Single-channel EEG Research Articles

Considering the effect of correlation between the channels in multivariate detectors for evoked responses in the electroencephalogram

Evoked potentials refer to the small responses in the electroencephalogram (EEG), resulting from repeated or continuous auditory, visual or other stimuli. The detection of these responses is most commonly based on visual inspection by experienced professionals. To overcome the inherent subjectivity, automated (usually statistical) methods can be used, which aim to find the statistical significance of some parameter(s) calculated from the recorded signals, based on the null-hypothesis of 'no stimulus response is present'. These tests need to rely on assumptions about the statistical distributions of the signals and their parameters which in many cases allows p-values to be calculated. For some convenient parameters, however, closed form mathematical solutions are intractable and simulations are often required to find p-values (or critical values). The use of multichannel recordings can increase the sensitivity of tests compared to single-channel EEG, permitting the detection of responses in shorter recordings. This can be achieved by simple extensions of originally single channel methods, such as magnitude-squared coherence, component synchrony measure or the spectral F-test. However, the correlation between EEG channels can lead to false-positive rates that are difficult to predict or control. In this paper we present a modified parametric bootstrap method that permits the accurate estimation of false positive rates (p-values and critical values) for multichannel extensions of univariate objective evoked response techniques, with concomitant increases in sensitivity over conventional single channel methods. Illustrative results with evoked potentials from repeated photic stimulation are provided, with multivariate magnitude-squared coherence demonstrating the most robust performance.

Identification of optimum segment in single channel EEG biometric system

Nowadays, biometric modalities have gained popularity in security systems. Nevertheless, the conventional commercial-grade biometric system addresses some issues. The biggest problem is that they can be imposed by artificial biometrics. The electroencephalogram (EEG) is a possible solution. It is nearly impossible to replicate because it is dependent on human mental activity. Several studies have already demonstrated a high level of accuracy. However, it requires a large number of sensors and time to collect the signal. This study proposed a biometric system using single-channel EEG recorded during resting eyes open (EO) conditions. A total of 45 EEG signals from 9 subjects were collected. The EEG signal was segmented into 5 second lengths. The alpha band was used in this study. Discrete wavelet transform (DWT) with Daubechies type 4 (db4) was employed to extract the alpha band. Power spectral density (PSD) was extracted from each segment as the main feature. Linear discriminant analysis (LDA) and support vector machine (SVM) were used to classify the EEG signal. The proposed method achieved 86% accuracy using LDA only from the third segment. Therefore, this study showed that it is possible to utilize single-channel EEG during a resting EO state in a biometric system.

Markers of Central Neuropathic Pain in Higuchi Fractal Analysis of EEG Signals From People With Spinal Cord Injury.

Central neuropathic pain (CNP) negatively impacts the quality of life in a large proportion of people with spinal cord injury (SCI). With no cure at present, it is crucial to improve our understanding of how CNP manifests, to develop diagnostic biomarkers for drug development, and to explore prognostic biomarkers for personalised therapy. Previous work has found early evidence of diagnostic and prognostic markers analysing Electroencephalogram (EEG) oscillatory features. In this paper, we explore whether non-linear non-oscillatory EEG features, specifically Higuchi Fractal Dimension (HFD), can be used as prognostic biomarkers to increase the repertoire of available analyses on the EEG of people with subacute SCI, where having both linear and non-linear features for classifying pain may ultimately lead to higher classification accuracy and an intrinsically transferable classifier. We focus on EEG recorded during imagined movement because of the known relation between the motor cortex over-activity and CNP. Analyses were performed on two existing datasets. The first dataset consists of EEG recordings from able-bodied participants (N = 10), participants with chronic SCI and chronic CNP (N = 10), and participants with chronic SCI and no CNP (N = 10). We tested for statistically significant differences in HFD across all pairs of groups using bootstrapping, and found significant differences between all pairs of groups at multiple electrode locations. The second dataset consists of EEG recordings from participants with subacute SCI and no CNP (N = 20). They were followed-up 6 months post recording to test for CNP, at which point (N = 10) participants had developed CNP and (N = 10) participants had not developed CNP. We tested for statistically significant differences in HFD between these two groups using bootstrapping and, encouragingly, also found significant differences at multiple electrode locations. Transferable machine learning classifiers achieved over 80% accuracy discriminating between groups of participants with chronic SCI based on only a single EEG channel as input. The most significant finding is that future and chronic CNP share common features and as a result, the same classifier can be used for both. This sheds new light on pain chronification by showing that frontal areas, involved in the affective aspects of pain and believed to be influenced by long-standing pain, are affected in a much earlier phase of pain development.

