Search

In this paper, we introduce a new variation of the Convolutional Neural Network Inception block, called Sinc, for sleep stage classification in premature newborn babies using electroencephalogram (EEG). In practice, there are many medical centres where only a limited number of EEG channels are recorded. Existing automated algorithms mainly use multi-channel EEGs which perform poorly when fewer numbers of channels are available. The proposed Sinc utilizes multi-scale analysis to place emphasis on the temporal EEG information to be less dependent on the number of EEG channels. In Sinc, we increase the receptive fields through Inception while by additionally sharing the filters that have similar receptive fields, overfitting is controlled and the number of trainable parameters dramatically reduced. To train and test this model, 96 longitudinal EEG recordings from 26 premature infants are used. The Sinc-based model significantly outperforms state-of-the-art neonatal quiet sleep detection algorithms, with mean Kappa 0.77 ± 0.01 (with 8-channel EEG) and 0.75 ± 0.01 (with a single bipolar channel EEG). This is the first study using Inception-based networks for EEG analysis that utilizes filter sharing to improve efficiency and trainability. The suggested network can successfully detect quiet sleep stages with even a single EEG channel making it more practical especially in the hospital setting where cerebral function monitoring is predominantly used. ispartof: IEEE JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS vol:26 issue:3 pages:1023-1033 ispartof: location:United States status: published

OBJECTIVE: Noise and disturbances hinder effective interpretation of recorded ECG. To identify the clean parts of a recording, free from such disturbances, various quality indicators have been developed. Previous instances of these indicators focus on human-defined desirable properties of a clean signal. The reliance on human-specified properties places an inherent limitation on the potential power of signal quality indicators. To move away from this limitation, we propose a data-driven quality indicator. METHODS: We use an unsupervised deep learning model, the auto-encoder, to derive the quality indicator. For different quality assessment settings we compare the performance of our quality indicator with traditional indicators. RESULTS: The data-driven method performs consistently strong across tasks while performance of the traditional indicators varies strongly from task to task. CONCLUSION: This strong performance indicates the potential of data-driven quality indicators for use in ECG processing, removing the reliance on expert-specified desirable properties. SIGNIFICANCE: The proposed methodology can easily be extended towards learning quality indicators in other data modalities. ispartof: Ieee Transactions On Biomedical Engineering vol:69 issue:2 pages:882-893 ispartof: location:United States status: published

Highlights • We demonstrate the feasibility of a fully automated REM sleep behaviour (RBD) screening tool using minimal sensors. • REM sleep is detected accurately and reliably in individuals with RBD, without EEG sensors. • Automated sleep staging was able to classify REM sleep with sufficient accuracy to allow for the accurate detection of RBD., Objective Rapid-Eye-Movement (REM) sleep behaviour disorder (RBD) is an early predictor of Parkinson’s disease, dementia with Lewy bodies, and multiple system atrophy. This study investigated the use of a minimal set of sensors to achieve effective screening for RBD in the population, integrating automated sleep staging (three state) followed by RBD detection without the need for cumbersome electroencephalogram (EEG) sensors. Methods Polysomnography signals from 50 participants with RBD and 50 age-matched healthy controls were used to evaluate this study. Three stage sleep classification was achieved using a random forest classifier and features derived from a combination of cost-effective and easy to use sensors, namely electrocardiogram (ECG), electrooculogram (EOG), and electromyogram (EMG) channels. Subsequently, RBD detection was achieved using established and new metrics derived from ECG and EMG channels. Results The EOG and EMG combination provided the optimal minimalist fully-automated performance, achieving 0.57 ± 0.19 kappa (3 stage) for sleep staging and an RBD detection accuracy of 0.90 ± 0.11, (sensitivity and specificity of 0.88 ± 0.13 and 0.92 ± 0.098, respectively). A single ECG sensor achieved three state sleep staging with 0.28 ± 0.06 kappa and RBD detection accuracy of 0.62 ± 0.10. Conclusions This study demonstrates the feasibility of using signals from a single EOG and EMG sensor to detect RBD using fully-automated techniques. Significance This study proposes a cost-effective, practical, and simple RBD identification support tool using only two sensors (EMG and EOG); ideal for screening purposes.

Leveraging consumer technology such as smartphone and smartwatch devices to objectively assess people with multiple sclerosis (PwMS) remotely could capture unique aspects of disease progression. This study explores the feasibility of assessing PwMS and Healthy Control's (HC) physical function by characterising gait-related features, which can be modelled using machine learning (ML) techniques to correctly distinguish subgroups of PwMS from healthy controls. A total of 97 subjects (24 HC subjects, 52 mildly disabled (PwMSmild, EDSS [0-3]) and 21 moderately disabled (PwMSmod, EDSS [3.5-5.5]) contributed data which was recorded from a Two-Minute Walk Test (2MWT) performed out-of-clinic and daily over a 24-week period. Signal-based features relating to movement were extracted from sensors in smartphone and smartwatch devices. A large number of features (n = 156) showed fair-to-strong (R 0.3) correlations with clinical outcomes. LASSO feature selection was applied to select and rank subsets of features used for dichotomous classification between subject groups, which were compared using Logistic Regression (LR), Support Vector Machines (SVM) and Random Forest (RF) models. Classifications of subject types were compared using data obtained from smartphone, smartwatch and the fusion of features from both devices. Models built on smartphone features alone achieved the highest classification performance, indicating that accurate and remote measurement of the ambulatory characteristics of HC and PwMS can be achieved with only one device. It was observed however that smartphone-based performance was affected by inconsistent placement location (running belt versus pocket). Results show that PwMSmod could be distinguished from HC subjects (Acc. 82.2 ± 2.9%, Sen. 80.1 ± 3.9%, Spec. 87.2 ± 4.2%, F 1 84.3 ± 3.8), and PwMSmild (Acc. 82.3 ± 1.9%, Sen. 71.6 ± 4.2%, Spec. 87.0 ± 3.2%, F 1 75.1 ± 2.2) using an SVM classifier with a Radial Basis Function (RBF). PwMSmild were shown to exhibit HC-like behaviour and were thus less distinguishable from HC (Acc. 66.4 ± 4.5%, Sen. 67.5 ± 5.7%, Spec. 60.3 ± 6.7%, F 1 58.6 ± 5.8). Finally, it was observed that subjects in this study demonstrated low intra- and high inter-subject variability which was representative of subject-specific gait characteristics. ispartof: Ieee Journal Of Biomedical And Health Informatics vol:25 issue:3 pages:838-849 ispartof: location:United States status: published

Heart disease is the leading cause of mortality worldwide, and it is of utmost importance that clinicians and researchers understand the dynamics of the heart. As an electrical measure of the heart’s activity, the electrocardiogram, or ECG, is the gold standard for recording the cardiac state, whether monitoring the structure of the traces that make up the ECG or indicating key metrics such as heart rate variability. Long-term monitoring of ECG is often required to identify cardiovascular issues but proves impractical; therefore, patients will remotely collect their data. However, ECG signals can become contaminated with various noise sources during data collection. This paper proposes a custom loss function capable of denoising electrode motion artefact in ECG data to a higher standard than other, more common loss functions. We implement our custom loss function with a convolutional neural network to return high-quality ECG, suitable for calculating the aforementioned key metrics from a previously unobtainable state. The proposed model improves ECG signals overall signal-to-noise ratio and preserves the R waves structure. The model outperforms a standard mean squared error loss function with an improvement of 0.5 dB in terms of signal to noise ratio and improves the heart rate estimation by 25%.

