Data augmentation for learning predictive models on EEG: a systematic comparison

Geometric neural network based on phase space for BCI-EEG decoding

Objective.The integration of Deep Learning (DL) algorithms on brain signal analysis is still in its nascent stages compared to their success in fields like Computer Vision. This is particularly true for Brain-computer interface (BCI), where the brain activity is decoded to control external devices without requiring muscle control. Electroencephalography is a widely adopted choice for designing BCI systems due to its non-invasive and cost-effective nature and excellent temporal resolution. Still, it comes at the expense of limited training data, poor signal-to-noise, and a large variability across and within-subject recordings. Finally, setting up a BCI system with many electrodes takes a long time, hindering the widespread adoption of reliable DL architectures in BCIs outside research laboratories. To improve adoption, we need to improve user comfort using, for instance, reliable algorithms that operate with few electrodes.Approach.Our research aims to develop a DL algorithm that delivers effective results with a limited number of electrodes. Taking advantage of the Augmented Covariance Method and the framework of SPDNet, we propose the Phase-SPDNet architecture and analyze its performance and the interpretability of the results. The evaluation is conducted on 5-fold cross-validation, using only three electrodes positioned above the Motor Cortex. The methodology was tested on nearly 100 subjects from several open-source datasets using the Mother Of All BCI Benchmark framework.Main results.The results of our Phase-SPDNet demonstrate that the augmented approach combined with the SPDNet significantly outperforms all the current state-of-the-art DL architecture in MI decoding.Significance.This new architecture is explainable and with a low number of trainable parameters.

A protocol for trustworthy EEG decoding with neural networks

Deep learning solutions have rapidly emerged for EEG decoding, achieving state-of-the-art performance on a variety of decoding tasks. Despite their high performance, existing solutions do not fully address the challenge posed by the introduction of many hyperparameters, defining data pre-processing, network architecture, network training, and data augmentation. Automatic hyperparameter search is rarely performed and limited to network-related hyperparameters. Moreover, pipelines are highly sensitive to performance fluctuations due to random initialization, hindering their reliability. Here, we design a comprehensive protocol for EEG decoding that explores the hyperparameters characterizing the entire pipeline and that includes multi-seed initialization for providing robust performance estimates. Our protocol is validated on 9 datasets about motor imagery, P300, SSVEP, including 204 participants and 26 recording sessions, and on different deep learning models. We accompany our protocol with extensive experiments on the main aspects influencing it, such as the number of participants used for hyperparameter search, the split into sequential simpler searches (multi-step search), the use of informed vs. non-informed search algorithms, and the number of random seeds for obtaining stable performance. The best protocol included 2-step hyperparameter search via an informed search algorithm, with the final training and evaluation performed using 10 random initializations. The optimal trade-off between performance and computational time was achieved by using a subset of 3-5 participants for hyperparameter search. Our protocol consistently outperformed baseline state-of-the-art pipelines, widely across datasets and models, and could represent a standard approach for neuroscientists for decoding EEG in a trustworthy and reliable way.

Non-Invasive Brain-Computer Interfaces: State of the Art and Trends

Brain-computer interface (BCI) is a rapidly evolving technology that has the potential to widely influence research, clinical and recreational use. Non-invasive BCI approaches are particularly common as they can impact a large number of participants safely and at a relatively low cost. Where traditional non-invasive BCIs were used for simple computer cursor tasks, it is now increasingly common for these systems to control robotic devices for complex tasks that may be useful in daily life. In this review, we provide an overview of the general BCI framework as well as the various methods that can be used to record neural activity, extract signals of interest, and decode brain states. In this context, we summarize the current state-of-the-art of non-invasive BCI research, focusing on trends in both the application of BCIs for controlling external devices and algorithm development to optimize their use. We also discuss various open-source BCI toolboxes and software, and describe their impact on the field at large.

MSVTNet: Multi-Scale Vision Transformer Neural Network for EEG-Based Motor Imagery Decoding

OBJECT

Transformer-based neural networks have been applied to the electroencephalography (EEG) decoding for motor imagery (MI). However, most networks focus on applying the self-attention mechanism to extract global temporal information, while the cross-frequency coupling features between different frequencies have been neglected. Additionally, effectively integrating different neural networks poses challenges for the advanced design of decoding algorithms.

