EEG-based classification of epilepsy and PNES: EEG microstate and functional brain network features

Area of Science:

• Neurology
• Neuroscience
• Biomedical Engineering

Background:

• Epilepsy and psychogenic non-epileptic seizures (PNES) present overlapping symptoms, complicating early diagnosis.
• Accurate differentiation often requires lengthy and costly long-term video-EEG monitoring.
• Electroencephalogram (EEG) patterns differ between epileptic seizures (epileptiform discharges) and PNES (normal brain activity).

Purpose of the Study:

• To analyze the classification of epilepsy and PNES using short-term EEG data.
• To identify key signal, functional network, and EEG microstate features for accurate differentiation.
• To evaluate the efficacy of beta-band frequency and microstate coverage in classification.

Main Methods:

• Analysis of short-term EEG data from epilepsy and PNES subjects.
• Feature extraction based on signal characteristics, functional brain networks, and EEG microstates.
• Classification modeling using identified features, with a focus on the beta-band frequency and microstate coverage.

Main Results:

• The beta-band frequency sub-band demonstrated the highest utility for classifying epilepsy and PNES.
• Incorporating the coverage feature of EEG microstate analysis within the beta-band yielded high classification accuracy and precision.
• Beta-band activity and microstate coverage emerged as critical features for patient classification.

Conclusions:

• Short-term EEG analysis, specifically leveraging beta-band features and microstate coverage, offers an efficient method for classifying epilepsy and PNES.
• This approach may reduce reliance on expensive and time-consuming long-term monitoring.
• The findings highlight the potential of targeted EEG analysis for improved diagnostic strategies.