the baby, thus improving her/his nutritive sucking performance. Feeding teats are instrumented with a device to assess nutritive sucking (the Feeding Assessment Monitor, FAM). The device records feeding pressures and a software extracts quantitative features already used and validated in clinical applications. Comparative cross-over analysis on 30 healthy newborns, demonstrates the appropriateness of the proposed metric to reveal differences in the teat models. In particular, our data confirm the better attachment of newborns when fed with the I model: they show a longer feeding, with higher level of depressurization, higher regularity, and higher number of sucking events.

Keywords: Global healthcare challenges, Medical technology - Design and development, Medical technology - Safety
Abstract: In the complex context of the medical device industry and healthcare systems, the reduction in cost may increase access to medical technologies moving towards global health equity. This paper is focused on the description of UBORA, an e-infrastructure based on a new concept of biomedical engineering which promotes the open-source approach for co-designing medical devices, fostering innovative ideas, needs-based, low-cost and safe technology. UBORA structures the entire design process using EN ISO 13485:2016, standard related to medical technology for inspiration. As a proof of concept, this paper shows an example of the development of an open source medical device for hand rehabilitation, designed using UBORA. We demonstrate the straightforward pathway to gather information on safety requirements. Finally, we describe a usability test of the e-infrastructure performed during the 4th WHO Global Forum on Medical Devices in India.

Keywords: Global healthcare challenges, Personalized medicine, Medical technology - Innovation
Abstract: Ocular prostheses are part of a substantial global market for ocular implants. However, demand for custom ocular prostheses (COPs) can outpace supply at clinics, particularly in emerging markets such as India, due to the slow pace of prosthesis manufacturing and limited supply of ocularists. Existing manufacturing methods for COPs involve multiple stages of casting and molds with limited quality control, resulting in time-intensive trial and error with patients to achieve a comfortable fit. Through collaboration with the LV Prasad Eye Institute (LVPEI) in India, the authors improved manufacturing process efficiency for COPs and without significantly increasing cost or decreasing customizability. A time study of the current process showed that no single step was a dominant contributor to process time, necessitating a holistic change to the manufacturing process. The modified process uses dip coating of a 3D printed internal body made from a scanned impression. Based on a timed experimental trial, the modified process has a projected daily COP production rate increase of 100% compared to the existing process. A study of produced COP quality showed accumulated error of critical dimensions within reasonable limits, with the greatest error being less than 65% of maximum acceptable error.

Keywords: Point of care - Evaluation and validation, Point of care - Home-based applications, Point of care - Detection and monitoring
Abstract: Balance quality measurement is a key element in the evaluation of numerous conditions, including frailty. Four parameters were extracted from the balance quality assessment for older subjects: Rising Rate (RR), Duration of the stabilization segment (ZD), Stabilogram Area (SA) and Average Velocity of the Trajectory (TV). These are then scored and weighted, thus creating an overall indicator of balance quality. The reliability, the absolute reliability and the minimum difference of the four parameters were evaluated using the intra-class correlation coefficient (ICC), the standard error measurement (SEM) and the Minimal Detectable Change (MDC), respectively. Reproducibility was very high, with ICC values of 0.83, 0.85, 0.88 and 0.95 for RR, ZD, SA and TV, respectively. These results revealed that the parameters are a reliable measure for evaluating balance quality measurement.

Keywords: Medical technology - Design and development
Abstract: This paper proposes design considerations that need to be followed in order to eliminate potential sources of artefact that could distort a recorded neural signal. The artefact that appears in a recorded signal has a combination of potential sources each of which contributes towards its formation. As such, these sources of artefact have been addressed in three main categories: a) electronics artefact, b) encapsulation artefact and c) interface artefact. Each source (component) is analyzed further and appropriate design techniques and considerations are suggested towards its mitigation.

Keywords: Neural interfaces - Bioelectric sensors, Neural interfaces - Implantable systems, Neural signal processing
Abstract: This paper presents an integrated biopotential preamplifier architecture targeting applications that simultaneously require high common-mode rejection ratio (CMRR), low noise, high input common-mode range (ICMR), and current-efficiency (low Noise Efficiency Factor or NEF). A biopotential preamplifier, which performs well in line with the state-of-the-art of the field while providing enhanced ICMR and CMRR performance, was fabricated in a 0.5 um CMOS process. Results from measurements show that the gain is 47 dB, the bandwidth ranges from 1 Hz to 7.7 kHz, the equivalent input noise is 1.8 uVrms, the CMRR is 100.5 dB, the ICMR is 1.7 V and the NEF is 3.2.

Keywords: Neural interfaces - Tissue-electrode interface, Neural interfaces - Implantable systems, Neural stimulation
Abstract: Direct current (DC) has the potential not only to excite but also to inhibit neurons. This property of DC stimulus has been used for generating peripheral nerve blocks. One translational challenge of DC-based neuromodulation technologies, especially for pain suppression, is that the commercially available cuff electrodes have metal-tissue interfaces that are incapable of delivering DC safely. Passing DC through any metal-tissue interface generates harmful electrochemical products which can damage the target nerve. To address this issue, we present a fabrication process for making self-curling silicone cuffs with paper/agar based, ionically conducting neural interface. We fabricate monopolar as well as bipolar cuffs and demonstrate that the electrode impedances can be easily controlled by modulating the paper/agar channel dimensions. Further, we perform in-vivo implantation of these electrodes on a rat sciatic nerve to qualitatively validate the self-curling action.

Keywords: Neural interfaces - Biomaterials, Neural interfaces - Implantable systems, Motor neuroprostheses - Prostheses
Abstract: Optically transparent encapsulation is presented with results from long-term reliability and light transmission tests. This technology is required in certain implantable neural prostheses that demand the transmission of optical signals through an encapsulating material, such as in retinal implants or in optogenetic applications. In this study, biocompatible film-type cyclic olefin polymers (COPs) with low moisture absorption (<0.01 %) and high light transmission (92 %) are utilized as encapsulating materials based on thermal lamination. The reliability of COP encapsulation is characterized through accelerated soak tests in a 75 °C saline solution to measure the leakage currents from encapsulated inter-digitated electrodes. These tests had been done for 211 days with the estimated lifetime of 8.05 years at 37 °C. In addition, the optical properties of a thermally laminated COP film sample in relation to its thickness are evaluated by an experimental setup which uses projected line patterns on an image sensor. The light transmittance of COP film samples thinner than 376 μm exceeded 91.69 %, and the minimum distinguishable line pitch was 47.6 µm at a thickness of 26 µm. These results validate the feasibility of optically transparent encapsulation using COPs and may contribute to its use in future implantable neural prostheses.

Keywords: Neural interfaces - Microelectrode technology, Neural stimulation - Deep brain, Brain physiology and modeling
Abstract: Microelectrode recording (MER) and intraoperative test stimulations are commonly used during stereotactic implantation of deep brain stimulation (DBS) electrodes but they can increase the risk of hemorrhage. The aim of the study is to present and evaluate a system combining laser Doppler flowmetry (LDF) and MER. An optical probe was designed with an inner metal tube for the microelectrode. Calibration of the MER-LDF probe in a standard microsphere solution showed expected LDF pattern. No interferences of the MER probe with the LDF signals could be observed. LDF was also acquired in one Parkinson patient undergoing DBS implantation. LDF data were obtained along the precalculated trajectory i.e. from cortex towards the target in the subthalamic nucleus. Results demonstrated the technical feasibility of the combined MER-LDF probe during in-vitro experiences and in one patient. The perfusion signal representing the microcirculation showed stable values with clear peaks from each heartbeat. This agreed with previous investigation using an optical probe without the MER function. Due to the forward-looking probe design, this new technology has a high potential to avoid vessels during MER recording. In addition, it could be possible to detect changes in microcirculatory blood flow during stimulation.

Keywords: Neural interfaces - Microelectrode technology, Neural stimulation, Neural interfaces - Biomaterials
Abstract: Only thin-film technology can satisfy the requirements of high spatial selectivity at high-channel-count electrode array designs by simultaneously good conformability to the targeted tissue through mechanical flexibility enriching future applications of functional neural stimulation. However, caused by the high impact of the microstructure on the mechanical and electro-chemical film properties, varying fabrication processes of the same thin-film makes the difference between acute and chronic long-term stable electrodes. The influence of standard clinical electrical pulsing on flexible polyimide-based thin-film platinum electrodes for neuroprostheses, either sputter deposited or evaporated, and different diameters was assessed and compared. The electrochemical and morphological analysis showed a higher corrosion susceptibility and electrochemical degrada-tion for the sputter deposited platinum electrodes with even total failures of smaller diameters. In contrast, the evaporated thin-films provided itself as more stable and reliable metallization with also smaller electrodes keeping their film integrity intact over the experimental period, -appearing to be the preferable material for improving thin-film electrodes’ longevity.

Keywords: Neural interfaces - Microelectrode technology, Neural interfaces - Implantable systems, Neural interfaces - Biomaterials
Abstract: Bioelectronic neural interfaces can fail in vivo due to water penetration and corrosion of the packaging technology used to protect sensitive portions of the device. Anisotropic conductive adhesive (ACA) is gaining popularity in the neural interface community to connect fabricated electrode arrays with back-end packages; however, ACA’s durability during chronic implants is largely unknown. We have designed a platform that uses an aggressive reactive-accelerated aging (RAA) environment to rapidly assess the ability of ACA and silicone rubber encapsulation to maintain electrical integrity in vitro. 24-day RAA tests at 77˚C with 10-20 mM H2O2 that approximates ~1 year in vivo showed that ACA rapidly fails (2-4 days RAA) due to water absorption through the silicone encapsulant. Electrical impedance spectroscopy (EIS) confirmed water penetration through the package and the resulting corrosion of the sensitive metallic components

Keywords: Sensor Informatics - Physiological monitoring, Sensor Informatics - Multi-sensor data fusion, Sensor Informatics - Sensors and sensor systems
Abstract: The early detection of occult bleeding is a difficult problem for clinicians because physiological variables such as heart rate and blood pressure that are measured with standard patient monitoring equipment are insensitive to blood loss. In this study, the pulse arrival time (PAT) was investigated as an easily recorded, non-invasive indicator of hypovolemia. A lower body negative pressure (LBNP) study with a step- wise increase of negative pressure was conducted to induce central hypovolemia in a study population of 30 subjects. PAT values were extracted from simultaneous recordings of the electrocardiogram (ECG) and photoplethysmographic (PPG) recordings both from the index finger and from within the outer ear canal. Stroke volume (SV) was recorded as a reference measure by transthoracic echocardiography. An inter- and intra-individual correlation analysis between changes in SV and the PAT measurements was performed. Furthermore, it was assessed if PAT measurements can indicate a diminished SV in this scenario. It could be demonstrated that the measured PAT values are significantly increased at the lowest LBNP pressure level. A very strong intra-individual correlation (ρ ≥ 0.8) and a moderate inter-individual correlation (ρ ≥ 0.5) between PAT and SV measurements were found. Thus, PAT measurements could be a viable tool to monitor patient specific volemic trends. Further research is needed to investigate if PAT information can be utilized for a more robust inter-subject quantification of the degree of hypovolemia.

Keywords: Sensor Informatics - Physiological monitoring
Abstract: Cuff-less blood pressure (BP) monitoring is increasingly being needed for cardiovascular events management in clinical. Many of the existing methods, however, are based on manual feature extraction, which cannot characterize the complex relationship between the physiological signals and BP. In this study, the 16-layer VGGNet was used to construct cuff-less BP from electrocardiogram (ECG) and pressure pulse wave (PPW) signals, with no need extract features from raw signals. The deep network architecture has the ability of automatic feature learning, and the learned features are the higher-level abstract description of low-level raw physiological signals. Eight-nine middle-aged and elderly subjects were enrolled to evaluate the performance of the proposed BP estimation method, with oscillometric technique-based BP as a reference. Experimental results indicate that the proposed method had a commendable accuracy in BP estimation, with a correlation coefficient of 0.91 and an estimation error of -2.06 ± 6.89 mmHg for systolic BP, and 0.89 and -4.66 ± 4.91 mmHg for diastolic BP. This study shows that the proposed method provided a potential novel insight for the cuff-less BP estimation.