METHODS

This study proposes a novel end-to-end Multi-Scale Vision Transformer Neural Network (MSVTNet) for MI-EEG classification. MSVTNet first extracts local spatio-temporal features at different filtered scales through convolutional neural networks (CNNs). Then, these features are concatenated along the feature dimension to form local multi-scale spatio-temporal feature tokens. Finally, Transformers are utilized to capture cross-scale interaction information and global temporal correlations, providing more distinguishable feature embeddings for classification. Moreover, auxiliary branch loss is leveraged for intermediate supervision to ensure the effective integration of CNNs and Transformers.

RESULTS

The performance of MSVTNet was assessed through subject-dependent (session-dependent and session-independent) and subject-independent experiments on three MI datasets, i.e., the BCI competition IV 2a, 2b and OpenBMI datasets. The experimental results demonstrate that MSVTNet achieves state-of-the-art performance in all analyses.

CONCLUSION

MSVTNet shows superiority and robustness in enhancing MI decoding performance.

SpeechBrain-MOABB: An open-source Python library for benchmarking deep neural networks applied to EEG signals

Deep learning has revolutionized EEG decoding, showcasing its ability to outperform traditional machine learning models. However, unlike other fields, EEG decoding lacks comprehensive open-source libraries dedicated to neural networks. Existing tools (MOABB and braindecode) prevent the creation of robust and complete decoding pipelines, as they lack support for hyperparameter search across the entire pipeline, and are sensitive to fluctuations in results due to network random initialization. Furthermore, the absence of a standardized experimental protocol exacerbates the reproducibility crisis in the field. To address these limitations, we introduce SpeechBrain-MOABB, a novel open-source toolkit carefully designed to facilitate the development of a comprehensive EEG decoding pipeline based on deep learning. SpeechBrain-MOABB incorporates a complete experimental protocol that standardizes critical phases, such as hyperparameter search and model evaluation. It natively supports multi-step hyperparameter search for finding the optimal hyperparameters in a high-dimensional space defined by the entire pipeline, and multi-seed training and evaluation for obtaining performance estimates robust to the variability caused by random initialization. SpeechBrain-MOABB outperforms other libraries, including MOABB and braindecode, with accuracy improvements of 14.9% and 25.2% (on average), respectively. By enabling easy-to-use and easy-to-share decoding pipelines, our toolkit can be exploited by neuroscientists for decoding EEG with neural networks in a replicable and trustworthy way.

Review of deep representation learning techniques for brain-computer interfaces

In the field of brain-computer interfaces (BCIs), the potential for leveraging deep learning techniques for representing electroencephalogram (EEG) signals has gained substantial interest.Objective: This review synthesizes empirical findings from a collection of articles using deep representation learning techniques for BCI decoding, to provide a comprehensive analysis of the current state-of-the-art.Approach: Each article was scrutinized based on three criteria: (1) the deep representation learning technique employed, (2) the underlying motivation for its utilization, and (3) the approaches adopted for characterizing the learned representations.Main results: Among the 81 articles finally reviewed in depth, our analysis reveals a predominance of 31 articles using autoencoders. We identified 13 studies employing self-supervised learning (SSL) techniques, among which ten were published in 2022 or later, attesting to the relative youth of the field. However, at the time being, none of these have led to standard foundation models that are picked up by the BCI community. Likewise, only a few studies have introspected their learned representations. We observed that the motivation in most studies for using representation learning techniques is for solving transfer learning tasks, but we also found more specific motivations such as to learn robustness or invariances, as an algorithmic bridge, or finally to uncover the structure of the data.Significance: Given the potential of foundation models to effectively tackle these challenges, we advocate for a continued dedication to the advancement of foundation models specifically designed for EEG signal decoding by using SSL techniques. We also underline the imperative of establishing specialized benchmarks and datasets to facilitate the development and continuous improvement of such foundation models.

Resting-State Electroencephalogram Depression Diagnosis Based on Traditional Machine Learning and Deep Learning: A Comparative Analysis

The global prevalence of Major Depressive Disorder (MDD) is increasing at an alarming rate, underscoring the urgent need for timely and accurate diagnoses to facilitate effective interventions and treatments. Electroencephalography remains a widely used neuroimaging technique in psychiatry, due to its non-invasive nature and cost-effectiveness. With the rise of computational psychiatry, the integration of EEG with artificial intelligence has yielded remarkable results in diagnosing depression. This review offers a comparative analysis of two predominant methodologies in research: traditional machine learning and deep learning methods. Furthermore, this review addresses key challenges in current research and suggests potential solutions. These insights aim to enhance diagnostic accuracy for depression and also foster further development in the area of computational psychiatry.

