Microstate-based brain network dynamics distinguishing temporal lobe epilepsy patients: A machine learning approach.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Epilepsy Research

Background:

• Temporal lobe epilepsy (TLE) is the most common focal epilepsy in adults, marked by abnormal brain dynamics.
• The exact mechanisms driving seizures in TLE patients are not fully understood.
• Understanding brain network alterations is crucial for diagnosing and treating TLE.

Purpose of the Study:

• To investigate brain dynamic differences between TLE patients and healthy controls using microstate analysis.
• To differentiate between drug-resistant epilepsy (DRE) and drug-sensitive epilepsy (DSE) patients based on brain network characteristics.
• To develop and validate machine learning models for classifying TLE and DRE/DSE status.

Main Methods:

• Microstate analysis was applied to EEG data from 116 TLE patients and 51 healthy controls.
• Dynamic functional connectivity networks were constructed and analyzed for spatial and temporal variability.
• Machine learning models were trained using spatiotemporal microstate network features.

Main Results:

• TLE patients exhibited accelerated temporal dynamics, increased synchronization, and network instability compared to controls.
• DRE patients showed reduced spatial variability in specific microstate networks (B, E, F) and increased temporal variability in others (E, G) compared to DSE patients.
• Machine learning models achieved high accuracy in distinguishing TLE from controls and DRE from DSE.

Conclusions:

• Accelerated microstate dynamics and altered sequences in TLE suggest highly unstable brain activity underlying seizures.
• Synchronized and unstable network activity in DRE patients indicates established epileptogenic networks resistant to medication.
• Spatiotemporal microstate network analysis provides a robust method for classifying epilepsy types and severity.