Access to medical data is highly regulated due to its sensitive nature, which can constrain communities’ ability to utilize these data for research or clinical purposes. Common de-identification techniques to enable the sharing of data may not provide adequate privacy in every circumstance. We investigate the ability of Generative Adversarial Networks (GANs) to generate synthetic, and more significantly, multichannel electrocardiogram signals that are representative of waveforms observed in patients to address these privacy concerns. Successful generation of high-quality synthetic time series data has the potential to act as an effective substitute for actual patient data. For the first time, we demonstrate a range of novel loss functions using our multivariate GAN architecture and analyse their effect on data quality and privacy. We also present the application of multivariate dynamic time warping as a means of evaluating generated time series. Quantitative evidence demonstrates that the inclusion of a penalisation coefficient (Dynamic Time Warping) in the loss function enables our GAN to outperform the other generative models and loss functions explored by 4.9% according to our metrics. This allows for the generation of data that is more representative of the training set and diverse across generated samples, all whilst ensuring sufficient privacy.

The emergence of digital technologies such as smartphones in healthcare applications have demonstrated the possibility of developing rich, continuous, and objective measures of multiple sclerosis (MS) disability that can be administered remotely and out-of-clinic. Deep Convolutional Neural Networks (DCNN) may capture a richer representation of healthy and MS-related ambulatory characteristics from the raw smartphone-based inertial sensor data than standard feature-based methodologies. To overcome the typical limitations associated with remotely generated health data, such as low subject numbers, sparsity, and heterogeneous data, a transfer learning (TL) model from similar large open-source datasets was proposed. Our TL framework leveraged the ambulatory information learned on human activity recognition (HAR) tasks collected from wearable smartphone sensor data. It was demonstrated that fine-tuning TL DCNN HAR models towards MS disease recognition tasks outperformed previous Support Vector Machine (SVM) feature-based methods, as well as DCNN models trained end-to-end, by upwards of 8-15%. A lack of transparency of "black-box" deep networks remains one of the largest stumbling blocks to the wider acceptance of deep learning for clinical applications. Ensuing work therefore aimed to visualise DCNN decisions attributed by relevance heatmaps using Layer-Wise Relevance Propagation (LRP). Through the LRP framework, the patterns captured from smartphone-based inertial sensor data that were reflective of those who are healthy versus people with MS (PwMS) could begin to be established and understood. Interpretations suggested that cadence-based measures, gait speed, and ambulation-related signal perturbations were distinct characteristics that distinguished MS disability from healthy participants. Robust and interpretable outcomes, generated from high-frequency out-of-clinic assessments, could greatly augment the current in-clinic assessment picture for PwMS, to inform better disease management techniques, and enable the development of better therapeutic interventions. ispartof: SCIENTIFIC REPORTS vol:11 issue:1 ispartof: location:England status: published

OBJECTIVE: Wearable seizure detection devices could provide more reliable seizure documentation outside the hospital compared to seizure self-reporting by patients, which is the current standard. Previously, during the SeizeIT1 project, we studied seizure detection based on behind-the-ear electroencephalography (EEG). However, the obtained sensitivities were too low for practical use, because not all seizures are associated with typical ictal EEG patterns. Therefore, in this paper, we aim to develop a multimodal automated seizure detection algorithm integrating behind-the-ear EEG and electrocardiography (ECG) for detecting focal seizures. In this framework, we quantified the added value of ECG to behind-the-ear EEG. METHODS: This study analyzed three multicenter databases consisting of 135 patients having focal epilepsy and a total of 896 seizures. A patient-specific multimodal automated seizure detection algorithm was developed using behind-the-ear/temporal EEG and single-lead ECG. The EEG and ECG data were processed separately using machine learning methods. A late integration approach was applied for fusing those predictions. RESULTS: The multimodal algorithm outperformed the EEG-based algorithm in two of three databases, with an increase of 11% and 8% in sensitivity for the same false alarm rate. SIGNIFICANCE: ECG can be of added value to an EEG-based seizure detection algorithm using only behind-the-ear/temporal lobe electrodes for patients with focal epilepsy. ispartof: EPILEPSIA vol:62 issue:10 pages:2333-2343 ispartof: location:United States status: published

Ischemic heart disease is the highest cause of mortality globally each year. This puts a massive strain not only on the lives of those affected, but also on the public healthcare systems. To understand the dynamics of the healthy and unhealthy heart, doctors commonly use an electrocardiogram (ECG) and blood pressure (BP) readings. These methods are often quite invasive, particularly when continuous arterial blood pressure (ABP) readings are taken, and not to mention very costly. Using machine learning methods, we develop a framework capable of inferring ABP from a single optical photoplethysmogram (PPG) sensor alone. We train our framework across distributed models and data sources to mimic a large-scale distributed collaborative learning experiment that could be implemented across low-cost wearables. Our time-series-to-time-series generative adversarial network (T2TGAN) is capable of high-quality continuous ABP generation from a PPG signal with a mean error of 2.95 mmHg and a standard deviation of 19.33 mmHg when estimating mean arterial pressure on a previously unseen, noisy, independent dataset. To our knowledge, this framework is the first example of a GAN capable of continuous ABP generation from an input PPG signal that also uses a federated learning methodology. ispartof: SENSORS vol:21 issue:18 ispartof: location:Switzerland status: published