Reliable sleep staging of unseen subjects with fusion of multiple EEG features and RUSBoost

Extensive experiments have been carried out in this study to classify sleep EEG from three different standard databases – Sleep EDF, DREAMS and Expanded sleep EDF databases. Both two-class (sleep-awake) and multiclass classifications have been performed using a fusion of various EEG features and an ensemble classifier called random undersampling with boosting technique (RUSBoost). The results achieved using a single channel EEG are comparable or better than the state-of-the-art methods in the literature for both types of classification, on all the databases. Two-class classification is useful to determine the preferred timings for sensory stimulation of patients with disorders of consciousness. 10-fold cross-validation accuracies of 92.6% and 97.9% have been obtained on Sleep EDF database for 6-class and 2-class problems, respectively. Using Expanded Sleep-EDF dataset, the accuracies improved to 96.3% for 6-state and 99.8% for 2-state classification. For DREAMS dataset, we achieved an accuracy of 96.6% for 2-state classification. Unlike most research in the literature where performance on unseen subjects is not considered, we report classification results on the data from unseen test subjects using both 50%-holdout and leave-one-out cross-validation approaches. Similar results were achieved using both validation techniques for different datasets emphasizing the reliability of our method. These results are very crucial for the method to be applicable for clinical use.

Deep long short term memory based minimum variance kernel random vector functional link network for epileptic EEG signal classification

In this paper, the efficiently extracted and reduced features using deep long short-term memory (DLSTM) of the epileptic EEG signal integrated with minimum variance kernel random vector functional link net (MVKRVFLN) classifier are used to identify the seizure and non-seizure productively. Our methodology uses Children Hospital Boston-Massachusetts Institute of Technology (CHB-MIT) multi-channel scalp EEG data, Bonn university single-channel intracranial EEG data, and two-channel Bern Barcelona intracranial EEG data recordings to assess the performance. The non-stationary, non-linear, complex, and chaotic type EEG signal is directly applied to DLSTM to obtain compressed significant features. The scatter plots of the DLSTM output features signify this compressed information are unique in nature. The excellent generalization ability, faster learning rate, simpler network-based MVKRVFLN classifier is formulated to well identify the seizure and non-seizure epochs precisely by applying the deep LSTM extracted discriminative features as input. The type of kernel function selection and choice of regularization coefficient are added information to improve the performance of the proposed approach. The suggested technique provides excellent classification accuracy, superior detection ability, faster speed, and insignificant false positive rate per hour, simpler structure, robustness to classify the seizure and non-seizure signals.

Automated Characterization of Cyclic Alternating Pattern Using Wavelet-Based Features and Ensemble Learning Techniques with EEG Signals.

