Deep learning-based electroencephalography analysis: a systematic review

By Yannick Roy, Hubert J. Banville, Isabela Albuquerque, Alexandre Gramfort, Tiago H. Falk, Jocelyn Faubert. Journal of Neural Engineering, 2019

1,358 citations recorded by the source corpus.

Context. Electroencephalography (EEG) is a complex signal and can require several years of training, as well as advanced signal processing and feature extraction methodologies to be correctly interpreted. Recently, deep learning (DL) has shown great promise in helping make sense of EEG signals due to its capacity to learn good feature representations from raw data. Whether DL truly presents advantages as compared to more traditional EEG processing approaches, however, remains an open question. Objective. In this work, we review 154 papers that apply DL to EEG, published between January 2010 and July 2018, and spanning different application domains such as epilepsy, sleep, brain–computer interfacing, and cognitive and affective monitoring. We extract trends and highlight interesting approaches from this large body of literature in order to inform future research and formulate recommendations. Methods. Major databases spanning the fields of science and engineering were queried to identify relevant studies published in scientific journals, conferences, and electronic preprint repositories. Various data items were extracted for each study pertaining to (1) the data, (2) the preprocessing methodology, (3) the DL design choices, (4) the results, and (5) the reproducibility of the experiments. These items were then analyzed one by one to uncover trends. Results. Our analysis reveals that the amount of EEG data used across studies varies from less than ten minutes to thousands of hours, while the number of samples seen during training by a network varies from a few dozens to several millions, depending on how epochs are extracted. Interestingly, we saw that more than half the studies used publicly available data and that there has also been a clear shift from intra-subject to inter-subject approaches over the last few years. About of the studies used convolutional neural networks (CNNs), while used recurrent neural networks (RNNs), most often with a total of 3–10 layers. Moreover, almost one-half of the studies trained their models on raw or preprocessed EEG time series. Finally, the median gain in accuracy of DL approaches over traditional baselines was across all relevant studies. More importantly, however, we noticed studies often suffer from poor reproducibility: a majority of papers would be hard or impossible to reproduce given the unavailability of their data and code. Significance. To help the community progress and share work more effectively, we provide a list of recommendations for future studies and emphasize the need for more reproducible research. We also make our summary table of DL and EEG papers available and invite authors of published work to contribute to it directly. A planned follow-up to this work will be an online public benchmarking portal listing reproducible results.

Subfields: Cognitive Neuroscience

View Paper: https://www.semanticscholar.org/paper/656bd33ebf7acd8887d3598e1513e53e8fb69d41

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Cognitive Neuroscience

Citations in the corpus

EEG-Based Brain-Computer Interfaces (BCIs): A Survey of Recent Studies on Signal Sensing Technologies and Computational Intelligence Approaches and Their Applications (2020, background): “Another recent survey [34] systematically reviewed articles published between 2010 and 2018 that applied deep learning to EEG in diverse domains by extracting and analysing various datasets to”
BENDR: Using Transformers and a Contrastive Self-Supervised Learning Task to Learn From Massive Amounts of EEG Data (2021, background): “The difﬁculty of learning both “lower-level” features and an expressive classiﬁer simultaneously may help explain why work using DNNs to classify raw BCI data has tended to prefer shallower networks [9, 10, 2, 11, 12].”
Machine Learning for Predicting Epileptic Seizures Using EEG Signals: A Review (2020, background): “adhoumi et al. [7] 2016 ES prediction p p p ˇ M.Iftikhar et al. [8] 2018 DL for EEG signals p ˇ p Acharya et al. [9] 2018 ES prediction p p ˇ p Kuhlmann et al. [9] 2018 ES prediction p ˇ p Roy et al. [10] 2019 DL for EEG p p p p Li et al. [11] 2019 DL for EEG ˇ p p p This paper 2019 ES prediction p ”
Data Augmentation for Deep-Learning-Based Electroencephalography. (2020, methodology): “This renders such task monitoring tractable using EEG [1].”
Cross-Modal Epileptic Signal Harmonization: Frequency Domain Mapping Quantization for Pre-training a Unified Neurophysiological Transformer (2025, unknown): “First, EpiNT consistently outperformed randomly initialized models, confirming that pre-training significantly enhances performance (Supplementary Tables 4, 5, 6, 7, and 8).”
The Neurostimulationist will see you now: prescribing direct electrical stimulation therapies for the human brain in epilepsy and beyond (2024, background): “…parameter space and its relationship to neural activity is indeed complex, ML provides an additional powerful set of tools, particularly to reverse engineer a stimulation response relative to stimulation parameters (Prato and Zanni, 2008; Roy et al., 2019; Hecker et al., 2021; Huang et al., 2022).”
Is Limited Participant Diversity Impeding EEG-based Machine Learning? (2025, unknown): “While prior works have acknowledged limited data as a challenge for EEG-based machine learning [5, 6], systematic studies are currently lacking.”