Sleep Stage Classification Based on EEG, EOG, and CNN-GRU Deep Learning Model

The most important part of sleep quality assessment is the classification of sleep stages, which helps to diagnose sleep-related disease. In the traditional sleep staging method, subjects have to spend a night in the sleep clinic for recording polysomnogram. Sleep expert classifies the sleep stages by monitoring the signals, which is time consuming and frustrating task and can be affected by human error. New studies propose fully automated techniques for classifying sleep stages that makes sleep scoring possible at home. Despite comprehensive studies have been presented in this field the classification results have not yet reached the gold standard due to the concentration on the use of a limited source of information such as single channel EEG. Therefore, this article introduces a new method for fusing two sources of information, including electroencephalogram (EEG) and electrooculogram (EOG), to achieve promising results in the classification of sleep stages. In the proposed method, extracted features from the EEG and EOG signals, are divided into two feature sets consisting of the EEG features and fused features of EEG and EOG. Then, each feature set transformed into a horizontal visibility graph (HVG). The images of the HVG are produced in a novel framework and classified by proposed transfer learning convolutional neural network for data fusion (TLCNN-DF). Employing transfer learning at the training stage of the model has accelerated the training process of the CNN and improved the performance of the model. The proposed algorithm is used to classify the Sleep-EDF and Sleep-EDFx benchmark datasets. The algorithm can classify the Sleep-EDF dataset with an accuracy of 93.58% and Cohen’s kappa coefficient of 0.899. The results show proposed method can achieve superior performance compared to state-of-the-art studies on classification of sleep stages. Furthermore, it can attain reliable results as an alternative to conventional sleep staging.

Poor quality and poor duration of sleep have been associated with cognitive decline, diseases, and disorders. Therefore, sleep studies are imperative to recapitulate phenotypes associated with poor sleep quality and uncover mechanisms contributing to psychopathology. Classification of sleep stages, vigilance state bout durations, and number of transitions amongst vigilance states serves as a proxy for evaluating sleep quality in preclinical studies. Currently, the gold standard for sleep staging is expert human inspection of polysomnography (PSG) obtained from preclinical rodent models and this approach is immensely time consuming. To accelerate the analysis, we developed a deep-learning-based software tool for automated sleep stage classification in rats. This study aimed to develop an automated method for classifying three sleep stages in rats (REM/paradoxical sleep, NREM/slow-wave sleep, and wakefulness) using a deep learning approach based on single-channel EEG data. Single-channel EEG data were acquired from 16 rats, each undergoing two 24 h recording sessions. The data were labeled by human experts in 10 s epochs corresponding to three stages: REM/paradoxical sleep, NREM/slow-wave sleep, and wakefulness. A deep neural network (DNN) model was designed and trained to classify these stages using the raw temporal data from the EEG. The DNN achieved strong performance in predicting the three sleep stages, with an average F1 score of 87.6% over a cross-validated test set. The algorithm was able to predict key parameters of sleep architecture, including total bout duration, average bout duration, and number of bouts, with significant accuracy. Our deep learning model effectively automates the classification of sleep stages using single-channel EEG data in rats, reducing the need for labor-intensive manual annotation. This tool enables high-throughput sleep studies and may accelerate research into sleep-related pathologies. Furthermore, we provide over 700 h of expert-scored sleep data, available for public use in future research studies.

Sleep stage classification is essential for clinical disease diagnosis and sleep quality assessment. Most of the existing methods for sleep stage classification are based on single-channel or single-modal signal, and extract features using a single-branch, deep convolutional network, which not only hinders the capture of the diversity features related to sleep and increase the computational cost, but also has a certain impact on the accuracy of sleep stage classification. To solve this problem, this paper proposes an end-to-end multi-modal physiological time-frequency feature extraction network (MTFF-Net) for accurate sleep stage classification. First, multi-modal physiological signal containing electroencephalogram (EEG), electrocardiogram (ECG), electrooculogram (EOG) and electromyogram (EMG) are converted into two-dimensional time-frequency images containing time-frequency features by using short time Fourier transform (STFT). Then, the time-frequency feature extraction network combining multi-scale EEG compact convolution network (Ms-EEGNet) and bidirectional gated recurrent units (Bi-GRU) network is used to obtain multi-scale spectral features related to sleep feature waveforms and time series features related to sleep stage transition. According to the American Academy of Sleep Medicine (AASM) EEG sleep stage classification criterion, the model achieved 84.3% accuracy in the five-classification task on the third subgroup of the Institute of Systems and Robotics of the University of Coimbra Sleep Dataset (ISRUC-S3), with 83.1% macro F1 score value and 79.8% Cohen's Kappa coefficient. The experimental results show that the proposed model achieves higher classification accuracy and promotes the application of deep learning algorithms in assisting clinical decision-making.