OBJECTIVE: Parkinson's disease (PD) is a neurodegenerative disorder that affects multiple neurological systems. Traditional PD assessment is conducted by a physician during infrequent clinic visits. Using smartphones, remote patient monitoring has the potential to obtain objective behavioral data semi-continuously, track disease fluctuations, and avoid rater dependency. METHODS: Smartphones collect sensor data during various active tests and passive monitoring, including balance (postural instability), dexterity (skill in performing tasks using hands), gait (the pattern of walking), tremor (involuntary muscle contraction and relaxation), and voice. Some of the features extracted from smartphone data are potentially associated with specific PD symptoms identified by physicians. To leverage large-scale cross-modality smartphone features, we propose a machine-learning framework for performing automated disease assessment. The framework consists of a two-step feature selection procedure and a generic model based on the elastic-net regularization. RESULTS: Using this framework, we map the PD-specific architecture of behaviors using data obtained from both PD participants and healthy controls (HCs). Utilizing these atlases of features, the framework shows promises to (a) discriminate PD participants from HCs, and (b) estimate the disease severity of individuals with PD. SIGNIFICANCE: Data analysis results from 437 behavioral features obtained from 72 subjects (37 PD and 35 HC) sampled from 17 separate days during a period of up to six months suggest that this framework is potentially useful for the analysis of remotely collected smartphone sensor data in individuals with PD. ispartof: Ieee Transactions On Biomedical Engineering vol:67 issue:12 pages:3491-3500 ispartof: location:United States status: published

Automatic sleep stage scoring is of great importance for investigating sleep architecture during infancy. In this work, we introduce a novel multichannel approach based on deep learning networks and hidden Markov models (HMM) to improve the accuracy of sleep stage classification in term neonates. The classification performance was evaluated on quiet sleep (QS) and active sleep (AS) stages, each with two sub-states, using multichannel EEG data recorded from sixteen neonates with postmenstrual age of 38–40 weeks. A comprehensive set of linear and nonlinear features were extracted from thirty-second EEG segments. The feature space dimensionality was then reduced by using an evolutionary feature selection method called MGCACO (Modified Graph Clustering Ant Colony Optimization) based on the relevance and redundancy analysis. A bi-directional long-short time memory (BiLSTM) network was trained for sleep stage classification. The number of channels was optimized using the sequential forward selection method to reduce the spatial space. Finally, an HMM-based postprocessing stage was used to reduce false positives by incorporating the knowledge of transition probabilities between stages into the classification process. The method performance was evaluated using the K-fold (KFCV) and leave-one-out cross-validation (LOOCV) strategies. Using six-bipolar channels, our method achieved a mean kappa and an overall accuracy of 0.71-0.76 and 78.9%-82.4% using the KFCV and LOOCV strategies, respectively. The presented automatic sleep stage scoring method can be used to study the neurodevelopmental process and to diagnose brain abnormalities in term neonates.

Generative adversarial networks (GAN) have recently been shown to be efficient for speech enhancement. However, most, if not all, existing speech enhancement GANs (SEGAN) make use of a single generator to perform one-stage enhancement mapping. In this work, we propose to use multiple generators that are chained to perform multi-stage enhancement mapping, which gradually refines the noisy input signals in a stage-wise fashion. Furthermore, we study two scenarios: (1) the generators share their parameters and (2) the generators' parameters are independent. The former constrains the generators to learn a common mapping that is iteratively applied at all enhancement stages and results in a small model footprint. On the contrary, the latter allows the generators to flexibly learn different enhancement mappings at different stages of the network at the cost of an increased model size. We demonstrate that the proposed multi-stage enhancement approach outperforms the one-stage SEGAN baseline, where the independent generators lead to more favorable results than the tied generators. The source code is available at http://github.com/pquochuy/idsegan., This letter has been accepted for publication in IEEE Signal Processing Letters

Change point detection (CPD) aims to locate abrupt property changes in time series data. Recent CPD methods demonstrated the potential of using deep learning techniques, but often lack the ability to identify more subtle changes in the autocorrelation statistics of the signal and suffer from a high false alarm rate. To address these issues, we employ an autoencoder-based methodology with a novel loss function, through which the used autoencoders learn a partially time-invariant representation that is tailored for CPD. The result is a flexible method that allows the user to indicate whether change points should be sought in the time domain, frequency domain or both. Detectable change points include abrupt changes in the slope, mean, variance, autocorrelation function and frequency spectrum. We demonstrate that our proposed method is consistently highly competitive or superior to baseline methods on diverse simulated and real-life benchmark data sets. Finally, we mitigate the issue of false detection alarms through the use of a postprocessing procedure that combines a matched filter and a newly proposed change point score. We show that this combination drastically improves the performance of our method as well as all baseline methods.

OBJECTIVE: To classify sleep states using electroencephalogram (EEG) that reliably works over a wide range of preterm ages, as well as term age. APPROACH: A convolutional neural network is developed to perform 2- and 4-class sleep classification in neonates. The network takes as input an 8-channel 30 s EEG segment and outputs the sleep state probabilities. Apart from simple downsampling of the input and smoothing of the output, the suggested network is an end-to-end algorithm that avoids the need for hand-crafted feature selection or complex pre/post processing steps. To train and test this method, 113 EEG recordings from 42 infants are used. MAIN RESULTS: For quiet sleep detection (the 2-class problem), mean kappa between the network estimate and the ground truth annotated by EEG human experts is 0.76. The sensitivity and specificity are 90% and 88%, respectively. For 4-class classification, mean kappa is 0.64. The averaged sensitivity and specificity (1 versus all) respectively equal 72% and 91%. The results outperform current state-of-the-art methods for which kappa ranges from 0.66 to 0.70 in preterm and from 0.51 to 0.61 in term infants, based on training and testing using the same database. SIGNIFICANCE: The proposed method has the highest reported accuracy for EEG sleep state classification for both preterm and term age neonates. ispartof: JOURNAL OF NEURAL ENGINEERING vol:17 issue:1 ispartof: location:England status: published

In neonatal intensive care units, there is a need for around the clock monitoring of electroencephalogram (EEG), especially for recognizing seizures. An automated seizure detector with an acceptable performance can partly fill this need. In order to develop a detector, an extensive dataset labeled by experts is needed. However, accurately defining neonatal seizures on EEG is a challenge, especially when seizure discharges do not meet exact definitions of repetitiveness or evolution in amplitude and frequency. When several readers score seizures independently, disagreement can be high. Commonly used metrics such as good detection rate (GDR) and false alarm rate (FAR) derived from data scored by multiple raters have their limitations. Therefore, new metrics are needed to measure the performance with respect to the different labels. In this paper, instead of defining the labels by consensus or majority voting, popular metrics including GDR, FAR, positive predictive value, sensitivity, specificity, and selectivity are modified such that they can take different scores into account. To this end, 353 hours of EEG data containing seizures from 81 neonates were visually scored by a clinical neurophysiologist, and then processed by an automated seizure detector. The scored seizures were mixed with false detections of an automated seizure detector and were relabeled by three independent EEG readers. Then, all labels were used in the proposed performance metrics and the result was compared with the majority voting technique and showed higher accuracy and robustness for the proposed metrics. Results were confirmed using a bootstrapping test. ispartof: IEEE Journal of Biomedical and Health Informatics vol:22 issue:4 pages:1114-1123 ispartof: location:United States status: published

