Hierarchical Harris hawks optimization for epileptic seizure classification.

Area of Science:

• * Neuroscience and Artificial Intelligence
• * Medical Signal Processing

Background:

• * Visual inspection of electroencephalogram (EEG) signals for epilepsy diagnosis is time-consuming and labor-intensive.
• * Existing automated methods often fail to fully capture discriminative EEG features, leading to suboptimal performance.
• * There is a need for more accurate and efficient automated systems for epilepsy diagnosis.

Purpose of the Study:

• * To propose an enhanced automatic epilepsy diagnosis method utilizing time-frequency analysis and an improved Harris Hawks Optimization (IHHO) algorithm.
• * To improve the accuracy and efficiency of epilepsy seizure classification from EEG signals.
• * To overcome the limitations of existing methods in handling discriminative features and avoiding local optima.

Main Methods:

• * Continuous wavelet transform for decomposing EEG signals into five rhythms.
• * Local binary pattern (LBP) and gray level co-occurrence matrix (GLCM) for extracting local and global features.
• * Improved Harris Hawks Optimization (IHHO) for feature selection and a hierarchical mechanism.
• * K-nearest neighbor (KNN) classifier for final recognition.

Main Results:

• * The IHHO algorithm demonstrated superior performance compared to classical meta-heuristic algorithms on 23 benchmark functions.
• * Achieved high accuracy rates: over 99.67% on the Bonn dataset and 99.06% on the CHB-MIT dataset.
• * Outperformed multiple state-of-the-art methods in epilepsy diagnosis.

Conclusions:

• * The proposed enhanced automatic epilepsy diagnosis method is effective and accurate.
• * The approach shows significant utility in the automatic diagnosis of epilepsy.
• * Publicly available datasets and source codes facilitate further research and application.