In this study, using the polysomnography data set, the classification of sleep stages was realized automatically with supervised learning method. In this study, sleep stage classification was carried out in three stages. In the first stage, the biomedical signal was divided into its independent components by Independent Component Analysis method. In the second stage, feature extraction was performed by using Mel Kepstrum Coefficient method. In the third stage, artificial neural networks were trained by using the extracted features and the sleep phases were estimated by using software architecture called Long ShortTerm Memory. As a result of the classification process performed in this way, the accuracy rate of the ten fold cross validation obtained for the binary classification (asleep / wake) was found to be 97,87%. For the five fold classification problem, the accuracy rate of the subject dependent algorithm and polysomnography data obtained using a single EEG channel was found to be 93,36% as a result of the ten-fold cross validation process.

Sleep plays a vital part in humans health. The correct classification of sleep stages provides clinical information that can be used to diagnose sleep disorders in patients. The gold standard for sleep evaluation is polysomnography. On the other hand, polysomnography has several drawbacks, such as expense, inconvenience, and a long wait time at sleep laboratories. Advancements in technology enable the adoption and use of single-channel EEG at in-home sleep monitoring; it provides better sleep-stage accuracy levels. Industry and academic researchers continuously improve automatic sleep-stage classification using machine and deep learning in multi-channel EEG and single-channel EEG. However, the performance evaluation of single-channel against multi-channel EEG was not established. The performance of sleep staging using XGBoost, LGBM and voting classifiers models was verified in multi-channel and single-channel EEG. A similar number of features and hyperparameters were employed for classification in multi-channel and single-channel EEG for sleep-stage classification. Each model used five healthy subjects data from the Sleep-EDFx database. K-fold validation was used for training and testing. XGBoost, LGBM, and voting classifiers classified the sleep stages and achieved an accuracy of 87.96%, 86.24%, 88.25% in multi-channel and 86.81%, 86.81%, 87.1% in single-channel, respectively. The sleep-stage classification results demonstrate that single-channel EEG gives almost similar accuracy to multi-channel EEG.

Automatic sleep stage monitoring is an essential tool for the diagnosis and treatment of sleep-related disorders effectively. Although extensive studies focusing on single source data or information, such as single channel electroencephalogram (EEG), electrooculogram (EOG), and electromyogram (EMG), with machine learning (ML) and deep learning (DL) have been presented in this field of automatic sleep staging, there are very limited literature's that have investigated the impact of multisource information fusion on ML or DL based schemes for the sleep stage classification. In addition, exploiting recurrent neural network (RNN) to learn temporal information from the sequences of multi-source inputs can bring significant improvement in classifying the sleep stage. Therefore, we aim to develop an multi-source data fusion, more specifically two source's signal fusion, enabled end-to-end convolutional recurrent neural network (CRNN) in order to perform automatic sleep stage classification in this article. In the proposed scheme, no hand crafted features is used to train the model and hence the proposed model can be called an end-to-end approach. The proposed CRNN model can extract time-invariant features from raw data of two input sources and can fuse them to learn temporally correlated features. Sleep-EDF expanded benchmark dataset is used in this study and we have employed Fpz-Cz EEG channel signal and EOG signal as input with 3 variants of RNN layer in model architecture separately in order to evaluate and investigate the performance. The proposed CRNN model with standard RNN layers can classify the Sleep-EDF dataset with maximum average accuracy of 90.30% and maximum Cohen's kappa coefficient of 86.86 compared other two RNN variant's layer (e.g., gated recurrent unit and long short term memory) based models which is very promising in the studies on classification of sleep stages as well as can be regarded as an alternative to conventional sleep staging.

Sleep stages classification by fusing the time-related synchronization analysis and brain activations

Sleep is a natural periodic state of rest for the body, in which the eyes usually close and consciousness is completely or partially lost. Consequently, there is a decrease in bodily movements and responsiveness to external stimuli. Slow wave sleep is of immense interest as it is the most restorative sleep stage during which the body recovers from weariness. During this sleep stage, electroencephalographic (EEG) and electro-oculographic (EOG) signals interfere with each other and they share a temporal similarity. In this investigation we used the EEG and EOG signals acquired from 10 patients undergoing overnight polysomnography with their sleep stages determined by certified sleep specialists based on RK rules. In this pilot study, we performed spectral estimation of EEG signals by Autoregressive (AR) modeling, and then used Itakura Distance to measure the degree of similarity between EEG and EOG signals. We finally calculated the statistics of the results and displayed them in an easy to visualize fashion to observe tendencies for each sleep stage. We found that Itakura Distance is the smallest for sleep stages 3 and 4. We intend to deploy this feature as an important element in automatic classification of sleep stages.

