Content area
Accurate and efficient detection of epileptic seizures from EEG signals remains a critical challenge due to high-dimensional data, class imbalance, and the limitations of standard classifiers. This thesis introduces two novel models to address these challenges.
The first model presents a new classifier based on the Hellinger Distance, specifically designed to enhance discriminative capability and robustness against imbalanced datasets. By integrating the Hellinger Distance Classifier with Particle Swarm Optimization (PSO) for feature selection, this model significantly improves classification performance while reducing computational complexity. Experimental evaluations on the Bonn dataset demonstrate an accuracy of 96.25%, an F1-score of 97.74%, a recall of 95.59%, and a precision of 100%, highlighting the classifier's effectiveness in seizure detection.
The second model introduces a hybrid feature selection approach that combines Hellinger Distance filtering with PSO to optimize feature selection and further improve seizure classification accuracy. The first phase employs Hellinger Distance to eliminate redundant and irrelevant features, reducing the feature space. The second phase applies PSO to identify the most informative feature subset, ensuring optimal classification performance. This approach enhances classification accuracy across multiple models, improving Logistic Regression (91% to 95%), Decision Tree (95% to 97%), Naïve Bayes (94% to 99%), and Random Forest (96% to 98%), while significantly reducing the dimensionality from 4047 features to a refined subset.
The integration of these two models provides a robust framework for epileptic seizure detection, enhancing both accuracy and computational efficiency. The findings contribute to advancing AI-driven medical diagnostics, offering a precise and efficient solution for real-time seizure detection, ultimately improving patient care and clinical decision-making.