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

2024-10-04

arXiv.org

AI Summary

Approximately one-third of patients with mesial temporal lobe epilepsy (MTLE) suffer from drug-resistant seizures, necessitating precise, automated seizure onset and offset detection to support evaluation of novel therapeutic interventions. This paper proposes an end-to-end framework for seizure detection directly from raw EEG signals—eliminating prior segmentation-based preprocessing—and introduces a reconfigurable post-processing module alongside a strict event-level evaluation paradigm. Crucially, it explicitly distinguishes seizure classification from temporal boundary detection. The architecture employs a CNN-Transformer hybrid encoder, integrated with temporally non-leaking sequence partitioning and event-aligned post-processing to ensure robust temporal localization. Evaluated on the Bonn dataset, the model achieves a 93% F1-score. Notably, it demonstrates, for the first time, cross-species generalizability—from animal to human EEG—validating its translational potential. The framework delivers an interpretable, robust, and clinically actionable tool for detecting seizures in pharmacoresistant epilepsy.

Technology Category

• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Machine Learning: Hardware-aware ML
• Computer Vision: Multi-modal Vision

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs

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.

Problem

• Developing deep learning pipeline for epileptic seizure detection in raw EEG signals
• Addressing differences between seizure classification and detection tasks in performance
• Creating cross-species generalization from animal to human EEG data analysis

Innovation

• Deep learning pipeline processes raw EEG signals
• Combines CNN and Transformer for cross-species generalization
• Integrates segmentation and reassembly for seizure detection