In recent years, more biomedical studies have begun to use multimodal data to improve model performance. Many studies have used ablation for explainability, which requires the modification of input data. This can create out-of-distribution samples and lead to incorrect explanations. To avoid this problem, we propose using a gradient-based feature attribution approach, called layer-wise relevance propagation (LRP), to explain the importance of modalities both locally and globally for the first time. We demonstrate the feasibility of the approach with sleep stage classification as our use-case and train a 1-D convolutional neural network with electroencephalogram (EEG), electrooculogram (EOG), and electromyogram (EMG) data. We also analyze the relationship of our local explainability results with clinical and demographic variables to determine whether they affect our classifier. Across all samples, EEG is the most important modality, followed by EOG and EMG. For individual sleep stages, EEG and EOG have higher relevance for awake and non-rapid eye movement 1 (NREM1). EOG is most important for REM, and EEG is most relevant for NREM2-NREM3. Also, LRP gives consistent levels of importance to each modality for the correctly classified samples across folds but inconsistent levels of importance for incorrectly classified samples. Our statistical analyses suggest that medication has a significant effect upon patterns learned for EEG and EOG NREM2 and that subject sex and age significantly affects the EEG and EOG patterns learned, respectively. Our results demonstrate the viability of gradient-based approaches for explaining multimodal electrophysiology classifiers and suggest their generalizability for other multimodal classification domains.

Electroencephalogram (EEG) signal is often used to assess sleep quality and treat sleep disorders. Many existing methods usually obtain high accuracy through a large number of feature preprocessing and feature extraction of EEG signals, which need a lot of prior knowledge as the basis. In this paper, a novel sleep stage classification framework, named FRL&S2SL, is proposed. The framework combines fast representation learning (FRL) and semantic-to-signal learning (S2SL) and uses single-channel EEG without any preprocessing of EEG signals. In the proposed framework, we utilize convolutional neural networks (CNN) to extract time-invariant features and bidirectional long short-term memory (BiLSTM) models to extract temporal features. Furthermore, auxiliary classifier generative adversarial network (ACGAN) is used to embed semantic features into signal features and to extract knowledge domain features of EEG signals for the first time. According to the American Academy of Sleep Medicine (AASM), sleep is divided into five stages: awake, rapid eye movement (REM) and three non-rapid eye movement (N1/N2/N3). We evaluated our framework using single-channel EEG (Fpz-Oz) from Sleep-EDF dataset, which is subject to the standards specified by AASM. The results show that our framework has achieved state-of-the-art in many evaluation metrics.

Accurate detection of sleep stages and sleep events is essential for the effective diagnosis and treatment of sleep disorders. However, current state-of-the-art methods often fall short in integrating these multiple tasks simultaneously. This study introduces a novel multi-task deep-learning approach for the joint detection of sleep events and hypnogram construction in a single pass. The proposed method adapts state-of-the-art single-shot object detection techniques to multi-channel time-series data, enabling simultaneous classification and detection of sleeping events. Experiments were conducted using diverse input channel combinations, including electroencephalogram (EEG), electrooculogram (EOG), electromyogram (EMG), oxygen saturation, airflow, and thoracic-abdominal movements, as well as multiple output configurations regarding sleep stages, EEG arousals, and respiratory events, both individually and combined. Results demonstrate improved accuracy in detecting respiratory events and sleep stages with additional channels. The multi-task approach enhanced overall performance, benefiting from shared knowledge across tasks.

In the pursuit of highly effective and efficient portable sleep classification systems, researchers have been testing a massive number of combinations of EEG features and classifiers. State of art sleep classification ensembles achieve accuracy in the order of 90%. However, there is presently no consensus regarding the best setof features for sleep staging with single channel EEG, leading researchers to modify feature selection according to the number of classification stages. This paper introduces a reduced set of frequency-domain features capable of yielding high classification accuracy (90.9%, 91.8%, 92.4%, 94.3% and 97.1%) for all 6- to 2-state sleep stages. The proposed system uses fast Fourier transform (FFT) to convert data from Pz-Oz EEG channel into the frequency domain. Afterwards, eight statistical features are extracted from specific frequency ranges and fed into a random forest classifier.

