Automatyczna analiza objętości podstruktur hipokampa przy użyciu metody AID-HS w napadach padaczki lobu skroniowego: porównanie z wynikami badań mikroskopowych

PURPOSE: To evaluate the diagnostic accuracy of the AID-HS pipeline in characterizing hippocampal subfield volumetric patterns and distinguishing HS subtypes, using histopathology as the reference standard in mesial temporal lobe epilepsy (mTLE). METHODS: Thirty patients with surgically treated mTLE underwent 3.0 T MRI. Volumetric subfield analysis was performed using the automated AID-HS tool.Hippocampal subfield volumes were compared between typical HS type 1 and atypical HS, including HS types 2 and 3, with validation against the 2013 ILAE histopathological criteria. Subfields of the sclerotic hippocampus were categorized as volume-reduced or volume-preserved based on the asymmetry index. RESULTS: HS type 1 was predominant, accounting for 83.3% of cases. In typical HS type 1, volume reduction in CA2 and CA3 was more frequently observed than in atypical HS on volumetric comparison, and MRI-detected volume reduction in CA2, CA3, and CA4 was significantly associated with histopathological sclerosis (p < 0.01) . Diagnostic accuracy showed fair to good classification performance for distinguishing HS subtypes, with AUC values ranging from 0.73 to 0.83. The CA3 subfield achieved the highest accuracy, with an AUC of 0.83 and 100% sensitivity. CONCLUSION: The AID-HS pipeline may provide quantitative support for subtype-oriented evaluation of HS.