Keywords: Neural signal processing, Human performance, Human performance - Cognition
Abstract: We propose a method for the automatic detection of mismatched feelings that occur in communication. As our first step, we examined the semantically anomalous feelings from EEGs when participants listened to spoken sentences. Previous studies have shown that the event-related potentials (ERP) of an electroencephalogram (EEG) are evoked in the auditory and visual modalities where a semantic anomaly occurs. We expand this knowledge and detect it from a single-trial ERP using machine learning techniques. We recorded the brain activity of eight participants as they listened to sentences that contained semantic anomalies and found that a combination of feature selection using linear discriminant analysis and linear kernel support vector machines achieved the highest accuracy that exceeded 60%. By applying this technique, we plan to detect other types of anomalies in practical situations.

Keywords: Neural signal processing, Motor learning, neural control, and neuromuscular systems, Brain-computer/machine interface
Abstract: Mirror neurons, which fire during both the execution and observation of movement, are believed to play an important role in motor processing and learning. However, much work still remains to understand the similarities and differences in how these neurons compute in the motor cortex during movement execution and observation. Here, we performed experiments where a monkey both executes and observes a center-out-and-back task within the same experimental session. By recording from putatively the same neural population, we were able to analyze and compare single neuron statistics between movement execution and observation. We found that a majority of neurons in the primary motor cortex (M1) and dorsal premotor cortex (PMd) have statistically different firing rate statistics between movement execution and observation. As a result of this difference, we then wondered if neurons during movement observation exhibited a similar characteristic to those during movement execution: changing of preferred directions as a function of movement speed. Interestingly, we found that while observed movement speed is encoded in the neural population, it only alters a small proportion of the neuron’s firing rate statistics. These results suggest that neural populations in M1 and PMd process information related to movement differently between execution and observation.

Keywords: Neural signal processing, Human performance - Cognition
Abstract: To gain knowledge of listening effort in adverse situations, it is important to know how the brain processes speech with different signal-to-noise ratios (SNR). To investigate this, we conducted a study with 33 hearing impaired individuals, whose electroencephalographic (EEG) signals were recorded while listening to sentences presented in high and low levels of background noise. To discriminate between these two conditions, features from the 64-channel EEG recordings were extracted using the power spectrum obtained by a Fast Fourier Transform. Features vectors were selected on an individual basis by using the statistical R2 approach. The selected features were then classified by a Support Vector Machine with a non-linear kernel, and the classification results were validated using a leave-one-out strategy, and presented an average classification accuracy over all 33 subjects of 83% (SD=6.4%). The most discriminative features were selected in the high-beta (19–30 Hz) and gamma (30-45 Hz) bands. These results suggest that specific brain oscillations are involved in addressing background noise during speech stimuli, which may reflect differences in cognitive load between the conditions of low and high background noise.

Keywords: Neural signal processing, Human performance - Sleep, Neurological disorders - Epilepsy
Abstract: Recent studies show that the rate of cortical high frequency oscillations (HFOs) differentiates epileptogenic tissue in individuals with epilepsy. However, HFO occurrence can vary widely with vigilance state. In this study we attempt to characterize this variation, which has implications for the choice of a suitable diagnostic baseline for spatiotemporal analysis of HFO activity. We analyzed simultaneous recordings of the scalp electroencephalogram (EEG) and the electrocorticogram (ECoG) to examine the correlation of HFO activity with vigilance state. We detected HFOs (80-500 Hz) from all bipolar ECoG derivations using the well-known Staba algorithm in ten seizure-free overnight recordings from five patients being evaluated for surgery. In addition, we classified EEG features using a linkage tree into four vigilance states representing gradations in sleep depth from wakefulness to slow wave sleep. Finally, we examined the correlation between vigilance state and HFO occurrence in the five channels with the most HFOs in each recording. The proportion of 30-s epochs containing HFOs was found to increase significantly with sleep depth (p < 0.01). Further analysis is necessary to examine the effects of epoch length and sample size in the choice of diagnostic baseline.

Keywords: Neural signal processing, Neural interfaces - Tissue-electrode interface, Neurological disorders - Epilepsy
Abstract: Propagating waves along the cortical surface have recently attracted significant attention by the neuroscience community. However, whether these propagating waves imply network connectivity changes for the neural circuits is not known. In this work, we employ a high density porous graphene microelectrode array and perform in vivo experiments with rodents to investigate network connectivity during cortical propagating waves. The spatial-temporal analysis of the cortical recordings reveals various types of propagating waves across the recording area. Network analysis results show that these propagating waves are consistent with the functional connectivity changes in the neural circuits, suggesting that the underlying network states are reflected by the cortical potential propagation patterns.

Keywords: Neural signal processing, Neural interfaces - Microelectrode technology, Brain physiology and modeling - Neuron modeling and simulation
Abstract: Classification of neurons from extracellular recordings is mainly limited to putatively excitatory or inhibitory units based on the spike shape and firing patterns. Narrow waveforms are considered to be fast spiking inhibitory neurons and broad waveforms excitatory neurons. The aim of this work is twofold. First, we intend to use the rich spatial information from high-density Multi-Electrode Arrays (MEAs) to make classification more robust; second, we hope to be able to classify sub-types of excitatory and inhibitory neurons. We first built, in simulation, a large dataset of action potentials from detailed neural models. Then, we extracted spike features from the simulated recordings on a high-density Multi-Electrode Array model. Finally, we used a Convolutional Neural Networks (CNN), to classify the different cell types. Compared with the ground truth data from the simulated dataset, the results show that this forward modelling/machine learning approach is very robust in recognizing excitatory and inhibitory spikes (accuracy ≥ 92.15 %). Additionally, the approach can, to a certain extent, correctly classify different cell sub-types. As the detail and fidelity of neural models increase and high-density recordings become available, this approach could become a viable and robust alternative for classification of neural cell types from in-vivo extracellular recordings.

Keywords: Parametric filtering and estimation, Connectivity measurements, Kalman filtering
Abstract: Owing to their millisecond-scale temporal resolution, magnetoencephalography (MEG) and electroencephalography (EEG) are well-suited tools to study dynamic functional connectivity between regions in the human brain. However, current techniques to estimate functional connectivity from MEG/EEG are based on a two-step approach; first, the MEG/EEG inverse problem is solved to estimate the source activity, and second, connectivity is estimated between the sources. In this work, we propose a method for simultaneous estimation of source activities and their dynamic functional connectivity using a Kalman filter. Based on simulations, our approach can reliably estimate source activities and resolve their time-varying interactions even at low SNR (<1). When applied on empirical MEG responses to simple visual stimuli, our approach could capture the dynamic patterns of the underlying functional connectivity changes between the lower (pericalcarine) and higher (fusiform and parahippocampal) visual areas. In conclusion, we demonstrate that our approach is capable of tracking changes in functional connectivity at the millisecond resolution of MEG/EEG

Keywords: Independent component analysis, Time-frequency and time-scale analysis - Time-frequency analysis
Abstract: Analysis of electroencephalographic (EEG) data requires cautious consideration of interfering artefacts such as ocular, muscular or cardiac noise. Independent component analysis (ICA) has proven to be a powerful tool for the detection and separation out of these contaminating components from brain activity. Yet thus far thorough investigation is lacking into how this pre-processing step might affect or even distort the information on brain connectivity inherent in the raw signals. The aim of this work is to address this question by systematically investigating and comparing three different strategies: first, analysis of all network nodes without eliminating contamination; second, removing the node which is contaminated by artefacts; third, using the ICA artefact removal method as an initial step prior to the analysis. Multivariate, time-variant autoregressive models are used to approximate the recorded data; the assessment of information flow within the modelled networks is carried out by means partial directed coherence, offering a frequency-selective estimation of connectivity.

Keywords: Physiological systems modeling - Multivariate signal processing, Connectivity measurements, Independent component analysis
Abstract: Alcohol and nicotine are substances that alter the functional connectivity of the brain. These changes have been observed after pinpointing particular brain areas as well as studying the overall brain wiring structure. One property of this wiring structure is the level of randomness. Evidence strongly agrees that brain connectivity is not random, but that chemical substances can affect the connectivity structure. This work aims at studying changes in resting state functional connectivity randomness in relation to the consumption of nicotine and alcohol. Results suggest that randomness in whole brain connectivity is not affected by used substance. However, connectivity among particular brain areas does show changed randomness linked to substance use. Abnormal randomness was found between salience and default mode functional domains. This dysfunction is in line with some postulates of the network model of addiction. The study provides new information on the effects of substance use on the brain.

Keywords: Coupling and synchronization - Coherence in biomedical signal processing, Connectivity measurements, Physiological systems modeling - Signal processing in physiological systems
Abstract: A professional quartet of saxophonists playing in ensemble provides a perfect scenario to study the eventual occurrence of synchronous oscillatory brain activity across subjects. Here, we applied hyperscanning methodologies for simultaneously recordings of electroencephalographic (EEG) signals from four professional saxophonists while they observ an audiovideo recording of their own previous musical performance. An ad-hoc musical composition was written for the study. At debriefing, the subjects were asked to answer two questionnaires to assess their empathy trait and the musical leadership. In order to estimate the hyperconnectivity of each musician we proposed a measure which combines phase synchronization index of brain oscillations and graph theory framework. The inter-connectivity level of each musician was statistically compared. Statistical results revealed a significant lower hyperconnectivity in the left Brodmann area 44 for the Soprano with respect to the other three members. Recent theories attributed this brain region (Broca's area) to music generation, empathy processes and communication. We hypothesize a relationship between brain-to-brain connectivity level and the musical role within the quartet.

Keywords: Connectivity measurements, Independent component analysis, Signal pattern classification
Abstract: The blood oxygen level dependent (BOLD) fMRI signal is influenced not only by neuronal activity but also by fluctuations in physiological signals, including respiration, arterial CO2 and heart rate/ heart rate variability (HR/HRV). Even spontaneous physiological signal fluctuations have been shown to influence the BOLD fMRI signal in a regionally specific manner. Consequently, estimates of functional connectivity between different brain regions, performed when the subject is at rest, may be confounded by the effects of physiological signal fluctuations. In addition, resting functional connectivity has been shown to vary with respect to time (DFC), with the sources of this variation not fully elucidated. The effect of physiological factors on dynamic (time-varying) resting-state functional connectivity has not been studied extensively, to our knowledge. In our previous study, we investigated the effect of heart rate (HR) and end-tidal CO2 (PETCO2) on the time-varying network degree of three well-described RSNs (DMN, SMN and Visual Network) using mask-based and seed-based analysis, and we identified brain-heart interactions which were more pronounced in specific frequency bands. Here, we extend this work, by estimating DFC and its corresponding network degree for the RSNs, employing a data-driven approach to extract the RSNs (low- and high-dimensional Independent Component Analysis (ICA)), which we subsequently correlate with the characteristics of simultaneously collected physiological signals. The results confirm that physiological signals have a modulatory effect on resting-state, fMRI-based DFC.

Keywords: Coupling and synchronization - Coherence in biomedical signal processing, Connectivity measurements, Time-frequency and time-scale analysis - Nonstationary processing
Abstract: Neural populations coordinate at fast sub-second time-scales during rest and task execution. As a result, functional brain connectivity assessed with different neuroimaging modalities (EEG, MEG, fMRI) may also change over different time scales. In addition to the more commonly used sliding window techniques, the General Linear Kalman Filter (GLFK) approach has been proposed to estimate time-varying brain connectivity. In the present work, we propose a modification of the GLFK approach to model time-varying connectivity. We also propose a systematic method to select the hyper-parameters of the model. We evaluate the performance of the method using MEG and EMG data collected from 12 young subjects performing two motor tasks (unimanual and bimanual hand grips), by quantifying time-varying cortico-cortical and cortico-muscular coherence (CCC and CMC). The CMC results revealed patterns in accordance with earlier findings, as well as an improvement in both time and frequency resolution compared to sliding window approaches. These results suggest that the proposed methodology is able to unveil accurate time-varying connectivity patterns with an excellent time resolution.

Keywords: Brain imaging and image analysis, Functional image analysis
Abstract: The combination of graph theoretical approaches and neuroimaging data provides a powerful way to explore the characteristics of brain network. Recently, the temporal variability of spontaneous brain activity and functional connectivity has attracted wide attention. Thus, it is essential to evaluate the reliability of functional network connectivity and properties from the dynamic perspective. However, previous test-retest (TRT) studies have explored this reliability with a static point of view. In this study, using a large rs-fMRI dataset from Human Connectome Project (HCP), we investigated TRT reliability of functional connectivity and graph metrics derived from the most commonly used method – sliding window at three time intervals (short: 72 seconds, middle: 15 minutes and long: >24 hours). The results revealed that reliable connectivities and related brain regions are mainly distributed in primary cortex, such as visual area and sensorimotor area and default mode network. Notably, connectivity strength and global efficiency have better reliability than other metrics. Finally, short scan time interval and long scan duration can increase the TRT reliability of metrics. Findings of present study provide important guidance for searching reliable network markers in future research.

Keywords: Brain imaging and image analysis, Magnetic resonance imaging - MR neuroimaging, Functional image analysis
Abstract: With the emergence of the dynamic functional connectivity analysis, and the studies relying on real-time neurological feedback, the need for rapid processing methods becomes even more critical. Seed-based Correlation Analysis (SCA) of fMRI data has been used to create brain connectivity networks. With close to a million voxels in a fMRI dataset, the number of calculations involved in SCA becomes high. This work aims to demonstrate a new approach which produces high-resolution brain connectivity maps rapidly. The results show that HPCME with four FPGAs can improve the SCA processing speed by a factor of 40 or more over that of a PC workstation with a multicore CPU.

Keywords: Brain imaging and image analysis, Functional image analysis, Magnetic resonance imaging - MR neuroimaging
Abstract: Transcranial direct current stimulation (tDCS) is an emerging non-invasive neuromodulation technique that serves as treatment tool to neurological disorders. However, the mechanism of how the stimulation modulates ongoing brain activity and connectivity is still not fully understood. Simultaneous acquisition of neuroimaging data together with brain stimulation could allow a noninvasive examination of the brain dynamic changes during the process. In this pilot study, concurrent tDCS and fMRI was conducted in a healthy subject. Dynamic functional connectivity and effective connectivity were used to reveal the information flow. The results demonstrated that tDCS duration has important effects on the brain connectivity and the causal relationships among the brain regions. These results might reflect the fundamental mechanism of brain processing under the external stimulation.

Keywords: Brain imaging and image analysis, Functional image analysis, Multivariate image analysis
Abstract: Motion-induced artifact detection has become a fixture in the assessment of functional magnetic resonance imaging (fMRI) quality control. However, the effects of other MR image quality (IQ) metrics on intrinsic connectivity brain networks are largely unexplored. Accordingly, we report herein the initial assessment of the effects of a comprehensive list of IQ metrics on resting state networks using a multivariate analysis of covariance (MANCOVA) approach based on high-order spatial independent component analysis (ICA). Three categories of MR IQ metrics were considered: (1) metrics for artifacts including the AFNI outlier ratio and quality index, framewise displacement, and ghost to signal ratio, (2) metrics for the temporal quality of MRI data including the temporal framewise change in global BOLD signals (DVARS), global correlation of time-series, and temporal signal to noise ratio, (3) metrics for the structural quality of MRI data including the entropy focus criterion, foreground-background energy ratio, full-width half maximum smoothness, and static signal to noise ratio. After FDR-correction for multiple comparisons, results showed significant effects of the static and temporal signal to noise ratios on the spatial map intensities of the basal ganglia, default-mode and cerebellar networks. AFNI outlier ratio, framewise displacement and DVARS exhibited significant effects on the BOLD power spectra of sensorimotor networks. The global correlation of time-series displayed wide-spread modulation of the spectral power in most networks. Further investigations of the effect of IQ metrics on the characteristics of intrinsic connectivity brain networks allow more accurate interpretation of the fMRI results.

Keywords: Brain imaging and image analysis, Functional image analysis, Multivariate image analysis
Abstract: Physiological noise corrections using RETROICOR algorithm has been shown to increase signal sensitivity in resting state networks such as the default-mode network. However, independent component analysis (ICA)-based network approach may suffer from such corrections especially if there is any overlap between two sources in the decomposition domain. To address the extent the physiological noise corrections may impact ICA derived intrinsic connectivity brain networks, we measured network features including functional network connectivity (FNC), power spectra, and network spatial maps in the resting state and task functional magnetic resonance imaging (fMRI) data that were acquired in the same visit from a group of healthy volunteers. Statistical analysis showed functional connectivity between several networks were significantly changed after RETROICOR corrections in both rest and task fMRI. Significant FNC alterations were found in the subcortical, basal ganglia, salience, and default-mode networks. Power spectra analysis showed a trend toward lower power spectra in the subcortical and salience networks at [0.20 and 0.24] Hz after RETROICOR corrections in both rest and task fMRI. Furthermore, physiological noise corrections led to volumetric decrease in the resting state networks that included the subcortical, basal ganglia, salience, and default-mode networks, and volumetric enlargement in the sensorimotor and cerebellar networks. In task fMRI data, physiological noise corrections generally resulted in the expansion of networks except for task-activated networks including the anterior salience, central executive, dorsal attention, and cerebellar networks. If confirmed with larger sample sizes, these results suggest that physiological noise corrections alter some network features, and that such alterations are different between resting state and task fMRI data.