An explainable deep-learning model to stage sleep states in children and propose novel EEG-related patterns in sleep apnea

What is the central question of this study? A transcranial magnetic stimulation (TMS)-induced twitch applied on isolated single breaths over the motor cortex somatotopic representation of the tongue briefly recruits submental muscles and improves airflow dynamics of flow-limited respiratory cycles without arousing sleep apnoea patients. However, the mechanical impact of the TMS-induced twitch applied during consecutive breathing cycles on airflow dynamics remains unknown. What is the main finding and what is its importance? Our results show that application of TMS with the stimulator output set at the sleep submental motor threshold intensity on consecutive respiratory cycles increases inspiratory flow and reduces the turbulent airflow component. These results indicate an improvement of airflow pattern after two single consecutive TMS-induced twitches without arousing sleep apnoea patients. Transcranial magnetic stimulation (TMS)-induced twitches applied on isolated breaths briefly recruit upper airway dilator muscles and improve airflow and inspiratory volume without arousing apnoeic patients from sleep, but the effects of applying such twitches consecutively on airflow dynamics is unknown. The objective of this study was to quantify the effects of five consecutive TMS-induced twitches applied on sleep-induced obstructive hypopnoeic breaths in 10 obstructive sleep apnoea patients. Submental muscle motor threshold (SUBMT) and motor-evoked potential were measured during wakefulness and sleep. The TMS-induced twitches were applied during stable non-rapid eye movement (NREM) sleep, at the beginning of inspiration of consecutive flow-limited respiratory cycles, with the stimulator output set at sleep SUBMT. Maximal inspiratory flow, inspiratory volume, inspiratory time, shifts of electroencephalogram frequency and pulse rate variability were assessed. During sleep, SUBMT increased (wakefulness, 25.3 ± 4.9%; NREM sleep, 27.0 ± 6.2%; P = 0.02). During each series of stimulations there was a rise in maximal inspiratory flow (from 306.7 ± 123.2 to 359.8 ± 154.1 ml s(-1); P = 0.0002) and in inspiratory volume (from 346.1 ± 128.1 to 414.9 ± 171.2 ml; P = 0.02) without differences in thoraco-abdominal efforts and inspiratory time. These responses were observed in the absence of arousals and ceased immediately after TMS interruption. Transcranial magnetic stimulation-induced cortical and/or autonomic arousal was observed in 30.2% of all series of stimulation. Consecutive twitch TMS of submental muscles may lead to arousals in a minority of patients but can be applied on consecutive respiratory cycles during sleep and can significantly improve maximal inspiratory flow and inspiratory volume of flow-limited cycles.

Sleep stage classification plays a pivotal role in effective diagnosis and treatment of sleep related disorders. Traditionally, sleep scoring is done manually by trained sleep scorers. The analysis of electroencephalogram (EEG) signals recorded during sleep by clinicians is tedious, time-consuming and prone to human errors. Therefore, it is clinically important to score sleep stages using machine learning techniques to get accurate diagnosis. Several studies have been proposed for automated detection of sleep stages. However, these studies have employed only healthy normal subjects (good sleepers). The proposed study focuses on the automated sleep-stage scoring of subjects suffering from seven different kind of sleep disorders such as insomnia, bruxism, narcolepsy, nocturnal frontal lobe epilepsy (NFLE), periodic leg movement (PLM), rapid eye movement (REM) behavioural disorder and sleep-disordered breathing as well as normal subjects. The open source physionet’s cyclic alternating pattern (CAP) sleep database is used for this study. The EEG epochs are decomposed into sub-bands using a new class of optimized wavelet filters. Two EEG channels, namely F4-C4 and C4-A1, combined are used for this work as they can provide more insights into the changes in EEG signals during sleep. The norm features are computed from six sub-bands coefficients of optimal wavelet filter bank and fed to various supervised machine learning classifiers. We have obtained the highest classification performance using an ensemble of bagged tree (EBT) classifier with 10-fold cross validation. The CAP database comprising of 80 subjects is divided into ten different subsets and then ten different sleep-stage scoring tasks are performed. Since, the CAP database is unbalanced with different duration of sleep stages, the balanced dataset also has been created using over-sampling and under-sampling techniques. The highest average accuracy of 85.3% and Cohen’s Kappa coefficient of 0.786 and accuracy of 92.8% and Cohen’s Kappa coefficient of 0.915 are obtained for unbalanced and balanced databases, respectively. The proposed method can reliably classify the sleep stages using single or dual channel EEG epochs of 30 s duration instead of using multimodal polysomnography (PSG) which are generally used for sleep-stage scoring. Our developed automated system is ready to be tested with more sleep EEG data and can be employed in various sleep laboratories to evaluate the quality of sleep in various sleep disorder patients and normal subjects.