Keywords: Sensor Informatics - Physiological monitoring, General and theoretical informatics - Machine learning, Sensor Informatics - Wearable systems and sensors
Abstract: High levels of cognitive workload decreases human's performance and leads to failures with catastrophic outcomes in risky missions. Today, reliable cognitive workload detection presents a common major challenge, since the workload is not directly observable. However, cognitive workload affects several physiological signals that can be measured non-invasively. The main goal of this work is to develop a reliable machine learning algorithm to identify the cognitive workload induced during rescue missions, which is evaluated through drone control simulation experiments. In addition, we aim to minimize the computing resources usage while maximizing the cognitive workload detection accuracy for a reliable real-time operation. We perform an experiment in which 24 subjects played a rescue mission simulator while respiration, electrocardiogram, photoplethysmogram, and skin temperature signals were measured. State-of-the-art feature-based machine learning algorithms are investigated for cognitive workload characterization using learning curves, data augmentation, and cross-validation techniques. The best classification algorithm is selected, optimized, and the most informative features are selected. Finally, the generalization power of the optimized model is evaluated on an unseen test set. We obtain an accuracy level of 86% on the new unseen datasets using the proposed and optimized eXtreme Gradient Boosting (XGB) algorithm. Then, we reduce the complexity of the machine learning model for future implementation on resource-constrained wearable embedded systems, by optimizing the model and selecting the 26 most important features. Overall, a generalizable and low-complexity machine learning model for cognitive workload detection based on physiological signals is presented for the first time in the literature.

Keywords: Sensor Informatics - Physiological monitoring, Health Informatics - Personal/consumer health informatics
Abstract: Heart rate and heart rate variability parameters provide important information on sympathetic and parasym- pathetic branches of autonomous nervous system. These pa- rameters are usually extracted from electrocardiograms often measured between two electrodes and called an ECG lead. Besides systems intended only for heart rate measurement, ECG measurement devices employ several well-known lead systems including the standard 12-lead system, EASI lead system and Mason-Likar systems. Therefore, the first step is to select the appropriate lead for heart rate variability analysis. The appropriate electrode locations for single-lead measurement systems or the preferred measurement lead in multi-lead measurement are choices that the user needs to make when the heart rate variability is of interest. However, it has not been addressed in the literature, if the lead selection has an effect on the obtained HRV parameters. In this work, we characterized the amount of deviation of heart rate and heart rate variability parameters extracted from nine ECG leads, six from EASI leads and three modified limb leads. The results showed a deviation of 2.04, 2.88, 2.06 and 3.45 ms in SDNN, rMSSD, SD1 and SD2, respectively. A relative difference up to 10% was observed in HRV parameters for single signal frames. Additionally, the discrimination of the R- peaks by amplitudes was evaluated. The A–S lead appeared to have the best performance in all the tests.

Keywords: Sensor Informatics - Low power, wireless sensing methods and systems, Sensor Informatics - Physiological monitoring, Sensor Informatics - Wireless sensors and systems
Abstract: Non-contact continuous respiratory monitoring during sleep is of high usability to early disease detection and daily health monitoring. This study introduces a novel microwave sensor prototype for real-time respiration measurement. The antennas of the sensor are placed below the bed sheet, and function by transmitting a series of microwave signals to detect the inhale-exhale body motions while breathing. Compared to other remote wireless monitors, our sensor is less interfered by environmental noises as well as without direct contact with the body. The received I/Q signals are merged into one output and process to detect the frequency of breathing. The performance is evaluated using overnight sleep data and compared with ground-truth data measured by standard PSG airflow sensor. Result achieves high detection rate of 98.88% with mean squared error (MSE) of 1.23 over 420 one-minute recordings. In addition, the sensor is able to detect respiration accurately regardless of a person’s sleep position. We demonstrate that our microwave sensor is robust and usable for real-time respiratory monitoring.

Keywords: Motor neuroprostheses - Epidural stimulation
Abstract: Intraspinal microstimulation and epidural spinal cord stimulation can be considered as the technique to restore function following spinal cord injury through further research. In this paper, the automatic brain stereotaxic instrument was used to electrically stimulate the lumbosacral spinal cord (T12-L2 spinal segments) in rats. The motor function regions under intraspinal microstimulation and epidural spinal cord stimulation were measured. Threshold currents and coordinate sites of related motions were recorded. Comparative analysis revealed that the threshold current required for epidural stimulation to induce hindlimb motion was greater. Although the distribution of motor function regions measured by these two methods differed in the type of motion, the segment distribution of each motion were roughly the same. Therefore, if conditions permit, epidural stimulation can be used instead of intraspinal microstimulation to reduce secondary damage to the spinal cord. This provides a reference for locating stimulation sites for epidural spinal cord stimulation.

Keywords: Motor neuroprostheses - Neuromuscular stimulation, Neurological disorders - Treatment methodologies
Abstract: Functional electrical stimulation has been used in rehabilitation programs for patients with chronic spinal cord injury. When used correctly it is able to improve the well-being of patients. However, when the stimulus is not adequate it can accelerate the process of fatigue, reducing the time available for training the programmed motor activity. To optimize the configuration of the stimulatory parameters, we developed a tool capable of simulating the muscle strength performance in response to different stimulatory profiles. The tool was able to reproduce the behavior of motoneurons in chronic spinal cord injury and to estimate the muscular strength resulting from the application of different stimuli. We consider that this FES Simulator is a promising tool to design and simulate different profiles of electrical stimulation, optimizing the decision process of the stimulation parameters.

Keywords: Motor neuroprostheses - Neuromuscular stimulation
Abstract: A Neuromuscular Electrical Stimulation (NMES) protocol that incorporates co-contraction of the quadriceps and hamstrings may provide greater functional benefits for knee rehabilitation. It is unclear if the addition of a co-contraction will affect the desired torque outputs of one or two of the involved muscle groups. Due to the proposed functional benefits of co-contraction, it may be beneficial to test the addition of a co-contraction electrical muscle stimulation. In this study we recruited 14 participants with whom we compared two NMES protocols; isolated quadriceps contraction (k-NMES) versus co-contraction of quadriceps and hamstrings (co-NMES). We examined peak knee extension evoked torque, current intensities required to produce given torque outputs, and self-reported discomfort levels at given torques. At maximum tolerable intensity peak torque output was similar in k-NMES versus co-NMES. To achieve specific submaximal levels of torque output as percentages of maximum voluntary contraction (MVC), a higher current intensity was required for co-NMES yet with no greater level of discomfort. Results suggest that clinicians who wish to achieve a co-contraction of quadriceps and hamstrings as part of a rehabilitation programme can use co-NMES without having to sacrifice the strength of contraction achieved in the quadriceps. This could lead to better functional outcomes, though more work is required to confirm this.

Keywords: Motor neuroprostheses - Prostheses, Human performance - Ergonomics and human factors, Human performance - Sensory-motor
Abstract: Usability of upper-limb prostheses remains to be a challenge due to the complexity of hand-object interactions in activities of daily living. Functional evaluation is critical for the optimization of prosthesis performance during device design and parameter tuning phase. Therefore, we implemented a low-cost physics-based virtual reality environment (VRE) capable of simulating wide range of object grasping and manipulation tasks to enable human-in-the-loop optimization. Importantly, our novel VRE can assess user performance quantitatively using movement kinematics and interaction forces. We present a preliminary experiment to validate our VRE. Four able-bodied subjects performed object transfer tasks with a simulated myoelectric one DoF soft-synergy prosthetic hand, while wearing braces to restrain different levels of wrist motion. We found that the task completion time was similar across conditions, however limited wrist pronation led to more shoulder compensatory motion whereas challenging object orientation caused more torso compensatory motion.

Keywords: Motor neuroprostheses - Prostheses, Motor neuroprostheses - Neuromuscular stimulation, Neurorehabilitation
Abstract: Although advances in technology promoted new physiotherapy approaches, there is still an urge for equipment and techniques to improve the quality of patients lives with motor disabilities. Functional electrical stimulation cycling (FES cycling) is an example of this type of technology, in which the control of stimulation parameters enables a spinal cord injured person to ride a bicycle. In this work, we aim to investigate the use of passive knee orthoses for FES cycling assistance. Hence, we compared the cycling cadence and quadriceps excitation using an FES cycling simulation platform for different spring torques and ranges. In this paper, we obtained spring parameters that increased cycling cadence by 10.60% while decreasing by 7.33% the quadriceps activity, which indicates that this type of passive orthosis may diminish fatigue caused by FES.

Keywords: Motor neuroprostheses - Neuromuscular stimulation, Motor learning, neural control, and neuromuscular systems
Abstract: The use of electrical stimulation to elicit single twitches and tetanic contractions of skeletal muscles has increased markedly in the last years, with applications ranging from basic physiology to clinical settings. Addressing all possible needs required by different applications with an electrical stimulator is challenging as it requires the device to be highly flexible in terms of stimulation configurations (number of channels and electrode location), and possibility to control the stimulation patterns (timing and stimulation profiles). This paper describes a new wireless, modular, and programmable electrical stimulator integrating the possibility to acquire and use biomechanical signals to trigger the stimulation output. A closed-loop FES Cycling setup has been presented to show a possible application of the system.

Keywords: Smart cochlear implants
Abstract: Cochlear Implant is used for patients with severe hearing loss. It is a neural-prosthesis that stimulates the nerve endings within the cochlea, which is the organ of hearing. The surgical technique involves inserting the electrode array of the implant into a very small ``snail-like'' spiral structure. During this insertion process, the surgeon's finger tip is not able to perceive the resistance from the contact of the implant and the cochlea's internal structure, below the internal rupture threshold. This can potentially damage vital structures and result in the worsening of residual hearing and poor speech perception. Currently, there is no clinically and commercially available intra-operative force feedback system. A custom made sensor is therefore proposed, integrated within the implant to enable real-time force readings. The device will provide surgeons with the vital force feedback information related to the implants' position within the cochlea. This paper concentrates on demonstrating that the proposed sensor is capable of measuring the contact force below the rupture threshold of the cochlea's internal structure.

Keywords: Other smart implanted systems
Abstract: We present a novel miniaturized multi sensor implant for monitoring hemodynamic parameters in cardiovascular regions. Pressure measurements are performed with a highly accurate capacitive pressure sensor. An additional acceleration and temperature sensor allows compensating the impact of patient’s inclination and temperature variations on the pressure measurement, respectively. A multi-functional transponder application-specific integrated circuit (ASIC) manages sensor signal processing, storage of ID, sensor calibration data, telemetric energy, and data transmission with an extracorporeal reading unit. Each component of the implant is assembled on a low temperature co-fired ceramics (LTCC) circuit board with an integrated antenna coil enabling an inductive near-field coupling at a frequency of 13.56 MHz. For a streamlined shape and reduction of thrombogenicity, the implant is encapsulated by biocompatible polymers.

Keywords: Smart implanted neurostimulation systems, Other smart implanted systems, Smart implanted drug delivery systems
Abstract: This paper presents a novel ultrasonic transmitter with the ability of focusing ultrasonic waves for maximum power transmission at different depths for brain neurostimulator implants. The most important advantages of the proposed multi-ring ultrasonic transducer (MRUT) is its simplicity and no requirement of any lens or air cavity for focusing the ultrasonic waves. Furthermore, adjusting the focal point compared to the conventional transducers is significantly easier, especially as the location of implants may vary due to, for example, head movement or the need of using these implants at different depths. By the use of multiple rings on a single piezoelectric disk in our transducer, not only more focused ultrasound beams can be achieved, but also the side lobes can be diminished by exciting each rings with different electrical signal. The proposed transmitter is envisioned to be used for optogenetic stimulation of neurons in freely-moving animals.

