Abstracts

Rationale: While medial temporal lobe epilepsy is relatively well detected by a variety of imaging approaches, neocortical epilepsy remains a challenge, particularly if the structural MRI is non-lesional. We recently demonstrated the substantially improved signal-to-noise ratio at ultra high field (7T), which may improve the detection of metabolic injury by MR spectroscopic imaging in neocortical epilepsy. Methods: We used a Varian/Magnex head only 7T MR system with a specialty built 8x1 transceiver array (9cm long, 23cm ellipitical diameter) for transmission and reception. Neocortical epilepsy patients whose seizure semiology were felt to be frontal were studied using a 3D spectroscopic imaging sequence, acquired with a 16x16x8 spherical encoding strategy at moderate echo sequence (TR/TE 3s/42ms), FOV 192x192x50mm, resulting in a nominal voxel size of 0.9cc. Including setup time the duration of the study ranged between 70 to 90min. Patients whose semiology were felt to be medial temporal lobe were studied in single slice mode, acquired with a 24x24 rectangular sampling at TR/TE 1.5s/40ms, FOV 192x192. In all acquisitions, data were processed by sum-of-squares summation between all 8 array elements and phased using a rapid water SI acquisition. All measurements target the detection of NAA/Creatine, as NAA has been extensively used to detect neuronal mitochondrial dysfunction while creatine is found in both astrocytes and neurons. It is worth noting that as a ratio, NAA/Cr is relatively independent of tissue loss and atrophy; thus declines are suggestive of neuronal injury and dysfunction. Results: Fig. 1 shows data from a neocortical epilepsy patient who underwent intracranial EEG monitoring to identify the R frontal lobe as the region of seizure onset, consistent with the area of NAA/Cr depression seen on the MRSI. On pathology, this patient was found to have a focal cortical dysplasia with balloon cells, and became seizure free after surgery. A MTLE patient is shown for comparison in Fig. 2. While the decline in NAA/Cr may seem somewhat milder in the neocortical patient than in the MTLE patient this may be attributable to partial volume effect, as the neocortical study was acquired with a larger voxel size (nominal size 0.9cc) than the MTLE patient (nominal size 0.64cc). Conclusions: These data show the feasibility of ultra high field spectroscopic imaging to detect the metabolic abnormalities in neocortical epilepsy. It is evident that the challenge in neocortical epilepsy lie in the needed large territory of coverage with potentially smaller metabolic changes. In this case, the high signal-to-noise available at 7T will be important for successful localization.