Independent Vector Analysis for Feature Extraction in Motor Imagery Classification

Independent vector analysis (IVA) can be viewed as an extension of independent component analysis (ICA) to multiple datasets. It exploits the statistical dependency between different datasets through mutual information. In the context of motor imagery classification based on electroencephalogram (EEG) signals for the brain-computer interface (BCI), several methods have been proposed to extract features efficiently, mainly based on common spatial patterns, filter banks, and deep learning. However, most methods use only one dataset at a time, which may not be sufficient for dealing with a multi-source retrieving problem in certain scenarios. From this perspective, this paper proposes an original approach for feature extraction through multiple datasets based on IVA to improve the classification of EEG-based motor imagery movements. The IVA components were used as features to classify imagined movements using consolidated classifiers (support vector machines and K-nearest neighbors) and deep classifiers (EEGNet and EEGInception). The results show an interesting performance concerning the clustering of MI-based BCI patients, and the proposed method reached an average accuracy of 86.7%.

Detection of Anxiety-Based Epileptic Seizures in EEG Signals Using Fuzzy Features and Parrot Optimization-Tuned LSTM

In humans, epilepsy is diagnosed through electroencephalography (EEG) signals. Epileptic seizures (ESs) arise due to anxiety. The detection of anxiety-based seizures is challenging for radiologists, and there is a limited availability of anxiety-based EEG signals. Data augmentation methods are required to increase the number of novel samples. An epileptic seizure arises due to anxiety, which manifests as variations in EEG signal patterns consisting of changes in the size and shape of the signal. In this study, anxiety EEG signals were synthesized by applying data augmentation methods such as random data augmentation (RDA) to existing epileptic seizure signals from the Bonn EEG dataset. The data-augmented anxiety seizure signals were processed using three algorithms-(i) fuzzy C-means-particle swarm optimization-long short-term memory (FCM-PS-LSTM), (ii) particle swarm optimization-long short-term memory (PS-LSTM), and (iii) parrot optimization LSTM (PO-LSTM)-for the detection of anxiety ESs via EEG signals. The predicted accuracies of detecting ESs through EEG signals using the proposed algorithms-namely, (i) FCM-PS-LSTM, (ii) PS-LSTM, and (iii) PO-LSTM-were about 98%, 98.5%, and 96%, respectively.

Reaching the ceiling? Empirical scaling behaviour for deep EEG pathology classification

Machine learning techniques, particularly deep convolutional neural networks (ConvNets), are increasingly being used to automate clinical EEG analysis, with the potential to reduce the clinical burden and improve patient care. However, further research is required before they can be used in clinical settings, particularly regarding the impact of the number of training samples and model parameters on their testing error. To address this, we present a comprehensive study of the empirical scaling behaviour of ConvNets for EEG pathology classification. We analysed the testing error with increasing the training samples and model size for four different ConvNet architectures. The focus of our experiments is width scaling, and we have increased the number of parameters to up to 1.8 million. Our evaluation was based on two publicly available datasets: the Temple University Hospital (TUH) Abnormal EEG Corpus and the TUH Abnormal Expansion Balanced EEG Corpus, which together contain 10,707 training samples. The results show that the testing error follows a saturating power-law with both model and dataset size. This pattern is consistent across different datasets and ConvNet architectures. Furthermore, empirically observed accuracies saturate at 85%-87%, which may be due to an imperfect inter-rater agreement on the clinical labels. The empirical scaling behaviour of the test performance with dataset and model size has significant implications for deep EEG pathology classification research and practice. Our findings highlight the potential of deep ConvNets for high-performance EEG pathology classification, and the identified scaling relationships provide valuable recommendations for the advancement of automated EEG diagnostics.

Machine learning of brain-specific biomarkers from EEG

BACKGROUND

Electroencephalography (EEG) has a long history as a clinical tool to study brain function, and its potential to derive biomarkers for various applications is far from exhausted. Machine learning (ML) can guide future innovation by harnessing the wealth of complex EEG signals to isolate relevant brain activity. Yet, ML studies in EEG tend to ignore physiological artefacts, which may cause problems for deriving biomarkers specific to the central nervous system (CNS).

