Evaluating the accuracy of perfusion/metabolism (SPET/PET) ratio in seizure localization

The uncoupling between brain perfusion and metabolism was evaluated as a potential tool for seizure localization by creating an interictal SPET divided by interictal PET functional ratio-image and by evaluating its sensitivity and specificity to areas subsequently surgically resected. The uncoupling between brain perfusion and metabolism was evaluated through the creation of a functional SPET/PET ratio-image relying on interictal single-photon emission computed tomography (SPET) and positron emission tomography (PET) scans in epilepsy patients. The uncoupling of these two physiological brain functions has been demonstrated to be a characteristic of epileptogenic tissue in temporal lobe epilepsy and could potentially serve as a diagnostic measure for localization of seizure onset areas in the brain. The accuracy of hemispheric localization, sensitivity, and specificity of perfusion to metabolism ratio-images were evaluated as compared to standard methods of PET reading.

Interictal HMPAO-SPET and FDG-PET scans were obtained from 21 patients who then went on to remain seizure free for a minimum of 1 year post surgical resection. Using Statistical Parametric Mapping (SPM2), the SPET and PET scans were spatially registered and spatially normalized to a standard template (geometric warping). A functional image was created by calculating the ratio of perfusion to metabolism. Discrete areas of uncoupling in the ratio-images were selected, quantified, and compared to visually interpreted PET readings as well as the actual site of subsequent surgical resection. Localization was determined by comparing the hemispheric location of these areas to sites of surgical resection. Sensitivity and specificity of ratio-images and PET readings were calculated by dividing the brains into four sections per hemisphere.

When compared to known sites of successful surgical resection, the pre-surgical visually interpreted PET readings had a correct hemispheric localization in 69.6% of cases, while the regions of uncoupling selected in the pre-surgical ratio-images had a correct hemispheric localization of 82.6%. In addition, the regional sensitivity of visually interpreted PET readings was 63.0% with a specificity of 95.7%, while the sensitivity of the ratio-images was 68.0% with a specificity of 96.0%.

Compared to the PET readings, the ratio-images yielded similar sensitivity and specificity measures, but had an improved hemispheric localization. Hence, ratio-images may be a valuable diagnostic tool in the hemispheric localization, which could enhance the use of PET readings alone.