Sleep is highly essential for maintaining metabolism of the body and mental balance for increased productivity and concentration. Often, sleep is analyzed using macrostructure sleep stages which alone cannot provide information about the functional structure and stability of sleep. The cyclic alternating pattern (CAP) is a physiological recurring electroencephalogram (EEG) activity occurring in the brain during sleep and captures microstructure of the sleep and can be used to identify sleep instability. The CAP can also be associated with various sleep-related pathologies, and can be useful in identifying various sleep disorders. Conventionally, sleep is analyzed using polysomnogram (PSG) in various sleep laboratories by trained physicians and medical practitioners. However, PSG-based manual sleep analysis by trained medical practitioners is onerous, tedious and unfavourable for patients. Hence, a computerized, simple and patient convenient system is highly desirable for monitoring and analysis of sleep. In this study, we have proposed a system for automated identification of CAP phase-A and phase-B. To accomplish the task, we have utilized the openly accessible CAP sleep database. The study is performed using two single-channel EEG modalities and their combination. The model is developed using EEG signals of healthy subjects as well as patients suffering from six different sleep disorders namely nocturnal frontal lobe epilepsy (NFLE), sleep-disordered breathing (SDB), narcolepsy, periodic leg movement disorder (PLM), insomnia and rapid eye movement behavior disorder (RBD) subjects. An optimal orthogonal wavelet filter bank is used to perform the wavelet decomposition and subsequently, entropy and Hjorth parameters are extracted from the decomposed coefficients. The extracted features have been applied to different machine learning algorithms. The best performance is obtained using ensemble of bagged tress (EBagT) classifier. The proposed method has obtained the average classification accuracy of 84%, 83%, 81%, 78%, 77%, 76% and 72% for NFLE, healthy, SDB, narcolepsy, PLM, insomnia and RBD subjects, respectively in discriminating phases A and B using a balanced database. Our developed model yielded an average accuracy of 78% when all 77 subjects including healthy and sleep disordered patients are considered. Our proposed system can assist the sleep specialists in an automated and efficient analysis of sleep using sleep microstructure.

EEG-Based Seizure detection using linear graph convolution network with focal loss

Background and Objectives: Epilepsy is a clinical phenomenon caused by sudden abnormal and excessive discharge of brain neurons. It affects around 70 million people all over the world. Seizure detection from Electroencephalography (EEG) has achieved rapid development. However, existing methods often extract features from single channel EEG while ignoring the spatial relationship between different EEG channels. To fill this gap, a novel seizure detection model based on linear graph convolution network (LGCN) was proposed to enhance the feature embedding of raw EEG signals during seizure and non-seizure periods. Method: Pearson correlation matrix of raw EEG signals was calculated to build the input graph of the graph neural network where the coefficients of the matrix models the spatial relations in EEG signals. The last softmax layer makes the final decision (seizure vs. non-seizure). In addition, focal loss was utilized to redefine the loss function of LGCN to deal with the data imbalance problem during seizure detection. Results: Experiments are conducted on the CHB-MIT dataset. The seizure detection accuracy, specificity, sensitivity, F1 and Auc are 99.30%, 98.82%, 99.43%, 98.73% and 98.57% respectively. Conclusions: The proposed approach yields superior performance over the-state-of-the-art in seizure detection tasks on the CHB-MIT dataset. Our method works in an end-to-end manner and it does not need manually designed features. The ability to deal with imbalanced data is also attractive in real seizure detection scenarios where the duration of seizures is much shorter than the lasting time of non-seizure events.

Ramanujan Periodic Subspace Based Epileptic EEG Signals Classification

Epilepsy is a chronic brain disorder that is characterized by intermittent epileptic seizures that can be identified in an electroencephalogram (EEG) signal. This letter proposes a low computational complex method to classify epileptic EEG signals by using a suitable Ramanujan periodic subspace (RPS). Initially, this method divides the given single-channel EEG signal into multiple nonoverlapping EEG blocks, which are projected onto a particular RPS. The energy of the projection is used as a feature to classify each block into epileptic or nonepileptic, using an SVM binary classifier. Here, in order to choose that particular RPS, a few sample blocks from the epileptic (ictal), interictal, and healthy EEG signal are projected onto the divisor RPSs of that block. The difference between the average subspace energy of healthy versus ictal or interictal versus ictal blocks is used as a measure to choose the suitable RPS. Finally, the class of each block of the EEG signal is combined using the majority voting scheme to classify the epileptic EEG signal. A publicly available benchmark EEG database from Bonn University, Germany, is used to evaluate the performance of the proposed method. Furthermore, the EEG signals are added with white Gaussian noise and ocular artifact for testing the robustness of the method against noise. Evaluation results in terms of accuracy, sensitivity, specificity, and F-score demonstrate that the proposed method is comparable with the state-of-the-art techniques and also robust against artifacts and noise.