Keywords: Brain imaging and image analysis, Functional image analysis, Multimodal imaging
Abstract: Mapping the brain alterations post stroke and post intervention is important for rehabilitation therapy development. Previous work has shown changes in functional connectivity based on resting-state fMRI, structural connectivity derived from diffusion MRI and perfusion as a result of brain-computer interface-assisted motor imagery (MI-BCI) and transcranial direct current stimulation (tDCS) in upper-limb stroke rehabilitation. Besides functional connectivity, regional amplitude of local low-frequency fluctuations (ALFF) may provide complementary information on the underlying neural mechanism in disease. Yet, findings on spontaneous brain activity during resting-state in stroke patients after intervention are limited and inconsistent. Here, we sought to investigate the different brain alteration patterns induced by tDCS compared to MI-BCI for upper-limb rehabilitation in chronic stroke patients using resting-state fMRI-based ALFF method. Our results suggested that stroke patients have lower ALFF in the ipsilesional somatomotor network compared to controls at baseline. Increased ALFF at contralesional somatomotor network and alterations in higher-level cognitive networks such as the default mode network (DMN) and salience networks accompany motor recovery after intervention; though the MI-BCI alone group and MI-BCI combined with tDCS group exhibit differential patterns.

Keywords: Neurorehabilitation, Neuromuscular systems - EMG processing and applications, Neural interfaces - Body interfaces
Abstract: Stroke, together with spinal cord injury (SCI), count up to 52% of the adult-onset disability. Enhancing the process of recovery of cognitive and motor functions after a neurological injury or disease is therefore widely recognized as a priority in healthcare. One of the opportunities in this direction lies in combining traditional approaches with robotic-based neurorehabilitation, thus improving the beneficial effects of the treatment. In this work, we present a novel exoskeleton device and propose its use in neurorehabilitation application scenarios.

Keywords: Brain functional imaging - EEG, Neural stimulation, Neurorehabilitation
Abstract: We developed a platform for investigating the effect of non-invasive brain stimulation performed during cognitive rehabilitation. We integrated the system with high-density electroencephalography (hdEEG) recordings, which are processed with novel computational techniques providing accurate source localization. The platform allows the investigation of brain electrophysiological activity across the cognitive network during working memory tasks, in order to highlight the plastic changes, induced by the stimulation.

Keywords: Neurorehabilitation, Neurological disorders - Stroke, Neurological disorders - Treatment methodologies
Abstract: Cortical stroke or neurodegenerative diseases such as Parkinson’s disease can severely impair fine motor function of the hand. Restoration or preservation of this function is a major effort during neurorehabilitation. Many of these patients also exhibit proprioceptive deficits, which negatively impacts on motor re-learning. We here demonstrate that robotic rehabilitation approaches that aim to improve proprioceptive function can aid motor recovery or preserve sensorimotor function in neurodegenerative disease.

Keywords: Motor learning, neural control, and neuromuscular systems, Neuromuscular systems - EMG processing and applications, Neurorehabilitation
Abstract: Body-machine interfaces (BMIs) are a viable option to control assistive devices and promote the recovery of movements after spinal cord injury, stroke or other neurological disorders that impair motor functions. BMIs translate signals from the body into the low dimensional control space of assistive and rehabilitative devices or of a personal computer. The mapping process allows us to select not only the subspace that is most comfortable for the user to act upon, but also the degrees of freedom and coordination patterns that the user has more difficulty engaging. Our goal is to develop BMI architectures that provide the user with the ability to identify and coordinate a convenient subset of movements and/or muscle signals for achieving task objectives with a flexible and adaptable motor behavior. Here we investigate methods to map muscle activities into control signals and to integrate activities from muscles and motion sensors to recover natural mobility.

Keywords: Neural stimulation, Neurorehabilitation, Neurological disorders - Stroke
Abstract: Non-invasive brain stimulation techniques, such as transcranial magnetic stimulation (TMS), can be used to both identify and manipulate recovery mechanisms post-stroke. In the latter case, non-invasive brain stimulation could be used as an adjuvant treatment that could enhance the efficacy of other rehabilitation therapies. The results of initial investigations into this possibility, however, have been largely inconclusive. There is a growing consensus that individual characteristics such as stroke location and severity are likely to be crucial considerations for identifying the stimulation parameters most likely to benefit recovery. To begin to address this, we studied the role of affected vs. unaffected hemisphere motor areas in movements of the paretic arm. We show that different motor areas contribute to paretic arm movements depending on the severity of corticospinal tract damage.

Keywords: Brain-computer/machine interface, Neural interfaces - Implantable systems, Neural interfaces - Microelectrode technology
Abstract: Over the last two decades we have seen significant advances in brain-computer interfaces (BCI) and neurostimulation technologies. Paralyzed users have been able to feel tactile sensations and control computer cursors, robotic and prosthetic arms, and even move their own arms and hands through their own thoughts again, thanks to advances in the fields of BCI, robotics, and neuromuscular stimulation. Specifically, new results have been achieved recently in restoring functional, rhythmic, and graded hand movements in a human. Neural decoding methods and computer processing power have also improved, yet we still are challenged to understand how the brain encodes complex movements and how those modulation patterns change over time. New neural decoding methods and technologies such as ultra-high density brain and nerve interfaces are under development and will be key to advancing this field and providing clinically-relevant solutions to patients.

Keywords: Sensory neuroprostheses - Somatosensory, Brain-computer/machine interface, Neural stimulation
Abstract: Intracortical microstimulation of the somatosensory cortex can be used to create a neuroprosthesis that restores tactile sensations even after long-term spinal cord injury. Evoked percepts have a range of qualities from pressure and touch, to tingling and buzzing, and an important area of investigation is to determine whether different stimulus parameters affect the perception of stimulation. Here we examine the effect of stimulus frequency and found that different electrodes produce more intense percepts in different frequency ranges and that these different groups of electrodes elicit sensations with different perceptual qualities.

Keywords: Sensory neuroprostheses, Neural interfaces - Implantable systems, Neural stimulation
Abstract: In Cleveland, peripheral nerve and intramuscular neurostimulation technologies have a nearly 50-year history of development and advancement to clinical implementation. Successful translation to chronic human implementation requires design that balances advanced technology, reliability, biocompatibility, and robustness. We present the current state of clinical peripheral nerve and intramuscular electrode technology in the context of sensorimotor restoration for limb prosthetics. We also present the iSens® neuromoduation system. iSens is a Bluetooth®-connected, modular system that can record or stimulate from up to 96 independent channels with real-time recording feature extraction; selectable bipolar electrode recording; multi-channel, asynchronous stimulation paradigms; and electrical field-shaping capability for improved nerve stimulation selectivity. We will present the results of implementation of latest clinical technology for more than five years in peripheral nerve stimulation for motor restoration in paralysis and sensory restoration and motor control in limb prosthetics. Finally, we discuss the technology capabilities of the iSens system and its potential as a platform for sensorimotor restoration.

Keywords: Neural interfaces - Bioelectric sensors
Abstract: Reliable chronic electrophysiological recordings remain elusive due to persistent biological failure affecting the electrode-tissue interface. We report on the development of conformal neural probe arrays microfabricated from thin film Parylene C. The use of a soft, flexible substrate seeks to minimize the immune response at interface and thereby improve the quality of chronic recordings. Acute implants of 64 electrode arrays were achieved and successfully targeted rat hippocampus. These technical advances seek to achieve high density arrays of polymer-based neural probes for chronic large scale recording of neural activity.

Keywords: Neural interfaces - Microelectrode technology, Neural stimulation, Motor neuroprostheses - Epidural stimulation
Abstract: Optogenetics is a powerful tool for relating brain function to behavior, as it enables cell-type specific manipulation of neurons with millisecond temporal precision. There has been major challenges for reliably implementing this technique in non-human primates (NHP). Here, we review the novel approaches that we have developed for large-scale, bi-directional manipulation of neural circuits in NHPs.

Keywords: Neural interfaces - Bioelectric sensors, Neural interfaces - Biomaterials, Neural interfaces - Implantable systems
Abstract: This talk highlights our recent work in flexible electronic and optoelectronic systems for advanced neural interface technologies. Examples range from thin, conformal sheets with actively multiplexed electronics for high density electrocorticography to narrow penetrating pins with cellular-scale LEDs for optogenetic stimulation/inhibition.

Keywords: Neural interfaces - Body interfaces
Abstract: Then neural dust platform uses ultrasonic power and communication to enable a scalable, wireless, and battery-less system for interfacing with the nervous system. Ultrasound offers several advantages over alternative wireless approaches, including a safe method for powering and communicating with sub mm-sized devices implanted deep in tissue. Our earlier studies demonstrated that neural dust motes could wirelessly transmit high-fidelity electrophysiological data in vivo, and that theoretically, this system could be miniaturized well below the mm-scale. I will review current work focused on further minimization of the platform, better encapsulation methods as a path towards truly chronic neural interfaces, improved delivery mechanisms, demonstration of in vivo stimulation capabilities, and modifications to the system that enable deployment of neural dust in the central nervous system.

Keywords: Wearable body sensor networks and telemetric systems, Smart textiles and clothings
Abstract: This paper briefly describes the rational behind and the aims of two ongoing projects addressing work related health problems using wearable systems.

Keywords: Wearable sensor systems - User centered design and applications, Wearable body sensor networks and telemetric systems, Smart textiles and clothings
Abstract: The aim of this paper is to present the need and potential for wearable sensors in occupational healthcare. In addition, it partially presents ongoing European and Swedish projects for developing personalized, and pervasive wearable systems for assessing the risk of developing musculoskeletal disorders and cardiovascular diseases at work. Occupational healthcare might benefit by preventing the diseases and disorders by providing the right feedback at the right time to the right person. Collected data from workers can provide evidence supporting the ergonomic and industrial tasks of redesigning the working environment to reduce the risks.

Keywords: Smart textiles and clothings, Wearable sensor systems - User centered design and applications, Integrated sensor systems
Abstract: Care cost for rehabilitation and treatment of work-related injuries and disorders are steadily increasing. Developments in sensing technologies, information and communication technologies allow for the development of novel p-health solutions for accurate assessment of risk exposure enabling prompt intervention for avoiding injuries. Combining sensorized garments, wearable electronics, mobile devices, cloud computing and machine learning, the We@work project has implemented an integral solution for preventive care of workers fostering a healthy and safe working life.

Keywords: Modeling and analysis
Abstract: Pocket mHealth, a patient-centered solution owned by ATOS, enables patient empowerment in the management of its own medical information, storing their Electronic Health Record (EHR) and health data coming from different Hospital Information Systems (HIS). Pocket mHealth solves the need of current healthcare organizations into the adoption of healthcare data openness that following a patient centered design, taking advantage of the potential benefits of this paradigm. Instead of having standardized repositories at the hospital facilities, the patient becomes the driver of the change bringing standardized pieces of EHR in the mobile, what we call “distributed interoperability”.

Keywords: Physiological monitoring - Novel methods, Wearable wireless sensors, motes and systems, Wearable sensor systems - User centered design and applications
Abstract: This paper presents strategies targeting prevention of work-related ill-health, and how the use of smart workwear (wearables) can facilitate these strategies.

Keywords: Smart textiles and clothings, Integrated sensor systems, Portable miniaturized systems
Abstract: The reliability of systematic observational risk assessment methods on biomechanical load was evaluated. The reliability was generally low, which shows a clear need for new technical objective easy-to-use risk assessment methods with significantly higher reliability.

Keywords: Physiological monitoring - Modeling and analysis, Optical and photonic sensors and systems
Abstract: Non-contact measurement of physiological parameters, like pulse rate variability (PRV), has numerous applications in medicine and affective computing. PRV is an informative measure of autonomic nervous system activity. Spectral estimation from unevenly sampled, non-stationary data is integral to pulse rate variability frequency-domain analysis. We present the first comparison of results of PRV computation using the Lomb-Scargle method and Bayesian Spectral Estimation. The Lomb-Scargle method performs well, even in the presence of missing beats. However, the Bayesian Spectral Estimation method has advantages when tracking changes in amplitude and frequency. We illustrate these characteristics with results from synthetic data and real non-contact imaging photoplethysmography measurements.

Keywords: Infra-red imaging, Image feature extraction, Novel imaging modalities
Abstract: Remote pulse oximetry (SpO2) measurements are based on photons remitted from the skin. As photon path-lengths are influenced by wavelength and depth-dependent skin properties, the SpO2 calibration may be susceptible to the skin’s pulsating profile; i.e., to relative variations of arteriolar pulsations at upper-to-lower dermal layers. We hypothesize that preferring red-infrared wavelengths ensures similar penetration depths and invariance to pulsatile variations. We show diffuse reflectance measurements and Monte Carlo simulations of the normal and compressed skin, which support that photoplethysmographic (PPG) signals in red-infrared originate mostly from the deep vasculature. Complementarily, we use a skin model and assess, numerically, the accuracy of the ratio-of-ratios (RR) method for estimating SpO2 under pulsatile profiles within estimated realistic ranges. We consider the RRs between red (R; 660 nm) and infrared (IR; 840 nm) and the 80-100% SpO2 range. Our results indicate the invariance of remote SpO2 to the skin’s pulsating profile in R and IR.

Keywords: Image analysis and classification - Machine learning / Deep learning approaches
Abstract: We investigated the ability of features extracted from noncontact Photo-Pletismography (PPG) signals in assessing Autonomic Nervous System (ANS) responses elicited by two sympathetic stimulations: i) rest-to-stand and ii) cold pressure tests. We observed that spectral parameters of Heart Rate Variability (HRV), obtained by post-processing of noncontact PPG, are able to highlight the expected parasympathetic withdrawal induced by the stimulations

Keywords: Optical imaging, Infra-red imaging
Abstract: This paper presents first results of a clinical study in which we tested reflective PPG sensors, which are integrated behind the fabric of a dentist chair’s backrest at three different locations, during dental treatment. Such sensors are able to measure through the fabric and the patient’s clothing. The system shows good results even during treatment phases where mechanical vibrations to the patient's body are induced, i.e. tooth cleaning with an ultrasonic scaler.

Keywords: Optical imaging, Infra-red imaging, Optical imaging and microscopy - Optical vascular imaging
Abstract: Remote assessment of vital, cardiovascular system parameters presents numerous opportunities in the areas of health, human machine interfaces, and affective computing. Imaging photoplethysmography (iPPG), utilizing cameras to measure light absorption changes corresponding with the cardiac cycle is one such promising method for remote physiological assessment. Previous work has demonstrated improvements resulting from optimizing hardware designs and signal processing using the spectral relationship of the measured blood volume pulse signal. To further understanding and methods in this area, we recently developed a multispectral iPPG testbed with a comprehensive array of conventional color cameras, multispectral camera array, and synchronized reference physiological measurement. We present the finalized testbed and study design and preliminary results from the study consisting of seven trials intended to affect sensor measurements similarly to as expected in real-world settings or participant physiological, cognitive, or affective states

Keywords: Micro- and nano-technology, Cellular force transduction - Mechanical stimuli and mechanotransduction
Abstract: Fusion of cancer cells has been found in tumors for more than a century, while the factors that regulate such process are poorly understood. Here we used several microengineered systems to depict that biomechanical environment can directly modulate the fusogenicity of cancer cells.

Keywords: Micro- and nano-sensors, Micro- and nano-technology
Abstract: In this study, a quantitative assessment method for evaluating the regularity of activity of daily living (ADL) has been proposed based on the previously reported activity index (AI) from the MEMS-based accelerometer wrist-worn device. With 24 hours’ continuously measured AI, an hourly AI pattern for that day can be generated and by finding the correlation coefficient between this hourly AI pattern and the previous day’s pattern, the regularity index (RI) of ADL can be quantitatively assessed. RI provides quantified numbers for the regularity of daily life and can be used in activity-based evaluation models for many medical-care applications, such as the risk assessment and readmission prediction for COPD patients.

Keywords: Micro- and nano-sensors, Microfluidic applications, Microfluidic techniques, methods and systems
Abstract: Tumor spheroid assay is generally used for the study of the development of early tumor and the effect of cytotoxic agents on tumors. In the current work, quantitative assessment of the growth of tumor spheroids was developed based on the impedance measurement technique. Result revealed that the impedance change could describe the growth of tumor spheroids during the culture. The technique owns the operation simplicity and quantitative assessment of tumor spheroids.

Keywords: Microfluidic applications, Microfluidic techniques, methods and systems, Micro- and nano-sensors
Abstract: In this presentation, EWOD (Electro-wetting on dielectric) system is used to be a dynamic culture platform for embryo formation until the status of BC (Blastocyst). The EWOD device is anticipated to provide a stable culture environment and improve the growth environment of mouse embryos based on its advantages of uniform hormone mixing in the culture medium. The development rate of embryos cultured by dynamic process is better than the one by static culture group. Other than digital microfluidic system, the continued microfluidic system will also be designed and used for insemination by sperms and oocyte and embryo culture. Either of microfluidic system would be used in clinical in the near future.

Keywords: Microfluidic techniques, methods and systems, Biomaterials - Chemical and electrochemical sensors, Microfluidic applications
Abstract: We report the use of a luminescent antibiotic-responsive bacteria strain for detecting the presence of an antibiotic, ciprofloxacin (CIP), in milk. A window-less CCD provided proximity detection of the bioluminescence and allowed highly efficient light collection by a lens-free optical design. The detection limit of CIP in milk was 8 ng/mL.

