Identification of Epileptic Spasms (ESES) Phases Using EEG Signals: A Vision Transformer Approach
Wei Gong, Yaru Li
TL;DR
This paper tackles automated detection of Epileptic Spasms during Sleep (ESES) from EEG signals. It introduces converting frequency-domain EEG representations into grayscale images and feeding them into a Vision Transformer (ViT) to exploit self-attention for capturing global patterns across channels and time. The key results show ViT achieving $97\%$ accuracy after $80$ epochs, outperforming a CNN baseline at $94\%$. The approach offers real-time, scalable deployment potential and can extend to multi-modal data fusion for enhanced clinical decision support.
Abstract
This work introduces a new approach to the Epileptic Spasms (ESES) detection based on the EEG signals using Vision Transformers (ViT). Classic ESES detection approaches have usually been performed with manual processing or conventional algorithms, suffering from poor sample sizes, single-channel-based analyses, and low generalization abilities. In contrast, the proposed ViT model overcomes these limitations by using the attention mechanism to focus on the important features in multi-channel EEG data, which is contributing to both better accuracy and efficiency. The model processes frequency-domain representations of EEG signals, such as spectrograms, as image data to capture long-range dependencies and complex patterns in the signal. The model demonstrates high performance with an accuracy of 97% without requiring intensive data preprocessing, thus rendering it suitable for real-time clinical applications on a large scale. The method represents a significant development in the advancement of neurological disorders such as ESES in detection and analysis.