BackgroundProgressive supranuclear palsy (PSP), a neurodegenerative conditions may be difficult to discriminate clinically from idiopathic Parkinson’s disease (PD). It is critical that we are able to do this accurately and as early as possible in order that future disease modifying therapies for PSP may be deployed at a stage when they are likely to have maximal benefit. Analysis of gait and related tasks is one possible means of discrimination.Research QuestionHere we investigate a wearable sensor array coupled with machine learning approaches as a means of disease classification.Methods21 participants with PSP, 20 with PD, and 39 healthy control (HC) subjects performed a two minute walk, static sway test, and timed up-and-go task, while wearing an array of six inertial measurement units. The data were analysed to determine what features discriminated PSP from PD and PSP from HC. Two machine learning algorithms were applied, Logistic Regression (LR) and Random Forest (RF).Results17 features were identified in the combined dataset that contained independent information. The RF classifier outperformed the LR classifier, and allowed discrimination of PSP from PD with 86% sensitivity and 90% specificity, and PSP from HC with 90% sensitivity and 97% specificity. Using data from the single lumbar sensor only resulted in only a modest reduction in classification accuracy, which could be restored using 3 sensors (lumbar, right arm and foot). However for maximum specificity the full six sensor array was needed.SignificanceA wearable sensor array coupled with machine learning methods can accurately discriminate PSP from PD. Choice of array complexity depends on context; for diagnostic purposes a high specificity is needed suggesting the more complete array is advantageous, while for subsequent disease tracking a simpler system may suffice.

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

Acoustic scenes are rich and redundant in their content. In this work, we present a spatio-temporal attention pooling layer coupled with a convolutional recurrent neural network to learn from patterns that are discriminative while suppressing those that are irrelevant for acoustic scene classification. The convolutional layers in this network learn invariant features from time-frequency input. The bidirectional recurrent layers are then able to encode the temporal dynamics of the resulting convolutional features. Afterwards, a two-dimensional attention mask is formed via the outer product of the spatial and temporal attention vectors learned from two designated attention layers to weigh and pool the recurrent output into a final feature vector for classification. The network is trained with between-class examples generated from between-class data augmentation. Experiments demonstrate that the proposed method not only outperforms a strong convolutional neural network baseline but also sets new state-of-the-art performance on the LITIS Rouen dataset., To appear at the 20th Annual Conference of the International Speech Communication Association (INTERSPEECH 2019)

Many sleep studies suffer from the problem of insufficient data to fully utilize deep neural networks as different labs use different recordings set ups, leading to the need of training automated algorithms on rather small databases, whereas large annotated databases are around but cannot be directly included into these studies for data compensation due to channel mismatch. This work presents a deep transfer learning approach to overcome the channel mismatch problem and transfer knowledge from a large dataset to a small cohort to study automatic sleep staging with single-channel input. We employ the state-of-the-art SeqSleepNet and train the network in the source domain, i.e. the large dataset. Afterwards, the pretrained network is finetuned in the target domain, i.e. the small cohort, to complete knowledge transfer. We study two transfer learning scenarios with slight and heavy channel mismatch between the source and target domains. We also investigate whether, and if so, how finetuning entirely or partially the pretrained network would affect the performance of sleep staging on the target domain. Using the Montreal Archive of Sleep Studies (MASS) database consisting of 200 subjects as the source domain and the Sleep-EDF Expanded database consisting of 20 subjects as the target domain in this study, our experimental results show significant performance improvement on sleep staging achieved with the proposed deep transfer learning approach. Furthermore, these results also reveal the essential of finetuning the feature-learning parts of the pretrained network to be able to bypass the channel mismatch problem., Accepted for 27th European Signal Processing Conference (EUSIPCO 2019)

Background Avoidance to look others in the eye is a characteristic symptom of Autism Spectrum Disorders (ASD), and it has been hypothesised that quantitative monitoring of gaze patterns could be useful to objectively evaluate treatments. However, tools to measure gaze behaviour on a regular basis at a manageable cost are missing. In this paper, we investigated whether a smartphone-based tool could address this problem. Specifically, we assessed the accuracy with which the phone-based, state-of-the-art eye-tracking algorithm iTracker can distinguish between gaze towards the eyes and the mouth of a face displayed on the smartphone screen. This might allow mobile, longitudinal monitoring of gaze aversion behaviour in ASD patients in the future. Results We simulated a smartphone application in which subjects were shown an image on the screen and their gaze was analysed using iTracker. We evaluated the accuracy of our set-up across three tasks in a cohort of 17 healthy volunteers. In the first two tasks, subjects were shown different-sized images of a face and asked to alternate their gaze focus between the eyes and the mouth. In the last task, participants were asked to trace out a circle on the screen with their eyes. We confirm that iTracker can recapitulate the true gaze patterns, and capture relative position of gaze correctly, even on a different phone system to what it was trained on. Subject-specific bias can be corrected using an error model informed from the calibration data. We compare two calibration methods and observe that a linear model performs better than a previously proposed support vector regression-based method. Conclusions Under controlled conditions it is possible to reliably distinguish between gaze towards the eyes and the mouth with a smartphone-based set-up. However, future research will be required to improve the robustness of the system to roll angle of the phone and distance between the user and the screen to allow deployment in a home setting. We conclude that a smartphone-based gaze-monitoring tool provides promising opportunities for more quantitative monitoring of ASD. Electronic supplementary material The online version of this article (10.1186/s12938-019-0670-1) contains supplementary material, which is available to authorized users.

Multi-source datasets often present the challenge of source-wise missing data which can render large portions of the dataset inaccessible. The applicability of traditional missing data techniques on multi-source datasets is poorly understood. We present the first quantitative evaluation of the state-of-the-art missing data techniques as applied to a freely available dataset of smart-phone recordings from Parkinsonian patients wherein source-wise missing data is simulated. The classification accuracy and imputation error of five missing data techniques, including a multi-modal autoencoder and multi-source ensemble learning, are compared at varying levels of missingness. These results demonstrate the relative applicability of each technique under different conditions and subsequently highlight the challenges of source-wise missing on remotely collected datasets. Specifically, multi-source ensemble learning proves to be a highly successful alternative to the traditional imputation techniques when a majority of observations possess missing data.