METHODS

We present a framework for conceptualising machine learning from CNS versus peripheral signals measured with EEG. A signal representation based on Morlet wavelets allowed us to define traditional brain activity features (e.g. log power) and alternative inputs used by state-of-the-art ML approaches based on covariance matrices. Using more than 2600 EEG recordings from large public databases (TUAB, TDBRAIN), we studied the impact of peripheral signals and artefact removal techniques on ML models in age and sex prediction analyses.

FINDINGS

Across benchmarks, basic artefact rejection improved model performance, whereas further removal of peripheral signals using ICA decreased performance. Our analyses revealed that peripheral signals enable age and sex prediction. However, they explained only a fraction of the performance provided by brain signals.

INTERPRETATION

We show that brain signals and body signals, both present in the EEG, allow for prediction of personal characteristics. While these results may depend on specific applications, our work suggests that great care is needed to separate these signals when the goal is to develop CNS-specific biomarkers using ML.

FUNDING

All authors have been working for F. Hoffmann-La Roche Ltd.

Temporally Local Weighting-Based Phase-Locked Time-Shift Data Augmentation Method for Fast-Calibration SSVEP-BCI

Various training-based spatial filtering methods have been proposed to decode steady-state visual evoked potentials (SSVEPs) efficiently. However, these methods require extensive calibration data to obtain valid spatial filters and temporal templates. The time-consuming data collection and calibration process would reduce the practicality of SSVEP-based brain-computer interfaces (BCIs). Therefore, we propose a temporally local weighting-based phase-locked time-shift (TLW-PLTS) data augmentation method to augment training data for calculating valid spatial filters and temporal templates. In this method, the sliding window strategy using the SSVEP response period as a time-shift step is to generate the augmented data, and the time filter which maximises the temporally local covariance between the original template signal and the sine-cosine reference signal is used to suppress the temporal noise in the augmented data. For the performance evaluation, the TLW-PLTS method was incorporated with state-of-the-art training-based spatial filtering methods to calculate classification accuracies and information transfer rates (ITRs) using three SSVEP datasets. Compared with state-of-the-art training-based spatial filtering methods and other data augmentation methods, the proposed TLW-PLTS method demonstrates superior decoding performance with fewer calibration data, which is promising for the development of fast-calibration BCIs.

Channel- and Label-Flip Data Augmentation for Motor Imagery-Based Brain-Computer Interfaces

Achieving high classification accuracy in motor-imagery-based brain-computer interfaces (BCIs) requires substantial amounts of training data. A challenge arises because of the impracticality of measuring large amounts of data from users. Data augmentation (DA) has emerged as a promising solution for this challenge. We propose a novel DA method called channel&label-flip DA that involves not only flipping channels but also flipping class labels. This method is based on the neuroscience finding that motor imageries of left- and right-hand movements are roughly symmetrical. The efficiency of the proposed method was evaluated using the OpenBMI dataset, which comprises electroencephalograms collected from 54 participants engaged in left- and right-hand motor imagery tasks. To compare the impact on classifiers, we employed three classical machine learning models utilizing filter bank common spatial pattern features, along with a deep learning-based model that uses raw signal input. As a result, the channel&label-flip DA improved the classification accuracy on average, whereas simple flipping of the channels reduced the classification accuracy compared to the case without DA.

Evaluating Augmentation Approaches for Deep Learning-based Major Depressive Disorder Diagnosis with Raw Electroencephalogram Data

While deep learning methods are increasingly applied in research contexts for neuropsychiatric disorder diagnosis, small dataset size limits their potential for clinical translation. Data augmentation (DA) could address this limitation, but the utility of EEG DA methods remains relatively underexplored in neuropsychiatric disorder diagnosis. In this study, we train a model for major depressive disorder diagnosis. We then evaluate the utility of 6 EEG DA approaches. Importantly, to remove the bias that could be introduced by comparing performance for models trained on larger augmented training sets to models trained on smaller baseline sets, we also introduce a new baseline trained on duplicate training data. We lastly examine the effects of the DA approaches upon representations learned by the model with a pair of explainability analyses. We find that while most approaches boost model performance, they do not improve model performance beyond that of simply using a duplicate training set without DA. The exception to this is channel dropout augmentation, which does improve model performance. These findings suggest the importance of comparing EEG DA methods to a baseline with a duplicate training set of equal size to the augmented training set. We also found that some DA methods increased model robustness to frequency (Fourier transform surrogates) and channel (channel dropout) perturbation. While our findings on EEG DA efficacy are restricted to our dataset and model, we hope that future studies on deep learning for small EEG datasets and on new EEG DA methods will find our findings helpful.