Research on automatic removal of ocular artifacts from single channel electroencephalogram signals based on wavelet transform and ensemble empirical mode decomposition

The brain-computer interface (BCI) systems used in practical applications require as few electroencephalogram (EEG) acquisition channels as possible. However, when it is reduced to one channel, it is difficult to remove the electrooculogram (EOG) artifacts. Therefore, this paper proposed an EOG artifact removal algorithm based on wavelet transform and ensemble empirical mode decomposition. Firstly, the single channel EEG signal is subjected to wavelet transform, and the wavelet components which involve EOG artifact are decomposed by ensemble empirical mode decomposition. Then the predefined autocorrelation coefficient threshold is used to automatically select and remove the intrinsic modal functions which mainly composed of EOG components. And finally the 'clean' EEG signal is reconstructed. The comparative experiments on the simulation data and the real data show that the algorithm proposed in this paper solves the problem of automatic removal of EOG artifacts in single-channel EEG signals. It can effectively remove the EOG artifacts when causes less EEG distortion and has less algorithm complexity at the same time. It helps to promote the BCI technology out of the laboratory and toward commercial application.

Probability mapping based artifact detection and removal from single-channel EEG signals for brain–computer interface applications

BackgroundDifferent types of artifacts in the electroencephalogram (EEG) signals can considerably reduce the performance of the later-stage EEG analysis algorithms for making decisions, such as those for brain–computer interfacing (BCI) classification. In this paper, we address the problem of artifact detection and removal from single-channel EEG signals. New methodWe propose a novel approach that maps the probability of an EEG epoch to be artifactual based on four different statistical measures: entropy (a measure of uncertainty), kurtosis (a measure of peakedness), skewness (a measure of asymmetry), and periodic waveform index (a measure of periodicity). Then, a stationary wavelet transform based artifact removal is proposed that employs a particular probability threshold provided by the user. ResultsWe have executed our experiments with both synthetic and real EEG data. It is observed that the proposed method exhibits a superior performance for suppressing the artifact contaminated from EEG with minimum distortion. Moreover, evaluation of the algorithm using EEG dataset for BCI experiments reveals that artifact removal can considerably improve the BCI output in both event-related potential and motor-imagery based BCI applications. Comparison with existing methodsThe proposed algorithm has been applied to both real and synthesized data testing and compared with other state-of-the-art automated artifact removal methods. Its superior performance is verified in terms of various performance metrics including computational complexity for justifying its use in BCI-like real-time applications. ConclusionOur work is expected to be useful for future research EEG signal processing and eventually to develop more accurate real-time EEG-based BCI applications.

Capturing the Development of Internal Representations in a High-Performing Deep Network for Sleep Stage Classification

Deep neural networks can provide accurate automated classification of human sleep signals into sleep stages that enables more effective diagnosis and treatment of sleep disorders. We develop a deep convolutional neural network (CNN) that attains state-of-the-art sleep stage classification performance on input data consisting of human sleep EEG and EOG signals. Nested cross-validation is used for optimal model selection and reliable estimation of out-of-sample classification performance. The resulting network attains a classification accuracy of 84.50 pm 0.13%; its performance exceeds human expert inter-scorer agreement, even on single-channel EEG input data, therefore providing more objective and consistent labeling than human experts demonstrate as a group. We focus on analyzing the learned internal data representations of our network, with the aim of understanding the development of class differentiation ability across the layers of processing units, as a function of layer depth. We approach this problem visually, using t-Stochastic Neighbor Embedding (t-SNE), and propose a pooling variant of Centered Kernel Alignment (CKA) that provides an objective quantitative measure of the development of sleep stage specialization and differentiation with layer depth. The results reveal a monotonic progression of both of these sleep stage modeling abilities as layer depth increases.

Epileptic Seizure Recognition Using Reduced Deep Convolutional Stack Autoencoder and Improved Kernel RVFLN From EEG Signals.