The goal of this study is to develop an automated algorithm to quantify background electroencephalography (EEG) dynamics in term neonates with hypoxic ischemic encephalopathy. The recorded EEG signal is adaptively segmented and the segments with low amplitudes are detected. Next, depending on the spatial distribution of the low-amplitude segments, the first part of the algorithm detects (dynamic) interburst intervals (dIBIs) and performs well on the relatively artifact-free EEG periods and well-defined burst-suppression EEG periods. However, on testing the algorithm on EEG recordings of more than 48 h per neonate, a significant number of misclassified and dubious detections were encountered. Therefore, as the next step, we applied machine learning classifiers to differentiate between definite dIBI detections and misclassified ones. The developed algorithm achieved a true positive detection rate of 98%, 97%, 88%, and 95% for four duration-related dIBI groups that we subsequently defined. We benchmarked our algorithm with an expert diagnostic interpretation of EEG periods (1 h long) and demonstrated its effectiveness in clinical practice. We show that the detection algorithm effectively discriminates challenging cases encountered within mild and moderate background abnormalities. The dIBI detection algorithm improves identification of neonates with good clinical outcome as compared to the classification based on the classical burst-suppression interburst interval.

Objective: Brain waves vary between people. This work aims to improve automatic sleep staging for longitudinal sleep monitoring via personalization of algorithms based on individual characteristics extracted from sleep data recorded during the first night. Approach: As data from a single night are very small, thereby making model training difficult, we propose a Kullback–Leibler (KL) divergence regularized transfer learning approach to address this problem. We employ the pretrained SeqSleepNet (i.e. the subject independent model) as a starting point and finetune it with the single-night personalization data to derive the personalized model. This is done by adding the KL divergence between the output of the subject independent model and it of the personalized model to the loss function during finetuning. In effect, KL-divergence regularization prevents the personalized model from overfitting to the single-night data and straying too far away from the subject independent model. Main results: Experimental results on the Sleep-EDF Expanded database consisting of 75 subjects show that sleep staging personalization with single-night data is possible with help of the proposed KL-divergence regularization. On average, we achieve a personalized sleep staging accuracy of 79.6%, a Cohen’s kappa of 0.706, a macro F1-score of 73.0%, a sensitivity of 71.8%, and a specificity of 94.2%. Significance: We find both that the approach is robust against overfitting and that it improves the accuracy by 4.5 percentage points compared to the baseline method without personalization and 2.2 percentage points compared to it with personalization but without regularization.

We present in this paper an efficient convolutional neural network (CNN) running on time-frequency image features for automatic sleep stage classification. Opposing to deep architectures which have been used for the task, the proposed CNN is much simpler However, the CNN's convolutional layer is able to support convolutional kernels with different sizes, and therefore, capable of learning features at multiple temporal resolutions. In addition, the 1-max pooling strategy is employed at the pooling layer to better capture the shift-invariance property of EEG signals. We further propose a method to discriminatively learn a frequency-domain filter bank with a deep neural network (DNN) to preprocess the time-frequency image features. Our experiments show that the proposed 1-max pooling CNN performs comparably with the very deep CNNs in the literature on the Sleep- EDF dataset. Preprocessing the time-frequency image features with the learned filter bank before presenting them to the CNN leads to significant improvements on the classification accuracy, setting the state- of-the-art performance on the dataset.

The diagnosis of cardiovascular diseases such as atrial fibrillation (AF) is a lengthy and expensive procedure that often requires visual inspection of ECG signals by experts. In order to improve patient management and reduce healthcare costs, automated detection of these pathologies is of utmost importance. In this study, we classify short segments of ECG into four classes (AF, normal, other rhythms or noise) as part of the Physionet/Computing in Cardiology Challenge 2017. We compare a state-of-the-art feature-based classifier with a convolutional neural network approach. Both methods were trained using the challenge data, supplemented with an additional database derived from Physionet. The feature-based classifier obtained an F1 score of 72.0% on the training set (5-fold cross-validation), and 79% on the hidden test set. Similarly, the convolutional neural network scored 72.1% on the augmented database and 83% on the test set. The latter method resulted on a final score of 79% at the competition. Developed routines and pre-trained models are freely available under a GNU GPLv3 license.

Automatic sleep staging has been often treated as a simple classification problem that aims at determining the label of individual target polysomnography (PSG) epochs one at a time. In this work, we tackle the task as a sequence-to-sequence classification problem that receives a sequence of multiple epochs as input and classifies all of their labels at once. For this purpose, we propose a hierarchical recurrent neural network named SeqSleepNet. At the epoch processing level, the network consists of a filterbank layer tailored to learn frequency-domain filters for preprocessing and an attention-based recurrent layer designed for short-term sequential modelling. At the sequence processing level, a recurrent layer placed on top of the learned epoch-wise features for long-term modelling of sequential epochs. The classification is then carried out on the output vectors at every time step of the top recurrent layer to produce the sequence of output labels. Despite being hierarchical, we present a strategy to train the network in an end-to-end fashion. We show that the proposed network outperforms state-of-the-art approaches, achieving an overall accuracy, macro F1-score, and Cohen's kappa of 87.1%, 83.3%, and 0.815 on a publicly available dataset with 200 subjects., This article has been published in IEEE Transactions on Neural Systems and Rehabilitation Engineering

Current sleep medicine relies on the supervised analysis of polysomnographic measurements, comprising amongst others electroencephalogram (EEG), electromyogram (EMG), and electrooculogram (EOG) signals. Convolutional neural networks (CNN) provide an interesting framework to automated classification of sleep based on these raw waveforms. In this study, we compare existing CNN approaches to four databases of pathological and physiological subjects. The best performing model resulted in Cohen's Kappa of $\kappa = 0 .75$ on healthy subjects and $\kappa = 0 .64$ on patients suffering from a variety of sleep disorders. Further, we show the advantages of additional sensor data (i.e., EOG and EMG). Deep learning approaches require a lot of data which is scarce for less prevalent diseases. For this, we propose a transfer learning procedure by pretraining a model on large public data and fine-tune this on each subject from a smaller dataset. This procedure is demonstrated using a private REM Behaviour Disorder database, improving sleep classification by 24.4%.

We propose in this work a feature learning approach using deep bidirectional recurrent neural networks (RNNs) with attention mechanism for single-channel automatic sleep stage classification. We firstly decompose an EEG epoch into multiple small frames and subsequently transform them into a sequence of frame-wise feature vectors. Given the training sequences, the attention-based RNN is trained in a sequence-to-label fashion for sleep stage classification. Due to discriminative training, the network is expected to encode information of an input sequence into a high-level feature vector after the attention layer. We, therefore, treat the trained network as a feature extractor and extract these feature vectors for classification which is accomplished by a linear SVM classifier. We also propose a discriminative method to learn a filter bank with a DNN for preprocessing purpose. Filtering the frame-wise feature vectors with the learned filter bank beforehand leads to further improvement on the classification performance. The proposed approach demonstrates good performance on the Sleep-EDF dataset.