Keywords: Smart implanted neurostimulation systems, Smart implanted neuromuscular stimulation systems
Abstract: Synchronized oscillations are a ubiquitous feature of neuronal circuits and can modulate online information transfer and plasticity between brain areas. The disruption of these oscillatory processes is associated with the symptoms of several brain disorders. While conventional therapeutic high-frequency deep brain stimulation can perturb neuronal oscillations, manipulating the timing of oscillatory activity between areas more precisely could provide a more efficient and effective method of modulating these activities. Here we describe a prototype circuit for synchronizing the clocks between an active implantable and an external sensing and stimulation system that could be used to achieve this goal. Our specific focus is on synchronizing the systems for paired-associative stimulation. The ability to repetitively drive two brain regions with a fixed latency has specific implications for neural plasticity. Furthermore, the general concept can be applied for many potential applications involving distributed neural interfaces.

Keywords: Smart implanted neurostimulation systems
Abstract: Neural stimulators have become more and more widely used as an effective tool in neural therapies. To address power supply and consumption issues in this application, an energy-efficient Implantable-Neural-Stimulator system composed of a pulse generator and a wireless charger is proposed and implemented in 0.8μm 40V Bipolar-CMOS-DMOS (BCD) process. By adopting a Single Ended Primary Inductor Converter (SEPIC) and optimizing the switching frequency and the gate width of its power MOSFET, the stimulating output voltage range can cover 0~12V with a maximum output ripple of 0.31%. The proposed charger can charge the implantable battery wirelessly by an inductively coupled resonance circuit. In addition, it can adjust the charging voltage to keep it constantly only a little higher than the battery voltage, which reduces the charging headroom voltage and greatly improves the charging efficiency. The measured maximum power efficiencies of these two modules reach as high as 78.04% and 70.67%, respectively.

Keywords: Smart implanted neurostimulation systems, Smart implanted neuromuscular stimulation systems, Smart pacemaker and implanted defibrillator
Abstract: Platinum is widely used as the electrode material for implantable devices. Owing to its high biostability and corrosion resistivity, platinum could also be used as the main metallization for tracks in active implants. Towards this goal, in this work we investigate the stability of parylene-coated Pt tracks using passive and active tests. The test samples in this study are Pt-on-SiO2 interdigitated comb structures. During testing all samples were immersed in saline for 150 days; for passive testing, the samples were left unbiased, whilst for active testing, samples were exposed to two different stress signals: a 5 V DC and a 5 Vp 500 pulses per second biphasic signal. All samples were monitored over time using impedance spectroscopy combined with optical inspection. After the first two weeks of immersion, delamination spots were observed on the Pt tracks for both passive and actively tested samples. Despite the delamination spots, the unbiased samples maintained high impedances until the end of the study. For the actively stressed samples, two different failure mechanisms were observed which were signal related. DC stressed samples showed severe parylene cracking mainly due to the electrolysis of the condensed water. Biphasically stressed samples showed gradual Pt dissolution and migration. These results contribute to a better understanding of the failure mechanisms of Pt tracks in active implants and suggest that new testing paradigms may be necessary to fully assess the long-term reliability of these devices.

Keywords: Neurological disorders - Diagnostic and evaluation techniques, Neurological disorders - Treatment methodologies, Human performance - Gait
Abstract: Muscular weakness is one of the main signs associated with the onset and progression of Duchenne Muscular Dystrophy. During motor functions, this disease also determines deviations in muscular activity, especially in terms of coordination and activation between muscles acting on the same joints. In this study, surface EMG activity of the lower limb muscles of 10 children with Duchenne Muscular Dystrophy at different times from disease onset were recorded along with kinematics during unconstrained gait. Muscular co-activation of muscle pairs was then evaluated by extracting different co-activation indicators, and linking them with kinematic markers of motor function. The combination of disease progression and pharmacological treatment resulted in a significant decrease in terms of co-activation indexes for two pairs of agonist-antagonist muscles, and for one of these two pairs the decrease in co-activation was correlated with a decrease in the motor function of gait.

Keywords: Neurological disorders - Epilepsy, Brain physiology and modeling - Cognition, memory, perception
Abstract: Temporal lobe epilepsy (TLE) is a medically refractory focal epilepsy associated with structural deficits. Considerable evidence has revealed that patients with TLE also exhibit deficits in functional connectivity. According to previous research, patients with TLE exhibited decreased performance in speech sound perception and auditory-motor integration for voice control, which might be related to the compromised brain network connectivity. However, the specific nature of functional connectivity within and across brain regions remains largely unknown. To answer this question, we extended previous research from examining the topological properties of the entire brain network to the intra- and inter-regional communications of different brain regions. Patients with TLE and healthy controls were recruited to perform a pitch reflex task, during which electroencephalograph (EEG) data were acquired to construct graphical brain networks. Compared with healthy controls, inter-regional and cross-hemispheric connections were reduced in patients with TLE, whose functional networks were primarily composed of intra-regional connections. Significant differences in network parameters (betweenness centrality, modularity, and functional integration) as well as network hubs between the two groups further supported our findings that TLE is associated with alterations in functional connectivity during auditory-motor integration.

Keywords: Neurological disorders - Epilepsy, Brain functional imaging - EEG, Brain functional imaging - Source localization
Abstract: Abstract— The study of brain waves propagation is of interest to understand the neural involvement in both physiological and pathological events, such as interictal epileptic spikes (IES). The possibility to track the trajectory of IESs could be useful to better characterize the role of the involved structures in the epileptic network, adding valuable information to the epileptic focus localization. Methods for the cortical traveling wave analysis (CTWA) have been proposed to trace the preferred propagation path of sleep slow waves, using scalp high-density EEG and reconstructing the trajectories both in the sensors and in the sources space. In this work, we propose a feasibility study of the application of these concepts to Stereo-EEG (SEEG) data for the analysis of IES. Through simulations, we selected the best performing Electrical Source Imaging inverse solution for our purpose and illustrate the CTWA procedure. We further show an exemplary application on real data and discuss advantages and pitfalls of the application of CTWA in SEEG.

Keywords: Neurological disorders - Epilepsy, Brain physiology and modeling, Brain functional imaging - EEG
Abstract: Epilepsy can be controlled by targeted treatment of the epileptogenic zone (EZ), the region in the brain where seizures originate. Identification of the EZ often requires visual inspection of invasive EEG recordings and thus relies heavily on placement of electrodes, such that they cover the EZ. A dense brain coverage would be ideal to obtain accurate boundaries of the EZ but is not possible due to surgical limitations. This gives rise to the "missing electrode problem", where clinicians desire to know what neural activity looks like between implanted electrodes. In this paper, we compare two methods for time series estimation of missing stereotactic EEG (SEEG) recordings. Specifically, we represent SEEG data as a sequence of Linear Time-Invariant (LTI) models. We then remove one signal from the data set and apply two different algorithms to simultaneously estimate the LTI models and the "missing" signal: (i) a Reduced-Order Observer in combination with Least Squares Estimation and (ii) an Expectation Maximization based Kalman Filter. The performance of each approach is evaluated in terms of (i) estimation error, (ii) sensitivity to initial conditions, and (iii) algorithm run-time. We found that the EM approach has smaller estimation errors and is less sensitive to initial conditions. However, the reduced-order observer has a run-time that is orders of magnitude faster.

Keywords: Neurological disorders - Epilepsy, Brain physiology and modeling - Neural dynamics and computation, Brain functional imaging - EEG
Abstract: We built a regression model to describe the progress of epileptogenesis in a rat intrahippocampal tetanus toxin (TeNT) epilepsy model by identifying informative neural features from hippocampal local field potentials (LFPs). The LFPs were recorded from the awake and freely behaving animals during the latent period and the active-seizure period. Frequency domain neural features including power spectral density, coherence and phase coherence were calculated from the hippocampal LFPs. A least angle regression with elastic net regularization (LARS-ENR) model successfully predicted a relative day from the first seizure in multiple rats (R2test = 0.724±0.025). By leveraging a characteristic of LARS-ENR which reduces unnecessary features, we identified the neural features related to epileptogenesis in a TeNT model.

Keywords: Neurological disorders - Epilepsy, Brain functional imaging - EEG, Neurological disorders - Treatment methodologies
Abstract: Seizures in patients with medically refractory epilepsy (MRE) cannot be controlled with drugs. For focal MRE, seizures originate in the epileptogenic zone (EZ), which is the minimum amount of cortex that must be treated to be seizure free. Localizing the EZ is often a laborious process wherein clinicians first inspect scalp EEG recordings during several seizure events, and then formulate an implantation plan for subsequent invasive monitoring. The goal of implantation is to place electrodes into the brain region covering the EZ. Then, during invasive monitoring, clinicians visually inspect intracranial EEG recordings to more precisely localize the EZ. Finally, the EZ is then surgically ablated, removed or treated with electrical stimulation. Unfortunately success rates average at 50%. Such grim outcomes call for analytical assistance in creating more accurate implantation plans from scalp EEG. In this paper, we introduce a method that combines imaging data (CT and MRI scans) with scalp EEG to derive an implantation distribution. Specifically, scalp EEG data recorded over a seizure event is converted into a time-gamma frequency map, which is then processed to derive a spectrally annotated implantation distribution (SAID). The SAID represents a distribution of gamma power in each of eight cortical lobe/hemisphere partitions. We applied this method to 4 MRE patients who underwent treatment, and found that the SAID distribution overlapped more with clinical implantations in success cases than in failed cases. These preliminary findings suggest that the SAID may help in improving EZ localization accuracy and surgical outcomes.

Keywords: Principal component analysis, Signal pattern classification, Data mining and processing - Pattern recognition
Abstract: In this paper, we extracted hand crafted features from the ECG signals and evaluated the performance of different combination of features for sleep apnoea detection. We calculated the ECG derived respiratory (EDR) signal using three methods (QRS area, amplitude demodulation and fast PCA methods) and then calculated the cardiopulmonary coupling (CPC) spectrum using each EDR method. We then extracted features from the CPC spectrums and the time and frequency representations of the heart rate variability (HRV) and EDR signals Then, we compared the performance results of different combinations of the features used for automated sleep apnoea detection. We also applied a temporal optimisation method by averaging the features of every three adjacent epochs. Two classifiers were used to detect sleep apnoea: the extreme learning machine (ELM), and linear discriminant analysis. The features were evaluated on the MIT PhysioNet Apnea-ECG database. Apnoea detection was evaluated with leave-one-record-out cross-validation. The PCA CPC features obtained the highest accuracy of 86.5% and AUC of 0.94 using LDA classifier. The performance results of the combined features (of PCA method) obtained the same results. We conclude that for this study, the CPC features using fast PCA method are our best feature set for sleep apnoea detection.

Keywords: Principal and independent component analysis - Blind source separation, Independent component analysis
Abstract: Analyses of motor unit activity provide a window to the neural control of motor output. In recent years, considerable advancements in surface EMG decomposition methods have allowed for the discrimination of dozens of individual motor units across a range of muscle forces. While these non-invasive methods show great potential as an emerging technology, they have difficulty discriminating a representative sample of the motor pool. In the present study, we investigate the distribution of recruitment thresholds and motor unit action potential waveforms obtained from high density EMG across four muscles: soleus, tibialis anterior, biceps brachii, and triceps brachii. Ten young and healthy control subjects generated isometric torque ramps between 10-50% maximum voluntary torque during elbow or ankle flexion and extension. Hundreds of motor unit spike trains were decomposed for each muscle across all trials. For lower contraction levels and speeds, surface EMG decomposition discriminated a large number of low-threshold units. However, during contractions of greater speed and torque level the proportion of low threshold motor units decomposed was reduced, resulting in a relatively uniform distribution of recruitment thresholds. The number of motor units decomposed decreased as the contraction level and speed increased. The decomposed units showed a wide range of recruitment thresholds and motor unit action potential amplitudes. In conclusion, although surface EMG decomposition is a useful tool to study large populations of motor units, results of such methods should be interpreted in the context of limitations in sampling of the motor pool.

