Dynamic assessment of high-resolution MRI with multi-planar reconstruction increases the yield of lesion detection in patients with partial epilepsy

Abstract

Objective: To investigate the presence and type of lesions associated with partial epilepsies by routine high resolution MRI and multi-planar reconstruction (MPR) and correlate the MRI abnormalities with semiology and EEG findings. Methods: We studied 100 consecutive patients followed in the epilepsy clinic of our Hospital with partial epilepsy who underwent MRI investigation. The MRI protocol included 6 mm sagittal T1-weighted, 3-4 mm axial T1 and T2-weighted, 3 mm coronal T1 inversion recovery and T2-weighted images that were printed on a radiographic film for routine analysis. In addition, all patients had a volume T1-gradient echo acquisition with isotropic voxels (1-1.5 mm) for multiplanar reconstruction (MPR). The MRIs were examined in two different occasions: first using only the images printed on films, without volume T1-gradiente echo acquisition and in a second occasion in a computer workstation when all the available images and MPR were analyzed blindly to the clinical information. The clinical and EEG findings were tabulated independently, and results were compared using Chi-square of Fisher exact test when appropriate. Results: The patients were divided into 10 groups according to their etiological classification (structural lesions) established by MRI. Mesial temporal sclerosis (MTS) was the largest group (40%). There were 65 women and 35 men. Mean age was 23.9 (± 5.7) years and mean age of onset of recurrent seizures was 9.9 (± 0.8) years. The most frequent risk factors were family history of seizures (23%), head trauma (10%), peri-natal anoxia (5%) and infection (9%). High resolution MRI including thin coronal slices, in addition to a "dynamic" analysis in a workstation with MPR, allowed a significant improvement in lesion detection compared to the traditional analysis with radiographic films (94% versus 80%) (p < 0.05). The lesions previously undetected were focal cortical dysplasia and subtle MTS. There was a good concordance between MRI lesions and clinical and EEG findings. Conclusion: High resolution MRI including thin coronal slices, in addition to a "dynamic" analysis in a workstation with MPR allowed a significative improvement in lesion detection compared to the traditional analysis with radiographic films (94% versus 80%). Patients with partial epilepsy and "normal" MRI need to be investigated further with thin slices and post-processing techniques using volume acquisitions that allow adequate multi-planar re-slicing.