The assessment of Parkinson's disease (PD) using wearable sensors in non-clinical environments presents an opportunity for objective disease classification and severity prediction on a high-frequency and longitudinal basis. However, many challenges exist in analysing remotely collected data due to many sources of data corruption. Using a cohort of 1,815 participants (866 controls and 949 with PD) we implement a range of classification algorithms on Alternate Finger Tapping test data collected on smart-phones in remote environments. We compare the disease classification ability of two traditional machine learning methods against two state-of-the-art deep learning approaches, determining if the latter is successful without the definition of an explicit feature set. We find the deep learning approaches capable of disease classification, often outperforming traditional methods. We show similarities between the participants successfully classified through use of a manually extracted feature set, and the features learnt by a convolutional neural network. Finally, we discuss the broader challenges of analysing remotely collected datasets whilst highlighting the suitability of deep learning for assessing PD when large participant numbers are available.

We propose a multi-label multi-task framework based on a convolutional recurrent neural network to unify detection of isolated and overlapping audio events. The framework leverages the power of convolutional recurrent neural network architectures; convolutional layers learn effective features over which higher recurrent layers perform sequential modelling. Furthermore, the output layer is designed to handle arbitrary degrees of event overlap. At each time step in the recurrent output sequence, an output triple is dedicated to each event category of interest to jointly model event occurrence and temporal boundaries. That is, the network jointly determines whether an event of this category occurs, and when it occurs, by estimating onset and offset positions at each recurrent time step. We then introduce three sequential losses for network training: multi-label classification loss, distance estimation loss, and confidence loss. We demonstrate good generalization on two datasets: ITC-Irst for isolated audio event detection, and TUT-SED-Synthetic-2016 for overlapping audio event detection., Comment: Accepted for the 44th International Conference on Acoustics, Speech, and Signal Processing (ICASSP 2019)

We revisit the multiple sparse priors (MSP) algorithm implemented in the statistical parametric mapping software (SPM) for distributed EEG source reconstruction (Friston et al., 2008). In the present implementation, multiple cortical patches are introduced as source priors based on a dipole source space restricted to a cortical surface mesh. In this note, we present a technique to construct volumetric cortical regions to introduce as source priors by restricting the dipole source space to a segmented gray matter layer and using a region growing approach. This extension allows to reconstruct brain structures besides the cortical surface and facilitates the use of more realistic volumetric head models including more layers, such as cerebrospinal fluid (CSF), compared to the standard 3-layered scalp-skull-brain head models. We illustrated the technique with ERP data and anatomical MR images in 12 subjects. Based on the segmented gray matter for each of the subjects, cortical regions were created and introduced as source priors for MSP-inversion assuming two types of head models. The standard 3-layered scalp-skull-brain head models and extended 4-layered head models including CSF. We compared these models with the current implementation by assessing the free energy corresponding with each of the reconstructions using Bayesian model selection for group studies. Strong evidence was found in favor of the volumetric MSP approach compared to the MSP approach based on cortical patches for both types of head models. Overall, the strongest evidence was found in favor of the volumetric MSP reconstructions based on the extended head models including CSF. These results were verified by comparing the reconstructed activity. The use of volumetric cortical regions as source priors is a useful complement to the present implementation as it allows to introduce more complex head models and volumetric source priors in future studies.

To study the dynamics of contour integration in the human brain, we simultaneously acquired EEG and fMRI data while participants were engaged in a passive viewing task. The stimuli were Gabor arrays with some Gabor elements positioned on the contour of an embedded shape, in three conditions: with local and global structure (perfect contour alignment), with global structure only (orthogonal orientations interrupting the alignment), or without contour. By applying JointICA to the EEG and fMRI responses of the subjects, new insights could be obtained that cannot be derived from unimodal recordings. In particular, only in the global structure condition, an ERP peak around 300ms was identified that involved a loop from LOC to the early visual areas. This component can be interpreted as being related to the verification of the consistency of the different local elements with the globally defined shape, which is necessary when perfect local-to-global alignment is absent. By modifying JointICA, a quantitative comparison of brain regions and the time-course of their interplay were obtained between different conditions. More generally, we provide additional support for the presence of feedback loops from higher areas to lower level sensory regions.

Identifying a core set of features is one of the most important steps in the development of an automated seizure detector. In most of the published studies describing features and seizure classifiers, the features were hand-engineered, which may not be optimal. The main goal of the present paper is using deep convolutional neural networks (CNNs) and random forest to automatically optimize feature selection and classification. The input of the proposed classifier is raw multi-channel EEG and the output is the class label: seizure/nonseizure. By training this network, the required features are optimized, while fitting a nonlinear classifier on the features. After training the network with EEG recordings of 26 neonates, five end layers performing the classification were replaced with a random forest classifier in order to improve the performance. This resulted in a false alarm rate of 0.9 per hour and seizure detection rate of 77% using a test set of EEG recordings of 22 neonates that also included dubious seizures. The newly proposed CNN classifier outperformed three data-driven feature-based approaches and performed similar to a previously developed heuristic method.

Objective Circadian and sleep dysfunction have long been symptomatic hallmarks of a variety of devastating neurodegenerative conditions. The gold standard for sleep monitoring is overnight sleep in a polysomnography (PSG) laboratory. However, this method has several limitations such as availability, cost and being labour-intensive. In recent years there has been a heightened interest in home-based sleep monitoring via wearable sensors. Our objective was to demonstrate the use of printed electrode technology as a novel platform for sleep monitoring. Approach Printed electrode arrays offer exciting opportunities in the realm of wearable electrophysiology. In particular, soft electrodes can conform neatly to the wearer's skin, allowing user convenience and stable recordings. As such, soft skin-adhesive non-gel-based electrodes offer a unique opportunity to combine electroencephalography (EEG), electromyography (EMG), electrooculography (EOG) and facial EMG capabilities to capture neural and motor functions in comfortable non-laboratory settings. In this investigation temporary-tattoo dry electrode system for sleep staging analysis was designed, implemented and tested. Main results EMG, EOG and EEG were successfully recorded using a wireless system. Stable recordings were achieved both at a hospital environment and a home setting. Sleep monitoring during a 6 h session shows clear differentiation of sleep stages. Significance The new system has great potential in monitoring sleep disorders in the home environment. Specifically, it may allow the identification of disorders associated with neurological disorders such as rapid eye movement (REM) sleep behavior disorder.