Keywords: Principal and independent component analysis - Blind source separation, Independent component analysis
Abstract: Study of white matter (WM) fiber-bundles is a crucial challenge in the investigation of neurological diseases like multiple sclerosis (MS). In this activity, the amount of data to process is huge, and an automated approach to carry out this task is in order. In this paper we show how tensor-based blind source separation (BSS) techniques can be successfully applied to model complex anatomical brain structures. More in detail, we show how through vector hankelization it is possible to formalize data extracted from WM fiber-bundles using a tensor model. Two main tensor factorization techniques, namely (Lr;Lr; 1) block term decomposition (BTD) and canonical polyadic decomposition (CPD), were applied to the generated tensor. The information extracted from the factorization was then used to differentiate between sets of fibers, within the bundle, affected by the pathology and normal appearing fibers. Performances of the proposed tensor-based model was evaluated on simulated data representing pathological effects of MS. Results show the capability of our tensor-based model to detect small pathological phenomena appearing along WM fibers.

Keywords: Principal component analysis, Nonlinear dynamic analysis - Biomedical signals
Abstract: Cognitive and behavioural impairments in early-onset epilepsy affect the children and families' quality of life. Our ability to detect these impairments is limited, and it requires laborious questionnaires. Here, we describe a pilot study exploring the fusion of resting-state EEG, volumetric MRI, and phenotypic scores of child development based on the Canonical Polyadic Decomposition, expanding the recently presented Joint EEG-Development Inference (JEDI) model. Pilot data fusion was performed on functional, structural and developmental brain features of 29 preschool children diagnosed with epilepsy. The results suggest that combining multimodal brain data towards a comprehensive analysis of brain development in young children is plausible.

Keywords: Data mining and processing in biosignals, Independent component analysis, Signal pattern classification
Abstract: This work presents an unsupervised mining strategy, applied to an independent component analysis (ICA) of segments of data collected while participants are answering to the items of the Halstead Category Test (HCT). This new methodology was developed to achieve signal components at trial level and therefore to study signal dynamics which are not available within participants' ensemble average signals. The study will be focused on the signal component that can be elicited by the binary visual feedback which is part of the HCT protocol. The experimental study is conducted using a cohort of 58 participants.

Keywords: Principal component analysis
Abstract: The heart rate is a fundamental measure which can be used to monitor an individual’s level of health or fitness, as well as a range of medical conditions. Conventional heart rate devices used in hospitals require continuous contact with specific points on the patient’s body, depending on the device being used. Such continuous contact could prove to be a risk for skin irritation or infections and may also be of inconvenience to the patients, potentially restricting movement. A contactless approach for measuring heart rate could thus prove significant benefits over conventional, contact-based devices. This paper presents a method for the contactless extraction of heart rate measurements from a video footage using principal component analysis, with no pre-defined region of interest being required. Three different ways of presenting the outcome from principal component analysis are presented and the results obtained are discussed.

Keywords: Optical imaging, Cardiac imaging and image analysis, Image feature extraction
Abstract: Heart rate (HR) is one of the most important vital signs for indicating the health condition of a person. Contactless camera-based HR measurement is particularly beneficial for sleep monitoring, as it is comfortable and convenient. However, compared with ambient light, the skin pulsation in near infrared range is much weaker and more susceptible to distortions (e.g. body motion, light changes). In this paper, we propose a method to expand the single-wavelength channel of a near infrared camera to multiple channels for illumination noise reduction, where the channel expansion is performed in the spatial domain using skin and non-skin pixels. The essence is using illumination changes of non-skin pixels to eliminate such a distortion on skin pixels and thus improve pulse extraction. On average, measurement coverage (MC) increased from 50% to 83% for the methods of substraction and Segment Principal Component Analysis (Seg-PCA), and Signal-to-Noise Ratio (SNR) is increased from -8.40 dB to -4.62 dB for the method of Segment Independent Component Analysis (Seg-ICA).

Keywords: Optical imaging, Image visualization
Abstract: Gastric endoscopy is a common clinical practice that enables medical doctors to diagnose the stomach inside a body. In order to identify a gastric lesion’s location such as early gastric cancer within the stomach, this work addressed to reconstruct the 3D shape of a whole stomach with color texture information generated from a standard monocular endoscope video. Previous works have tried to reconstruct the 3D structures of various organs from endoscope images. However, they are mainly focused on a partial surface. In this work, we investigated how to enable structure-from-motion (SfM) to reconstruct the whole shape of a stomach from a standard endoscope video. We specifically investigated the combined effect of chromo-endoscopy and color channel selection on SfM. Our study found that 3D reconstruction of the whole stomach can be achieved by using red channel images captured under chromo-endoscopy by spreading indigo carmine (IC) dye on the stomach surface.

Keywords: Optical imaging
Abstract: Light field imaging technology has been attracting increas- ing interest because it enables capturing enriched visual infor- mation and expands the processing capabilities of traditional 2D imaging systems. Dense multiview, accurate depth maps and multiple focus planes are examples of different types of visual information enabled by light fields. This technology is also emerging in medical imaging research, like derma- tology, allowing to find new features and improve classifica- tion algorithms, namely those based on machine learning ap- proaches. This paper presents a contribution for the research community, in the form of a publicly available light field im- age dataset of skin lesions (named SKINL2). This dataset contains 250 light fields, captured with a focused plenoptic camera and classified into seven clinical categories, according to the type of lesion. Each light field is comprised of 81 differ- ent views of the same lesion. The database also includes the dermatoscopic image of each lesion. A representative subset of 17 central view images of the light fields is further char- acterised in terms of spatial information (SI), colourfulness (CF) and compressibility. This dataset has high potential for advancing medical imaging research and development of new classification algorithms based on light fields, as well as in clinically-oriented dermatology studies.

Keywords: Optical imaging, Image feature extraction
Abstract: Surgical navigation systems can enhance surgeon vision and form a reliable image-guided tool for complex interventions as spinal surgery. The main prerequisite is successful patient tracking which implies optimal motion compensation. Nowadays, optical tracking systems can satisfy the need of detecting patient position during surgery, allowing navigation without the risk of damaging neurovascular structures. However, the spine is subject to vertebrae movements which can impact the accuracy of the system. The aim of this paper is to investigate the feasibility of a novel approach for offering a direct relationship to movements of the spinal vertebra during surgery. To this end, we detect and track patient spine features between different image views, captured by several optical cameras, for vertebrae rotation and displacement reconstruction. We analyze patient images acquired in a real surgical scenario by two gray-scale cameras, embedded in the flat-panel detector of the C-arm. Spine segmentation is performed and anatomical landmarks are designed and tracked between different views, while experimenting with several feature detection algorithms (e.g. SURF, MSER, etc.). The 3D positions for the matched features are reconstructed and the triangulation errors are computed for an accuracy assessment. The analysis of the triangulation accuracy reveals a mean error of 0.38 mm, which demonstrates the feasibility of spine tracking and strengthens the clinical application of optical imaging for spinal navigation.

Keywords: Optical imaging, Novel imaging modalities
Abstract: Photoplethysmography Imaging (PPGI) is a camera-based and non-contact technology for measurement of physiological signals. It has been shown that important physiological parameters such as heart rate, heart rate variability and respiratory rate can be derived from PPGI. However, as is the case with most non-contact measurement techniques, motion artefacts present a major challenge. Various algorithms for application to both the 2D PPGI video frames as well as the resulting 1D PPGI waveforms have been developed in order to enhance robustness against motion. In this paper, we focus on the aspect of feature point tracking in the 2D PPGI video sequences. We present an experimental setup, where we used a motion capture system in order to obtain a reference for motion during the recording of PPGI video sequences. In a laboratory experiment, PPGI video sequences were recorded from ten healthy volunteers, who were asked to perform various movements during the recording. The KLT tracking algorithm was applied to the recorded sequences and results compared with the reference values from the motion capture system. The results indicate, that tracking of measurement regions in PPGI video sequences is only one element towards motion robust PPGI. In most scenarios, tracking is not sufficiently precise, requiring further processing of the PPGI waveforms in order to reduce motion artefacts in PPGI signals. These indications were confirmed by further analysis when we looked into the effects of tracking on PPGI heart rate extraction.

Keywords: Optical imaging, Optical imaging and microscopy - Optical vascular imaging, Optical imaging and microscopy - Neuroimaging
Abstract: Laser speckle contrast imaging (LSCI) is a high-resolution full-field optical technique for measuring blood flow, which has been widely used in clinical and biomedical research. However, most of current LSCI instruments are bulky, limiting their application settings. In this work, we proposed a portable laser speckle imager. Unlike the desktop laser speckle systems that utilize personal computer (PC), our system was designed with embedded GPU system (Jetson TX2, NVIDIA, USA) and a LCD touch screen (16.5×12.4 cm in size, 380 g in weight). In-vivo experiments showed that the portable GPU-based system had comparable performance with our laboratory LSCI system. Such a portable LSCI imager could be potentially used in different applications such as intraoperative monitoring or bedside diagnosis.

Keywords: Neural networks and support vector machines in biosignal processing and classification, Signal pattern classification
Abstract: Automated seizure warning systems can improve the quality of life of epilepsy patients. One option to detect epileptic seizures in a home environment is to use the heart rate. Patient-independent detection algorithms however lead to a low performance due to the large inter-patient variability. Personalized algorithms are required to obtain a better performance, but straight-forward methods are difficult to implement in practice. Therefore, different personalization options for heart rate-based seizure detection are reviewed.

Keywords: Connectivity measurements, Physiological systems modeling - Signal processing in physiological systems, Time-frequency and time-scale analysis - Empirical mode decomposition in biosignal analysis
Abstract: Investigations into brain-heart interactions are gaining importance in epilepsy research. Convergent Cross Mapping (CCM) has been employed to quantify such interactions from 10 min recordings of children with TLE in the past. For evaluation of the interactions in longer recordings an adapted CCM approach is required. This study presents a proof-of-principle application of the adapted CCM to sequential segments of concurrently recorded EEG and HRV from children and adult epilepsy patients.

Keywords: Causality, Physiological systems modeling - Multivariate signal processing, Connectivity measurements
Abstract: We apply information decomposition methods to elicit unique, redundant and synergistic causal contributions of ipsilateral and contralateral EEG activity to heart rate dynamics in epileptic children. We find that information flows from brain to heart according to opposite lateralization effects for the delta and alpha rhythms, suggesting that different neuroautonomic mechanisms take place in the pre- and post-ictal phases of temporal lobe seizures.

Keywords: Physiological systems modeling - Signal processing in physiological systems, Connectivity measurements, Nonlinear dynamic analysis - Biomedical signals
Abstract: Epilepsy patients have 20 times greater risk than the general population of dying suddenly and unexpectedly. The underlying pathophysiological causes of sudden unexpected death in epilepsy (SUDEP) are currently not well-understood. In this study, we analyzed long-term concurrent EEG and ECG recordings in the genetic Kcna1-knockout mouse model of SUDEP to investigate the directional brain->heart and heart->brain functional interactions in SUDEP-prone subjects in periods with and without seizures. Multivariate directional linear analysis of the EEG and ECG in the frequency domain (0-200Hz) showed impairment of the feedback (afferent) branch from the heart to the brain in SUDEP-prone compared to healthy animals. It was also found that it was the same directional branch ECG->EEG in the high frequency spectrum (130-170Hz) that was most affected during seizures in the epileptic animals. These results give credence to a joint analysis of ECG and EEG in the quest of SUDEP’s mechanisms in particular, and ictogenesis in general.

Keywords: Brain-computer/machine interface, Neural signals - Coding, Sensory neuroprostheses - Signal and vision processing
Abstract: We proposed a novel visual stimulus for brain-computer interface (BCI). The stimulus is in the form gaiting sequence of a human which is the action observation (AO) that could activate the mirror neuron system. Different from traditional focus, the response in sensorimotor cortex in AO, this study mainly focused on the response in the occipital cortex. The results showed that the proposed visual gaiting stimulus-induced steady-state motion visual evoked potential, with classification accuracies of 88.9% ± 12.0% in a four-class scenario. Thus, the proposed AO-based BCI provided new insight into rehabilitation.