In this paper, reduced deep convolutional stack autoencoder (RDCSAE) and improved kernel random vector functional link network (IKRVFLN) are combined to recognize the epileptic seizure using both the multichannel scalp and single-channel electroencephalogram (EEG) signals. The novel RDCSAE structure is designed to extract the most discriminative unsupervised features from EEG signals and fed into the proposed supervised IKRVFLN classifier to train efficiently by reducing the mean-square error cost function for recognizing the epileptic seizure activity with promising accuracy. The proposed RDCSAE-IKRVFLN algorithm is tested over the benchmark Boston Children's Hospital multichannel scalp EEG (sEEG) and Boon University, Germany single-channel EEG databases. The less computational complexity, higher learning speed, better model generalization, accurate epileptic seizure recognition, remarkable classification accuracy, negligible false positive rate per hour (FPR/h) and short event recognition time are the main advantages of the proposed RDCSAE-IKRVFLN method over reduced deep convolutional neural network (RDCNN), RDCSAE and RDCSAE-KRVFLN methods. The proposed RDCSAE-IKRVFLN method is implemented in a high-speed reconfigurable field-programmable gate array (FPGA) hardware environment to design a computer-aided-diagnosis (CAD) system for automatic epileptic seizure diagnosis. The simplicity, feasibility, and practicability of the proposed method validate the stable and reliable performances of epilepsy detection and recognition.

Eye-blink artifact removal from single channel EEG with k-means and SSA

In recent years, the usage of portable electroencephalogram (EEG) devices are becoming popular for both clinical and non-clinical applications. In order to provide more comfort to the subject and measure the EEG signals for several hours, these devices usually consists of fewer EEG channels or even with a single EEG channel. However, electrooculogram (EOG) signal, also known as eye-blink artifact, produced by involuntary movement of eyelids, always contaminate the EEG signals. Very few techniques are available to remove these artifacts from single channel EEG and most of these techniques modify the uncontaminated regions of the EEG signal. In this paper, we developed a new framework that combines unsupervised machine learning algorithm (k-means) and singular spectrum analysis (SSA) technique to remove eye blink artifact without modifying actual EEG signal. The novelty of the work lies in the extraction of the eye-blink artifact based on the time-domain features of the EEG signal and the unsupervised machine learning algorithm. The extracted eye-blink artifact is further processed by the SSA method and finally subtracted from the contaminated single channel EEG signal to obtain the corrected EEG signal. Results with synthetic and real EEG signals demonstrate the superiority of the proposed method over the existing methods. Moreover, the frequency based measures [the power spectrum ratio (Gamma ) and the mean absolute error (MAE)] also show that the proposed method does not modify the uncontaminated regions of the EEG signal while removing the eye-blink artifact.

A novel convolutional neural network method for subject-independent driver drowsiness detection based on single-channel data and EEG alpha spindles.

A significant number of fatal accidents are caused by drowsy drivers worldwide. Driver drowsiness detection based on electroencephalography (EEG) signals has high accuracy and is known as a reference method for evaluating drowsiness. Among brain waves, EEG alpha spindle activity is a silent feature of decreasing alertness levels. In this paper, based on the detection of EEG alpha spindles, a novel driver drowsiness detection method is presented. The EEG spindles were detected using Continuous Wavelet Transform (CWT) analysis and the Morlet function. To do so, the signal is divided into 30-s epochs, and the observer rating of drowsiness determines the drowsiness level in each epoch. Tests were conducted on 17 healthy males in a driving simulator with a monotonous driving scenario. The Convolutional Neural Network (CNN) is used for classifying EEG signals and automatically learns features of the early drowsy state. The subject-independent classification results for single-channel P4 show 94% accuracy.

