A Novel Frame Funnel-Shaped CNN with SE Block for Epilepsy Detection in EEG Signals

Epilepsy is a common neurological disorder characterized by sudden and unpredictable seizures, posing serious threats to patient safety and affecting diagnostic efficiency. To improve the accuracy of epilepsy detection, this paper proposes a method named FFS-CNN-SEB. The model employs frame processing to enhance features and extract local patterns under multi-scale temporal windows. A funnel-shaped 1D-CNN with a progressively compressed architecture is used to aggregate deep features, while the SE block is introduced to recalibrate channel-wise features through weighted modulation, thereby enhancing the model’s ability to identify key epileptic patterns. Experimental results on the Bonn EEG dataset demonstrate that the proposed model achieves 100% accuracy, sensitivity, and specificity, and exhibits superior performance in multiple comparative and ablation studies.