The decomposition of high-density surface EMG (HD-sEMG) interference patterns into the contribution of motor units is still a challenging task. We introduce a new, fast solution to this problem. The method uses a data-driven approach for selecting a set of electrodes to enable discrimination of present motor unit action potentials (MUAPs). Then, using shapes detected on these channels, the hierarchical clustering algorithm as reported by Quian Quiroga et al. (Neural Comput 16:1661-1687, 2004) is extended for multichannel data in order to obtain the motor unit action potential (MUAP) signatures. After this first step, more motor unit firings are obtained using the extracted signatures by a novel demixing technique. In this demixing stage, we propose a time-efficient solution for the general convolutive system that models the motor unit firings on the HD-sEMG grid. We constrain this system by using the extracted signatures as prior knowledge and reconstruct the firing patterns in a computationally efficient way. The algorithm performance is successfully verified on simulated data containing up to 20 different MUAP signatures. Moreover, we tested the method on real low contraction recordings from the lateral vastus leg muscle by comparing the algorithm's output to the results obtained by manual analysis of the data from two independent trained operators. The proposed method showed to perform about equally successful as the operators.

ispartof: Machiine Learning for Healthcare Technologies pages:83-109 ispartof: IET Healthcare Technologies Series vol:2 pages:83-109 status: published

© 2016 IEEE. Electroencephalography (EEG) and functional magnetic resonance imaging (fMRI) are two complementary modalities capturing a mixture of various underlying neural sources. The fusion of these modalities promises the best of both worlds, i.e. A better resolution in time and space, respectively. Assuming that EEG and fMRI observations are generated by the same mixing system in both modalities, their fusion can be achieved by joint blind source separation (BSS). We solve the joint BSS problem using different variants of joint independent component analysis (jointICA) and coupled matrix-tensor factorization (CMTF). We demonstrate that EEG-fMRI fusion provides a detailed spatio-temporal characterization of an EEG-fMRI dataset recorded in epilepsy patients, leading to new insights in epileptic network behaviour. ispartof: pages:240-244 ispartof: Proc. of the European signal processing conference vol:2016-November pages:240-244 ispartof: EUSIPCO 2016 location:Budapest, Hungary date:Aug - Aug 2016 status: published

To decode auditory attention from electroencephalography (EEG) recordings in a cocktail-party scenario with two competing speakers a least-squares method has recently been proposed, showing a promising decoding accuracy. This method however requires the clean speech signals of both the attended and the unattended speaker to be available as reference signals, which is difficult to achieve from the noisy recorded microphone signals in practice. In addition, optimizing the parameters involved in the spatio-temporal filter design is of crucial importance in order to reach the largest possible decoding performance. In this paper, the influence of noisy acoustic reference signals and the spatio-temporal filter and regularization parameters on the decoding performance is investigated. The results show that to some extent the decoding performance is robust to noisy acoustic reference signals, depending on the noise type. Furthermore, we demonstrate the crucial influence of several parameters on the decoding performance, especially when the acoustic reference signals used for decoding have been corrupted by noise.

OBJECTIVE: One of the major drawbacks in EEG brain-computer interfaces (BCI) is the need for subject-specific training of the classifier. By removing the need for a supervised calibration phase, new users could potentially explore a BCI faster. In this work we aim to remove this subject-specific calibration phase and allow direct classification. APPROACH: We explore canonical polyadic decompositions and block term decompositions of the EEG. These methods exploit structure in higher dimensional data arrays called tensors. The BCI tensors are constructed by concatenating ERP templates from other subjects to a target and non-target trial and the inherent structure guides a decomposition that allows accurate classification. We illustrate the new method on data from a three-class auditory oddball paradigm. MAIN RESULTS: The presented approach leads to a fast and intuitive classification with accuracies competitive with a supervised and cross-validated LDA approach. SIGNIFICANCE: The described methods are a promising new way of classifying BCI data with a forthright link to the original P300 ERP signal over the conventional and widely used supervised approaches. journal_title: Journal of Neural Engineering article_type: paper article_title: Tensor-based classification of an auditory mobile BCI without a subject-specific calibration phase copyright_information: © 2016 IOP Publishing Ltd date_received: 2015-07-17 date_accepted: 2015-12-23 date_epub: 2016-02-01 ispartof: Journal of Neural Engineering vol:13 issue:2 pages:026005-026005 ispartof: location:England status: published

Electrical artifacts caused by the cochlear implant (CI) contaminate electroencephalographic (EEG) recordings from implanted individuals and corrupt auditory evoked potentials (AEPs). Independent component analysis (ICA) is efficient in attenuating the electrical CI artifact and AEPs can be successfully reconstructed. However the manual selection of CI artifact related independent components (ICs) obtained with ICA is unsatisfactory, since it contains expert-choices and is time consuming. We developed a new procedure to evaluate temporal and topographical properties of ICs and semi-automatically select those components representing electrical CI artifact. The CI Artifact Correction (CIAC) algorithm was tested on EEG data from two different studies. The first consists of published datasets from 18 CI users listening to environmental sounds. Compared to the manual IC selection performed by an expert the sensitivity of CIAC was 91.7% and the specificity 92.3%. After CIAC-based attenuation of CI artifacts, a high correlation between age and N1–P2 peak-to-peak amplitude was observed in the AEPs, replicating previously reported findings and further confirming the algorithm's validity. In the second study AEPs in response to pure tone and white noise stimuli from 12 CI users that had also participated in the other study were evaluated. CI artifacts were attenuated based on the IC selection performed semi-automatically by CIAC and manually by one expert. Again, a correlation between N1 amplitude and age was found. Moreover, a high test–retest reliability for AEP N1 amplitudes and latencies suggested that CIAC-based attenuation reliably preserves plausible individual response characteristics. We conclude that CIAC enables the objective and efficient attenuation of the CI artifact in EEG recordings, as it provided a reasonable reconstruction of individual AEPs. The systematic pattern of individual differences in N1 amplitudes and latencies observed with different stimuli at different sessions, strongly suggests that CIAC can overcome the electrical artifact problem. Thus CIAC facilitates the use of cortical AEPs as an objective measurement of auditory rehabilitation.

Since several years, neuroscience research started to focus on multimodal approaches. One such multimodal approach is the combination of electroencephalography (EEG) and functional magnetic resonance imaging (fMRI). However, no standard integration procedure has been established so far. One promising data-driven approach consists of a joint decomposition of event-related potentials (ERPs) and fMRI maps derived from the response to a particular stimulus. Such an algorithm (joint independent component analysis or JointICA) has recently been proposed by Calhoun et al. (2006). This method provides sources with both a fine spatial and temporal resolution, and has shown to provide meaningful results. However, the algorithm's performance has not been fully characterized yet, and no procedure has been proposed to assess the quality of the decomposition. In this paper, we therefore try to answer why and how JointICA works. We show the performance of the algorithm on data obtained in a visual detection task, and compare the performance for EEG recorded simultaneously with fMRI data and for EEG recorded in a separate session (outside the scanner room). We perform several analyses in order to set the necessary conditions that lead to a sound decomposition, and to give additional insights for exploration in future studies. In that respect, we show how the algorithm behaves when different EEG electrodes are used and we test the robustness with respect to the number of subjects in the study. The performance of the algorithm in all the experiments is validated based on results from previous studies. ispartof: NeuroImage vol:60 issue:2 pages:1171-1185 ispartof: location:United States status: published

Independent Component Analysis (ICA) aims at blindly decomposing a linear mixture of independent sources. It has lots of applications in diverse research areas. In some applications, there is prior knowledge on the sources and/or the mixing vectors. This prior knowledge can be incorporated in the computation of the independent sources. In this paper we provide an algorithm for so-called spatially constrained ICA (scICA). The algorithm deals with the situation when one mixing vector is exactly known. Also the generalization to more mixing vectors is discussed. Numerical experiments are reported that allow us to assess the improvement in accuracy that can be achieved with these algorithms compared to fully blind ICA and to a previously proposed constrained algorithm. We illustrate the approach with a biomedical application.