Keywords: Computational modeling - Biological networks, Model building - Network modeling, Models of organ physiology
Abstract: Motor evoked potentials are a commonly used monitoring methodology during surgeries that can put the motor system components at risk. Application of current to the motor tracts in the brain, via transcranial electrical stimulation, activates axons in the cortical spinal tract at different areas and affects the alpha motor neurons in the spine based on various stimulation parameters. Computational modeling has been used to show the likely location in the brain of stimulation activation as well as the optimal parameters for transmission of the signal through the corticospinal tract and within the spinal cord.

Keywords: Modeling of cell, tissue, and regenerative medicine - 2d and 3d cell modeling, Data-driven modeling
Abstract: Improvements in the efficacy of retinal neuroprostheses can stem from investigating more sophisticated neural stimulation strategies which enable selective activation of specific retinal ganglion cells (RGCs). Computational models are particularly well suited for these investigations. The electric field can be accurately described by mathematical formalisms, and the neurons can be ‘probed’ at resolutions well beyond those achievable by today’s state-of-the-art biological techniques. In this study, we used computational models to explore the ability of high frequency electrical stimulation (HFS) to differentially activate ON and OFF RGCs. Performance of a wide range of electrical stimulation amplitudes and frequencies on functional RGC responses were evaluated.

Keywords: Systems modeling - Clinical applications of biological networks, Model building - Algorithms and techniques for systems modeling, Models of medical devices
Abstract: The mechanisms of action in ultra high frequency (10kHz) dorsal column stimulation for pain remain unclear. There are generally two hypotheses: direct effects of the stimulation field on the dorsal horn or indirect effects via dorsal column axons on dorsal horn processing. While there is limited support for either hypothesis experimentally, prior computational efforts from our lab show the potential for differential suppression and stimulation of axon diameters in the dorsal column, in favor of the indirect hypothesis. We provide more support using bipolar biphasic stimuli with characteristics similar to what is used clinically, in multiple axonal membrane models and with detailed ionic channel simulations to show these findings are plausible.

Keywords: Computational modeling - Analysis of high-throughput systems biology data, Systems modeling - Clinical applications of biological networks, Systems modeling - Decision making
Abstract: Gliomas are associated with poor outcome and significant neurological morbidity. Key biological features of these malignant brain tumors include mitosis, brain invasion, and angiogenesis (new blood vessel formation). Mathematical modeling of has made important contributions to the biology and clinical care of gliomas. The equations model the growth of grades 2-4 gliomas at the scales of pathology and magnetic resonance imaging; simulations yield computational trials (in silico clinical trials) that accurately predict the survival times of GBM patients. The results identify the rates of angiogenesis, migration, and replication as fundamental properties that predict survival times, response to treatment, and advancement to higher grades. The findings also uncover novel clinical behaviors, like evolution to higher grade, inability to transform, and new recurrence patterns of GBM treated by anti-angiogenic therapy. These results support the recent announcement by the Food and Drug Administration to use modeling and simulations in predicting effectiveness and clinical outcomes.

Keywords: Systems modeling - Clinical applications of biological networks, Model building - Algorithms and techniques for systems modeling, Computational modeling - Biological networks
Abstract: The electroceutical paradigm envisions modulating central and peripheral nervous system activity at will to treat disease and disorder. This effort requires the identification of fibers responsible for efficacy and side effects, their sources and targets, and the stimulation protocols that maximize efficacy and minimize side effects. In this context we present our modeling of vagus nerve stimulation for epilepsy. provide an example of electroceutical modeling coupled to known results.

Keywords: Computational modeling - Biological networks, Model building - Algorithms and techniques for systems modeling
Abstract:

Keywords: Neuromuscular systems - Locomotion, Neuromuscular systems - Peripheral mechanisms, Neuromuscular systems - Computational modeling
Abstract: The mechanical impedance of our ankles plays a significant role in human locomotion. Recent studies have begun to characterize ankle impedance during movement tasks. However, it is difficult to determine how impedance varies with position, torque and muscle activation under the dynamic conditions of movement, as all three variables vary continuously. In this study we systematically investigated how dynamically varying position, torque and their interaction affect ankle impedance. We found that impedance during dynamic changes in either position or torque is much different than that estimated at matched isometric position and torques. However, most interesting was that dynamically and simultaneously varying position and torque produced ankle impedance estimates that could not be predicted based on the estimates produced when position and torque varied independently. These results highlight the inherently non-linear behavior of the ankle during movement tasks, and emphasize caution when trying to predict impedance during natural movement tasks.

Keywords: Human performance - Gait
Abstract: Biomechanical devices are shaped by the ways they try to mimic the natural body, including geometric, structural, and control properties. Different conceptual models lead to different solutions. We present a new perspective and method to describe ankle control across walking and running, and an adaptive foot-ankle prosthesis that mimics the behavior.

Keywords: Human performance - Sensory-motor, Neuromuscular systems - Locomotion, Motor neuroprostheses - Robotics
Abstract: Current models of human biomechanics consisting only of joint torques and joint angles are insufficient to completely describe lower-limb gait dynamics. The agonist-antagonist neuromuscular geometry permits regulation of joint impedance, in addition to joint torques and angles. Joint impedance governs the instantaneous response to a perturbation, and is implicated in energy storage, dissipation, and exchange. Despite knowledge of how joint impedance changes throughout the gait cycle, little is known about the role of these properties in neuromotor control. The purpose of this study was to quantify the accuracy and precision of ankle stiffness control, and compare these data to the regulation of ankle torque and position. Our results showed that both the accuracy and precision of matching ankle stiffness to a desired target were not significantly different from analogous torque or position target-matching tasks.

Keywords: Neuromuscular systems - Locomotion, Neurorehabilitation, Neural signal processing
Abstract: The nervous system simplifies neuromuscular control by using muscle synergies, thus organizing muscle activity into a small number of coordinative modules. In the present study we investigated how muscle modularity is structured across a broad range of walking conditions including different speeds and ground elevations. This descriptive analysis of muscle modularity can then be translated into a predictive model to estimate how motor components modulate across locomotion speeds and ground elevations, which in turn, can inform human to WR interfaces for a robust and natural operation.

Keywords: Neuromuscular systems - Computational modeling, Neuromuscular systems - Locomotion, Human performance - Gait
Abstract: It is well understood that the whole body mechanics of human locomotion are characterized by compliant leg behavior in stance. While in some animals this behavior is realized by passive leg mechanics, it requires control in humans. In the past, several suggestions have been made about what challenges this control faces and how solutions may be realized by the human nervous system. In current work, we are interested in understanding whether mimicking the proposed human control in powered prostheses can improve the stability and robustness of amputee gait. In initial results, we find that it stabilizes gait after slipping events substantially more often than impedance control.

Keywords: Neuromuscular systems - Locomotion, Motor neuroprostheses - Prostheses, Neuromuscular systems - Peripheral mechanisms
Abstract: Efference copy is a neural signal mirroring the command from the motor cortex to the periphery that activates the muscles. It provides information necessary for the nervous system to predict expected sensory feedback in the control of movement. Most robotic lower limb prostheses decouple or alter the relationship between efference copy and the predicted sensory feedback. As a result, the use of robotic lower limb prostheses and exoskeletons can be difficult to control and may provide a sensation of perturbation rather than coordinated motion. Control approaches that integrate efferent signals can reduce perturbation perceptions and may generate a more naturalistic movement pattern when merging machines with humans.

Keywords: Physiological systems modeling - Signal processing in physiological systems, Physiological systems modeling - Signals and systems
Abstract: We applied a recently proposed multiscale complexity (MSC) method to heart period (HP) variability to verify whether complexity of the cardiac control at temporal scales typical of sinus node neural regulation exhibits a circadian rhythm. MSC analysis was performed on 24h Holter HP variability recordings during daytime and nighttime in 12 healthy humans (age: 34-55 yr). We found that HP complexity in the range of frequencies below the respiratory band increased during nighttime. We conclude that the increase of cardiac control complexity during nighttime is mainly due to sympathetic withdrawal more than vagal activation.

Keywords: Nonlinear dynamic analysis - Biomedical signals
Abstract: Fuzzy entropy (FuzEn), instead of sample entropy (SampEn), was implemented in Refined Multiscale Entropy (RMSE) applied to EEG signals, in order to predict pain responses in patients under sedation-analgesia. EEG segments were classified according to Ramsay sedation score (RSS) values in two groups: with response (2≤RSS≤5) and without response (RSS=6) to noxious stimuli. FuzEn had better nociception prediction in long scales than SampEn.

Keywords: Nonlinear dynamic analysis - Biomedical signals
Abstract: Multiscale entropy (MSE) has long been recognized as a complexity measurement. Recently, MSE was also applied as a discriminant statistic for a nonlinearity test of heart rate variability series. Here, we applied MSE and refined MSE as both complexity and nonlinearity measurements of arterial pressure (AP) oscillations in a mice model of baroreflex denervation. Results showed that AP contains reasonable levels of nonlinearity, but they are independent of the baroreflex. Moreover, complexity decreased when the baroreflex was removed, pointing that the two concepts might bring complementary information.

Keywords: Physiological systems modeling - Signal processing in physiological systems
Abstract: This study demonstrated the ability of newly developed linear multiscale entropy analysis (LMSE) to detect the changes in heart rate and blood pressure oscillations complexity on various time scales related to orthostatic and mental stress.

Keywords: Physiological systems modeling - Multivariate signal processing, Directionality, Physiological systems modeling - Signal processing in physiological systems
Abstract: We present the extension to multiscale analysis of Partial Information Decomposition, a recently proposed information-theoretic framework to dissect the information transferred within a network of interacting signals. The extension is based on the theory of state-space models, and leads to assess unique, redundant and synergistic information transfer as a function of the temporal scale at which the signals are observed. The usefulness of this new framework for the clinical study of physiological networks is demonstrated in the analysis of brain dynamics and cardiovascular control.

Keywords: Causality, Physiological systems modeling - Multivariate signal processing
Abstract: Beat-to-beat variability of the QT interval in ECG is closely linked to heart rate variability. Here, we explore this relationship on multiple time scales, and the effect of ageing by estimating Granger causality. We observe a bidirectional coupling, predominately from heart rate to QT variability that is strongest on the original time scale and weakens with age.

Keywords: General and theoretical informatics - Big data analytics, Health Informatics - Clinical information systems, Health Informatics - Electronic health records
Abstract: The identification of patient deterioration is a critical factor in providing timely and effective medical care; major physiological events are typically preceded by derangement of those physiological data that are now routinely collected in many hospitals. However, the parsing of continuous multivariate time-series data acquired minute-by-minute for very large cohorts of hospital patients is not a task that is well-suited for humans. We outline advances in the field of machine learning for modelling time-series physiological data acquired from hospital patients, and describe how such a system can provide a continuous screening facility that enables human expertise to be brought to the bedside in a timely manner, for those patients most at risk of imminent physiological deterioration.

Keywords: General and theoretical informatics - Algorithms, General and theoretical informatics - Artificial Intelligence, General and theoretical informatics - Big data analytics
Abstract: The probabilistic modelling of vital-signs is a key tool to understand the variability in both patient physiology and the measurement of that physiology. Vari- ability in vital-sign measurements may correlate across time and across other vital signs, or may result from measurement noise. Current clinical practice in patient monitoring (e.g., the Nation Early Warning Score) ignore the noise, time-dependence, and inter-vital-sign covariance. Modelling methods of time-series tend to either (i) model time-dependent correlation directly, or (ii) derive a set of descriptive IID features from the time-series. Popular meth- ods to model multiple vital-sign time-series (e.g. Kalman filtering, vector auto-regression, or Gaussian processes) model both time and inter-vital-sign correlation, but do so by making unreasonable assumptions about the marginal distributions of the individual vital signs (e.g. Gaussian noise or log-Gaussian noise, after a data transformation). This paper describes how copulas may serve as a simple modelling tool to create principled estimates of the joint density between vital signs. Using heart rate, respiratory rate, and SpO2 as exemplars, we begin by learning an optimal marginal distribution family to describe IID noise for each vital sign. Time-series properties of IID correlation is then examined.

Keywords: Health Informatics - Patient tracking, Sensor Informatics - Body sensor networks, Sensor Informatics - Wearable systems and sensors
Abstract: This explorative study investigates the effect of multitasking for patients with Parkinson’s disease using a body sensor network. Initial results indicate that a cumulative measurement of speech can differentiate between patients and controls.

Keywords: Physiological monitoring - Novel methods, Physiological monitoring - Instrumentation, Wearable sensor systems - User centered design and applications
Abstract: New cuffless blood pressure (BP) monitoring devices have been developed to monitor BP continuously without the requirement for a cuff. While several international companies are interested in developing cuffless BP monitors, these medical devices require approval from government agencies and the International Standards Organization (ISO). In this mini-symposium, we summarize the current status of such devices in Japan and approaches taken toward gaining ISO approval for their use.

Keywords: Optical and photonic sensors and systems, Sensor systems and Instrumentation, Wearable body sensor networks and telemetric systems
Abstract: Pulse wave velocity (PWV) has a correlation with blood pressure (BP) and has been reported to be suitable for cuffless and continuous BP monitoring. On the other hand, BP monitors are extremely sensitive to measurement site position. For this reason, PWV may also be affected by the arm-position change. The aim of this study was to verify the effect of the arm-position in BP estimation by using PWV. The results showed that the movement of the elbow joint had a little effect on PAT and PTT measurements.

Keywords: Physiological monitoring - Modeling and analysis
Abstract: Pulse transit tine (PTT) is the surrogate parameter for cuffless measurement of arterial blood pressure (BP). The association between BP and PTT is determined by the pressure dependency of arterial stiffness measured as pulse wave velocity (PWV = distance/PTT). The sensitivity of PWV to BP is not uniform and varies with BP. Due to this characteristic, similar changes in PTT will result in different errors in BP estimation depending on the level of BP. This may have implication for accuracy of cuffless BP measurement when PTT is measured by skin surface sensors.

Keywords: Cardiovascular, respiratory, and sleep devices - Wearables, Cardiac electrophysiology - Atrial fibrillation, Cardiac mechanics, structure & function - Atrial Fibrillation
Abstract: This paper briefly presents main applications of new non-invasive multi-dimensional heart monitoring systems for e-health, telemonitoring, and medical imaging.

Keywords: Cardiovascular and respiratory signal processing - Cardiovascular signal processing, Coronary artery disease
Abstract: This work investigates the relative goodness of seismocardiography (SCG) to ECG in predicting cardiac quiescence. As a baseline, quiescence derived from patient-specific echocardiography is used for comparing quiescence derived from ECG- and combined ECG-SCG-based methods. Numerically derived as an SCG index and validated by the baseline echocardiography, our results indicate that the SCG index is an effective measure in distinguishing subjects who would benefit from a multimodal (ECG and SCG) quiescence prediction, compared with an ECG-only-based prediction. The development of SCG index enables a more comprehensive multimodal framework for cardiac gating of imaging modalities such as computed tomography (CT) and magnetic resonance imaging.

Keywords: Cardiovascular, respiratory, and sleep devices - Wearables, Cardiovascular, respiratory, and sleep devices - Sensors, Cardiovascular, respiratory, and sleep devices - Smart systems
Abstract: We observed that the seismocardiogram (SCG) fiducial points associated with the opening and closure of aortic and mitral valves may be sometimes characterized by inflection points rather than clear peaks or valleys as traditionally considered. This aspect should be taken into account in the design of algorithms for the SCG processing. In this paper we report on the analysis of this pattern in a large SCG dataset collected during sleep in microgravity, namely in a condition where behavioral influences are minimized.

Keywords: Cardiovascular and respiratory signal processing - Cardiovascular signal processing
Abstract: Gyrocardiography (GCG) is a non-invasive method for measuring the angular velocity of the chest caused by heart motion. In this study, we assessed the reliability and accuracy of GCG as a promising technology for measuring the cardiac timing intervals. We recorded the 3-axial GCG signals from 50 healthy volunteers. In addition, the electrocardiogram (ECG) and the tissue Doppler Imaging (TDI) were recorded, simultaneously. We identified several cardiac fiducial points on the GCG x- and y-axis signals and estimated the cardiac time intervals from ECG Q to the delineated points. We labelled mitral and aortic valve opening and closure on TDI images and measured the time intervals from ECG Q to the TDI points as the gold reference. We assessed the agreement between the estimated and reference time intervals using the Bland-Altman technique. The analysis of results showed that delineation of the fiducial points on the GCG x- and y-axis was not available for 30% and 20% of cycles, respectively. In comparison to GCG x-axis, the GCG y-axis time intervals had smaller bias and variability with reference to the TDI measurements, which shows that the GCG y-axis estimates provide better approximations for cardiac intervals. However, the finding of this study does not confirm that CGC is an accurate tool for measuring the cardiac timing intervals.

Keywords: Cardiovascular, respiratory, and sleep devices - Sensors
Abstract: This minisymposium contribution focuses on the ongoing studies where we have provided ballistocardiogram (BCG) and electrocardiogram (ECG) sensing weighing scales to patients with heart failure (HF) for serial measurements at home. The BCG is a measurement of the mechanical aspects of cardiovascular function. The ultimate goal of the study is to assess whether BCG and ECG signals can provide value in predicting exacerbations for patients with HF at home, and thus be used to titrate care remotely. Thus far, we have recruited a total of 16 patients with HF for a total of 425 patient-days of recordings. Of these recordings, 365 were found to be usable based on signal quality indices (i.e., signal-to-noise ratio). Moreover, in one subject, the variability of BCG signal parameters over time was high at the beginning of the 30-day recording period, then stabilized at the end – this subject had inadvertently placed himself on a beta-blocker for the period of variability. We will present the findings thus far from our study, and discuss key lessons we have learned.