Recruiting neural field theory for data augmentation in a motor imagery brain-computer interface

We introduce a novel approach to training data augmentation in brain-computer interfaces (BCIs) using neural field theory (NFT) applied to EEG data from motor imagery tasks. BCIs often suffer from limited accuracy due to a limited amount of training data. To address this, we leveraged a corticothalamic NFT model to generate artificial EEG time series as supplemental training data. We employed the BCI competition IV '2a' dataset to evaluate this augmentation technique. For each individual, we fitted the model to common spatial patterns of each motor imagery class, jittered the fitted parameters, and generated time series for data augmentation. Our method led to significant accuracy improvements of over 2% in classifying the "total power" feature, but not in the case of the "Higuchi fractal dimension" feature. This suggests that the fit NFT model may more favorably represent one feature than the other. These findings pave the way for further exploration of NFT-based data augmentation, highlighting the benefits of biophysically accurate artificial data.

A Data Augmentation Method for Motor Imagery EEG Signals Based on DCGAN-GP Network

Motor imagery electroencephalography (EEG) signals have garnered attention in brain-computer interface (BCI) research due to their potential in promoting motor rehabilitation and control. However, the limited availability of labeled data poses challenges for training robust classifiers. In this study, we propose a novel data augmentation method utilizing an improved Deep Convolutional Generative Adversarial Network with Gradient Penalty (DCGAN-GP) to address this issue. We transformed raw EEG signals into two-dimensional time-frequency maps and employed a DCGAN-GP network to generate synthetic time-frequency representations resembling real data. Validation experiments were conducted on the BCI IV 2b dataset, comparing the performance of classifiers trained with augmented and unaugmented data. Results demonstrated that classifiers trained with synthetic data exhibit enhanced robustness across multiple subjects and achieve higher classification accuracy. Our findings highlight the effectiveness of utilizing a DCGAN-GP-generated synthetic EEG data to improve classifier performance in distinguishing different motor imagery tasks. Thus, the proposed data augmentation method based on a DCGAN-GP offers a promising avenue for enhancing BCI system performance, overcoming data scarcity challenges, and bolstering classifier robustness, thereby providing substantial support for the broader adoption of BCI technology in real-world applications.

Amplifying pathological detection in EEG signaling pathways through cross-dataset transfer learning

Pathology diagnosis based on EEG signals and decoding brain activity holds immense importance in understanding neurological disorders. With the advancement of artificial intelligence methods and machine learning techniques, the potential for accurate data-driven diagnoses and effective treatments has grown significantly. However, applying machine learning algorithms to real-world datasets presents diverse challenges at multiple levels. The scarcity of labeled data, especially in low regime scenarios with limited availability of real patient cohorts due to high costs of recruitment, underscores the vital deployment of scaling and transfer learning techniques. In this study, we explore a real-world pathology classification task to highlight the effectiveness of data and model scaling and cross-dataset knowledge transfer. As such, we observe varying performance improvements through data scaling, indicating the need for careful evaluation and labeling. Additionally, we identify the challenges of possible negative transfer and emphasize the significance of some key components to overcome distribution shifts and potential spurious correlations and achieve positive transfer. We see improvement in the performance of the target model on the target (NMT) datasets by using the knowledge from the source dataset (TUAB) when a low amount of labeled data was available. Our findings demonstrated that a small and generic model (e.g. ShallowNet) performs well on a single dataset, however, a larger model (e.g. TCN) performs better in transfer learning when leveraging a larger and more diverse dataset.