Keywords: Sensory neuroprostheses - Visual, Brain functional imaging - EEG, Brain functional imaging - Evoked potentials
Abstract: As one determinant of the efficacy of mirror therapy in stroke patients, the perception of embodiment needs to be sustainable and can be enhanced, which requires various strategies. We aimed to explore the effect of combining haptic stimulation and mirror visual feedback (MVF) on the perception of the embodiment. The present experiment showed that the latency time significantly decreased with the integration. Moreover, EEG analysis indicated significant deactivations in C3 regions with the integration. In conclusion, the combination of haptic stimulation and MVF could strengthen the perception of the embodiment, which might be used as a strategy in rehabilitation.

Keywords: Brain-computer/machine interface, Neural signal processing, Neural stimulation
Abstract: Building computational models to explain and mimic complex brain functions is one of the most challenging goals in science and engineering. In this article, we describe a specific form of input-output model of brain functions termed sparse generalized Laguerre-Volterra model. In this approach, input and output signals are spike trains a brain region receives from and sends out to other brain regions. Brain function is defined as its input-output transformational properties that can be represented by a multi-input, multi-output nonlinear dynamical model. Using regularized estimation and basis functions, sparse form of the model can be derived to reduce model complexity and better capture the sparse connectivities in the brain. This approach has been successfully applied to the human hippocampus. The resulting hippocampal CA3-CA1 model accurately predicts the CA1 (output) spike trains based on the ongoing CA3 (input) spike trains and provides a computational basis for developing hippocampal memory prostheses.

Keywords: Brain-computer/machine interface
Abstract: The goal of brain-machine interface (BMI) is to help the disabled people brain control the external device to accomplish a variety of tasks. Most of the existing algorithms are only trained for a single task. When facing a new task, the decoder needs to be re-trained from scratch, which is not efficient. We propose a kernel reinforcement learning algorithm to transfer the weights from the previously well-trained task to a similar but different task. Compared with the same algorithm re-trained from scratch, the proposed algorithm achieves a faster learning speed to the new task on the synthetic neural data.

Keywords: Neurorehabilitation, Brain-computer/machine interface, Neurological disorders - Stroke
Abstract: A variety of adjuvant protocols have been tested to enhance the spontaneous biological recovery in the acute and subacute phases following a stroke. However, no significant benefits have been reported with respect to classic therapeutic interventions. Here I will present a paradigm where a peripheral activation of afferent fibers, with the resulting afferent volley timed to arrive at the peak negativity of the movement-related cortical potential, induces significant cortical plasticity in healthy individuals and chronic stroke patients. The main focus will be on data of a randomized clinical control trial, where this associative brain-computer interface paradigm was applied to sub-acute stroke patients three times per week over a four-week period. Compared to a sham group, the associative intervention group had substantial increases in corticospinal excitability to the target muscle (tibialis anterior) as well as significant improvements in clinical functional scores. Because this intervention can be individually tailored to the patients’ current brain state and all patients were able to perform the intervention with minimal training, this approach is a feasible and effective tool for restoring motor function in stroke patients both in the clinic and at home.

Keywords: Micro- and nano-technology, Biomaterial-cell interactions - Functional biomaterials, Biomaterials - Chemical and electrochemical sensors
Abstract: Glassy carbon (GC) neural electrodes have recently gained visibility thanks to their great resistance to corrosion combined to their ability to record and stimulate neuronal activity. To enhance their electrochemical performance, GC electrodes are often subjected to activation, either through electrical or chemical means. In this study, we have compared the activation of GC electrodes performed using electrical biphasic pulses to chemically-induced activation. Because the GC electrodes used for this research are made by pyrolysing SU-8 photoresist - and thus they undergo massive shrinkage during carbonization - 2 electrode diameters were investigated (300 and 50 µm) with the aim of understanding if their surface composition and their ability to get activated change with their geometry. Chemical activation was induced by immersing the electrodes in 2 solutions: A1 and A2, 30 and 150 mM H2O2/PBS (hydrogen peroxide in phosphate buffered saline) respectively. The comparison between activation methods was done by measuring GC electrodes impedance, charge storage capacity (CSC) and by performing surface analysis, before and after the treatments. Results show that impedance drops in all the cases, especially at low frequencies (< 1 kHz) and that there is always an increase in CSC. Raman spectra and relative intensities of disorder are very similar for both electrode diameters and before and after every treatment. X-Ray photoelectron spectroscopy (XPS) interestingly shows graphite content only on the 300 µm electrodes and a high percentage of graphite only on the pristine one. Apart from oxygen and nitrogen, no other species were present on the electrodes surface. In conclusion, both electrically and chemically-induced activation help improving the electrochemical performance of GC electrodes without harming them.

Keywords: Micro- and nano-technology, Biomaterial-cell interactions - Surface modification of biomaterials
Abstract: The next generation of flexible, electrically active implants, such as brain implants or retina chips require a flexible, biostable as well as biocompatible passivation, ensuring a degradation-free usage for long time periods on the order of several years. Until today, these passivations are prepared mostly by polyimides or parylene, both of which are water vapor permeable to a certain degree. To remedy this deficiency, Atomic Layer Deposited (ALD) thin films are characterized regarding their electrical passivating features under conditions of accelerated aging, such as elevated temperatures in a liquid environment. The initial electrical passivation by various ALD deposited multilayers, combining alternating thin Al2O3 and TiO2 layers is the goal of this research as well as the stability of these layers under induced degradation. Such layers, in combination with a parylene passivation, would ensure a water vapor impermeable and biocompatible coating.

Keywords: Nano-bio technology design, Biomaterial-cell interactions - Functional biomaterials, Biomaterials - Chemical and electrochemical sensors
Abstract: Polymer-derived carbon as neural electrode material has recently been subject of interest due to its ability to act as a multimodal platform for simultaneous recording and stimulation of neural activity and neurotransmitter detection. Its mechanical properties and the inverse fabrication protocol commonly used for the manufacturing of pyrolyzed carbon thin-film devices, however, only allow for its use as the electrode material and not as material for interconnects and other conductive components. In this study for the first time a process to fabricate flexible neural devices entirely made of carbon fibers (CFs) and polyimide was developed. The devices consist of carbonized polyacrylonitrile (PAN) fiber mats embedded in polyimide (PI), which were patterned into the desired shapes using reactive ion etching (RIE). This method allowed for the fabrication of miniaturized, flexible and conductive carbon components with critical dimensions of 12.5 µm. Tensile tests were performed to evaluate the mechanical stability of the CF/PI composite, to detect potential electrical resistance changes due to bending and to study the adhesion of different PI layers onto each other. A strong mechanical interlock between CFs and PI was demonstrated and no significant change in the resistance of the CFs was detected after 100k cycles of tensile bending (r = 3 mm). The fabrication approach proposed here successfully yielded entirely metal-free and entirely flexible electrodes. It opens the door to further studies with the guarantee of highly stable electrodes, both mechanically and electrically.

Keywords: Nano-bio technology design, Micro- and nano-technology
Abstract: Liposomes are amongst the most effective delivery vehicles developed to date. Despite many advantages including biocompatibility, biodegradability, and the ability to carry both hydrophilic and lipophilic compounds, liposomes suffer from low physical stability. This limitation can be effectively addressed by inclusion of a polymeric scaffold within the core of liposomes. Given the versatility of poly(ethylene glycol) (PEG) hydrogels, these polymers have a great potential for the use in liposomal core. As a step towards the development of a robust liposomal delivery platform, here we aim to develop a simple and reliable technique for the fabrication of liposomes with PEG gel cores. We assess the resultant nanoparticles using scanning electron microscopy and dynamic light scattering and demonstrate that the presented approach can successfully produce gel-liposome nanoparticles with spherical shape and 150-200 nm size. These nanoparticles are further evaluated for colloidal stability in physiological solution. Moreover, we demonstrate the versatility of this method by studying the effect of changing (A) the membrane composition in liposomes, and (B) the hydrogel concentration in liposomal core, on the formation of gel-liposome particles.

Keywords: Scaffolds in tissue engineering - Biofabrication
Abstract: Poor proliferation and migration of fibroblast, keratinocyte and endothelial cells delay the wound healing in diabetic patients and results into chronicity of wounds. Slow or decreased formation of blood vessels is another issue that increases the chronicity of non-healing wounds. These chronic wounds turn into an ulcer that may lead to limb amputation. Recently, nitric oxide (NO) has emerged as a potential agent for accelerating cell migration and proliferation to enhance wound healing. It increases the expression of necessary angiogenic growth factors which stimulates the proliferation and migration of major cell types involved in wound repair. Here we report the synthesis of chitosan (CS), polyvinyl alcohol (PVA) and a NO donor S-nitroso-N-acetyl-DL-penicillamine (SNAP) to enhance the wound healing activities in chronic wounds. A three-fold increase in the proliferation of 3T3 cells was observed with NO-releasing CS-PVA hydrogels. In vitro cell migration assay demonstrated a four-fold faster migration of cells to the scratched area compared to the control group. The results depict that the use of CS-PVA hydrogel impregnated with the NO donor (SNAP) can be a promising material for promoting cell migration and subsequent accelerated healing of the chronic wounds in burns and diabetic patients.

Keywords: Scaffolds in tissue engineering - Biofabrication
Abstract: Chronic wound or slow healing of a wound is one of the serious complications in diabetic patients. The decrease in the proliferation and migration of cells such as keratinocytes and fibroblasts is the major reason for the development of such chronic wounds in a diabetic patient. Therefore, designing a wound dressing patch using a biodegradable hydrogel, which can provide a sustained release/delivery of active agents that can support cell proliferation and cell migration, will be highly beneficial for promoting diabetic wound healing. Multiple evidences from both in-vitro and in-vivo studies have shown that graphene oxide (GO) and reduced graphene oxide promote wound healing by promoting migration and proliferation of keratinocyte cells. In addition, GO possesses angiogenic property. Gelatin methacrylate (GelMA) based hydrogels display excellent hydrophilic properties due to the presence of hydrophilic amino, amido, carboxyl, and hydroxyl groups in the polymer chains, which gives them highly porous, soft and flexible structure. In this work, we report the development of hydrogel dressing incorporated with GO to improve wound healing by increasing the proliferation and migration of cells.

Keywords: General and theoretical informatics - Decision support systems, General and theoretical informatics - Predictive analytics, Health Informatics - Personal health systems
Abstract: Cardiac function deterioration of heart failure patients is fre-quently manifested by the occurrence of decompensation events. One relevant step to adequately prevent cardiovascular status degradation is to predict decompensation episodes in order to allow preventive medical interventions. In this paper we introduce a methodology with the goal of finding onsets of worsening progressions from multiple physio-logical parameters which may have predictive value in decom-pensation events. The best performance was obtained for the model composed by only two features using a telemonitoring dataset (myHeart) with 41 patients. Results were achieved by applying leave-one-subject-out validation and correspond to a geometric mean of 83.67%. The obtained performance suggests that the methodology has the potential to be used in decision support solutions and assist in the prevention of this public health burden.

Keywords: General and theoretical informatics - Decision support systems, General and theoretical informatics - Statistical data analysis, Health Informatics - Mobile health
Abstract: mHealth technologies are increasingly utilized in various medical contexts. Mobile crowdsensing is such a technology, which is often used for data collection scenarios related to questions on chronic disorders. One prominent reason for the latter setting is based on the fact that powerful Ecological Momentary Assessments (EMA) can be performed. Notably, when mobile crowdsensing solutions are used to integrate EMA measurements, many new challenges arise. For example, the measurements must be provided in the same way on different mobile operating systems. However, the newly given possibilities can surpass the challenges. For example, if different mobile operating systems must be technically provided, one direction could be to investigate whether users of different mobile operating systems pose a different behaviour when performing EMA measurements. In a previous work, we investigated differences between iOS and Android users from the TrackYourTinnitus mHealth crowdsensing platform, which has the goal to reveal insights on the daily fluctuations of tinnitus patients. In this work, we investigated differences between iOS and Android users from the TrackYourHearing mHealth crowdsensing platform, which aims at insights on the daily fluctuations of patients with hearing loss. We analyzed 3767 EMA measurements based on a daily applied questionnaire of 84 patients. Statistical analyses have been conducted to see whether these 84 patients differ with respect to the used mobile operating system and their given answers to the EMA measurements. We present the obtained results and compare them to the previous mentioned study. Our insights show the differences in the two studies and that the overall results are worth being investigated in a more in-depth manner. Particularly, it must be investigated whether the used mobile operating system constitutes a confounder when gathering EMA-based data through a crowdsensing platform.