Keywords: Neural interfaces - Bioelectric sensors, Neural interfaces - Tissue-electrode interface, Neural interfaces - Body interfaces
Abstract: Ear-EEG is a recording method where EEG signals are acquired from electrodes placed on an earpiece inserted into the ear. Previously reported ear-EEG recordings have been performed with wet electrodes, and the objective of this study was to develop and evaluate dry-contact electrode ear-EEG. A novel dry-contact ear-EEG platform, comprising electrodes embedded in a soft-earpiece, was developed. The platform was evaluated in a study of four EEG paradigms: auditory steady-state response, steady-state visual evoked potential, mismatch negativity, and alpha band modulation. With both the measuring electrode and the reference electrode located within the ear, statistically significant (p<0.05) responses were measured for all paradigms, although for mismatch negativity it was necessary to use a reference located in the opposite ear, to obtain a statistically significant response. The prototyped dry-contact ear-EEG platform represents an important technological advancement of the method in terms of user-friendliness and suitability for long-term use.

Keywords: Neuromuscular systems - Locomotion, Neural interfaces - Bioelectric sensors, Neural signal processing
Abstract: We used high-density electroencephalography (EEG) to evaluate electrocortical dynamics during obstacle navigation using a novel dual-layer electrode approach for noise cancellation. After validating the technique on a phantom head, we collected data from subjects walking and running on a treadmill as they stepped over obstacles. Data revealed event related synchronizations in premotor, primary motor, and posterior parietal cortices tied to obstacles appearing on the treadmill. The results show it is possible to document the timing of brain network activity synchronized to reactive adjustments in human locomotion using mobile EEG.

Keywords: Neural signal processing
Abstract: This paper looks at the existing state-of-the-art use of wearable EEG in the area of sleep monitoring and the future challenges for the development of EEG-based wearable systems for the diagnosis of neurological sleep disorders.

Keywords: Neural interfaces - Body interfaces, Neural signal processing, Neural stimulation
Abstract: Closed loop EEG systems monitor the EEG and apply real-time signal processing to trigger state dependent auditory, visual or electrical stimulation. This talk will overview the technical platforms we are creating for delivering such interventions for personalized brain therapies.

Keywords: Neural interfaces - Bioelectric sensors, Brain-computer/machine interface, Brain functional imaging - EEG
Abstract: Currently available dry EEG electrodes are uncomfortable with prolonged use and can pose a safety hazard in many real-world scenarios. In order to overcome these practical challenges we have developed and tested soft, pliable electrodes and novel electrode form factors from existing materials.

Keywords: Brain physiology and modeling, Human performance, Neural interfaces - Body interfaces
Abstract: Novel methods of EEG data acquisition, cleaning and interpretation are notoriously difficult to validate due to a lack of ground truth. We advocate development and use of EEG “phantom devices” as a way to address this gap, and describe some initiatives towards this end.

Keywords: Diagnostic devices - Physiological monitoring, Cardiovascular assessment and diagnostic technologies
Abstract: We describe the design of a low-power, wireless sensor for biomedical applications, taking a Bluetooth-enabled oximeter and heart rate monitor as example. The two-chip solution consists of a battery-operated, radio-enabled SoC for signal processing and communication, and an optoelectronic front end for sensing. The whole system fits in an ear lobe clip.

Keywords: Clinical laboratory, assay and pathology technologies, Diagnostic devices - Physiological monitoring
Abstract: A biosensor system has been developed and tested which uses microfabricated gold electrodes which are chemically modified. Gold nanoparticles were chemically bound to the electrodes and a resulting electrical impedance change was measured. This type of test can be modified with biological components to create a versatile biosensor system for medical diagnostics, environmental monitoring, and research applications.

Keywords: Clinical laboratory, assay and pathology technologies, Diagnostic devices - Physiological monitoring
Abstract: 3rd generation nucleic acid sensors allow raw DNA/RNA measurements to be gathered at high rates in small, mobile platforms. However the computing resources needed to accurately predict nucleotide sequences from the raw measurements "at speed" are substantial and preclude portability. As a step towards addressing this issue, we present a recurrent neural network hardware design intended for embedded nucleic acid basecalling within the sequencing pipeline.

Keywords: Diagnostic devices - Physiological monitoring, Wearable or portable devices for vital signal monitoring
Abstract: Printed electronics has become of great interest over the last few years due to its advantages of a reduction of fabrication cost and the possibility to print on flexible substrates. Here we present the fabrication of temperature and humidity sensors by using thermal transfer printed interdigitated electrodes covered by graphene oxide. Graphene oxide acts as the temperature and humidity sensing material by changing its electrical properties as a function of the ambient conditions. These type of printed sensors could easily be integrated into biomedical applications such as smart bandages.

Keywords: Diagnostic devices - Physiological monitoring, Clinical laboratory, assay and pathology technologies, Neuromodulation devices
Abstract: We present preliminary work on the detection of neurotransmitters (NTs) in a microfluidic channel by visible light spectroscopic means. NTs are not active in the visible spectrum in general, and that complicates their detection by direct spectroscopic methods. Nevertheless, some NTs interfere with the absorption spectrum of gold nanoparticles (AuNPs) by shifting their 524 nm maximum absorbance wavelength. This opens the door to NT detection with visible light optical means, in addition to the possibility to integrate the whole system on a single chip for continuous measurement in real time.

Keywords: Medical devices interfacing with the brain or nerves, Wireless technologies for interrogation of implantable therapeutic devices, Diagnostic devices - Physiological monitoring
Abstract: This paper presents the recent advances in CMOS capacitive sensors that operate based on a charge based capacitance measurement (CBCM) technique. In this paper, the author briefly discusses the challenges in the design and implementation of core-CBCM capacitive sensors. Furthermore, the advantages of these high throughput sensors are put forward by demonstrating experimental results for monitoring the cellular activities,

Keywords: Sleep - Sleep apnea therapy, Sleep - Obstructive sleep apnea, Sleep - Periodic breathing & central apnea
Abstract: Connected devices for the treatment of sleep and respiratory disorders continue to evolve and enable telemonitoring, remote personalization and enriched patient engagement; elements of the P4 medicine paradigm. Recent database analyses and efforts in patient phenotyping have shown that personalized treatment is possible with the potential for improved compliance to chronic therapy.

Keywords: Sleep - Obstructive sleep apnea, Sleep - Sleep apnea therapy
Abstract: The sensitivity/stability of the ventilatory control system (often described by loop gain) is important in the pathophysiology of both central sleep apnea (CSA) and obstructive sleep apnea (OSA). In this presentation, we will review: (1) the current methods available to assess an individual’s loop gain; (2) interventions that can employed to manipulate it; and (3), whether knowledge of an individual’s underlying ventilatory sensitivity can help predict responses to commonly recommended treatments for these sleep-related breathing disorders.

Keywords: Sleep - Periodic breathing & central apnea, Sleep - Obstructive sleep apnea, Sleep - Cardiovascular & Metabolic consequences of sleep disorders
Abstract: Plant gain quantifies the extent and rapidity with which arterial blood gases change following hypopneic or hyperpneic events. Thus, it plays an important role in determining the total duration of apnea and/or hypopnea. In this paper, we report recent observations of elevated plant gain in obese adolescents and preterm-born children. We propose that elevated plant gain may be a manifestation of abnormal lung gas exchange in these subpopulations using a mathematical model.

Keywords: Sleep - Obstructive sleep apnea, Sleep - Sleep apnea therapy, Sleep - Snoring
Abstract: This last time-slot for the series of 3 back-to-back minisymposia covering the theme covering the theme: TOWARDS “P4 MEDICINE” IN SLEEP THERANOSTICS is dedicated to a general discussion of highlights from the presentations made by the minisymposia speakers, as well as the most recent advances in the field not covered in the previous papers.

Keywords: Image analysis and classification - Machine learning / Deep learning approaches
Abstract: This paper presents a case for inclusion of non-image data to extend the effectiveness of biomedical image analysis algorithms, and conversely the inclusion of elements of information derived from image data to improve “big data” approaches to the personalization of health care. It has been argued that this expansion of current practice could yield performance improvements as well as allowing differentiation of patient cohorts on a more systematic basis.

Keywords: Optical imaging - Coherence tomography, Ophthalmic imaging and analysis
Abstract: Optical coherence tomography allows in vivo, non-invasive imaging of tissue. We present two systems developed for the analysis of OCT images. AGARPLUS is developed for anterior segment ophthalmic imaging to automatically segment, measure and differentiate glaucoma subtypes. ASHIMA is developed for dermatological imaging to identify and separate the components of a skin OCT image for enhanced visualization and assessment of skin layers. Such systems could assist clinicians in the analysis of OCT images, with potential applications in screening and workflow improvement.

Keywords: Ophthalmic imaging and analysis, Image registration, segmentation, compression and visualization - Machine learning / Deep learning approaches, Image segmentation
Abstract: The fundus examination is effective for early detection of systemic hypertension, arteriosclerosis, etc. in the medical check-up or mass screenings. Many papers for blood vessels extraction and abnormalities detection have been published. In the recent years, deep convolution neural network (DCNN) has been applied to such studies instead of image features base techniques. This paper describes about retinal blood vessels extraction and microaneurysms detection by using patch based CNN.

Keywords: Image analysis and classification - Machine learning / Deep learning approaches, Ophthalmic imaging and analysis, Image registration, segmentation, compression and visualization - Machine learning / Deep learning approaches
Abstract: Early detection of glaucoma is important to slow down progression of the disease and to prevent total vision loss. We have been studying an automated scheme for detection of retinal nerve fiber layer defect (NFLD), which is one of the earliest signs of glaucoma on retinal fundus images. In our previous study, we investigated the use of a convolutional neural network (CNN) with deconvolutional layers for segmentation of NFLD regions. In this study, we investigated the use of CNN for detection of rectangular boxes enclosing NFLD regions. We also investigated the use of two networks in a cascade form to improve detection performance.

Keywords: Image feature extraction, CT imaging applications
Abstract: To identify prognostic imaging biomarkers in hepatocellular carcinoma (HCC) with biological interpretations by associating imaging features and gene modules. For the patients with CECT imaging data and gene expression profiles, intra-tumor partition was performed resulting in three spatially distinct subregions. Quantitative imaging features were extracted from each subregion. Prognostic gene modules were obtained, and their biological functions were annotated. The imaging features that significantly correlated with prognostic gene modules were selected, and their prognostic capabilities for overall survival (OS) were evaluated. The volume fraction of subregion, which was significantly correlated with all prognostic gene modules representing cancer-related interpretation, was predictive of OS. The texture feature cluster prominence in subregion, which was correlated with the prognostic gene module representing lipid metabolism and complement activation, also had the ability to predict OS. Imaging features of subregions have potentials to be predictors of OS with interpretable biological meaning.

Keywords: CT imaging, CT imaging applications, Image analysis and classification - Machine learning / Deep learning approaches
Abstract: Deep Learning is particularly suitable for medical imaging analysis. We aim at the characteristics of 3D low dose CT image volume data, and design an automatic segmentation algorithm for three-dimensional lung parenchyma to determine the range of lung cancer screening at high speed. We investigate the traditional solution for CT scan analysis, by which we apply 2D CT slice images and the state-of-the-art deep neural network architecture “Faster RCNN” for lung nodule detection. Moreover, on the basis of the multi task deep neural network architecture, the effective three-dimensional convolution operator is designed and the 3D deep neural network of multi task learning is constructed for this task.

Keywords: Health technology - Verification and validation, Clinical engineering, Neuromodulation devices
Abstract: This paper describes an electric field probe that could be used both in air and in a tissue to measure quasi-static electric fields of biomedical instruments from 100Hz to 300kHz. The probe uses a small dipole antenna connected via a long shaft with a simplified Dyson balun to an optical isolation amplifier with low common-mode gain. Induced electric field measured in air are presented for a quasi-static solenoidal field.

Keywords: Muscle stimulation, Neural stimulation (including deep brain stimulation), Computer modeling for treatment planning
Abstract: One major contributor to the degradation of quality of life for a significant portion of the adult population is chronic pain. In many cases, pharmacological solutions are limited in efficacy and alternative treatments, such as Transcutaneous Electrical Nerve Stimulation (TENS), are pursued. This paper describes simulation results that estimate the electric fields generated within an accurate computational human phantom placed within a full-body birdcage resonator operating at 100-200 kHz. The device may operate as a novel and effective TENS device, capable of deep stimulation over large body areas.

Keywords: Computer model-based assessments for regulatory submissions, Wearable or portable devices for vital signal monitoring
Abstract: Radio frequency wave propagation near the surface of a human body is highly sensitive to a number of items, including skin geometry, material composition and proximity to internal air-filled cavities. This study develops an anatomically realistic model of a human ear canal integrated into a full body computational phantom to address this final factor and examines, through numerical simulation, the impact of this level of detail on the power transmission of two-port and larger networks operating near or on the human body in the UHF band.

Keywords: Wireless technologies for interrogation of implantable therapeutic devices
Abstract: Comprehensive study of radio waves propagation for ingestible electronics is a very challenging task. Obtaining physical measurements is nearly impossible. And, although limited experimentation on animals or liquid phantom measurements are possible; the results are not quite reflective of the complex and inhomogeneous human body environment. Computational phantoms is another alternative; however, it requires sophisticated human body models along with appropriate transmitter & receiver antenna models. To initiate the study on UWB propagation from a wireless capsule endoscopy, a flexible and interactive immersive platform containing and an enhanced 3D body model has been developed. This platform enables researchers to conduct a comprehensive study of the UWB propagation channel for WCE applications.

Keywords: Computer model-based assessments for regulatory submissions, Image-guided devices - MRI-compatible instrumentation and device management, Cochlear implant
Abstract: Implant safety studies typically require a large number of simulations to test the compliance in various positions of the human body model (HBM). Applying the Huygens Box concept allows a significant speed-up of the process.

An equivalent field source (EFS) approach is used to efficiently investigate RF safety of various implants in an MRI system. The approach allows to first design the MRI system based on the full simulation domain and then replacing the MRI coil with the equivalent fields on a surface enclosing the patient for following calculations with variations in terms of implant type and location or uncertain tissue properties.

SAR distribution shows a good agreement between EFS and full run. A speed-up of a factor of 10 and more depending on the complexity of the full run can be reached for one implant evaluation at one position. Speed-up increases vastly with the number of evaluations.

The presentation or final paper will also include investigations on changes in the HBM like variations of tissue properties or breathing.

Keywords: Image-guided devices - RF and microwave ablation, Image-guided devices - Interstitial thermal therapy, Computer modeling for treatment planning
Abstract: Image-guided microwave ablation is clinically used to treat tumors in the liver and other organs. When treating large tumors, physicians may use multiple antennas simultaneous to rapidly create large volume ablation zones Pre-clinical experimental and simulation studies to characterize ablation patterns created by multiple antennas often presume parallel antenna insertion, which may not be possible in clinical practice. We employed coupled electromagnetic – heat transfer simulations, incorporating temperature dependent tissue dielectric properties, to assess the impact of antenna misalignment on ablation zone profiles. Modeling results were validated with experiments in ex vivo liver tissue. For inter-antenna spacing in the range of 10 – 20 mm, the Dice Similarity Coefficient between parallel and non-parallel ablation zones ranged between 0.73 and 0.95.

Keywords: Brain-computer/machine interface, Neural signal processing, Neurorehabilitation
Abstract: Insufficient training data is a serious problem in all domains related to bioinformatics. Transfer learning is a promising tool to solve this problem, which relaxes the hypothesis that training data must be independent and identically distributed with the test data. We construct a sophisticated electroencephalography (EEG) signal representation and obtain an efficient EEG feature extractor through manifold constraints-based joint adversarial training with training data from other domains. EEG signal is more easily distinguished in the feature space mapped by the feature extractor. Negative transfer is one of the most challenging problems in transfer learning. In our approach, we apply manifold constraints to overcome this problem, which can avoid the geometric manifolds in the target domain being destroyed. The experiments demonstrate that our approach has many advantages when applied to EEG classification tasks.

Keywords: Brain-computer/machine interface, Human performance - Cognition, Brain functional imaging - EEG
Abstract: In the field of brain-machine interface (BMI) research, the development of cognitive BMI is a hot topic because it may lead to more intuitive and goal-directed findings than existing BMI technology. In this study, we devised a “visual-imagery method,” which enables visual imaging of the operation of a target. We also investigated an “inner-speech method,” which comprised internal pronunciation of words without emitting sounds, and an “inner-speech + visual-imagery method,” which combined the two methods. When only the high  band (60-120 Hz) power in the prefrontal cortex was used, the average accuracy of the 15 participants, with 20-fold cross-validation, was 81.3% in inner speech, 84.6% in visual imagery, and 83.2% in inner speech + visual imagery. This study also found that the frontal pole was the most useful region in the prefrontal cortex.