Multimodal approaches are of growing interest in the study of neural processes. To this end much attention has been paid to the integration of electroencephalographic (EEG) and functional magnetic resonance imaging (fMRI) data because of their complementary properties. However, the simultaneous acquisition of both types of data causes serious artifacts in the EEG, with amplitudes that may be much larger than those of EEG signals themselves. The most challenging of these artifacts is the ballistocardiogram (BCG) artifact, caused by pulse-related electrode movements inside the magnetic field. Despite numerous efforts to find a suitable approach to remove this artifact, still a considerable discrepancy exists between current EEG-fMRI studies. This paper attempts to clarify several methodological issues regarding the different approaches with an extensive validation based on event-related potentials (ERPs). More specifically, Optimal Basis Set (OBS) and Independent Component Analysis (ICA) based methods were investigated. Their validation was not only performed with measures known from previous studies on the average ERPs, but most attention was focused on task-related measures, including their use on trial-to-trial information. These more detailed validation criteria enabled us to find a clearer distinction between the most widely used cleaning methods. Both OBS and ICA proved to be able to yield equally good results. However, ICA methods needed more parameter tuning, thereby making OBS more robust and easy to use. Moreover, applying OBS prior to ICA can optimize the data quality even more, but caution is recommended since the effect of the additional ICA step may be strongly subject-dependent. ispartof: NeuroImage vol:50 issue:3 pages:920-934 ispartof: location:United States status: published

Essential information about early brain maturation can be retrieved from the preterm human electroencephalogram (EEG). This study proposes a new set of quantitative features that correlate with early maturation. We exploit the known early trend in EEG content from intermittent to continuous activity, which changes the line length content of the EEG. The developmental shift can be captured in the line length histogram, which we use to obtain 28 features; 20 histogram bins and 8 other statistical measurements. Using the mutual information, we select 6 features with high correlation to the infant's age. This subset appears promising to detect deviances from normal brain maturation. The presented data-driven index holds promise for developing into a computational EEG index of maturation that is highly needed for overall assessment in the Neonatal Intensive Care Units.

Objective. Recent studies have provided evidence that temporal envelope driven speech decoding from high-density electroencephalography (EEG) and magnetoencephalography recordings can identify the attended speech stream in a multi-speaker scenario. The present work replicated the previous high density EEG study and investigated the necessary technical requirements for practical attended speech decoding with EEG. Approach. Twelve normal hearing participants attended to one out of two simultaneously presented audiobook stories, while high density EEG was recorded. An offline iterative procedure eliminating those channels contributing the least to decoding provided insight into the necessary channel number and optimal cross-subject channel configuration. Aiming towards the future goal of near real-time classification with an individually trained decoder, the minimum duration of training data necessary for successful classification was determined by using a chronological cross-validation approach. Main results. Close replication of the previously reported results confirmed the method robustness. Decoder performance remained stable from 96 channels down to 25. Furthermore, for less than 15 min of training data, the subject-independent (pre-trained) decoder performed better than an individually trained decoder did. Significance. Our study complements previous research and provides information suggesting that efficient low-density EEG online decoding is within reach.

Electroencephalography (EEG) allows the study of the brain–behavior relationship in humans. Most of what we have learned with EEG was through observing the brain–behavior relationship under well-controlled laboratory conditions. However, by reducing “normal” behavior to a minimum the ecological validity of the results can be limited. Recent developments toward mobile EEG solutions allow to study the brain–behavior relationship outside the laboratory in more natural situations. Besides mobility and robustness with respect to motion, mobile EEG systems should also interfere as little as possible with the participant's behavior. For example, natural interaction with other people could be hindered when it is obvious that a participant wears an EEG cap. This study evaluates the signal quality obtained with an unobtrusive solution for EEG monitoring through the integration of miniaturized EEG ton-electrodes into both a discreet baseball cap and an individualized ear piece. We show that such mini electrodes located at scalp and ear locations can reliably record event related potentials in a P300 brain–computer–interface application.

Goal: This paper deals with the problems that some EEG signals have no good sparse representation and single channel processing is not computationally efficient in compressed sensing of multi-channel EEG signals. Methods: An optimization model with L0 norm and Schatten-0 norm is proposed to enforce cosparsity and low rank structures in the reconstructed multi-channel EEG signals. Both convex relaxation and global consensus optimization with alternating direction method of multipliers are used to compute the optimization model. Results: The performance of multi-channel EEG signal reconstruction is improved in term of both accuracy and computational complexity. Conclusion: The proposed method is a better candidate than previous sparse signal recovery methods for compressed sensing of EEG signals. Significance: The proposed method enables successful compressed sensing of EEG signals even when the signals have no good sparse representation. Using compressed sensing would much reduce the power consumption of wireless EEG system., Comment: 11 pages, 3 figures; accepted by IEEE Transactions on Biomedical Engineering

Subspace methods have been applied in various application fields to obtain robust results. Using multilinear algebra, they can also be applied on structured tensorial data. This work combines this principle with the power of non-linear kernels to investigate its merits in single trial classification for a mobile BCI ERP classification task. The accuracy difference with regard to more conventional vector kernels is evaluated for sitting and walking condition, increasing training data set and averaging over multiple trials. The study concludes that in general, the tensorial approach does not yield any advantage, though it might for specific subjects.

In the interest of the neonatal comfort, the need for noncontact respiration monitoring increases. Moreover, home respiration monitoring would be beneficial. Therefore, the goal is to extract the respiration rate from video data included in a polysomnography. The presented method first uses Eulerian video magnification to amplify the respiration movements. A respiration signal is obtained through the optical flow algorithm. Independent component analysis and principal component analysis are applied to improve the signal quality, with minor enhancement of the signal quality. The respiratory rate is extracted as the dominant frequency in the spectrograms obtained using the short-time Fourier transform. Respiratory rate detection is successful (94.12%) for most patients during quiet sleep stages. Real-time monitoring could possibly be achieved by lowering the spatial and temporal resolutions of the input video data. The outline for successful video-aided detection of the respiration pattern is shown, thereby paving the way for improvement of the overall assessment in the NICU and application in a home-friendly environment.