Keywords: Health Informatics - Computer-aided decision making, Health Informatics - Decision support methods and systems, Health Informatics - Informatics for chronic disease management
Abstract: This paper explores the use of ensemble classification methods in the context of the diabetes disease. An analysis was carried out that formulates and answers seven research questions: publication trends, channels and venues; medical tasks undertaken; ensemble types proposed; single techniques used to construct the ensemble methods; rules used to draw the output of the ensemble; datasets used to build and evaluate the ensemble methods; and tools used. A total of 107 papers were chosen after a study selection process. Ensemble methods were applied to diabetes in 2003 for the first time. All medical tasks related to the diabetes disease were investigated, and the diagnosis task was the most frequently addressed activity by means of ensemble methods. The homogeneous ensembles were the most common in the literature. Moreover, decision trees and support vector machines were the most used techniques to build homogeneous and heterogeneous ensembles, respectively. The most frequently found combiner was the majority voting rule. Our findings suggest that ensemble classification methods yield better accuracy than single classifiers. This statement, however, requires an aggregation of the evidence reported in the literature by means of a systematic literature review.

Keywords: Health Informatics - Computer-aided decision making, Health Informatics - Decision support methods and systems, General and theoretical informatics - Artificial Intelligence
Abstract: Opioids are the preferred medications for the treatment of pain in the intensive care unit. While undertreatment leads to unrelieved pain and poor clinical outcomes, excessive use of opioids puts patients at risk of experiencing multiple adverse effects. In this work, we present a sequential decision making framework for opioid dosing based on deep reinforcement learning. It provides real-time clinically interpretable dosing recommendations, personalized according to each patient’s evolving pain and physiological condition. We focus on morphine, one of the most commonly prescribed opioids. To train and evaluate the model, we used retrospective data from the publicly available MIMIC-3 database. Our results demonstrate that reinforcement learning may be used to aid decision making in the intensive care setting by providing personalized pain management interventions.

Keywords: Health Informatics - Decision support methods and systems, General and theoretical informatics - Machine learning, General and theoretical informatics - Algorithms
Abstract: Sleep apnea syndrome (SAS) is a prevalent disorder which causes daytime fatigue with the increased risk of lifestyle diseases. A large number of patients are undiagnosed and untreated partly because of the difficulty in performing its gold standard test, polysomnography (PSG). In this research, we propose a simple screening method utilizing heart rate variability (HRV) and long short-term memory (LSTM) which is a kind of neural network techniques. The result of applying this algorithm to clinical data demonstrates that it can discriminate between patients and healthy people with high sensitivity (100%) and specificity (100%).

Keywords: Health Informatics - Clinical information systems, Health Informatics - Computer-aided decision making, Health Informatics - Decision support methods and systems
Abstract: Parkinson’s Disease (PD) is the second most prevalent progressive neurological disorder around the world with high incidence rates for seniors. Since most symptoms are exposed in the later stages of the disease, early diagnosis of PD is essential for more effective treatment. The motivation of this research is early automatic assessment of PD using clinical information, not only for disease diagnosis but also for monitoring progression. After preprocessing the data, feature selection is done by the Mean Decrease Impurity (MDI) method. In the classification step, Random Forest (RF) is used as a classifier model for two tasks, including (1) classifying the subjects to PD and Healthy Control (HC), and (2) determining the disease severity level by Hoehn & Yahr (H&Y) scale. The clinical data used is taken from the Parkinson’s Progression Markers Initiative (PPMI) database, which is the most prominent source of data for PD. Experimental results show promising performance of the proposed model for assessment of PD by incorporating clinical properties.

Keywords: Data-driven modeling, Models of organ physiology, Systems modeling - Decision making
Abstract: Estimating cardiac output (CO) and thoracic fluid content (TFC) by non-invasive measurement of thoracic electrical bioimpedance (TEB) is on the rise of becoming clinically established. Dynamic fluid management and detecting and trending excess quasi-static thoracic fluids are of particular interest to benefit the critically ill patient. While the advantages such as easy application and non-invasive assessment are intriguing, there are some challenges. In addition to artifacts due to the patient’s movement, instability of the electrode-skin interfaces, the accuracy of the applied current and varying cable capacity because of motion, exact placement of electrodes - or lack thereof - must be considered. In particular the robustness of the electrode placement, i.e., the insensitivity to inaccuracy of electrode placement (actual sensor from the nominal). We propose a new technique for evaluation of electrode placement by simulation with MRI based human phantom with FEM based quasi-static solver for bioimpedance measurement. Results: We identified alternative electrode positions which lead to an increase in robustness of bioimpedance measurements of up to 9.5 times compared to the standard electrode placement. Simulations suggested an improvement in ECG signal quality which was confirmed by a subject measurement.

Keywords: Data-driven modeling, Models of organ physiology, Systems modeling - Decision making
Abstract: Transthoracic electrical bioimpedance (TEB) measurement for acquiring of hemodynamic parameters e.g. stroke volume (SV) and cardiac output (CO) becomes commonly used. For precise and reliable measurement, a good understanding of the measurement system is needed to provide robust electrode placement, reproducible results, and large signal amplitudes. We propose an evaluation by MRI based human phantoms with FEM based quasi-static solver for electrode placement in bioimpedance measurement. Placements according to Osypka, Cheetah and Bernstein et al. were simulated and compared with measurements taken from a (real) human subject. As a parameter for evaluation of signal quality, the percentage of current passing through the descending aorta from the overall injected current is measured. The placement according to Osypka results in 1.46 % of current flow through the descending aorta (Cheetah placement 1.12 %, Bernstein et al. placement 0.877 %) for the male human phantom DUKE. The simulation was compared with a real human subject (with comparable age, height, weight) by calculating the baseline impedance (Z0). The simulation seems to fit at best with Osypka placement, were the deviation between simulation baseline impedance and real human subject is 19.9 % (deviation for Cheetah 34.1 %, deviation for Bernstein 62.5 %). The simulation with MRI based human phantoms seems to be a very good basis for further investigations of current injection and bioimpedance measurement comprehension.

Keywords: Organ modeling, Organs and medical devices - Multiscale modeling and the physiome
Abstract: In Western countries, stroke is the third-most widespread cause of death. 80% of all strokes are ischemic and show a mortality rate of about 25 %. Furthermore, 35-55% of affected patients retain a permanent disability. Therapeutic hypothermia (TH) could decrease inflammatory processes and the stroke-induced cerebral damage. Currently, the standard technique to induce TH is cooling of the whole body, which can cause several side effects. A novel cooling sheath uses intracarotid blood cooling to induce local TH. Unfortunately, the control of the temporal and spatial cerebral temperature course requires invasive temperature measurements. Computational modeling could be used to predict the resulting temperature courses instead. In this work, a detailed 1D hemodynamics model of the cerebral arterial system was coupled with an energetic temperature model. For physiological conditions, 50% and 100% M1-stenoses, the temperatures in the supply area of the middle cerebral artery (MCA) and of the systemic body was analyzed. A 2K temperature decrease was reached within 10 min of cooling for physiological conditions and 50% stenosis. For 100% stenosis, a significant lower cooling effect was observed, resulting in a maximum cerebral temperature decrease of 0.7K after 30 min of cooling. A significant influence of collateral flow rates on the cooling effect was observed. However, regardless of the stenosis degree, the temperature decrease was strongest within the first 20 min of cooling, which demonstrates the fast and effective impact of intra-carotid blood cooling.

Keywords: Models of medical devices, Organs and medical devices - Multiscale modeling and the physiome, Models of organs and medical devices - Inverse problems in biology
Abstract: A new miniaturized figure-of-eight coil (µCoil) for TMS applications has been developed taking advantage of the Flex circuit technology. First experiments on volunteers demonstrated the ability of the µCoil to elicit sensorial action potentials of the peripheral nervous system. The necessity of reducing the size of standard TMS stimulator arises from the poor spatial resolution of the latter. This study aims to model the µCoil and study the electromagnetic fields induced inside the arm during peripheral nerve stimulation. Results confirmed that the µCoil is capable of inducing a focalized electric field inside the tissues with amplitudes up to 70V/m consistent with the observed peripheral nerve stimulation in healthy volunteers.

Keywords: Models of medical devices
Abstract: In vitro and in vivo evaluation of magnetic nanoparticles in relation to magnetic fluid hyperthermia (MFH) treatment is an on-going quest. This current paper demonstrates the design, fabrication, and evaluation of an in vivo coil setup for real-time, whole body thermal imaging. Numerical calculations estimating the flux densities, and in silico analysis suggest that the proposed in vivo coil setup could be used for real-time thermal imaging during MFH experiments (within the limitations due to issues of penetration depth). Such in silico evaluations provide insights into the design of suitable AMF applicators for AC magnetic field-mediated in vivo MNP heating as demonstrated in this study.

Keywords: Modeling of cell, tissue, and regenerative medicine - Cells, Modeling of cell, tissue, and regenerative medicine - Ionic modeling
Abstract: Due to the inevitable drawbacks of the implantable electrical pacemaker, the biological pacemaker was believed to be an alternative therapy for heart failure. Previous experimental studies have shown that biological pacemaker could be produced by genetically manipulating non-pacemaking cardiac cells by suppressing the inward rectifier potassium current (IK1) and expressing the hyperpolarization- activated current (If). However, the role of If in such bio-engineered pacemaker is not clear. In this study, we simulated the action potential of biological pacemaker cells by manipulating If-IK1 parameters (i.e., inhibiting IK1 as well as incorporating If) to analyze possible mechanisms by which different If densities control pacemaking action potentials. Our simulation results showed different pacing mechanism between the bioengineered pacemaking cells with and without If. In addition, it was shown that a greater If density might result in a slower pacing frequency, and excessive of it might produce an early-afterdepolarizations-like action potential due to a sudden release of calcium from sarcoplasmic reticulum into the cytoplasm. This study indicated that when IK1 was significantly suppressed, incorporating If may not enhance the pacing ability of biological pacemaker, but lead to abnormal dynamics of intracellular ionic concentration, increasing risks of dysrhythmia in the heart.

Keywords: Medical devices interfacing with the brain or nerves
Abstract: Identifying the target speaker in hearing aid applications is an essential ingredient to improve speech intelligibility. To identify the target speaker from single-trial EEG recordings in an acoustic scenario with two competing speakers, a least-squares-based auditory attention decoding (AAD) method has been recently proposed. While the performance of this method has been mainly studied for noiseless and anechoic acoustic conditions, it is important to fully understand its performance in realistic noisy and reverberant acoustic conditions. In this contribution, we investigate AAD for different acoustic conditions (anechoic, reverberant, noisy, and reverberant-noisy). Furthermore, aiming at enhancing the target speaker in a noisy and reverberant environment with two competing speakers, we propose a cognitive-driven speech enhancement system based on binaural beamforming.

Keywords: Cochlear implant, Clinical engineering, Medical devices interfacing with the brain or nerves
Abstract: Although cochlear implants are a great success story, there is still room for improvement for a better acoustical representation. We propose a closed loop system that enables to adapt the stimulation pattern for each patient individually. The pattern is iteratively evaluated using electrophysiological measures.