Keywords: Brain-computer/machine interface, Brain functional imaging - MEG, Human performance - Cognition
Abstract: Recently, brain computer interface (BCI) technologies that control external devices with human brain signals have been developed. However, most of the BCI systems, such as P300-speller, can only discriminate among options that have been given in advance. Therefore, the ability to decode the state of a person's perception and recognition, as well as that person's fundamental intention and emotions, from cortical activity is needed to develop a more general-use BCI system. In this study, two experiments were conducted. First, articulations were measured for Japanese monosyllabic utterances masked by several levels of noise. Second, auditory brain magnetic fields evoked by the monosyllable stimuli used in the first experiment were recorded, and neuronal current sources were localized in regions associated with speech perception and recognition — the auditory cortex (BA41), the Wernicke's area (posterior part of BA22), Broca's area (BA22), motor (BA4), and premotor (BA6) areas. Although the source intensity did not systematically change with SNR, the peak latency changed along SNR in the posterior superior temporal gyrus in the right hemisphere. The results suggest that the information associated with articulation is processed in this area.

Keywords: Brain-computer/machine interface, Neural signals - Coding
Abstract: Advances in neurotechnology are expected to provide access to thousands of neural channel recordings including neuronal spiking, multiunit activity and local field potentials. In addition, recent studies have shown that deep learning, in particular recurrent neural networks (RNNs), provide promising approaches for decoding of large-scale neural data. These approaches involve computationally intensive algorithms with millions of parameters. In this context, an important challenge in the application of neural decoding to next generation brain-computer interfaces for complex human tasks is the development of low-latency real-time implementations. We demonstrate a Field-Programmable Gate Array (FPGA) implementation of Long Short-Term Memory (LSTM) RNNs for decoding 10,000 channels of neural data on a mobile low-power embedded system platform called "NeuroCoder". We provide a proof of concept in the context of decoding 20-dimensional spectrotemporal representation of spoken words from simulated 10,000 neural channels. In this particular case, the LSTM model included 4,042,420 parameters. In addition to providing multiple communication interfaces for the BCI system, the NeuroCoder platform can achieve sub-millisecond real-time latencies.

Keywords: Brain-computer/machine interface, Motor neuroprostheses, Neurological disorders
Abstract: Access to the brain to implant recording electrodes has conventionally required a craniotomy. To mitigate risks of open brain surgery, we previously developed a stent-electrode array that can be delivered to the cortex via cerebral vessels. Following implantation of a stent-electrode array (Stentrode) in a large animal model, we investigated the longevity of high-quality signals, by measuring bandwidth in animals implanted for up to six months; no signal degradation was observed. We also investigated whether bandwidth was influenced by implant location with respect to the superior sagittal sinus and branching cortical veins; it was not. Finally, we assessed whether electrode orientation had an impact on recording quality. There was no significant difference in bandwidths from electrodes facing different orientations. Interestingly, electrodes facing the skull (180°) were still able to record neural information with high fidelity. Consequently, a minimally invasive surgical approach combined with a stent-electrode array is a safe and efficacious technique to acquire neural signals over a chronic duration.

Keywords: Brain-computer/machine interface, Motor neuroprostheses, Neural signal processing
Abstract: Non-invasive brain computer interface (BCI) has been successfully used to control cursors, helicopters and robotic arms. However, this technology is not widely adopted by people with late-stage amyotrophic lateral sclerosis (ALS) due to poor effectiveness. In this study, we attempt to assess the cognitive state of a completely locked-in ALS subject, and her ability to use motor imagery-based BCI for control. The subject achieves above chance level accuracies for both open loop (62.2%) and closed-loop (68.7%) 2-class movement vs. idle decoding. We also observe a prominent theta oscillation with peak frequency at 4.5 Hz during the experiments. Quantification shows that the theta oscillatory power increases during motor imagery tasks compared to idle tasks for both open-loop as well as closed-loop BCI tasks. Furthermore, for closed-loop sessions, theta oscillation power correlates positively with feedback accuracy during movement tasks, and negatively with feedback accuracy during idle tasks. Our study demonstrates the feasibility of motor imagery-based BCI for late-stage ALS subjects, and highlights the importance of feedback during BCI implementation.

Keywords: Data mining and processing - Pattern recognition
Abstract: Steady-state Visual Evoked Potentials, SSVEP), as the most commonly used communication paradigm for non-implantable Brain-Computer Interface (BCI), boasts the advantages of no training, noise immunity and obvious periodicity. The traditional SSVEP extraction methods can effectively identify the target frequency contained in original EEG, however, the required data length usually lasts a few seconds. In this paper, bistable stochastic resonance (BSR) is applied to SSVEP extraction. BSR is very sensitive to amplitude mutation and frequency fluctuation of the input signal, making the output difference can be used for the detection of the target frequency. The processing results illustrate that the proposed method not only has a high recognition accuracy, but also effectively shortens the recognition time, thus improving the calculating speed. Therefore, SSVEP Extraction has a higher information transfer rate (ITR) based on bistable stochastic resonance, which is more suitable for the real-time BCI system.

Keywords: Data mining and processing in biosignals
Abstract: A mono-feature fuzzy index that evaluates the stress level from one feature extracted from ECG or GSR is presented. It is build using several measures of the feature recorded when the subject is at rest. The mono-feature fuzzy index can be merged in a multi-feature stress index without any tuning. It can be used to select relevant features and to detect stress. The performance of the stress index is analyzed on a data set made of 160 time periods of time when 20 subjects had to perform stressful tasks and corresponding control tasks. The stress was induced by 4 different tasks. The performances reached are 72% of correctly classified time periods in stress and no stress situations. Interesting conclusions could also be made on the tasks ability to induce stress.

Keywords: Data mining and processing in biosignals, Signal pattern classification, Physiological systems modeling - Signal processing in physiological systems
Abstract: Hot flashes (HF) are intense, transient feelings of heat usually accompanied with flushed skin and sweating that are experienced by women around the time of menopause. HFs are associated with poor quality of life and increased cardiovascular risk. Automatic detection of HF occurrence and precise timing of HF onset could provide unique insight into the physiology of the HF and its effect on the cardiovascular system. A novel automatic algorithm is proposed for the detection of HFs occurrence and timing from the sternal skin conductance signal that is robust to noise and artifacts. The method is based on the gold standard rule (2 μS rise in skin conductance within 30 s) and considers several conditions based on the skin conductance level and its derivative to reject unwanted events. ECG-derived heart rate pattern variations are studied prior to the detected HF onset. The algorithm is validated against expert detected HFs over 200 hours of sleep data collected from 12 perimenopausal women. It achieved a total accuracy of 93% and a total error of 3% in HF detection. It was observed that heart rate increased before the onset of 80% of the HFs occurring in undisturbed sleep. Application of this algorithm along with fusion of other simultaneously recorded physiological measures has the potential to advance understanding of the HF.

Keywords: Data mining and processing - Pattern recognition, Neural networks and support vector machines in biosignal processing and classification, Signal pattern classification
Abstract: Electromyographic activities (EMG) generated during contraction of upper limb muscles can be mapped to distinct hand gestures and movements, posing them as a promising modality for prosthetic and cybernetic applications. This paper presents a comparative analysis between different recurrent neural network (RNN) configurations for EMG-based hand gesture classification. In particular, RNNs with recurrent units of long short-term memory (LSTM) and gated recurrent unit (GRU) are evaluated. Furthermore, the effects of an attention mechanism and varying learning rates are evaluated. Results show a classifier 1)~with a bidirectional recurrent layer composed of LSTM units, 2)~that applies the attention mechanism, and 3)~trained with step-wise learning rate outperforms all other tested RNN classifiers.

Keywords: Data mining and processing in biosignals, Signal pattern classification
Abstract: The aim of this study is to investigate the validity of an entropy-based objective assessment of cerebellar ataxia patients performing rhythmic tapping. Previous research conducted, particularly in time and frequency domains, tested the adherence of patients to more stringent experimental requirements. These requirements may inadvertently cause higher level brain functions to influence the performance and possibly obscure the cerebella related disabilities in the data stream. In this study, a multiscale entropy-based learning process that overcomes this practical limitation was considered. In particular, assessment techniques with less restrictions on the tapping duration were considered. Thirty-three patients were engaged in the test, with three levels of severity 0 (normal), 1 (moderate) and 2 (severe) ranked by specialist clinicians. The performance of each model was evaluated using leave-one-out cross validation. Results from both time-frequency features and entropy features extracted and characterized the cerebellar condition captured during the finger and foot tapping tests (with over 80% accuracy). Strong correlations with clinical assessment-based scoring were observed with the entropy based approach for both tests, although the correlation with time-frequency features were less convincing.

Keywords: Data mining and processing in biosignals
Abstract: In this paper an analysis of compression schemes based on compressed sensing (CS) and predictor techniques for neural signals is presented. The focus is on how much a compression algorithm can reduce data while not affecting the subsequent signal processing. Since neural signals are processed by means of spike sorting algorithms the evaluation is not trivial and not well defined, since there exists in fact many different ways to detect and cluster the spikes. Evaluating how much a compression scheme affects the result of spike sorting programs is a crucial step before implementing such compression technique. In the analysis two use cases are evaluated: in the first, spikes are detected and extracted and only thereafter compressed. In the second case, no information on the spikes is available and the whole raw signal is compressed. When dealing only with spike frames CS offers great compression at almost no loss, in the case of the whole recording its performances are greatly impaired and delta compression outperforms it in terms of data reduction and spike sorting results. In this case the reduction rates are modest but significant, ≈3-4 times data reduction and the whole signal is preserved avoiding big permanent losses of information.

Keywords: Image reconstruction - Performance evaluation, Image reconstruction - Fast algorithms, Iterative image reconstruction
Abstract: A novel region-based method to track beating heart is proposed. Sparse statistical pose modeling is used to reconstruct the region of interest (ROI) on beating heart surface. Firstly, a high-complexity thin plate spline is employed to pre-reconstructed the ROI of a series of frames. The 3D pose data of the ROI from the pre-reconstructed results are extracted to train a low-complexity model based on the sparse statistical analysis. The new trained low-complexity model is robust and efficient for ROI reconstruction of the following frames. The proposed model significantly reduces the redundant degrees of freedom to fit the surface of the heart. A constraint item is added to the objective function which describes the 3D tracking problem to avoid erroneous convergence of the efficient second-order minimization (ESM) optimization algorithm. The proposed method is evaluated on the phantom heart video and the in vivo video obtained by the da Vinci surgical system.

Keywords: Cardiac imaging and image analysis, Magnetic resonance imaging - Cardiac imaging, Magnetic resonance imaging - Dynamic contrast-enhanced MRI
Abstract: The use of implantable cardiac devices has in- creased in the last 30 years. Cardiac resynchronisation therapy (CRT) is a procedure which involves implanting a coin sized pacemaker for reversing heart failure. The pacemaker electrode leads are implanted into cardiac myocardial tissue. The optimal site for implantation is highly patient-specific. Most implanters use empirical placement of the lead. One region identified to have a poor response rate are myocardial tissue with transmural scar. Studies that precisely measure transmurality of scar tissue in the left ventricle (LV) are few. Most studies lack proper validation of their transmurality measurement technique. This study presents an image analysis technique for computing scar transmurality from late-gadolinium enhancement MRI. The technique is validated using phantoms under a CRT image guidance system. The study concludes that scar transmurality can be accurately measured in certain situations and validation with phantoms is important.

Keywords: Magnetic resonance imaging - Cardiac imaging, Magnetic resonance imaging - Pulse sequence, Cardiac imaging and image analysis
Abstract: Four-dimensional phase-contrast (PC) velocity-encoded flow magnetic resonance imaging (4D flow MRI) is a potentially valuable tool for studying cardiovascular hemodynamics for disease monitoring and/or treatment planning. In this study we compared the performance of two 4D flow MRI pulse sequences - echo-planar imaging (EPI) and segmented gradient-echo (turbo-field-echo or TFE on vendor’s platform) - on a clinical 3T system in 6 human subjects including 3 patients with Tetralogy of Fallot (TOF). For aortic flow rate, the coefficients of variation (COV) between 2D and 4D EPI were 7.0% and 7.7% for controls and patients respectively. The corresponding COV between 2D and 4D TFE were 19.0% and 18.3% for controls and patients respectively. The COV between 4D TFE and 4D EPI were larger than 18.7% in kinetic energy analysis. 4D EPI demonstrated acceptable accuracy of intra-cardiac flow quantification, which was also shown in the ex-vivo phantom measurements.

Keywords: Image segmentation, Ultrasound imaging - Cardiac, Cardiac imaging and image analysis
Abstract: Segmentation of the left ventricle (LV) in temporal 3D echocardiography sequences poses a challenge. However, it is an essential component in generating quantitative clinical measurements for the diagnosis and treatment of various cardiac diseases. Identifying the endocardial borders of the left ventricle can be difficult due to the inherent properties of ultrasound. This study proposes a 4D segmentation algorithm that segments over temporal 3D volumes that has minimal user interaction and is based on a diffeomorphic registration approach. In contrast to several existing algorithms, the proposed method does not depend on training data or make any geometrical assumptions. The algorithm was evaluated on seven patients obtained from the Mazankowski Alberta Heart Institute, Edmonton, Canada in comparison to expert manual segmentation. The proposed approach yielded Dice scores of 0.94 (0.01), 0.91 (0.03) and 0.92 (0.02) at end diastole, at end systole and over the entire cardiac cycle, respectively. The corresponding Hausdorff distance values were 4.49 (1.01) mm, 4.94 (1.41) mm, and 5.05 (0.85) mm, respectively. These results demonstrate that the proposed 4D segmentation approach for the left ventricle is robust and can potentially be used in clinical practice.

Keywords: Image registration, segmentation, compression and visualization - Machine learning / Deep learning approaches, Cardiac imaging and image analysis, Magnetic resonance imaging - Cardiac imaging
Abstract: Accurate delineation of heart substructures is a prerequisite for abnormality detection, for making quantitative and functional measurements, and for computer-aided diagnosis and treatment planning. Late Gadolinium-Enhanced Cardiac MRI (LGE-CMRI) is an emerging imaging technology for myocardial infarction or scar detection based on the differences in the volume of residual gadolinium distribution between scar and healthy tissues. While LGE-CMRI is a well-established non-invasive tool for detecting myocardial scar tissues in the ventricles, its application to left atrium (LA) imaging is more challenging due to its very thin wall of the LA and poor quality images, which may be produced because of motion artefacts and low signal-to-noise ratio. As the LGE-CMRI scan is designed to highlight scar tissues by altering the gadolinium kinetics, the anatomy among different heart substructures has less distinguishable boundaries. An accurate, robust and reproducible method for LA segmentation is highly in demand because it can not only provide valuable information of the heart function but also be helpful for the further delineation of scar tissue and measuring the scar percentage. In this study, we proposed a novel deep learning framework working on LGE-CMRI images directly by combining sequential learning and dilated residual learning to delineate LA and pulmonary veins fully automatically. The achieved results showed accurate segmentation results compared to the state-of-the-art methods. The proposed framework leads to an automatic generation of a patient-specific model that can potentially enable an objective atrial scarring assessment for the atrial fibrillation patients.

Keywords: Ultrasound imaging - Cardiac, Image enhancement, Cardiac imaging and image analysis
Abstract: Analysis of wall motion abnormality using echocardiography is an established method for detecting myocardial ischemia. We describe a hybrid approach of enhancing 2D+T echo datasets with border detection and Eulerian motion magnification to improve the visual assessment of wall motion. We implemented a local phase-based approach using the monogenic signal and its derived features, either feature asymmetry (FA) or oriented feature symmetry (OFS), to detect boundaries of the heart structure. We enhanced the 2D+T datasets using either an intensity-based or phase-based Eulerian Motion Magnification (EMM) video processing technique, and identified among eight different types of enhancements the best performing method as OFS with an accuracy of 78% versus the original B-Mode with an accuracy of 71%.

Keywords: Clinical engineering, Diagnostic devices - Physiological monitoring, Health technology management and assessment
Abstract: Localized electrical bioimpedance measurements quantify the passive electrical properties of a tissue. Recently bioimpedance methods have been investigated as a technique to monitor localized tissues for changes from the contraction and fatigue of skeletal muscle. The positive results from these studies support the further investigation and application of this technique to quantify exercise-induced changes in bioimpedance as a marker for fatigue status and recovery from fatigue.

Keywords: Diagnostic devices - Physiological monitoring, Image-guided devices - Biopsy
Abstract: Electrical Impedance Tomography (EIT) maps the electrical properties of biological systems. While the resolution of this modality is somewhat limited, it provides high levels of sensitivity and specificity in terms of pathology identification and assessment. By coupling EIT with ultrasound (US) as a multi-modal imaging system one can achieve higher resolution electrical property imaging with more sensitivity and specificity than US alone. Multiple approaches to implementing EIT-coupled US imaging are possible including geometric constrained, hard-prior, and soft-prior imaging.

Keywords: Diagnostic devices - Physiological monitoring, Wearable or portable devices for vital signal monitoring
Abstract: Electrical impedance tomography (EIT) has shown a great potential as a new portable real-time imaging modality in lung imaging. Most previous studies have focused on regional lung ventilation imaging during mechanical ventilation. We proposed two new clinical applications of EIT in pulmonary medicine including obstructive sleep apnea (OSA) diagnosis and pulmonary function test (PFT). In this paper, we report our latest experimental results of these two topics. The application of EIT for OSA diagnosis was tested in a series experimental studies where lung EIT and polysomnography (PSG) were simultaneously performed. For the application in PFT, we combined an existing spirometer with our lung EIT device as a spirotomometer. Preliminary experimental results are promising and suggest further clinical studies.

