Epileptic seizure focus detection from interictal electroencephalogram: a survey

Area of Science:

• Neurology
• Artificial Intelligence
• Biomedical Engineering

Background:

• Electroencephalogram (EEG) is a key diagnostic tool for localizing epileptic foci.
• Artificial intelligence (AI) is increasingly used for automated analysis of EEG signals to detect seizure focus.
• Existing AI solutions often rely on interictal EEGs, analyzing biomarkers or statistical features.

Purpose of the Study:

• To review and categorize recent AI solutions for epileptic focus detection using EEG.
• To provide a comprehensive overview of AI methodologies applied to interictal EEG analysis.
• To identify research gaps, evaluation criteria, and future challenges in AI-driven epilepsy diagnosis.

Main Methods:

• Categorization of AI solutions into three main groups: biomarker-based, feature-extraction-based, and neural network-based (end-to-end).
• Review of clinical diagnosis methods for seizure focus identification.
• Summary of publicly available EEG datasets for epilepsy research.
• Analysis of recent performance evaluation criteria for AI systems.

Main Results:

• AI solutions for epileptic focus detection are broadly classified into three categories.
• Biomarker-based methods utilize signals like high-frequency oscillations and interictal epileptiform discharges.
• Feature-extraction and end-to-end neural network approaches represent alternative AI strategies.

Conclusions:

• The review provides a structured overview of AI techniques for epileptic focus localization.
• It highlights the importance of public datasets and novel evaluation metrics.
• Future research should focus on overcoming challenges to develop more efficient computer-aided diagnostic tools for epilepsy.