From Epilepsy Seizures Classification to Detection: A Deep Learning-based Approach for Raw EEG Signals

Davy Darankoum,Manon Villalba,Clelia Allioux,Baptiste Caraballo,Carine Dumont,Eloise Gronlier,Corinne Roucard,Yann Roche,Chloe Habermacher,Sergei Grudinin,Julien Volle

2024-10-04

Abstract:Epilepsy represents the most prevalent neurological disease in the world. One-third of people suffering from mesial temporal lobe epilepsy (MTLE) exhibit drug resistance, urging the need to develop new treatments. A key part in anti-seizure medication (ASM) development is the capability of detecting and quantifying epileptic seizures occurring in electroencephalogram (EEG) signals, which is crucial for treatment efficacy evaluation. In this study, we introduced a seizure detection pipeline based on deep learning models applied to raw EEG signals. This pipeline integrates: a new pre-processing technique which segments continuous raw EEG signals without prior distinction between seizure and seizure-free activities; a post-processing algorithm developed to reassemble EEG segments and allow the identification of seizures start/end; and finally, a new evaluation procedure based on a strict seizure events comparison between predicted and real labels. Models training have been performed using a data splitting strategy which addresses the potential for data leakage. We demonstrated the fundamental differences between a seizure classification and a seizure detection task and showed the differences in performance between the two tasks. Finally, we demonstrated the generalization capabilities across species of our best architecture, combining a Convolutional Neural Network and a Transformer encoder. The model was trained on animal EEGs and tested on human EEGs with a F1-score of 93% on a balanced Bonn dataset.

Machine Learning,Neurons and Cognition

What problem does this paper attempt to address?

The paper attempts to address the problem of developing a deep learning-based method for detecting epileptic seizures from raw electroencephalogram (EEG) signals. Specifically, the paper focuses on the following key issues:

1. **Improving the accuracy of seizure detection**: Currently, seizure detection mainly relies on manual analysis of EEG signals, which is prone to misjudgment, inefficiency, and subjectivity. Therefore, the paper proposes an automated deep learning method to improve the accuracy and reliability of seizure detection.

2. **Distinguishing between classification and detection tasks**: The paper emphasizes the fundamental difference between seizure classification tasks and detection tasks. Classification tasks typically involve classifying EEG signals within known time periods, whereas detection tasks require real-time identification of the start and end times of seizures in continuous EEG signals. The paper demonstrates the performance differences between these two tasks and proposes new methods suitable for detection tasks.

3. **Cross-species generalization ability**: The paper also explores the model's generalization ability across different species. Specifically, the model is trained on EEG data from animals (mice) and then tested on human EEG data. The results show that the model performs well in cross-species applications, with a high F1 score (93%).

4. **Handling the complex characteristics of EEG signals**: EEG signals have complex characteristics such as low signal-to-noise ratio, high frequency dimensions, non-stationarity, non-linearity, and the presence of artifacts. The paper proposes a new preprocessing technique that can effectively segment continuous raw EEG signals without prior distinction between seizure and non-seizure activities. Additionally, a post-processing algorithm is developed to recombine the segmented EEG fragments to identify the start and end times of seizures.

Through these methods, the paper aims to provide more accurate tools for the development of anti-seizure medications (ASM), thereby improving the evaluation and optimization of treatment efficacy.