Functional anomaly mapping lateralizes temporal lobe epilepsy with high accuracy in individual patients

Area of Science:

• Neuroscience
• Machine Learning
• Medical Imaging

Background:

• Mesial temporal lobe epilepsy (mTLE) presents with diverse functional deficits extending beyond the temporal lobe.
• Accurate lateralization of seizure onset is crucial for effective mTLE management.

Approach:

• Functional anomaly mapping (FAM), a machine learning technique, was employed to analyze resting-state fMRI data.
• Multivariate analysis measured cortical and subcortical functional aberrations in mTLE patients compared to controls.
• Individualized FAMs were generated to assess functional changes in each patient.

Key Points:

• Group-level analysis revealed significant functional differences in limbic and somatomotor networks between mTLE subtypes and controls.
• Individual FAMs highlighted anomalies in bilateral mesial temporal and medial parietooccipital regions in most mTLE patients.
• A classification model trained on FAMs achieved high accuracy in lateralizing the hemisphere of seizure onset.

Conclusions:

• Functional anomaly mapping effectively detects widespread functional aberrations in mTLE.
• FAM demonstrates significant potential as a non-invasive method for localizing seizure onset in mTLE.
• Further research will explore FAM's utility in larger cohorts and its correlation with clinical factors.