Classification of EEG abnormalities in partial epilepsy with simultaneous EEG-fMRI recordings.

Area of Science:

• Neuroscience
• Medical Imaging
• Signal Processing

Background:

• Scalp electroencephalogram (EEG) and interictal epileptiform discharge (IED) classification are crucial for pre-surgical epilepsy evaluation.
• Current methods rely on time-consuming expert visual analysis, leading to inter-observer variability.
• Defining the irritative zone (IZ) is key for surgical planning in refractory epilepsies.

Observation:

• A novel approach using the Wave_clus automatic spike sorting algorithm was applied to classify IEDs.
• Simultaneous EEG-fMRI data from 8 refractory epilepsy patients undergoing pre-surgical evaluation were analyzed.
• Two fMRI analyses were performed: one based on visual classification and one based on algorithmic sorting.

Findings:

• The automated approach identified more IED classes (29) compared to visual identification (26).
• EEG-fMRI analysis showed improved performance with algorithmic classification (72%) versus visual (50%).
• Hemodynamic maps derived from algorithmic IED classification localized within the presumed IZ for more IED classes and patients.

Implications:

• Automated spike sorting offers an efficient tool for mapping IED-related fMRI changes.
• This method enhances the clinical value of EEG-fMRI for pre-surgical assessment in severe epilepsy.
• The algorithm provides critical and reliable information for IZ localization, surpassing previous approaches.