Keywords: Cochlear implant
Abstract: In cochlear implants (CIs), different signal processing techniques are used at various stages of a CI user’s hearing journey. Auditory nerve responses such as the cochlear microphonic are recorded during surgery, and electrically compound action potentials during and after surgery help with fitting. Each requires different techniques in terms of noise reduction and artifact rejection. For electrical hearing, a sound coding strategy transforms the incoming microphone signals to a series of electrical pulses to drive the intra-cochlear electrodes hence representing the frequency and energy of the audio signal. Additionally, techniques such as directional microphones exploit the spatial distribution of sound sources while other speech enhancement schemes remove as much of the interfering background noise as possible without introducing excessive speech distortion. Both approaches substantially reduce background noise, leading to improved speech perception.

Keywords: Cochlear implant
Abstract: This paper mentions the major trends in cochlear implants today from the author's perspective. Further, two recent approaches – individualization of cochlear implant treatment and incorporation of top-down processing in cochlear implant systems – to further improve user performance with cochlear implants are briefly discussed.

Keywords: Medical devices interfacing with the brain or nerves
Abstract: The integration of brain sensing capabilities into hearing devices promises to improve the acceptance of hearing devices. EEG (electroencephalography) sensing using behind-the-ear or in-ear electrodes can help to improve our understanding of the neural processes underlying aided hearing in every life situations. We combine hearing devices with behind-the-ear EEG to study auditory processing. The long-term goal is to have a brain-feedback adaptive hearing device.

Keywords: Cochlear implant, Medical devices interfacing with the brain or nerves, Neural stimulation (including deep brain stimulation)
Abstract: Electrochemical microsensors based on noble metals can give vital information on their microenvironment in high spatio-temporal resolution. Our approach is to utilize unmodified noble metal electrodes (Pt, Pt/Ir) already present in many neural implants as in situ chemical sensors at the electrode/tissue interface, e.g. in the cochlear implant. We developed a new chronoamperometric/potentiometric sensor method to measure oxygen, redox active species and electrode potential. Long-term stable sensor performance was demonstrated in the presence of proteins.

Keywords: Cochlear implant, Neural stimulation (including deep brain stimulation), Neuromodulation devices
Abstract: Cochlear implants (CIs) are often highly effective at enabling many deaf patients to understand speech, but they fail to restore normal hearing capabilities, including sensitivity to fine interaural time differences (ITDs) for spatial hearing. A lack of early binaural experience is often blamed for the inability of CI users to hear ITDs. However, recently we were able to show that neonatally deafened animals without any early hearing experience localize ITDs perfectly well when given microsecond synchronized binaural electrical stimuli from the beginning of CI stimulation. The representation of ITD remains present in the inferior colliculus and primary auditory cortex in the early deaf animals. This suggests that improved ITD perception in bilateral CI patients may require a new generation of CIs delivering precisely synchronized bilateral input.

Keywords: Neural stimulation (including deep brain stimulation), Medical devices interfacing with the brain or nerves, Neuromodulation devices
Abstract: One major goal of my research is to treat deafness and to provide hearing solutions that can overcome performance limitations associated with cochlear implants and hearing aids. My lab is developing two new types of implantable auditory prostheses that directly target the auditory nerve or the auditory midbrain with penetrating electrode arrays to achieve a greater transmission of frequency channels of information required for more natural hearing compared to cochlear implants. Each project involves pre-clinical and clinical studies towards implanting 3 to 5 patients. More recently, we discovered the ability to apply ultrasound stimulation to the head to vibrate the brain tissue and fluids, which in turn can noninvasively vibrate the cochlea and activate the ascending auditory pathway. This ultrasound hearing approach could potentially improve hearing aid performance by overcoming acoustic feedback issues and degraded transmission in patients with damaged outer or middle ears. An ultrasound-based hearing device can also be placed on the head without occluding the ear canal, opening up opportunities for more comfortable and versatile hearing aid and consumer audio technologies. In addition to hearing loss applications, my lab is developing wearable neuromodulation technologies to treat neurological disorders, such as tinnitus. An emerging field of noninvasive neuromodulation is the use of electrical stimulation of peripheral nerves, such as the trigeminal nerve, to modulate neural networks and drive therapeutic plasticity. Through animal research in my lab and large-scale clinical trials (>500 patients) sponsored by a start-up company (Neuromod Devices), my colleagues and I have demonstrated the ability to electrically stimulate the tongue and trigeminal pathways combined with customized sound stimulation to treat tinnitus, which is a phantom auditory sensation that is bothersome or debilitating for about 5% of the population.

Keywords: Sleep - Periodic breathing & central apnea, Sleep - Snoring, Cardiovascular and respiratory signal processing - Cardiovascular signal processing
Abstract: Sleep-disordered breathing (SDB) is a health problem that includes different respiratory issues, ranging from simple snoring, through prolonged partial obstruction (PPO), to obstructive sleep apnea (OSA). New materials and technologies have led to the evolution of unobtrusive sensors such as electromechanical film transducer (Emfit) mattress, to complement the polysomnography in clinical and ambulatory environments. However, the Emfit signal has inherent problems, mainly related to the sensitivity to noise and high variability. We have researched state of the art signal processing methods to infer relevant clinical information from the Emfit sensor. Signal characteristics of the Emfit signal in SDB and its most bothersome symptom, snoring, haven been studied. We have used this knowledge to design methods and machine learning techniques to detect SDB events and epochs, improving overall sensitivity in SDB diagnostics.

Keywords: Sleep - Obstructive sleep apnea, Sleep - Sleep apnea therapy, Cardiovascular, respiratory, and sleep devices - Therapeutics
Abstract: Digital health applications offer the promise of improved adherence to therapies for chronic respiratory disorders such as sleep apnea. This paper reports on insights gleaned from a database analysis of telemonitored therapy devices and a small pilot study using a digital health application for insomnia and sleep apnea.

Keywords: Sleep - Obstructive sleep apnea
Abstract: Continuous quantitative monitoring of tidal volume and respiration rate has been tried using time-difference electrical impedance tomography (EIT) imaging of the chest. Computing a minute ventilation signal from the EIT-derived tidal volume signal and respiration rate, we suggest the chest EIT imaging method for continuous hypoventilation monitoring during polysomnography or home sleep tests. In this paper, we report the results of its feasibility study through animal experiments. We collected chest EIT data simultaneously with EtCO2 and SpO2 signals from eight mechanically ventilated pigs. Real-time EIT images were reconstructed and processed to produce tidal volume, respiration rate and minute ventilation signals. The minute ventilation signal accurately and immediately quantified changes in the amount of supplied air volume from the mechanical ventilator. The EtCO2 signal showed exponential increases in response to reduced air volumes. The oxygen desaturation response observed from the SpO2 signal was quite slow unless the amount of the supplied air volume was reduced down to 40 to 30% of its normal value. We propose future studies of clinical trials.

Keywords: Sleep - Cardiovascular & Metabolic consequences of sleep disorders, Sleep - Obstructive sleep apnea
Abstract: Large sleep datasets, data linkage, modern signal processing and deep learning methods provide an exciting opportunity to predict much more than the AHI from polysomnography studies. These methods are potential enablers for P4 medicine approaches to sleep disordered breathing management.

Keywords: Magnetic resonance imaging - Cardiac imaging
Abstract: Cardiac Magnetic Resonance Imaging (CMR) is a central tool for diagnosis of various ischemic and non-ischemic cardiomyopathies. CMR protocols commonly comprise assessment of functional properties using cardiac phase-resolved CINE MRI and characterization of myocardial viability using late gadolinium enhancement (LGE) imaging. Conventional LGE imaging requires inversion recovery preparation with a specific inversion time to null the healthy myocardium, which restricts the acquisition to a single cardiac phase. In turn, this necessitates separate scans for cardiac function and viability. In this work, we develop a new method for functional LGE imaging in a single breath-hold using a three-step approach: 1) ECG-triggered multi-contrast data is acquired for each cardiac phase, 2) semi-quantitative relaxation maps are generated, 3) LGE imaging contrast is synthesized based on the semi-quantitative maps. The proposed functional LGE method is evaluated in four healthy subject and 20 patients at 1.5T and 3T. Thorough suppression of the healthy myocardium, as well as 40-80ms temporal resolution are achieved, with no visually apparent temporal blurring at tissue interfaces. Functional LGE in patients with focal scar demonstrates robust hyperenhancement in the scar area throughout all cardiac phases, allowing for visual assessment of scar motility. The proposed technique bears the potential to simplify and speed-up common cardiac imaging protocols, while enabling improved data fusion of functional and viability information for improved evaluation of CMR.

Keywords: Image reconstruction - Fast algorithms, Image registration, segmentation, compression and visualization - Volume rendering, Magnetic resonance imaging - Other organs
Abstract: Dynamic MRI has made it possible to non-invasively capture the moving human joints in vivo. Real-time Fast Field Echo (FFE) sequences have the potential to reduce the effect of motion artifacts by acquiring the image data within a few milliseconds. However, the short acquisition times affect the temporal resolution of the acquired sequences. In this paper, we propose a post-processing technique to reconstruct the missing frames of the sequence given the reduced amount of acquired data, which leads to recover the entire joint trajectory outside the MR scanner. To do this, we generalize the Log-Euclidean polyrigid registration framework to deal with dynamic three-dimensional articulated structures by adding the time as fourth dimension : we first estimate the rigid motion of each bone from the acquired data using linear intensity-based registration. Then, we fuse these local transformations to compute the non-linear joint deformations between successive images using a temporal log-euclidean polyrigid framework. The idea is to reconstruct the missing time frames by interpolating the realistic joint deformation fields in the domain of matrix logarithms assuming the motion to be consistent over a short period of time. The algorithm has been applied and validated using dynamic data from five children performing passive ankle dorsi-plantar flexion.

Keywords: Magnetic resonance imaging - Cardiac imaging, Regularized image Reconstruction, Image reconstruction and enhancement - Compressed sensing / Sampling
Abstract: Mapping the longitudinal relaxation time constant (T1) of the myocardium using Magnetic Resonance Imaging (MRI) is an emerging technique for quantitative assessment of the morphology and viability of the myocardium. However, three-dimensional (3D) T1 mapping of the heart is challenging due to the high dimensionality of the signal and the presence of cardiac and respiratory motions. We propose a subspace-based method for free-breathing 3D T1 mapping of the heart without respiratory gating. The image function is represented as a high-order partially separable (PS) function to explore the inherent spatiotemporal correlations of the underlying signal. A special data acquisition scheme enabled by the high-order PS model is used for sparse sampling of the (k,t)-space, where complementary sparse datasets are acquired, each covering only a small portion of the (k,t)-space to characterize a single subspace (spatial or temporal). High-resolution dynamic MR images are reconstructed from the highly undersampled (k,t)-space using low-rank tensor and sparsity constraints. We demonstrate the feasibility of our proposed method using in vivo data obtained from healthy subjects on a 3T MR scanner. The proposed method can enable new clinical applications of T1 mapping in cardiac MR.

Keywords: Magnetic resonance imaging - Other organs, Cardiac imaging and image analysis, Image segmentation
Abstract: Objective: We investigate the basic feasibility of estimating the brachial artery area-pressure relationship from MRI data obtained during pressure cuff inflations in-vivo. Methods: We acquired cross-sectional real-time MR images and cardiac-gated CINE MR images from the upper arm of a single male subject at rest during supra-systolic pressure cuff inflations and deflations. We estimate from the MR images the lumen area changes of the brachial artery, and, simultaneously, from the cuff pressure the systemic blood pressure of the subject. We reconstruct the area-pressure curve from two real-time and three CINE independent measurements. Results: The area-pressure curve can be reconstructed, and it is plausible and appears largely consistent with the literature using other methods. Conclusion: MR imaging during pressure cuff inflations is an easy to use, non-invasive candidate method to estimate the brachial artery pressure-area curve.

