CPRSCA-ResNet: a novel ResNet-based model with Channel-Partitioned Resolution Spatial-Channel Attention for EEG-based...

Area of Science:

• Neurology
• Artificial Intelligence
• Biomedical Engineering

Background:

• Epilepsy is a chronic neurological disorder characterized by recurrent seizures, impacting cognitive function and increasing mortality risk.
• Current diagnosis relies on manual electroencephalogram (EEG) interpretation, which is time-consuming, labor-intensive, and prone to errors.
• There is a critical need for automated, accurate, and efficient seizure detection models to aid clinical diagnosis.

Purpose of the Study:

• To develop and validate a novel automatic seizure detection model for epilepsy using EEG data.
• To introduce a Channel-Partitioned Resolution Spatial-Channel Attention (CPRSCA) mechanism to enhance EEG feature representation.
• To evaluate the model's performance on diverse datasets for both patient-dependent and patient-independent scenarios.

Main Methods:

• Developed a CPRSCA-ResNet model based on the ResNet-34 architecture, integrating fine-grained channel partitioning and multi-dimensional attention mechanisms.
• Incorporated multi-scale feature fusion to capture complex EEG patterns.
• Conducted experiments on the public CHB-MIT dataset and two local hospital datasets (JHCH, JHMCHH).

Main Results:

• The CPRSCA-ResNet model achieved high accuracies in patient-dependent experiments (99.12% ± 2.09%, 96.88% ± 4.64%, 98.84% ± 1.75%) across datasets.
• Patient-independent experiment accuracies were also strong (78.71% ± 13.06%, 87.15% ± 15.32%, 89.23% ± 7.87%).
• The proposed model consistently outperformed existing state-of-the-art methods, demonstrating its effectiveness and generalizability.

Conclusions:

• The novel CPRSCA mechanism and CPRSCA-ResNet model offer an efficient, robust, and highly generalizable solution for automatic seizure detection.
• This approach has the potential to significantly reduce the burden of manual EEG interpretation in clinical settings.
• The study highlights a promising technical advancement for improving the diagnostic efficiency and auxiliary clinical diagnosis of epilepsy.