Pseudo-online framework for BCI evaluation: a MOABB perspective using various MI and SSVEP datasets

Objective. BCI (Brain-Computer Interfaces) operate in three modes:online,offline, andpseudo-online. Inonlinemode, real-time EEG data is constantly analyzed. Inofflinemode, the signal is acquired and processed afterwards. Thepseudo-onlinemode processes collected data as if they were received in real-time. The main difference is that theofflinemode often analyzes the whole data, while theonlineandpseudo-onlinemodes only analyze data in short time windows.Offlineprocessing tends to be more accurate, whileonlineanalysis is better for therapeutic applications.Pseudo-onlineimplementation approximatesonlineprocessing without real-time constraints. Many BCI studies beingofflineintroduce biases compared to real-life scenarios, impacting classification algorithm performance.Approach. The objective of this research paper is therefore to extend the current MOABB framework, operating inofflinemode, so as to allow a comparison of different algorithms in apseudo-onlinesetting with the use of a technology based on overlapping sliding windows. To do this will require the introduction of a idle state event in the dataset that takes into account all different possibilities that are not task thinking. To validate the performance of the algorithms we will use the normalized Matthews correlation coefficient and the information transfer rate.Main results. We analyzed the state-of-the-art algorithms of the last 15 years over several motor imagery and steady state visually evoked potential multi-subjects datasets, showing the differences between the two approaches from a statistical point of view.Significance. The ability to analyze the performance of different algorithms inofflineandpseudo-onlinemodes will allow the BCI community to obtain more accurate and comprehensive reports regarding the performance of classification algorithms.

Evaluating Augmentation Approaches for Deep Learning-based Major Depressive Disorder Diagnosis with Raw Electroencephalogram Data

While deep learning methods are increasingly applied in research contexts for neuropsychiatric disorder diagnosis, small dataset size limits their potential for clinical translation. Data augmentation (DA) could address this limitation, but the utility of EEG DA methods remains relatively underexplored in neuropsychiatric disorder diagnosis. In this study, we train a model for major depressive disorder diagnosis. We then evaluate the utility of 6 EEG DA approaches. Importantly, to remove the bias that could be introduced by comparing performance for models trained on larger augmented training sets to models trained on smaller baseline sets, we also introduce a new baseline trained on duplicate training data to better. We lastly examine the effects of the DA approaches upon representations learned by the model with a pair of explainability analyses. We find that while most approaches boost model performance, they do not improve model performance beyond that of simply using a duplicate training set without DA. The exception to this is channel dropout augmentation, which does improve model performance. These findings suggest the importance of comparing EEG DA methods to a baseline with a duplicate training set of equal size to the augmented training set. We also found that some DA methods increased model robustness to frequency (Fourier transform surrogates) and channel (channel dropout) perturbation. While our findings on EEG DA efficacy are restricted to our dataset and model, we hope that future studies on deep learning for small EEG datasets and on new EEG DA methods will find our findings helpful.

An extended clinical EEG dataset with 15,300 automatically labelled recordings for pathology decoding

Automated clinical EEG analysis using machine learning (ML) methods is a growing EEG research area. Previous studies on binary EEG pathology decoding have mainly used the Temple University Hospital (TUH) Abnormal EEG Corpus (TUAB) which contains approximately 3,000 manually labelled EEG recordings. To evaluate and eventually even improve the generalisation performance of machine learning methods for EEG pathology, decoding larger, publicly available datasets is required. A number of studies addressed the automatic labelling of large open-source datasets as an approach to create new datasets for EEG pathology decoding, but little is known about the extent to which training on larger, automatically labelled dataset affects decoding performances of established deep neural networks. In this study, we automatically created additional pathology labels for the Temple University Hospital (TUH) EEG Corpus (TUEG) based on the medical reports using a rule-based text classifier. We generated a dataset of 15,300 newly labelled recordings, which we call the TUH Abnormal Expansion EEG Corpus (TUABEX), and which is five times larger than the TUAB. Since the TUABEX contains more pathological (75%) than non-pathological (25%) recordings, we then selected a balanced subset of 8,879 recordings, the TUH Abnormal Expansion Balanced EEG Corpus (TUABEXB). To investigate how training on a larger, automatically labelled dataset affects the decoding performance of deep neural networks, we applied four established deep convolutional neural networks (ConvNets) to the task of pathological versus non-pathological classification and compared the performance of each architecture after training on different datasets. The results show that training on the automatically labelled TUABEXB dataset rather than training on the manually labelled TUAB dataset increases accuracies on TUABEXB and even for TUAB itself for some architectures. We argue that automatically labelling of large open-source datasets can be used to efficiently utilise the massive amount of EEG data stored in clinical archives. We make the proposed TUABEXB available open source and thus offer a new dataset for EEG machine learning research.

