An Efficient Deep Learning Framework for Automated Epileptic Seizure Detection: Toward Scalable and Clinically...

Area of Science:

• * Neuroscience
• * Machine Learning
• * Medical Technology

Background:

• * Epilepsy diagnosis relies heavily on electroencephalogram (EEG) analysis.
• * Conventional methods often struggle with complex spatiotemporal feature extraction.
• * There is a need for efficient and accurate automated seizure detection systems.

Purpose of the Study:

• * To develop an efficient epileptic seizure detection framework using a graph convolutional neural network (GCNN).
• * To leverage GCNNs for capturing comprehensive spatiotemporal features from EEG data.
• * To reduce computational complexity and enhance clinical applicability of EEG-based diagnostics.

Main Methods:

• * Implementation of a GCNN model for direct processing of EEG electrode spatial dependencies.
• * Minimal preprocessing including bandpass filtering and segmentation.
• * Validation on the CHB-MIT and SH-SDU EEG databases.

Main Results:

• * Achieved high segment-based accuracy (98.64% on CHB-MIT, 95.23% on SH-SDU) and event-based sensitivity (96.81% on CHB-MIT, 94.11% on SH-SDU).
• * Demonstrated low computational overhead with an average testing time of 3.89 s per hour of EEG.
• * High specificity (98.64% on CHB-MIT, 95.25% on SH-SDU) indicates robust performance.

Conclusions:

• * The GCNN framework offers an efficient and accurate method for epileptic seizure detection.
• * Its ability to capture spatiotemporal features and low computational cost make it suitable for clinical settings.
• * This approach has the potential to advance EEG-based epilepsy diagnostics and improve patient care.