Epileptic Seizure Diagnosis using Bayesian Belief Network and T-CNN

Epilepsy, a neurological disorder characterized by recurring seizures, presents considerable challenges in diagnosis. This study proposes a novel diagnostic approach integrating Bayesian Belief Network (BBN) and Temporal Convolutional Neural Network (T-CNN). The system begins with EEG feature selection, identifying crucial discriminative features for seizure detection. A BBN models complex feature relationships, capturing causal dependencies and uncertainties. Additionally, T-CNN extracts temporal patterns from EEG signals, facilitating robust seizure detection in time and frequency domains. The fusion of BBN and T-CNN enhances diagnostic accuracy by leveraging probabilistic modeling and deep learning strengths. Experimental results on benchmark datasets demonstrate the methodology’s effectiveness in accurately detecting seizures with high sensitivity and specificity. The proposed system offers clinicians reliable diagnostic support, insights into seizure mechanisms, and easy integration into existing clinical workflows, potentially improving epilepsy management and patient care.