Keywords: Magnetic resonance imaging - Cardiac imaging, Image segmentation, Cardiac imaging and image analysis
Abstract: Cardiac segmentation is the first most important step in assessing cardiac diseases. However, it still remains challenging owing to the complicated information of myocardium’s boundary. In this work, we investigate approaches based on deep learning for fully automatic segmentation of the left ventricular (LV) endocardium using cardiac magnetic resonance (CMR) images. The deep convolutional neural network architectures, specifically, GoogleNet and U-Net, are modified and deployed to extract the features and then classify each pixel into either endocardium or background. Since adjacent frames for a given slice are imaged over a short time period across a cardiac cycle, the LV endocardium exhibit strong temporal correlation. To utilize the temporal information of heart motion to assist segmentation, we propose to construct multi-channel cardiac images by combining adjacent frames together with the current frame, which are used as the inputs for deep learning models. This allows the deep learning models to automatically learn spatial and temporal information. The performance of our constructed networks is evaluated by using the Dice metric to compare the segmented areas with the manually segmented ground truth. The experiments show that the multi-channel approaches converge more rapidly and achieve higher segmentation accuracy compared to the single channel approach.

Keywords: Magnetic resonance imaging - Cardiac imaging, Image segmentation, Image registration, segmentation, compression and visualization - Machine learning / Deep learning approaches
Abstract: The assessment of right ventricular (RV) function is essential in the diagnosis of many cardiac diseases. Magnetic resonance imaging (MRI) offers an excellent solution to image right ventricle non-invasively with high contrast and temporal resolution. Manual assessment of the RV function from MRI sequences is tedious and time-consuming and automating the process is of great interest. This study proposes a convolutional neural network-based machine learning approach to automate the delineation of the RV from a sequence of MRI. The architecture of the neural network differs from that of a widely-known U-Net approach. Additionally, the proposed approach used image concatenation to create and utilize 3D spatial information in the segmentation process. Quantitative evaluations were performed over 256 images acquired from 16 patients in publicly available data in comparison to manual delineations. Comparisons with the results by U-Net demonstrated that the proposed method outperforms the prior state-of-the-art method.

Keywords: Bio-electric sensors - Sensing methods, Implantable sensors, Integrated sensor systems
Abstract: This paper presents a design and implementation of an ultrasonic wireless communication link for an injectable biomedical implanted device. The results address how the ultrasound link encounter from the multiple paths propagation effect. The ultrasound link characterized in term of channel impulse response and power transmission losses against the depth of the implant, the achieved data transmission rate was 70 Kbps and the signal to noise ratio was (30, 35 and 47) dB at a transmission voltage of (1.8, 3.3 and 20) V peak to peak in 12 cm depth. The transmission loss increases as the depth of the implant increases. The ultrasound link represented by two piezoelectric transducers that operate in 320 KHz radial resonance frequency.

Keywords: Wearable antennas and in-body communications, Implantable technologies, New sensing techniques
Abstract: In this study, we investigated a highly hydrated gel phantom with electrical anisotropy that can be used at 18.375 MHz to 23.625 MHz. This is one of the frequency bands used for human body communication. To achieve the communication, the electrical characteristics of the quadriceps femoris muscle of the rat were measured immediately after sacrifice. These were used to obtain an indicator of electrical characteristics to be satisfied by the phantom. Electrical anisotropy was realized by adding carbon fiber to the phantom and controlling its direction. We were able to develop a high hydrated gel phantom for human body communication with a maximum error of 8.1% assuming its use at 18.735 MHz to 23.625 MHz.

Keywords: Wearable antennas and in-body communications, Implantable systems
Abstract: For wireless capsule endoscopy, high quality images need to be transmitted from inside the digestive tract to an on-body receiver. Ultra wideband transmission offers the possibility to achieve much larger data rates than achievable with today's technology. To design such an ultra wideband transmission system a comprehensible channel model is needed for simulation of the propagation behavior through the human abdomen. In this paper we present a stochastic channel model, that includes the variation of the radio propagation depending on the location of a receive antenna on the body as well as on the physiological properties of different human body models.

Keywords: Wearable body sensor networks and telemetric systems, Integrated sensor systems, Physiological monitoring - Instrumentation
Abstract: This paper presents a self-synchronizing, low-power, low-complexity body-coupled communication (BCC) transceiver using the recently proposed Pulsed-Index Communication (PIC) techniques. The unique features of these techniques are used to simplify the BCC transceiver hardware and reduce its power consumption by eliminating the need for circuitries dedicated to clock and data recovery (CDR) and to duty cycle correction. The self-synchronizing feature of the transceiver is achieved by exploiting the edge-coding property of PIC which consists of using pulse edges for encoding and detecting transmitted pulses rather thanbit times or duty cycles. A working prototype of the proposed BCC transceiver using off-the-shelf components is developed and used to test, for the first time, a full, bi-directional BCC link by transmitting arbitrary 16-bit data words through the human body over a range of 150cm with zero bit-error rate and sub-1nJ/bit energy efficiency.

Keywords: Wearable low power, wireless sensing methods
Abstract: Human Body Communication (HBC) is an attractive low complexity technology with promising applications in wearable biomedical sensors. In this paper, a simple parametric model based on the finite-element method (FEM) using a full human body model is developed to virtually emulate and examine the HBC channel. FEM allows better modeling and quantification of the underlying physical phenomena including the impact of the human body for the desired applications. By adjusting the parameters of the model, a good match with the limited measurement results in the literature is observed. Having a flexible and customizable simulation platform could be very helpful to better understand the communication medium for capacitively coupled electrodes in HBC. This knowledge, in turn, leads to better transceiver design for given applications. The platform presented here can also be extended to study communication channel characteristics when the HBC mechanism is used by an implant device.

Keywords: Wearable body sensor networks and telemetric systems, Wearable sensor systems - User centered design and applications, Wearable wireless sensors, motes and systems
Abstract: Human body communication (HBC) has become one of the most energy-efficient candidates for wireless body area network (WBAN) as it uses higher conductivity of human body as transmission media to reduce transmission loss. The use of medical electrodes instead of bulky antennas makes it suitable for biomedical sensors. The IEEE 802.15.6 standard has reserved only one communication channel for HBC centering at 21 MHz with 5.25 MHz bandwidth, which enlarges the probability of the collision and interference when multiple wearable devices locate on the human body. As a result, the error vector magnitude (EVM) of HBC will be affected by the various states of the interference sources. However, none of the previous works has studied the effects of interferences in HBC. In this paper, the interference factors in HBC WBAN systems assisted by actual measurement on human body are investigated. The HBC signal EVM under different interference states are studied, including frequency, power, modulation scheme, symbol rate, and apart distance between interference source and receiver of HBC. Based on the result and analysis, we propose a possible communication scheme for other body nodes to avoid the collision with HBC.

Keywords: Physiological systems modeling - Multivariate signal processing, Causality, Coupling and synchronization - Nonlinear coupling
Abstract: A causal algorithmic framework quantifying cross-channel phase-amplitude transfer entropy was proposed to measure long-range transmission dynamics between frontal and occipital brain areas during sleep. To this end, a noise-assisted multivariate empirical mode decomposition method was used to guarantee the consistent scales across multivariate signals. On the other side, transfer entropy was applied to measure information transfers from a low-frequency phase to a high-frequency amplitude across different brain regions. Our results showed δ phase may modulate either θ or α amplitude. The frontal cortex transferred information to the occipital brain area more than its inverse direction during Awake and N3 sleep stages, whereas N1 was more likely of serving as a transition state. On the other side, the information flow transferred from the occipital area to the frontal cortex surpassed its inverse flow in the N2 sleep stage. The proposed causal algorithmic framework facilitated identifying information flow and driving force across brain regions in sleep.

Keywords: Physiological systems modeling - Closed loop systems, Nonlinear dynamic analysis - Biomedical signals, Physiological systems modeling - Signal processing in physiological systems
Abstract: Recent evidence has shown that enhancing slow-wave activity (SWA) during sleep has positive effects on cognitive, metabolic, and autonomic function. We have developed a consumer, integrated device that automatically detects sleep stages from a single electroencephalogram (EEG) signal and delivers auditory stimulation in a closed-loop manner. The stimulation was delivered in 15-auditory tone blocks separated from each other by at least 15 seconds. The first tone in a block was synchronized to the up-state of a detected slow-wave while subsequent ones were separated from each other by a constant 1-second inter-tone interval. The system was tested in a study involving 22 participants and SWA enhancement (average 45.8%; p=0.0027) was found in 19/22 participants.

Keywords: Data mining and processing in biosignals, Time-frequency and time-scale analysis - Wavelets, Signal pattern classification
Abstract: Typically, two symmetrical EEG channels are recorded during polysomnography (PSG). As a rule, only the recommended channel is used for sleep stage scoring or sleep apnea detection, and the other for backup. Concurrently, there are many works demonstrating the asymmetry in brain activity. The aim of this work was to compare the accuracy of sleep apnea detection with the use of features obtained from one (C3-A2 or C4-A1) versus these two symmetrical EEG channels. To this end, the relevant data from the PhysioBank database (25 whole-night PSGs) were used. The same methodology of feature extraction and selection was applied for one and combined EEG channels. Automated classification was performed using the k-nearest neighbors algorithm (kNN) with k = 12 and cityblock metric for the three classes of EEG epochs, representing normal breathing, obstructive apnea and hypopnea, and central apnea and hypopnea. The accuracy of kNN-based classification was 63.8 %, 64.3 % and 70.3 % for C3-A2, C4-A1 and both EEG channels, respectively. The statistical tests have indicated that the accuracy of classification based on two combined symmetrical EEG channels is significantly higher compared to the single-channel cases.

Keywords: Signal pattern classification, Data mining and processing in biosignals, Data mining and processing - Pattern recognition
Abstract: This paper presents a method for classification of microsleep (MS) from baseline utilizing linear and non-linear features derived from electroencephalography (EEG), which is recorded from five brain regions: frontal, central, parietal, occipital, and temporal. The EEG is acquired from sixteen commercially-rated pilots during the window of circadian low (2:00 am-6:00 am). MS events are annotated using the Driver Monitoring System and further verified using electrooculogram. A total of 55 features are extracted from EEG. A subset of these features are then selected using a wrapper-based method. The selected features are fed into a linear or quadratic discriminant analysis (LDA or QDA) classifier to automatically differentiate baseline from MS states. The overall classification performance of the best-proposed algorithm is 87.11% in terms of F1 score. This preliminary result highlights the potential of proposed method towards automatic drowsiness detection which could assist mitigating aviation accidents in the future, pending hardware development to record such EEG signals from the confines of the aviation headset.

Keywords: Time-frequency and time-scale analysis - Empirical mode decomposition in biosignal analysis
Abstract: In this paper we propose a novel methodology for investigating pathological sleep patterns through network neuroscience approaches. It consists of initial identification of statistically significant alterations in cortical functional connectivity patterns. The resulting sub-network is then analyzed by employing graph theory for estimating both global performance metrics (integration and specialization) as well as the significance of specific network nodes and their hierarchical organization. So, nodes with important role in network structure are recognized and their functionality is correlated with adenosine biomarker which is important in sleep regulation and promotion. The aforementioned pipeline is applied in a dataset of sleep data gathered during a microgravity simulation experiment. The analysis was performed on cortical resting-state networks involved in sleep physiology. It demonstrated the detrimental effects of microgravity which were more prominent for the group which did not perform reactive sledge jumps as a countermeasure.

Keywords: Image classification, Image analysis and classification - Digital Pathology
Abstract: Specialists may gauge the severity of sickle cell disease crisis by quantifying the number of abnormal-looking and sickle-shaped erythrocytes in blood smears. State-of-the-art integral geometry-based descriptors for automatic classification of erythrocytes as normal cells, sickle cells or cells with other deformations have achieved excellent results. Unfortunately, they are computationally expensive, requiring powerful desktop computers and a great deal of memory to run. We propose two new integral geometry-based descriptors for the shape of erythrocytes. Like state-of-the-art techniques, the overall sensitivity of our solutions is above 94%. Nevertheless, our descriptors are designed to avoid a great amount of computation in comparison to similar solutions and to present a lower memory footprint. Our descriptors offer a high specificity of normal cells and a high sensitivity of deformed cells, making them a good alternative in clinical applications.