Automated normalized FLAIR imaging in MRI-negative patients with refractory focal epilepsy

Patients with focal epilepsy that is refractory to medical treatment are often considered candidates for resective surgery. Magnetic resonance imaging (MRI) has a very important role in the presurgical work-up of these patients, but is unremarkable in about one-third of cases. These patients are often deferred from surgery or have a less positive outcome if surgery is eventually undertaken. The aim of this study was to evaluate our recently described voxel-based technique using routine T2-FLAIR (fluid-attenuated inversion-recovery) scans in MRI-negative patients and to compare the results with video-EEG (electroencephalography) telemetry (VT) findings.

We identified 70 epilepsy patients with refractory focal seizures who underwent VT and had a normal routine MRI. T2-FLAIR scans were bias-corrected, and intensity and spatially normalized (nFSI) using Statistical Parametric Mapping 5 (SPM5) as previously described. Individual scans were then compared against a set of 25 normal controls using a voxel-based method.

SPM5 identified 10 patients with suprathreshold clusters (14.3%). In 50% of these there was concordance between the lobe of the most significant cluster and the presumed lobe of seizure onset, as defined by VT. All cases were concordant with respect to lateralization of the putative focus.

Using nFSI we identified focal structural cerebral abnormalities in 11.4% of patients with refractory focal seizures, and normal conventional MRI, that were fully or partially concordant with scalp VT. This voxel-based analysis of FLAIR scans, which are widely available, could provide a useful tool in the presurgical evaluation of epilepsy patients. Ongoing work is to compare these imaging findings with the results of intracranial EEG and histology of surgical resections.