Multiresolution directed transfer function approach for segment-wise seizure classification of epileptic EEG signal

Area of Science:

• Artificial Intelligence in Healthcare
• Cognitive Computing for Neurological Disorders
• Biomedical Signal Processing

Background:

• Epileptic seizures are chronic brain diseases requiring accurate detection.
• Existing AI methods for seizure detection have limitations in precision and localization.
• Human-centered cognitive computing offers potential for advanced diagnostic tools.

Purpose of the Study:

• To propose a novel Human-Centered Cognitive Computing (HCCC) method for segment-wise seizure classification.
• To introduce the Multiresolution Directed Transfer Function (MDTF) approach for enhanced epileptic seizure detection.
• To improve the accuracy and efficiency of diagnosing chronic brain diseases like epilepsy.

Main Methods:

• Utilized Multiresolution Adaptive Filtering (MRAF) to extract seizure signal information.
• Employed Directed Transfer Function (DTF) to compute information flow in high-frequency bands.
• Applied complexity measures (Approximate Entropy, Sample Entropy) and k-NN/SVM classifiers for EEG signal classification.

Main Results:

• Achieved high performance metrics: 98.31% sensitivity, 96.13% specificity, and 98.89% accuracy with SVM.
• The MDTF approach demonstrated an average detection rate of 97.72%, surpassing existing methods.
• Successfully validated on standard and local hospital datasets.

Conclusions:

• The MDTF approach accurately locates seizure activity and information drift within EEG signals.
• This method assists neuro-specialists in precise, automated epileptic seizure localization.
• Reduces the time and effort required for analyzing complex epileptic seizure data.