Abstracts

Accurate localization of the epileptogenic zone prior to epilepsy surgery is often not possible with existing noninvasive methods. Novel noninvasive methods of localization are needed for such situations. We investigated the value of resting functional MRI (fMRI) with temporal clustering analysis for correct localization of the epileptogenic zone.
We investigated two patients groups: 6 patients with temporal lobe epilepsy, who subsequently became seizure-free or almost seizure-free after temporal resection, and 3 patients with refractory suspected extratemporal epilepsy, who also had a complete presurgical evaluation. Functional MRI scanning was performed on baseline antiepileptic drugs. Images were acquired over several minutes with the patients awake but with eyes closed. This resulted in a 4-D dataset of 200 image volumes over time. A histogram of the number of voxels in the volume whose signal intensity reached a maximum at each time point was created (temporal clustering technique). Points in the histogram over 100 voxels reaching maximum were considered relevant to interictal or subclinical ictal activation of the epileptogenic zone. These times were used to create an activation map using SPM99 software. For patient set 1, the hemisphere of the hippocampus with highest peak activation (T-value) was compared to the side of resection. For patient set 2, the activation maps were compared to results of standard presurgical assessments.
The histograms for each of the TLE patients demonstrated at least one peak. In all patients the hippocampus on the side of resection showed the highest correlation with the peaks of the histogram (p=0.005). The extratemporal patients demonstrated activations in regions consistent with the clinical semiology and the overall conclusions of the presurgical evaluations. All patients had extratemporal frontal seizure semiology and poorly localizing or misleading ictal and interictal EEG data. The first two patients had nocturnal sleep related seizures with tonic posturing and no alteration of awareness. The fMRI analysis indicated frontal regions of activation consistent with seizure semiology and not demonstrated by other tests. The third patient had nocturnal hypermotor seizures. MRI was normal and EEG, PET, and ictal SPECT data favored right temporal localization. The fMRI analysis demonstrated several extratemporal areas of activation that invasive recordings confirmed to be an epileptic network activated with every seizure. These regions were also not demonstrated by other noninvasive tests.
Resting fMRI with temporal clustering analysis demonstrates excellent potential for localization of the epileptogenic zone. It can be used to evaluate the whole brain with relatively high, 3-dimensional spatial resolution in less than one hour. This non-invasive test deserves more investigation to potentially replace invasive procedures in the presurgical evaluation for some patients.