Hybrid algorithm for multi artifact removal from single channel EEG

Electroencephalogram (EEG) signals recorded from the ambulatory systems are mostly contaminated by various artifacts like, electrooculogram (EOG), motion artifacts (MA) and electrocardiogram (ECG) artifacts. These artifacts limit the accuracy in further analysis of EEG in practise. So far, several existing methods have been proposed with the combination of decomposition techniques and independent component analysis (ICA) to remove single artifacts and only few methods to remove multiple artifacts from the single channel EEG. As improperly denoised EEG signals can result in wrong diagnosis, in this work, Singular Spectrum Analysis (SSA) and ICA are jointly combined with Generalized Moreau Envelope Total Variation (GMETV) technique to simultaneously remove combinations of different artifacts from single channel EEG. In this work, the SSA is used to decompose the contaminated single channel EEG, while the ICA is employed to separate the various hidden sources as independent components (ICs). Although the ICA is adequate in source separation, there is still, some essential EEG signal data appearing as artifact in the IC. Hence, eliminating this would allow EEG signal information to be lost. The GMETV approach is proposed in this paper, to estimate the actual artifacts in order to address these issues. The estimated actual artifacts are subtracted from the artifact ICs providing the residue of wanted component of EEG. This residue is added back to the remaining ICs, to obtain the denoised EEG. Simulation results demonstrated that the proposed technique performs better compared to the existing techniques. The Relative Root Mean Square Error (RRMSE) is reduced by 12.02% and 7.22% compared to SSA-ICA and SSA-ICA-thresholding respectively. Similarly, the Correlation Coefficient (CC) is increased by 21.48% and 8.25% with respect to SSA-ICA and SSA-ICA-thresholding respectively.

Validation of an EEG seizure detection paradigm optimized for clinical use in a chronically implanted subcutaneous device

BackgroundMany electroencephalography (EEG) based seizure detection paradigms have been developed and validated over the last two decades. The majority of clinical approaches use scalp or intracranial EEG electrodes. Scalp EEG is limited by patient discomfort and short duration of useful EEG signals. Intracranial EEG involves an invasive surgical procedure associated with significant risk making it unsuitable for widespread use as a practical clinical biometric. A less invasive EEG monitoring approach, that is between invasive intracranial procedures and noninvasive methods, would fill the need of a safe, accurate, chronic (ultra-long term) and objective seizure detection method. We present validation of a continuous EEG seizure detection paradigm using human single-channel EEG recordings from subcutaneously placed electrodes that could be used to fulfill this need. MethodsTen-minute long sleep, awake and ictal EEG epochs obtained from 21 human subjects with subscalp electrodes and validated against simultaneous iEEG recordings were analyzed by three experienced clinical neurophysiologists. The 201subscalp EEG time series epochs where classified as diagnostic for awake, asleep, or seizure by the clinicians who were blinded to all other information. Seventy of the epochs were classified in this way as representing seizure activity. A subject specific seizure detection algorithm was trained and then evaluated offline for each patient in the data set using the expert consensus classification as the gold standard. ResultsThe average seizure detection performance of the algorithm across 21 subjects exceeded 90 % accuracy: 97 % sensitivity, 91 % specificity, and 93 % accuracy. For 19 of 21 patient datasets the algorithm achieved 100 % sensitivity. For 15 of 21 patients, the algorithm achieved 100 % specificity. For 13 of 21 patients the algorithm achieved 100 % accuracy. ComparisonNo comparable published methods are available for subgaleal EEG seizure detection. ConclusionsThese findings suggest that a simple seizure detection algorithm using subcutaneous EEG signals could provide sufficient accuracy and clinical utility for use in a low power, long-term subcutaneous brain monitoring device. Such a device would fill a need for a large number of people with epilepsy who currently have no means for accurately quantifying their seizures thereby providing important information to healthcare providers not currently available.

Sleep structure and electroencephalographic spectral power of middle-aged or older adults: Normative values by age and sex in the Korean population.

The fine structure of sleep electrocortical activity reflects health and disease. The current study provides normative data for sleep structure and electroencephalography (EEG) spectral power measures derived from overnight polysomnography (PSG) and examines the effect of age and sex among Korean middle-aged and older adults with or without obstructive sleep apnea (OSA). We analysed home PSG data from 1,153 adult participants of an ongoing population-based cohort study, the Korean Genome and Epidemiology Study. Sleep stages were visually scored and spectral power was measured on a single-channel EEG (C4-A1). We computed spectral power for five frequency ranges. The EEG power was reported in relative (%) and log-transformed absolute values (µV2 ). With ageing, the proportion of N1 sleep increased, whereas N3 decreased, which is more noticeable in men than in women. The amount of N3 was relatively low in this cohort. With ageing, relative delta power decreased and alpha and sigma power increased for the whole sleep period, which was more pronounced during REM sleep in non-OSA. For men compared with women, relative theta power was lower during REM and sigma and beta were higher during N1 sleep. The differences of relative powers by age and sex in OSA were comparable to those in non-OSA. In a community-based Korean population, we present normative data of sleep structure and spectral power for middle-aged or older adults of a non-Caucasian ethnicity. The values varied with age and sex and were not influenced by sleep apnea.