Keywords: Wearable or portable devices for vital signal monitoring, Diagnostic devices - Physiological monitoring
Abstract: This minisymposium contribution focuses on our ongoing studies focused on using electrical bioimpedance (EBI) to detect edema in joints. The ultimate goal is to incorporate EBI sensing into a wearable brace, such that joint edema levels can be longitudinally monitored by patients at home. This could provide benefit, for example, to patients rehabilitating acute knee injuries, delivering quantitative feedback on the healing progress such that therapies and rehabilitation protocols can be adjusted based on changing patient needs. We have performed experiments in human subjects with acute knee injuries and compared the EBI measures between the injured and contralateral side, demonstrating that significant changes exist due to fluid accumulation on the injured side. In this contribution, we also present results of a study performed in a pig limb, with small amounts of saline (2-10 mL) infused into the limb and detected effectively by the EBI hardware.

Keywords: Magnetic resonance imaging - Cardiac imaging, Image reconstruction and enhancement - Machine learning / Deep learning approaches
Abstract: Deep learning approaches have shown great success in many medical imaging applications. Here, we will demonstrate that deep learning approaches can be used to solve a range of tasks in cardiac Magnetic Resonance Imaging (MRI) ranging from image reconstruction to image super-resolution enhancement.

Keywords: Image registration, segmentation, compression and visualization - Machine learning / Deep learning approaches, Image analysis and classification - Machine learning / Deep learning approaches
Abstract: In this talk, we will provide an overview of machine learning techniques with application to cancer imaging, for a range of imaging modalities. One application of interest is one the most common cancers in both women and men: Colorectal cancer. We demonstrate how one can study tumour characteristics by using different machine learning approaches, including deep learning concepts.

Keywords: Image analysis and classification - Machine learning / Deep learning approaches, Image registration, segmentation, compression and visualization - Machine learning / Deep learning approaches, Multivariate image analysis
Abstract: This talk will showcase some of the work carried out at the BCN Medtech research unit of Universitat Pompeu Fabra in Barcelona, Spain. In particular, the focus will be on machine learning approaches applied to medical applications, including the use of manifold learning and multiple kernel learning for cardiology, learning approaches for label fusion in brain image segmentation for the study of Alzheimer’s disease and perinatal brain development, and the use of advanced neural networks for fetal surgery planning.

Keywords: Cardiac imaging and image analysis, Image segmentation, CT imaging
Abstract: In this presentation, we will demonstrate applications of deep learning methods for cardiovascular CT image analysis. We will focus on segmentation and quantification for cardiovascular risk determination in CT scans as a requested or unrequested finding. Furthermore, we will show an application of generative deep learning that allows determination of cardiovascular risk in CT images acquired with very low radiation dose. Finally, we will show how a combination of conventional machine learning and deep learning can be used to detect functionally significant coronary artery stenosis in CT angiography scans.

Keywords: Image analysis and classification - Machine learning / Deep learning approaches
Abstract: Machine learning techniques generally assume that the training data that is used to learn model parameters is representative of the target data to analyze. If this assumption does not hold, performance will deteriorate. For machine learning in medical imaging this means that accuracy may decrease when training data is acquired at another hospital, with a different scanner, with slightly different scan parameters, or from a different patient population. In this talk, we will provide an overview of techniques to compensate for differences between train and target data. We demonstrate how these transfer learning techniques improve accuracy across scanners for a range of applications in medical image analysis.

Keywords: Sensory neuroprostheses - Visual, Neural interfaces - Implantable systems, Neural interfaces - Tissue-electrode interface
Abstract: Photovoltaic subretinal prosthesis converts light into electric current, stimulating the nearby inner retinal neurons. Visual information is projected onto the implant by video goggles using pulsed near-infrared (880-915nm) light. Such design avoids the need for bulky electronics and wires, thereby greatly reducing the surgical complexity, and allows scaling the number of pixels to thousands. Like in normal vision, retinal response to prosthetic stimulation exhibits flicker fusion at high frequencies (>20 Hz), adaptation to static images, antagonistic center-surround organization, and non-linear summation of subunits in the RGCs receptive fields, providing much higher spatial resolution than the average size of receptive fields. Such implants elicited responses to both onset and offset of light, with approximately 1/6th of the natural contrast sensitivity. Photovoltaic arrays with 55um pixels implanted in blind rats restored a grating visual acuity up to the pixel pitch. If these results translate to human retina, such implants could restore visual acuity up to 20/200. Higher resolution may be achieved utilizing 3-dimensional electro-neural interfaces.

Keywords: Neural stimulation, Sensory neuroprostheses - Visual
Abstract: Electric stimulation of the retina via retinal implants is currently the only commercially available method to restore vision in patients suffering from a wide range of outer retinal degenerations. To improve the quality of retinal implants, it is desirable to better understand how different retinal cell classes and types respond to electric stimuli so that more effective stimulation strategies can be developed. Here, we measured the response of seven major types of retinal ganglion cells to electric stimulation. A simple series of light stimuli were used to classify cells into known types. Electric stimulation produced unique responses in most ganglion cell types and the responses typically matched elements of the corresponding light responses.

Keywords: Neural stimulation, Sensory neuroprostheses - Visual
Abstract: Recent retinal studies have demonstrated the ability to differentially recruit ON and OFF retinal ganglion cells (RGCs), using high frequency stimulation (HFS). By a combination of in vitro experiments and computational models we further investigated whether preferential excitation of ON and OFF RGCs can be optimized through a combination of stimulation parameters and stimulus electrode location relative to the axon initial segment (AIS) and axon direction. Stimulation parameters investigated included HFS pulse amplitude and HFS pulse train frequency.

Keywords: Sensory neuroprostheses - Visual, Neural stimulation, Neural interfaces - Tissue-electrode interface
Abstract: Retinal prostheses use a multi-electrode array to apply a pattern of electrical stimulation to the retina. Targeting specific cellular or subcellular structures (i.e. axon, soma) can potentially result in better visual acuity. In vitro mammalian retina studies can be used to investigate possible mechanisms for targeting.

Keywords: Sensory neuroprostheses - Visual, Neural stimulation, Sensory neuroprostheses
Abstract: In recent years, we have been utilizing the voltage-sensitive dye (VSD) imaging technique to measure the neural responses to microstimulation in the visual cortical areas of rodents in vivo, and in the slice preparations in vitro. These imaging experiments enable us to estimate, for example, the threshold and saturation charges for a single stimulus pulse to induce neural excitation in the immediate vicinity of the stimulating electrode, or the spatial extent of population spike initiated within a few milliseconds after the stimulus onset. Therefore, the VSD imaging is thought to be useful to gain insight into the design of safe and efficacious stimulation strategy for neural prostheses.

Keywords: Sensory neuroprostheses - Visual, Neural interfaces - Implantable systems, Neural stimulation
Abstract: Deaf children fitted with cochlear implants now attend mainstream schools and lead lives as part of the hearing world; artificial limbs have achieved Olympic caliber performance; and, more recently, the profoundly blind are capturing their first glimpses of what the future will bring through retinal neuroprosthesis technologies. While it is tempting to say, ‘mission accomplished’, there remains an enormous gap between natural and artificially-modulated neural activity. A key element in our quest towards making artificial neuromodulation more natural is the unlocking of the neural code – that is, the elicitation of neural firing that describes what we hear, how we should move, and how we see. The spatial and temporal constituents of this code are only now being exploited to improve outcomes. Here we describe our work towards unlocking of the neural code for the restoration of vision.

Keywords: Rehabilitation robotics and biomechanics - Exoskeleton robotics, Biomechanics and robotics - Clinical evaluation in rehabilitation and orthopedics, Assistive and cognitive robotics in rehabilitation
Abstract: Abstract— This talk will provide a short overview of Robotic exoskeletons and how they are used in different fields for enhancing human locomotion.

Wearable robotics, specifically robotic exoskeletons research that has gained substantial attention in recent years. Interestingly, the acceptance of these devices into everyday clinical practice and home use is still limited by past and present research. Currently, there exists a void between the research evidence and clinical evidence on the predicated utility and actual usability of wearable robotic systems. Our talk will discuss on how engineering and clinical science can be combined and performed simultaneously and sequentially to gain insight to clinical utility of wearable robots to specific clinical populations and to understand the needed continued adaptations in the hardware and controller mechanisms of these devices when they are used in disabled populations who struggle with differential muscle weakness, range of motion limitations, cognitive disabilities, variable balance and altered neuromotor control. We will discuss how specific wearable robots can be used to target specific populations and how they provide clinical benefits. Additionally, we will talk about training strategies and complexities to consider when training individuals with varying disabilities to take these devices as personal mobility devices home. Finally, we will discuss how robotic exoskeletons can be transitioned to being used as recreational or performance augmentation devices in everyday life. This information will help clinicians and scientists gain some additional insight into how these eloquent technologies can be seamlessly transitioned further into the field of rehabilitation.

Keywords: Therapeutic robotics in rehabilitation, Wearable robotic systems - Orthotics
Abstract: Sensorimotor impairment after stroke presents a multifaceted challenge for recovery. By combining robotic technologies with other treatment modalities, a holistic approach can be advanced to address a range of impairment mechanisms. We are conducting an ongoing clinical trial in which a pharmacological agent, cyproheptadine, is administered in conjunction with training of muscle activation patterns through custom, electromyographically controlled orthoses and computer games.

Keywords: Biomechanics and robotics - Clinical evaluation in rehabilitation and orthopedics, Rehabilitation robotics and biomechanics - Devices and methods for assessment and therapy in infancy
Abstract: How can rehabilitation clinicians take advantage of new technologies to help individuals improve movement function after neurologic injury? People recovering from nervous system disease or injury are faced with many constraints in terms of how to use remaining neural resources to recover functional movement. In particular, people post-stroke are faced with loss of corticofugal fibers that can recruit and drive interneuronal and motoneuronal circuits in a coordinated and effective manner that results in successful movement tasks. Remaining circuits may allow for alternate pathways to be recruited, so the potential exists for improved movement patterns, even under the constraints imposed by lost input pathways. Collaborative robots allow an interaction between the patient and the machine so that the patient can be challenged by the machine and the machine can respond to the patient’s movements to provide clear feedback about success or failure of a given task. We have used a collaborative robotic system, The KineAssist, to develop an approach to clinical testing and training that takes advantage of the specific biomechanical requirements of individual movement tasks in order to measure the highest performance capacity of a task and then use that measurement to personalize a training regimen designed to expand that highest performance capacity.

Keywords: Hardware and control developments in rehabilitation robotics, Wearable robotic systems - Orthotics, Robotic prosthetics
Abstract: This contribution advocates the principle of complementarity for the design of devices that support human locomotion, whereby a fully complementary design avoids any duplication of human mechanical or control functionalities. Several examples are provided.

Keywords: Rehabilitation robotics and biomechanics - Devices and methods for assessment and therapy in infancy, Hardware and control developments in rehabilitation robotics, Neural-robotic interfaces
Abstract: Making use of visual display technology and human-robotic interfaces, many researchers have illustrated various opportunities to distort visual and physical realities. Judicial application of these leads to training situations that enhance the learning process, and can restore movement ability after neural injury. We will trace out clinical studies that have employed such technologies to improve the health and function, as well as share the remaining needs future insights.

Keywords: Neural interfaces - Bioelectric sensors
Abstract: Among functional brain scanning methods, electroencephalography (EEG) is the most promising method for non-invasive brain monitoring in real-life environments. Ear-EEG is a method where signals are recorded from electrodes placed on an earpiece inserted into the ear. The compact and discreet nature of ear-EEG devices makes it suitable for long-term real-life recordings. In this study, EEG were recorded from 6 subjects with conventional scalp EEG and dry-contact ear-EEG. All recordings were performed with the same instrumentation and paradigms in both a lab setting and a real-life setting. The study comprised four paradigms: auditory steady-state response (ASSR), steady-state visual evoked potential (SSVEP), auditory onset response, and alpha band modulation. For both settings, the investigated responses were observable and statistically significant (p<0.05) in recordings from ear-electrodes referenced to a scalp electrode (Cz). Statistically significant ASSR and SSVEP were measured in the lab setting by ear-electrodes referenced to an electrode within the same ear. In the real-life setting, only the ASSR was statistically significant for a reference within the same ear. The results demonstrates that ear-EEG recordings can be performed with dry-contact electrodes in real-life.

Keywords: Brain functional imaging - EEG, Human performance - Activities of daily living, Human performance - Attention and vigilance
Abstract: Ear-centered EEG has the potential to open up new applications for research, diagnostic and therapy. The combination of ear-electrodes, miniaturized EEG amplification and mobile signal acquisition allows brain monitoring beyond the lab in real life situations. However, ear-EEG solutions come at the cost of a limited coverage of the head. There is trade-off between the minimal disturbance of the user and the sensitivity to the signals of interest. Placing electrodes around as well as in the ear canal provides the optimal solution.

Keywords: Brain-computer/machine interface, Sensory neuroprostheses - Auditory, Neural signal processing
Abstract: Recent technological advances in the design of concealable miniature electroencephalography (mini-EEG) devices are paving the way towards 24/7 neuromonitoring applications in daily life. However, such mini-EEG devices only cover a small area and record EEG over much shorter inter-electrode distances than in traditional EEG headsets. These drawbacks can potentially be mitigated by deploying a multitude of such mini-EEG devices. In this study, we investigate the effect of using such multi-node EEG recordings with short inter-electrode distances and galvanic separation between the nodes for a use-case in auditory attention detection (AAD). We demonstrate that the AAD performance using galvanically separated short-distance EEG measurements is comparable to using long-distance EEG measurements if in both cases the electrodes are optimally placed on the scalp using a group-LASSO optimization.

Keywords: Neurological disorders - Diagnostic and evaluation techniques, Brain-computer/machine interface, Brain functional imaging - EEG
Abstract: Flexible electrodes offer exciting opportunities in the realm of wearable electrophysiology. In particular, tattoo electrodes are markedly soft and conform neatly to the wearer skin allowing nearly artifact-free facial sEMG recordings. As such, tattoo electrodes offer a unique opportunity to combine EEG and facial EMG capabilities to capture neural and motor functions. Here we report the design, implementation and testing of wireless tattoo-based electrodes for REM sleep behavior disorder (RBD) identification. EMG, EOG and EEG were successfully recorded on a smart phone demonstrating the potential use in the home environment.

Keywords: Neural signals - Information theory
Abstract: In this paper, we study canonical correlation analysis (CCA), which is an attractive way to investigate the correlations between multidimensional datasets, to address the cocktail party problem. Our main contribution is a new approach that utilizes CCA for the classification of the attended sound in acoustically highly complex, cocktail-party settings. Extensive test results show that single-trial attention classification accuracy rates, based on  60 second long batches of data from full-scalp EEG, are satisfactorily high ( 95%), and remain equally high for the subset of only 24 electrodes close to the temporal lobe. These results suggest that CCA is a promising tool in the auditory attention classification in different cocktail-party scenarios, compared with other related approaches. These findings suggest the full utility of CCA in the development of intelligible attention-steered brain-computer interfaces (BCIs) and hearing aids (HAs).

Keywords: Human performance - Sleep, Neural signals - Blind source separation (PCA, ICA, etc.), Brain functional imaging - EEG
Abstract: Polysomnography (PSG) with head-mounted EEG electrodes represents the gold standard in sleep research, but can be cumbersome to mount and is not ideal to sleep with. Here we report the exploration of a new type of superthin electrode, the cEEGrid, for discrete and user-friendly ear-EEG recordings for sleep–wake assessment. Expected modulations of alpha and delta power across the night were confirmed in the data recorded with the new cEEGrid electrode. We also computed correlations between sleep parameters, on the one hand derived from manually annotated PSG data and on the other hand from fully automated cEEGrid data analysis. Obtained correlation values were large and support the idea that the cEEGrid concept is a viable tool for sleep research.

Keywords: Fetal and Pediatric Imaging, CT imaging applications, Image feature extraction
Abstract: Imaging-based diagnosis and treatment planning in pediatric craniofacial diseases must incorporate growth knowledge. Here, we will review our past efforts to design and disseminate methods that account for the dynamic growth happening in the different skull structures (i.e. bone, brain) during childhood. We believe this is crucial to accelerate the pace of reproducible research in pediatric craniofacial analysis.

Keywords: Ultrasound imaging - Prenatal, Image analysis and classification - Machine learning / Deep learning approaches, Fetal and Pediatric Imaging
Abstract: This work presents two alternative deep convolutional architectures for the reconstruction of the fetal skull in partially cut 3D ultrasound volumes: a conditional variational autoencoder (CVAE), and a TL-embedding network. In our experiments, we simulate different levels of corruption (10%-50%), obtaining accurate reconstructions (+80% Dice coef.) from both networks for cuts up to 40%.

Keywords: Fetal and Pediatric Imaging, CT imaging applications
Abstract: Craniosynostosis is a birth defect in which one or more of the cranial sutures fuse prematurely, thus changing the shape and growth pattern of the skull. We have developed an automatic, personalized and quantitative imaging technology for the diagnosis of craniosynostosis, and the surgical planning of cranial vault reconstruction for its treatment. In addition, we introduce a novel framework to evaluate the surgical outcome that does not involve radiation or sedation in children.

