Abstracts

Rationale: Ultra-high field MR (7T and above) holds significant promise for MR spectroscopy, expected to improve because of the field-dependent (at least) linear increase signal to noise (SNR). In epilepsy, 31P MRSI at 3 and 4T has been limited by SNR and thus performed with typically 15-30cc resolution (or greater), lateralizing medial temporal lobe epilepsy with ~70% accuracy. Thus for epilepsy, 31P MRSI may be expected to particularly benefit from the increased field strength. We describe initial 7T studies evaluating 31P MRSI to localize the region of seizure onset. Methods: We used a Varian-Magnex 7T Direct drive human head only MR system with a double tuned 31P-1H volume TEM coil for detection. Pulse acquire spectroscopic imaging was performed with a cosine weighted spherical sampling strategy to produce a full width half maximum volume of ~7.5cc. Studies were performed with angulation along the planum temporale in order to maximize consistency over the temporal lobes. 3D inversion recovery gradient echo images (TR/TE 3000/4) were acquired to correlate with MRSI. Image guided single voxel reconstruction and visual inspection of the entire SI was used for identification of abnormalities. Spectral curve fitting was performed using nine 31P metabolite resonances. Eight control volunteers and seven patient volunteers (n=3 suspected MTLE, n=4 neocortical or dual pathology) undergoing pre-surgical evaluation were recruited. Only patients who had not had a seizure 48hrs prior to study were included. Results: By using a combined measure of PCr/ATP and an asymmetry index in comparison to controls, we identified regions of abnormal PCr/ATP in 5/7 of the epilepsy patients studied thus far that is concordant with other data (Table). The remaining 2/7 patient data is of borderline significance primarily because of limited available control data. An example is shown in the figure, a patient with possible dual pathology. In this case the medial temporal lobes were not significantly abnormal (although they were mildly depressed), but the L temporal region was significantly abnormal. Conclusions: Ultra-high field 31P MRSI provides a whole brain energetic assessment which may be informative for localizing the region of seizure onset. The higher SNR available at 7T (which gives sensitivity comparable to 1H at 3T, although the 31P metabolite concentrations are much lower) can be used towards improved volume resolution. We have implemented 7T 31P MRSI in controls and epilepsy patients, and find that it may be informative particularly for neocortical epilepsy.