Abstracts

Rationale: Voxel-wise analysis using Statistical Parametric Mapping (SPM) is becoming an increasingly important tool in functional imaging for localizing the seizure onset zone. A major improvement of SPECT has been in the techniques of subtraction ictal SPECT co-registered to MRI (e.g. SISCOM). We have developed a technique for determining the seizure onset zone by using SPM to analyze the subtraction ictal SPECT of intractable temporal lobe epilepsy patients (TLE), based on the normal expected variance in control subjects. Methods: 106 (53 females) consecutive patients who had undergone pre-operative 99mTc-ECD SPECT imaging and temporal lobectomy were identified from our TLE database from 1998 to 2005. Voxel-wise SPM analysis (SPM2, Wellcome Department of Neurology, UK) was performed on each patient’s subtraction SPECT image (Analyze-Mayo Foundation) and compared with a SPECT dataset of control subjects who each had two baseline SPECT studies. (We termed this technique 'STATISCOM' - i.e. Statistical Ictal Subtraction SPECT Co-registered to MRI). Two blinded reviewers then compared these images with those of SISCOM for each patient. The following were determined: interobserver reliability, rate of localization for TLE, and rate of correct localization for TLE subtypes (i.e. mesial vs. neocortical). Results: Based on clinical history, preoperative MRI, prolonged scalp & intracranial EEG, we identified 48 mesial TLE patients, 45 neocortical TLE patients, and 13 with indeterminate TLE sub-type. Median age of patients was 36 years (range 13-72) and latency of ictal SPECT injection was 26 seconds (range 7-166). Interobserver agreement for STATISCOM-generated results were superior to those of SISCOM (85%, kappa 0.77 vs. 60.2%, kappa 0.36; p<0.0001). STATISCOM localized abnormal perfusion in the temporal lobe in 78 patients, 75 (97%) of whom were correctly localizing. With SISCOM, only 59 patients had localized temporal lobe perfusion changes, of whom 48 (81%) were correct. However, STATISCOM-generated results correctly distinguished between mesial and neocortical subtypes of TLE in 67% of patients, compared with 35% by SISCOM (p<0.005). In all patients (n=8) with no region of STATISCOM-identified hyperperfusion, the post-lobectomy outcome was poor.Conclusions: ‘STATISCOM’ is a novel method that may be superior to SISCOM in localizing the temporal lobe seizure onset zone in medically intractable epilepsy patients. Moreover, STATISCOM can also identify patterns of ictal SPECT perfusion that may have prognostic value for temporal lobectomy outcome. (Sources of funding - NHMRC Postgraduate Scholarship, RACS Foundation Award)