Keywords: Fetal and Pediatric Imaging, Brain imaging and image analysis, Image registration, segmentation, compression and visualization - Machine learning / Deep learning approaches
Abstract: Existing computational tools developed for adult brain MRI are not suitable for processing of infant brain MRI typically with low tissue contrast and region-specific dynamic development. In this paper, we present a set of infant-tailored, cortical surface-based computational tools and atlases dedicated for computational neuroanatomy of baby brains.

Keywords: Fetal and Pediatric Imaging, Image analysis and classification - Machine learning / Deep learning approaches
Abstract: This talk will showcase some of the work carried out at the BCN Medtech research unit of Universitat Pompeu Fabra in Barcelona, Spain. In particular, the focus will be on perinatal and fetal applications, including: imaging and image reconstruction in fetal MRI, machine learning approaches for brain image segmentation and classification for the study of perinatal brain development, and a computer-assisted surgery system for fetal surgery.

Keywords: Image reconstruction and enhancement - Machine learning / Deep learning approaches, Fetal and Pediatric Imaging
Abstract: Measurement of head biometrics from fetal ultrasonography images is of key importance in monitoring the healthy development of fetuses. However, the accurate measurement of relevant anatomical structures is subject to large inter-observer variability depending on user expertise and attention fatigue. To address this issue, an automated method utilizing Fully Convolutional Networks (FCN) is proposed to determine measurements of fetal head circumference (HC) and biparietal diameter (BPD). An FCN was trained on approximately 2000 2D ultrasound images of the head with annotations provided by 45 different clinicians during routine screening examinations to perform semantic segmentation of the head. An ellipse is fitted to the resulting segmentation to mimic the annotation typically produced by a clinician. To assess the model's performance, an intra- and inter-observer variability study was performed, where two experts manually annotated 100 test images. Inter-observer variability was slightly higher than the mean expert-to-model variability for HC (2.16mm vs 1.61mm) and BPD (0.59mm vs 0.50mm) measurements, while Dice coefficient was the same (0.980 vs 0.980). Our results demonstrate that the model performs at a level similar to a human expert but reduces user-specific bias by learning from a large dataset annotated by many sonographers. Additionally, measurements are generated in near real-time at 15fps on a GPU, which could speed up clinical workflow for both skilled and trainee sonographers.

Keywords: Microfluidic techniques, methods and systems, Cellular force transduction - Mechanical stimuli and mechanotransduction, Micro- and nano-sensors
Abstract: Organ-on-a-chip technology has a high potential to create the physiologic environment that can be found in the tissue of interest in vivo. In this project, a bone-on-a-chip system is developed that allows for the application and quantification of mechanical load and the regulation of oxygen levels.

Keywords: Electromagnetic field effects and cell membrane
Abstract: The two terms magnetic and electromagnetic had been defined by J.C.Maxwell in 1855. The first is a DC or low frequent-the second a high frequent field application. In the high frequent application, material constants cannot be neglected and herewith a certain heat generation, which is intended in biomed applications, e.g. Hyperthermia. Here however we avoid to heat up the treated cells and work magnetically (low frequently) with tagged magnetic particles like SPIONs.

Keywords: Electromagnetic field effects and cell membrane
Abstract: Heavy Duty industrial machines are operated with large electric fields safely controlled by a suitable insulation, e.g. a long bushing. Living cells are efficient organic heavy duty systems with a membrane capable to insulate gigantic electric fields higher as such devices, mentioned above. That brought up the idea to apply dynamic (superposed, locally displaced) heavy duty field techniques into the environment of living cells. In one case we move magnetic particles individually (each one of the bulk separately field powered) above the surface of the cell membrane, in order to insert them endocytotically. This application is cold meaning negligible, nearly non-detectible, heat is produced inside the particles.

Keywords: Biomaterials - Chemical and electrochemical sensors, Translational issues in tissue engineering and biomaterials, Micro- and nano-sensors
Abstract: In the presented work a system for monitoring of living cells is described. Its different modules for cell culture, sensor readout and data processing are presented and examples for applications in toxicology and environmental monitoring are enlighten.

Keywords: Physiological systems modeling - Signal processing in physiological systems, Time-frequency and time-scale analysis - Time-frequency analysis
Abstract: One of the hallmarks of the impaired auditory system is a difficulty segregating clean and noisy segments of a speech signal. This results in a decrease in speech understanding for affected individuals in noisy environments. This problem is complex, largely due to the myriad of deficits associated with hearing impairment. Here we will review these deficits, and why effective noise reduction techniques in hearing aids and cochlear implants are essential.

Keywords: Physiological systems modeling - Multivariate signal processing, Time-frequency and time-scale analysis - Nonstationary processing, Data mining and processing in biosignals
Abstract: The societal need for assistive hearing devices for users with hearing loss has increased exponentially over the past two decades; however, actual human speech recognition performance with such devices has only seen modest gains relative to the pace of actual digital signal processing (DSP) technology. A major challenge with clinical hearing technologies is the limited ability to run complex signal processing algorithms which would require powerful DSPs at the source. The CCi-MOBILE platform, developed by our center at UT-Dallas, provides the research community with an open-source, software-flexible, and powerful computing research interface to conduct listening studies with either/both hearing-aids and cochlear implants. The platform uses commercially available smartphone/tablet devices as portable sound processors, and is able to provide bilateral electric and acoustic stimulation.

Keywords: Adaptive filtering, Parametric filtering and estimation, Physiological systems modeling - Signals and systems
Abstract: Proposed smartphone-based real-time speech signal processing algorithms and their test results for hearing study and improving hearing aid applications are reviewed.

Keywords: Time-frequency and time-scale analysis - Time-frequency analysis, Adaptive filtering, Neural networks and support vector machines in biosignal processing and classification
Abstract: With advances in wireless technology and sensor miniaturization, more non-audio sensors are integrated into ear-level hearing devices. These sensors help improve speech understanding and sound quality and enhance usability. However, the introduction of these sensors also present a new set of challenges to researchers and engineers. Compared with traditional audio sensors, these new sensor inputs come from different modalities and often have different scales and different sampling frequencies. In some cases, they are not even linear and not synchronized to each other at all. In this paper, we will review these challenges in details in the context of hearing devices. Furthermore, we will demonstrate how multimodal signal processing and machine learning can be used to overcome these challenges and bring greater degree of satisfactions to the end users. Finally, how to choose an appropriate approach in different situations will be discussed

Keywords: Cardiovascular assessment and diagnostic technologies
Abstract: Monitoring fetal heart rate in an important aspect in evaluating fetal well being. Maternal fetal interaction has shown evolution during fetal maturation. In this work, we studied maternal-fetal heart rate synchronization in early and late gestation fetuses. We also evaluated variations in the synchronization due to congenital heart defect (CHD).

Keywords: Cardiovascular assessment and diagnostic technologies, Diagnostic devices - Physiological monitoring
Abstract: Several different studies demostrated that an certain influence of maternal heart rate on fetal heart rate at short and larger time scales exist while other failed to identify such relationships. Therefore, the aim of this study was to investigate if maternal-fetal cardiac couplings in 22 mid gestation healthy fetuses are changing over time. To quantify the time-variant short-term maternal–fetal cardiac couplings we applied the normalized short time partial directed coherence approach on maternal and fetal ecg recordings. Short-term maternal-fetal cardiac couplings (strength and direction) are changing over time. While the causal influence of maternal (driver) on fetal heart rate was stronger at time lags of τ=1-20 beats this behavoiur changed significantly after a lag τ>20 beats. This finding pointing to an altering maternal-fetal cardiac coupling process over time.

Keywords: Magnetic resonance imaging - Diffusion tensor, diffusion weighted and diffusion spectrum imaging, Brain imaging and image analysis, Image segmentation
Abstract: Neuroimaging and neuropsychology are employed to investigate the pathological features and clinical characteristics of Alzheimer’s disease (AD) in order to find a method for the precise treatment. Diffusion tensor imaging (DTI) provides a non-intrusive examination of cranial nerve diseases which can help us observe the microstructure of neuron fibers. Building the brain network provides a chance to reveal the significance of specific brain region and the relevance among different regions. In this study, we propose a completely novel method to analyze AD. First whole brain network is built on the basis of a novel segmentation atlas, and global graph theoretical parameters are calculated to evaluate the characteristic of whole brain. Then graph theoretical parameters of specific brain regions are extracted based on whole brain network. Finally neuropsychology scale are employed and we analyze the correlation between graph theoretical parameters of specific regions and scale scores. Our results illustrate the connection between neuroimaging data and neuropsychological scores, and provide a reasonable explanation for the potential connection between clinical performance and physiological brain lesions of AD patients.

Keywords: Magnetic resonance imaging - MR neuroimaging, Magnetic resonance imaging - MR spectroscopy, Multimodal imaging
Abstract: Despite the high incidence of major depressive disorder (MDD) in females, the detailed neurobiology mechanism remains not fully understood. Increasing evidence showed that MDD was associated with hippocampal volumetric abnormality with different subfields demonstrating various alteration features. However, the linkage between hippocampal atrophy with its biochemical information remains unclear. In this study, we aim to investigate the relationship between bilateral hippocampal subfields volumetric and metabolic information in female MDD patients, using a combined T1-weighted magnetic resonance imaging (MRI) and proton magnetic resonance spectroscopy (1H-MRS) technology. There are 15 female MDD patients and 12 matched healthy controls involved in the study. We found a significant decrease in left fimbria volume in MDD group, which was negatively correlated with left hippocampal choline level. In addition, the left hippocampal creatine concentration in patients was negatively correlated with left fimbria volume. Moreover, the NAA level in left hippocampus was negatively correlated with MDD clinical symptomology including anxiety and depression scores. Our findings suggest that the altered coupling between hippocampal structural and metabolic features might contribute to the etiology of female MDD patients.

Keywords: Magnetic resonance imaging - MR neuroimaging, Magnetic resonance imaging - Diffusion tensor, diffusion weighted and diffusion spectrum imaging, Multivariate image analysis
Abstract: The connectome-wide association studies exploring the association between brain connectome and disease phenotypes have suffered from a massive number of comparisons. In this paper, we propose to apply a multivariate distance-based analytic framework on brain white matter (WM) structural networks invaded by Alzheimer’s disease (AD). Eighty-three subjects including patients with AD, amnestic mild cognitive impairment (aMCI) and healthy subjects were scanned with dMRI. By constructing WM structural network for each individual, we used both multivariate and traditional univariate statistical models to complimentarily analyze network pattern and fiber strength changes due to AD. WM connections linked with several brain structures were found significantly changed between AD group and normal controls. No significant findings were observed between aMCI group and normal controls. Our results demonstrate the sensitivity of the combined connectome-based analytic framework in detecting abnormalities of structural brain networks.

Keywords: Magnetic resonance imaging - MR neuroimaging, Functional image analysis, Brain imaging and image analysis
Abstract: A number of resting-state functional magnetic resonance imaging (rs-fMRI) studies indicate dysfunction of large-scale brain networks underlying the pathophysiology of obsessive-compulsive disorder (OCD). Recent epidemiology studies show that the prevalence of female OCD patients is higher than that of males. However, the underlying neurobiology mechanism for female OCD patients remains not fully understood. In this study, we are aimed to explore the spontaneous brain neuronal activity in unmedicated female OCD patients using rs-fMRI methodology and fractional amplitude of low frequency fluctuations (fALFF) analysis. Additionally, we examine the relationship between fALFF changes and female OCD symptomatology. Increased fALFF values in right brainstem, right rectus, left middle temporal gyrus and right angular were found in OCD females. And decreased fALFF values in right cerebellum, left middle occipital gyrus, left insula, postcentral gyrus and left precentral gyrus were shown in female OCD patients. Moreover, the fALFF values in left precentral gyrus and left middle temporal gyrus were positively associated with patients YBOCS-Obsessions scores and HAMD scores, respectively. Our findings bring additional insights in understanding the pathophysiology of female OCD patients.

Keywords: Magnetic resonance imaging - MR neuroimaging
Abstract: Autism Spectrum Disorder (ASD) is a neurodevelopmental disorder that is associated with atypical connectivity within and between brain regions. In this study, we attempted to classify functional Magnetic Resonance Images (fMRI) of Typically Developing (TD) and ASD participants using conditional random forest and random forest. Resting-state fMRI images of TD and ASD participants (N=320 for training and N=80 for validation) were obtained from the Autism Imaging Data Exchange; ABIDE-I, ABIDE-II. Images were preprocessed using a standard pipeline. A Functional Connectivity (FC) matrix was calculated using 237 cortical, subcortical, and cerebellar Regions of Interest (ROIs). The dimensionality of the FC matrix was reduced using conditional random forests and at each dimension classification accuracy was tested using random forests. Results suggest that in the current dataset, the random forest is able to classify the TD and ASD with a peak accuracy of 65% using 143 features. Remarkably, the Cingulo-Opercular Task Control (COTC) region contributed the highest number of features linked to more accurate classification, and connectivity between COTC and the dorsal attention network distinguished ASD and TD participants.

Keywords: Magnetic resonance imaging - Diffusion tensor, diffusion weighted and diffusion spectrum imaging, Brain imaging and image analysis, Image feature extraction
Abstract: Due to the high resistance (35%) to the current treatment methods in adolescent Major Depressive Disorder (MDD) and its tragic outcomes, the discovery of treatment-related responders is critical to developing effective treatments. In this paper, the permutation test is performed to identify statistically significant changes in anatomical characteristics during pairwise comparisons among the control group (n=27), treated MDD group (n=37), and untreated MDD group (n=15). The anatomical characteristics include: 1) anatomical connectivity defined using DTI metrics between a pair of brain regions, and 2) topological measurements of anatomical networks. With the Bonferroni correction for multiple-comparison, significant alterations in community structure and local topology were identified as the p-value < 5%, which include: 1) a reduced nodal centrality (degree and strength) on right hippocampus for treated compared to untreated group, 2) an elevated clustering coefficient and local efficiency on right lateral orbitofrontal cortex for untreated compared to the combination of control and treated groups, 3) an increased participation coefficient for untreated patients on left insula cortex in the mean-diffusivity network compared to the combination of control and treated groups, and 4) a degraded module degree z-score on right caudate nucleus for all the patients compared to the control group. Two connections, hippocampus-insula in the right hemisphere and parahippocampal-insula in the left hemisphere, were found significantly altered in TR, AD, and FA due to MDD.

Keywords: Physiological monitoring - Instrumentation, Physiological monitoring - Novel methods
Abstract: In this symposium, noncontact-based vital sign measurements for daily monitoring is discussed. In particular, our system remotely estimates pulse transit time (PTT) using camera and microwave sensor. A microwave sensor detects heartbeats from subject’s chest movement, and a camera sensor obtains pulse wave signals from change of skin’s brightness. By accurately synchronizing these sensors, the system calculates the time difference between a heartbeat and a pulse, thus PTT is estimated. An experiment conducted by twelve subjects resulted in physically measurement PTT (around 180 ms).

Keywords: Optical and photonic sensors and systems, Physiological monitoring - Novel methods, New sensing techniques
Abstract: In a previous study, our group showed the pulse wave transit time obtained from two areas in a video image positively correlated with blood pressure changes. However, this result does not agree with the principle of the pulse wave transit time. In this study, we show not only the reason for this result but also the possibility of estimating blood pressure changes from one area in a video image.

Keywords: Physiological monitoring - Instrumentation, Optical and photonic sensors and systems
Abstract: The aim of this study is to propose an extraction method of video plethysmograpy (VPG) from not only face but also the skin with a few capillary vessels. The extraction algorithm is based on periodic component analysis. The results indicated that the proposed method can extract valid pulse wave which has the information of local vascular regulations from a region of interest (ROI) located in a part other than a face.

Keywords: Ultrasound imaging - High-frequency technology
Abstract: Evaluation of cell mechanics plays a vital role in predicting and evaluating diseases in medical and clinical fields. Moreover, the mechanical stiffness level indicates the metastasis potential in tumor cells. In this study, the biomechanical properties of breast cancer cells were investigated by single-beam acoustic tweezers (SBAT), a non-destructive assessment tool. We demonstrated that SBAT is a new promising tool for quantifying the mechanical phenotype of cells at the single-cell level without labeling.

Keywords: Ultrasound imaging - Breast, Ultrasound imaging - Elastography, Ultrasound imaging - Interventional
Abstract: Ultrasound is a powerful imaging modality available to clinicians, engineers and researchers today. In recent years, advances in computing, electronics, transducer, contrast imaging and signal/imaging processing algorithms have fueled the growth of US devices in many clinical applications. Three main new technology and applications discussed in this presentation are: tomographic imaging of breast, fibrosis staging of liver, and use of contrast agent for differentiating benign from malignant lesions. Current state-of-the-art technology in each of these applications will be discussed along with the associated clinical advantages. We will also outline limitations and future challenges of each of these technologies.

Keywords: Ultrasound imaging - Breast, Ultrasound imaging - Elastography, Ultrasound imaging - Photoacoustic/Optoacoustic/Thermoacoustic
Abstract: PA imaging has been explored for the early diagnosis of diseases by imaging tumor angiogenesis and blood oxygen saturation. We have developed a 3D photoacoustic mammography system for breast cancer diagnosis and demonstrated that 3D imaging of tumor-related vessels can be realized with a resolution of less than 0.5mm using the prototype with hemispherical sensors. In addition, this device was applied to analyze peripheral blood vessels, such as those in the palm. Results indicated that the PA imaging system enabled visualization of the 3D features of blood vessels in the palm and noninvasive analysis of arterial tortuousness.