Automatic Detection of Abnormal EEG Signals Using WaveNet and LSTM

Neurological disorders have an extreme impact on global health, affecting an estimated one billion individuals worldwide. According to the World Health Organization (WHO), these neurological disorders contribute to approximately six million deaths annually, representing a significant burden. Early and accurate identification of brain pathological features in electroencephalogram (EEG) recordings is crucial for the diagnosis and management of these disorders. However, manual evaluation of EEG recordings is not only time-consuming but also requires specialized skills. This problem is exacerbated by the scarcity of trained neurologists in the healthcare sector, especially in low- and middle-income countries. These factors emphasize the necessity for automated diagnostic processes. With the advancement of machine learning algorithms, there is a great interest in automating the process of early diagnoses using EEGs. Therefore, this paper presents a novel deep learning model consisting of two distinct paths, WaveNet-Long Short-Term Memory (LSTM) and LSTM, for the automatic detection of abnormal raw EEG data. Through multiple ablation experiments, we demonstrated the effectiveness and importance of all parts of our proposed model. The performance of our proposed model was evaluated using TUH abnormal EEG Corpus V.2.0.0. (TUAB) and achieved a high classification accuracy of 88.76%, which is higher than in the existing state-of-the-art research studies. Moreover, we demonstrated the generalization of our proposed model by evaluating it on another independent dataset, TUEP, without any hyperparameter tuning or adjustment. The obtained accuracy was 97.45% for the classification between normal and abnormal EEG recordings, confirming the robustness of our proposed model.

Sleep-Energy: An Energy Optimization Method to Sleep Stage Scoring

Sleep is essential for physical and mental health. Polysomnography (PSG) procedures are labour-intensive and time-consuming, making diagnosing sleep disorders difficult. Automatic sleep staging using Machine Learning (ML) - based methods has been studied extensively, but frequently provides noisier predictions incompatible with typical manually annotated hypnograms. We propose an energy optimization method to improve the quality of hypnograms generated by automatic sleep staging procedures. The method evaluates the system's total energy based on conditional probabilities for each epoch's stage and employs an energy minimisation procedure. It can be used as a meta-optimisation layer over the sleep stage sequences generated by any classifier that generates prediction probabilities. The method improved the accuracy of state-of-the-art Deep Learning models in the Sleep EDFx dataset by 4.0% and in the DRM-SUB dataset by 2.8%.

Impact of dataset size and long-term ECoG-based BCI usage on deep learning decoders performance

Introduction

In brain-computer interfaces (BCI) research, recording data is time-consuming and expensive, which limits access to big datasets. This may influence the BCI system performance as machine learning methods depend strongly on the training dataset size. Important questions arise: taking into account neuronal signal characteristics (e.g., non-stationarity), can we achieve higher decoding performance with more data to train decoders? What is the perspective for further improvement with time in the case of long-term BCI studies? In this study, we investigated the impact of long-term recordings on motor imagery decoding from two main perspectives: model requirements regarding dataset size and potential for patient adaptation.

Methods

We evaluated the multilinear model and two deep learning (DL) models on a long-term BCI & Tetraplegia (ClinicalTrials.gov identifier: NCT02550522) clinical trial dataset containing 43 sessions of ECoG recordings performed with a tetraplegic patient. In the experiment, a participant executed 3D virtual hand translation using motor imagery patterns. We designed multiple computational experiments in which training datasets were increased or translated to investigate the relationship between models' performance and different factors influencing recordings.

Results

Our results showed that DL decoders showed similar requirements regarding the dataset size compared to the multilinear model while demonstrating higher decoding performance. Moreover, high decoding performance was obtained with relatively small datasets recorded later in the experiment, suggesting motor imagery patterns improvement and patient adaptation during the long-term experiment. Finally, we proposed UMAP embeddings and local intrinsic dimensionality as a way to visualize the data and potentially evaluate data quality.

Discussion

DL-based decoding is a prospective approach in BCI which may be efficiently applied with real-life dataset size. Patient-decoder co-adaptation is an important factor to consider in long-term clinical BCI.