Lightweight Convolution Transformer for Cross-patient Seizure Detection in Multi-channel EEG Signals

Salim Rukhsar,Anil K. Tiwari

2023-05-08

Abstract:Background: Epilepsy is a neurological illness affecting the brain that makes people more likely to experience frequent, spontaneous seizures. There has to be an accurate automated method for measuring seizure frequency and severity in order to assess the efficacy of pharmacological therapy for epilepsy. The drug quantities are often derived from patient reports which may cause significant issues owing to inadequate or inaccurate descriptions of seizures and their frequencies. Methods and materials: This study proposes a novel deep learning architecture based lightweight convolution transformer (LCT). The transformer is able to learn spatial and temporal correlated information simultaneously from the multi-channel electroencephalogram (EEG) signal to detect seizures at smaller segment lengths. In the proposed model, the lack of translation equivariance and localization of ViT is reduced using convolution tokenization, and rich information from the transformer encoder is extracted by sequence pooling instead of the learnable class token. Results: Extensive experimental results demonstrate that the proposed model of cross-patient learning can effectively detect seizures from the raw EEG signals. The accuracy and F1-score of seizure detection in the cross-patient case on the CHB-MIT dataset are shown to be 96.31% and 96.32%, respectively, at 0.5 sec segment length. In addition, the performance metrics show that the inclusion of inductive biases and attention-based pooling in the model enhances the performance and reduces the number of transformer encoder layers, which significantly reduces the computational complexity. In this research work, we provided a novel approach to enhance efficiency and simplify the architecture for multi-channel automated seizure detection.

Signal Processing,Machine Learning

What problem does this paper attempt to address?

The paper aims to address the automation and accuracy issues in seizure detection from electroencephalogram (EEG) signals in epilepsy patients. Specifically, the research objectives include:

1. **Improving the accuracy of seizure detection**: By developing a novel deep learning architecture—Lightweight Convolution Transformer (LCT), to accurately detect seizures in short segments of multi-channel EEG signals.

2. **Addressing the issues of existing methods**: Current methods often fail to simultaneously consider spatial and temporal correlations. Simplified deep learning features may lose information, and developing cross-patient detectors is challenging because seizure patterns can vary significantly between different patients and even within the same patient over different periods.

3. **Optimizing computational efficiency**: By introducing inductive biases and attention-based pooling, LCT can reduce the number of parameters, thereby lowering computational complexity.

4. **Achieving cross-patient learning**: The proposed method can effectively detect seizures across different patients, indicating that the model has good generalization capabilities.

In summary, the main contribution of this research is the proposal of a new Lightweight Convolution Transformer model. This model not only learns spatial and temporal information from multi-channel EEG signals simultaneously but also improves detection performance and computational efficiency through a series of innovative architectural improvements.