Diagnostic Accuracy of Epilepsy-dedicated MRI with Post-processing

Horst Urbach1, Christian Scheiwe2, Muskesh J Shah2

  • 1Dept. of Neuroradiology, Medical Center, University of Freiburg, Breisacher Str. 64, 79106, Freiburg, Germany. horst.urbach@uniklinik-freiburg.de.

Abstract

Insights

This study shows that 3 Tesla MRI with MP2RAGE is highly accurate for diagnosing drug-resistant focal epilepsy. While some subtle lesions may be missed, the protocol significantly aids in surgical planning and predicting seizure outcomes.

Area of Science:

  • Neurology
  • Radiology
  • Pathology

Background:

  • Drug-resistant focal epilepsy poses significant challenges in diagnosis and treatment.
  • Accurate pre-surgical localization of epileptogenic foci is crucial for successful surgical outcomes.

Purpose of the Study:

  • To assess the diagnostic accuracy of 3 Tesla MRI, utilizing MP2RAGE sequence and MAP post-processing, in patients with drug-resistant focal epilepsy.
  • To correlate MRI findings with histopathology and postsurgical seizure outcomes.

Main Methods:

  • 116 consecutive patients with drug-resistant focal epilepsy underwent 3 Tesla MRI with MP2RAGE sequence.
  • Morphometric analysis program (MAP) was used for post-processing MRI data.
  • MRI findings were compared with histopathology reports and postsurgical seizure outcomes.

Main Results:

  • MRI and histopathology concordance was observed in 101 out of 116 patients.
  • Three lesions (hippocampal sclerosis/gliosis only) were MRI-negative.
  • Focal cortical dysplasia (FCD) type II was accurately identified in most cases, with one case misclassified as glial scar.
  • Surgery led to seizure freedom (Engel class IA) in 71% of patients.

Conclusions:

  • The 3 Tesla MRI protocol with MP2RAGE and MAP post-processing demonstrates high diagnostic accuracy for drug-resistant focal epilepsy.
  • Subtle hippocampal sclerosis/gliosis lesions may not be detected by MRI.
  • Discrepancies between MRI and histopathology for FCD may be attributed to tissue sampling errors.