Keywords: Ultrasound imaging - Interventional, Multimodal imaging
Abstract: Resistant hypertension is a refractory heart disease in which blood pressure cannot be controlled using medication. Catheter-based radiofrequency ablation has been identified as a potential method to treat the resistant hypertension by inactivating the sympathetic nerves along the renal artery. However, a recent clinical trial using this method could not show the efficacy. Here, we demonstrate a hypothesis of failure of catheter based method and propose an alternative surgical approach to the treatment of resistant hypertension. We describe a surgical instrument design and validate the feasibility of the method using in vitro and ex vivo studies.

Keywords: Ultrasound imaging - Photoacoustic/Optoacoustic/Thermoacoustic, Image enhancement, Ultrasound imaging - High-frequency technology
Abstract: Introducing a medical ultrasound course at the undergraduate level in biomedical engineering programs will be beneficial to students. However, addition of such a course in BME programs encounters several constraints. The objective of this present initiative is to propose an undergraduate elective course on diagnostic ultrasound for BME students. Two models of diagnostic ultrasound course are proposed. The first is designed with lecture-based pedagogy. The second is an innovative model with complimentary lab modules based on clinical collaboration. Preliminary feedback lends support to the feasibility of the proposed models.

Keywords: Scaffolds in tissue engineering
Abstract: Traditional tissue engineering models use sophisticated instrumentation or costly set-ups for fabrication of 3D scaffolds, require extensive optimization procedures, and do not provide physiologically relevant size structures without mass transport limitations. Therefore, it is challenging to fabricate biocompatible scaffolds for personalized medicine. To tackle these hurdles, we developed paper-based cell culture platforms for a range of applications that involve the use of different types of cells.

Keywords: Translational issues in tissue engineering and biomaterials - Osseointegration, Translational issues in tissue engineering and biomaterials
Abstract: Functionally graded materials (FGM) are a material whose composition, structure, and properties spatially, gradually change to seamlessly integrate different functions together. We have been using this concept in our biomaterial and medical device developments. Here we will present two cases of functionally graded biomaterials to improve bone and tendon regeneration.

Keywords: Biomimetic materials, Biomaterial-cell interactions - Functional biomaterials, Biomaterial-cell interactions - Surface modification of biomaterials
Abstract: Reciprocal interactions of cells with their microenvironment are fundamental to multiple cellular processes necessary for tissue development, homeostasis, and regeneration. To this end, biomimetic systems emulating the physical and chemical properties of tissue specific extracellular matrices are being developed at a rapid pace. Mineral environment of the bone tissue plays a key role in maintaining the bone health and tissue homeostasis. In this talk, I will discuss our efforts to delineate the role of the mineralized extracellular matrix on cellular responses relevant to bone tissue repair, stem cell differentiation, and bone degenerative diseases.

Keywords: Stem cells - Engineered matrices for stem cells, Stem cells - Tissue morphogenesis, Scaffolds in tissue engineering - Fabrication of cell seeded scaffolds
Abstract: The objective of this work was to investigate the phenotypic fate of human mesenchymal stem cells (hMSCs) in model 3D matrices simulating the developmentally-inspired physio-mechanical and biochemical properties of the superficial, middle and calcified zones of articular cartilage. Results demonstrate that gradients in cell density, matrix stiffness, and zone-specific growth factors can potentially regenerate zonal structure of the articular cartilage.

Keywords: Implantable sensors, Modeling and analysis, Physiological monitoring - Modeling and analysis
Abstract: An insertable cardiac monitor (ICM) is a small medical device that is placed under the chest muscle to continuously monitor electrical heart activities and record electrocardiograms (ECGs). Clinicians use it to diagnose and manage abnormal heart activities in patients with unexplained syncope, palpitations, cryptogenic stroke, lightheadedness, dizziness, and seizures. However, clinical studies showed that false detection rates of abnormal heart rhythms are still high, and key variables can cause the ICM to detect heart signals inappropriately by altering amplitudes and morphologies of ECGs. The objective of this paper is to investigate the effects of these key variables on ICM sensing by using computer simulations and virtual human family. This study uses Sim4Life finite element analysis (FEA) software to simulate cardiac propagation and ICM electrode sensing inside three members of a virtual human family. The ICM CAD model was used and placed in each member at various locations and with electrodes facing up or down. The propagation vectors in the hearts are approximated by fields created along the heart axis going in the direction from the left atrium (LA) and right atrium (RA) disc pair to apex disc. After solving the models, the voltage difference on the electrodes is obtained from each simulation with a model placement. In the adult male model, the results at default 45 degree was scaled to the clinical data with the similar BMI range, and the scaling factor was applied to all the simulation results. We observed in these simulations that sensing amplitudes can vary greatly depending on device flipping, orientation/rotation, and migration; change significantly due to respiration effect; and are most sensitive to body mass with the similar trend from clinical data. Those findings support identification of the key variables impacting clinical false detections.

Keywords: Modeling and analysis, Integrated sensor systems, Mechanical sensors and systems
Abstract: Evaluation of lung mechanics is the primary component for designing lung protective optimal ventilation strategies. This paper presents a machine learning approach for bedside assessment of respiratory resistance (R) and compliance (C). We develop machine learning algorithms to track flow rate and airway pressure and estimate R and C continuously and in real-time. An experimental study is conducted, by connecting a pressure control ventilator to a test lung that simulates various R and C values, to gather sensor data for validation of the devised algorithms. We develop supervised learning algorithms based on the decision tree, decision table, and Support Vector Machine (SVM) techniques to predict R and C values. Our experimental results demonstrate that the proposed algorithms achieve 90.3%, 93.1%, and 63.9% accuracy in assessing respiratory R and C using decision table, decision tree, and SVM, respectively. These results along with our ability to estimate R and C with 99.4% accuracy using a linear regression model demonstrate the potential of the proposed approach for constructing a new generation of ventilation technologies that leverage novel computational models to control their underlying parameters for personalized healthcare and context-aware interventions.

Keywords: Wearable sensor systems - User centered design and applications, Sensor systems and Instrumentation, Modeling and analysis
Abstract: Recent advancements in mobile devices, data analysis, and wearable sensors render the capability of in-place health monitoring. Supervised machine learning algorithms, the core intelligence of these systems, learn from labeled training data. However, labeling vast amount of data is time-consuming and expensive. Moreover, sensor data can often contain personal information that a user may not be comfortable sharing. Therefore, there is a strong need to develop methods for generating realistic labeled sensor data. In this paper, we propose a supervised generative adversarial network architecture that learns from feedback from both a discriminator and a classifier in order to create synthetic sensor data. We demonstrate the effectiveness of the architecture on a publicly available human activity dataset. We show that our generator learns to output diverse samples that are similar but not identical to the training data.

Keywords: Physiological monitoring - Instrumentation
Abstract: The importance of home blood pressure (BP) monitoring has been emphasized for effective hypertension management. Currently, the most popular non-invasive BP monitors for home use are the upper arm cuff-style oscillometric devices which determine BP from the cuff pressure oscillations during the cuff inflation/deflation induced by the pulsatile blood flow in the compressed arteries. However, the large size of the upper arm cuff is not favorable for attachment in daily life for ambulatory BP monitoring. Therefore, the miniaturization of home BP monitors is in demand to improve their portability for frequent measurements. This work examined the oscillometric measurement of mean blood pressure(MBP) at upper arm (UA), middle forearm (MA), wrist (WR), finger proximal phalanx (FP) and finger distal phalanx (FD) on 14 young adults. The experimental results showed that the mean and standard deviation of the differences from the oscillometric MBP at UA are 8.86±6.28 mm Hg at MA, 14.43±5.52 mm Hg at WR, 9.80±6.57 mm Hg at FP and -0.77±6.37 mm Hg at FD, respectively. Based on hand checking and literature data, the order of the ratios of the bone volume to the surrounding tissue volume from large to small is WR>MA≈FP>FD≈UA. Together with the experimental results, we infer that a larger bone-tissue ratio could result in a larger oscillometric MBP reading. Since the applied cuff pressure are supposed to be less effectively absorbed by the soft-tissue surrounding a larger rigid bone, it is more difficult to occlude the arteries buried in the pressure-absorbing tissue at a bonier site by the inflatable cuffs, which leads to a higher measured MBP than the real MBP. In conclusion, it is promising to develop finger oscillometric BP monitors worn on the finger distal phalanx which have a compact size and provide consistent measurement results with the UA measurements.

Keywords: Physiological monitoring - Novel methods, Wearable sensor systems - User centered design and applications, Physiological monitoring - Modeling and analysis
Abstract: This exploratory study examined the effects of varying g-forces, including feelings of weightlessness, on an individual’s physiology during parabolic flight. Specifically, we collected heart rate, accelerometer, and skin conductance measurements from 16 flyers aboard a parabolic flight using wearable, wireless sensors. The biosignals were then correlated to participant reports of nausea, anxiety, and excitement during periods of altered g-forces. Using linear mixed-effects models, we found that (1) heart rate was positively correlated to individuals' self-reported highest/lowest periods of both anxiety and excitement, and (2) bilateral skin conductance asymmetry was positively correlated to individuals' self-reported highest/lowest periods of nausea.

Keywords: Wearable low power, wireless sensing methods, Physiological monitoring - Modeling and analysis, Bio-electric sensors - Sensor systems
Abstract: Validation of biosensor algorithms is paramount for regulated medical devices applied to patient monitoring. We present validation of breathing rate (BR) measurement using a patch medical device via a novel synthetic simulation platform, in-hospital data collection and controlled laboratory study. Single-lead ECG and triaxial body acceleration signals with variability and noise are synthetically generated and quantized for a constellation according to the input parameters of heart rate (HR) as a fundamental frequency(fc) of ECG and reference BR as a modulating frequency (fr). Synthetic signals are input to the BR algorithms and the performance of output BRs are evaluated for a region-of-interest of the constellation (fc/fr ≥ 3 & fc/fr ≤ 8) accounting the Nyquist and physiological varability. The performances of patch sensor’s BR are also evaluated in 13 post-operative patients with reference to a clinical bedside monitor and in 57 subjects carrying out a controlled laboratory protocol with reference to capnography. The synthetic simulations revealed mean absolute error (MAE) of 0.8±0.6 brpm and standard deviation of absolute error of 0.3±0.2 brpm for the BR algorithms of patch sensor. The controlled laboratory testing revealed MAE of 1.7±0.7 brpm (n=57) for stationary conditions. The proposed simulation platform can be useful for developmental refinement or validation of BR measurement prior to testing in humans at clinical or laboratory conditions and applicable for testing other patient monitoring devices with modular modifications.

Keywords: Wearable low power, wireless sensing methods, Bio-electric sensors - Sensor systems, Optical and photonic sensors and systems
Abstract: Wearable devices for body-status monitoring require various signal processing such as ambient noise filtering and signal quality evaluation. While wearable devices are very limited in resource, conventional noise filtering and signal quality evaluating methods consume considerable amount of processing power. Moreover, these conventional methods are not suitable for processing bio-signals in terms of its performance. In this paper, we propose a novel method of preprocessing bio-signals. This preprocessing method includes distortionless noise filtering and signal quality estimation, where both parts are basically based on a simple combination of multiple low pass IIR filters.

Keywords: Wearable power and on-body energy harvesting, Wearable low power, wireless sensing methods, Wearable sensor systems - User centered design and applications
Abstract: Wearable devices have the potential to improve healthcare, but suffer from significant barriers to adoption, including the need for constant recharging. Harvesting energy from the ambient environment to top-up batteries can overcome this, but the actual energy available is very small, and hence it is critical that the whole system is highly optimized. This paper presents an investigation into the optimization of inertial energy harvesters for placement at the human foot. Lower body locations have previously been shown to be very energy dense, however previous energy harvester modeling has focused on the lower leg rather than the foot itself for ease of device placement. We show that the typical energy density can be almost double at the foot compared with lower leg positions, with substantially more energy concentrated in a smaller bandwidth. There is thus a dual benefit of placing a harvester at the foot: there is more energy due to the larger movement of the foot, and more efficient (higher Q) harvesters can be used to increase the collected energy. We place these results in context by analyzing the power demands of a typical wearable and identify that with appropriate harvester tuning the peak current requirements of the electronics can be fitted into the energy peaks generated from each footstep.

Keywords: Sensor systems and Instrumentation, Novel methods, Modeling and analysis
Abstract: This study presents an approach for instrumenting a spinal motion measurement protocol using two inertial measurement units (IMU)s affixed at the posterior pelvis and superior trunk. The accuracy of the inertial motion measurement instrumentation in tracking the relative orientation of the trunk with respect to the pelvis in three spinal motions (flexion-extension, side bending, and rotation) is compared to that of a concurrent optical motion capture (mocap) system. Six healthy adults (31.5±11.2; 2 females) were recruited to perform the spinal motions. The results show minimal deviations of the IMU measurements from those of the mocap system (RMSE<2deg, r>0.84 in all cases) and demonstrate the efficacy of the proposed instrumentation approach for spinal motion measurement.

Keywords: Wearable sensor systems - User centered design and applications, Wearable wireless sensors, motes and systems, Modeling and analysis
Abstract: Human activity recognition (HAR) is an important component in health-care systems. For example, it can enable context-aware applications such as elderly care and patient monitoring. Relying on a set of training data, supervised machine learning algorithms form the core intelligence of most existing HAR systems. Meanwhile, the accuracy of an HAR model highly depends on the similarity between the training and the operating context. Therefore, there is a need for developing machine learning algorithms that can easily adapt to the operating context at hand. In this paper, we propose a cross-subject transfer learning algorithm that links source and target subjects by constructing manifolds from feature-level representation of the source subject(s). Our algorithm assigns labels to the unlabeled data in the current context using the manifold learned from the source subject(s). The newly labeled data is used to develop a personalized HAR model for the current context (i.e. target subject). We demonstrate the efficacy of the algorithm using a publicly available dataset on HAR. We show that the proposed framework improves the accuracy of activity recognition by up to 24%.

Keywords: Wearable body sensor networks and telemetric systems, Wearable low power, wireless sensing methods, Wearable sensor systems - User centered design and applications
Abstract: Age related spinal deformity is an becoming increasingly prevalent problem, resulting in decreased quality of life. While spinal deformity can be corrected via surgical intervention, a large number of people with spinal fusions require follow-up surgery due to further degeneration. The identification of changes to a subject's kinematics and kinetics post-surgery are limited by a lack of methods to collect patient-specific motion data over the course of surgical recovery. This paper introduces an Instrumented Spine Orthosis (ISO) that can capture the motions of the subject torso without requiring the use of a control computer or other dedicated motion capture equipment. This system is used to collect the peak torso angles and velocities for a single subject performing sit-to-stand actions. The accuracy of the ISO is evaluated using motion capture, during different sit-to-stand protocols designed to highlight motion changes that have been seen in subjects with reduced mobility. This system was found to provide reliable measurements of these kinematic and kinetic torso measures across all tested motions, demonstrating the potential for the use of Instrumented Spine Orthotics to provide quantitative measures during the surgical recovery process.

Keywords: Wearable sensor systems - User centered design and applications
Abstract: Clinical balance assessments often rely on functional tasks as a proxy for balance (e.g., Timed Up and Go). In contrast, analyses of balance in research settings incorporate quantitative biomechanical measurements (e.g., whole-body angular momentum, H) using motion capture techniques. Fully instrumenting patients in the clinic is not feasible, and thus it is desirable to estimate biomechanical quantities related to balance from measurements taken from a subset of the body segments. Machine learning algorithms are well-suited for this type of low- to high-dimensional mapping. Thus, our objective was to develop and validate an artificial neural network for estimating contributions to H from 12 body segments using only five inertial measurement units. The network was trained, tested and validated on data from five able-bodied individuals performing forty trials each of a circuit involving complex walking tasks, including stairs, ramp, and direction changes. The network was also separately tested on four trials of an individual with Parkinson’s disease walking on the circuit. The output of the network was strongly correlated with the segment contributions to H in both able-bodied (R=0.997) and Parkinson’s disease (R=0.998) subjects. The estimated values also had low error relative to the signal magnitude, with the largest mean±SD root-mean-squared errors of 8.04±1.76% peak signal magnitude in able-bodied individuals and 7.96±0.91% in the individual with Parkinson’s disease. These promising results establish the feasibility of using a sparse set of inertial measurement units to provide quantitative data to clinicians for targeted balance rehabilitation across different patients.

Keywords: Image analysis and classification - Machine learning / Deep learning approaches, Image analysis and classification - Digital Pathology, Image classification
Abstract: In this paper, we will discuss some of the issues specifically associated to using deep convolutional neural network (CNN) in the medical image analysis setting. Recently, research shows that CNN is achieving good performance in classifying and analyzing medical images. Contrary to conventional machine learning approaches, CNN does not require the construction of hand-crafted features nor field-knowledge about the dataset. Typically, CNNs have a large number of parameters that need to be optimized in the training phase and are trained using a large number of examples (supervised learning). The requirement of a large training dataset poses a problem in medical image analysis research areas as large (balanced) disease specific dataset is difficult to achieved. This paper will discuss a few issues related to medical image analysis research including overfitting, imbalanced dataset, data augmentation and color normalization/ standardization when H&E stained histopathological images are involved.