AID-HS

AID-HS detects and lateralizes hippocampal sclerosis in temporal lobe epilepsy by analyzing T1-weighted MRI scans and producing clinically-relevant patient reports.

Key Features:

• Morphological Feature Extraction: HippUnfold extracts surface-based morphological features and volumes of the hippocampus from T1-weighted MRI scans.
• Normative Comparison: Pathological hippocampi are characterized by comparison to normative growth charts derived from healthy controls.
• Asymmetry Analysis: Within-subject feature asymmetries are analyzed to quantify hippocampal differences between hemispheres.
• HS Classification: A logistic regression classifier trained on feature asymmetry scores detects and lateralizes hippocampal sclerosis.
• Patient Reporting: Individual patient reports contextualize hippocampal features against normative growth trajectories, visualize feature asymmetries, and report classifier predictions.

Scientific Applications:

• Presurgical Evaluation: Aids presurgical evaluation of children and adults with suspected temporal lobe epilepsy by localizing hippocampal sclerosis.
• MRI-Negative HS Detection: Detects and lateralizes hippocampal sclerosis in patients with MRI-negative but histopathologically-confirmed disease.
• Multicenter Generalization: Validated for generalization to new data across independent multicenter patient cohorts.

Methodology:

HippUnfold extracts morphological surface-based features and volumes from T1-weighted MRI scans. Feature asymmetry scores train a logistic regression classifier for detection and lateralization, validated on an independent multicenter cohort.