Revisiting ICEEMDAN and EEG rhythms

A new procedure is proposed to choose the optimum noise amplitude in the Improved Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (ICEEMDAN) method. This new procedure is referred to as the Performance Improved Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (PICEEMDAN). The intrinsic mode functions (IMF) obtained using ICEEMDAN depend on the chosen noise amplitude. Depending on the noise amplitude, the number of IMF’s can vary and also its properties. The paper explores the best choice of the noise amplitude that allows minimal mode mixing amongst its components. This involves the use of the marginal Hilbert spectrum, to compute the spectral energy over the mode’s bandwidth for each IMF and a parameter defined as the total index of overlap in the frequency domain (TIOF). The procedure is demonstrated initially using synthetic signals and later applied to electroencephalogram (EEG) signals. The results indicate a marked reduction in mode mixing amongst in components. The IMF’s obtained from single channel EEG data using this technique showed remarkable correspondence with the beta, alpha, theta and delta EEG rhythms. This was demonstrated using eyes open, eyes closed and seizure data and provides an alternate methodology to process EEG data.

Pilot study of a single-channel EEG seizure detection algorithm using machine learning.

Seizures are one of the most common emergencies in the neonatal intensive care unit (NICU). They are identified through visual inspection of electroencephalography (EEG) reports and treated by neurophysiologic experts. To support clinical seizure detection, several feature-based automatic neonatal seizure detection algorithms have been proposed. However, as they were unsuitable for clinical application due to their low accuracy, we developed a new seizure detection algorithm using machine learning for single-channel EEG to overcome these limitations. The dataset applied in our algorithm contains EEG recordings from human neonates. A 19-channel EEG system recorded the brain waves of 79 term neonates admitted to the NICU at the Helsinki University Hospital. From these datasets, we selected six patients with conformational seizure annotations for the pilot study and allocated four and two patients for our training and testing datasets, respectively. The presence of seizures in the EEGs was annotated independently by three experts through visual interpretation. We divided the data into epochs of 5 s each and further defined a seizure block to label the annotations from each expert recorded every second. Subsequently, to create a balanced dataset, any data point with a non-seizure label was moved to the training and test dataset. The developed principal component feature-extracted machine learning algorithm used 62.5% of the relative time (only 5 s for decision) of the baseline, reaching an area under the ROC curve score of 0.91. The effect of diversified parameters was meticulously examined, and 100 principal components were extracted to optimize the model performance. Our machine learning-based seizure detection algorithm exhibited the potential for clinical application in NICUs, general wards, and at home and proved its convenience by requiring only a single channel for implementation.

Joint Singular Spectrum Analysis and Generalized Moreau Envelope Total Variation for motion artifact removal from single channel EEG signals

Recorded electroencephalogram (EEG) signals are commonly interfered by several artifacts. It is very crucial to remove the artifacts for automatic detection of brain disorders. In this work, the joint use of Singular Spectrum Analysis (SSA) and Generalized Moreau Envelope Total Variation (GMETV), namely SSA-GMETV technique is proposed to remove the motion artifacts from the single channel EEG signals. In this work, initially, the interfered EEG is decomposed into several bands by SSA. The last subband of the interfered signal is applied to GMETV filter to extract the artifact. This is then subtracted from the last subband signal and added to the remaining SSA decomposed bands, to obtain the denoised EEG signal. Relative Root Mean Square Error (RRMSE) and difference in Signal to Noise Ratio (ΔSNR) are considered to quantify the denoised EEG signals. Experimental results demonstrate that the proposed SSA-GMETV technique performs much better than the existing methods in terms of the performance metrics, average RRMSE and average ΔSNR by 47.41% and 31.81 dB respectively.