Abstracts

Rationale: Single-Photon Emission Computed Tomography (SPECT) has become an important non-invasive method for localizing the seizure onset zone in epilepsy surgery candidates. Subtraction ictal SPECT with coregistration to MRI (SISCOM) has been demonstrated to be superior to standard visual analysis, and is now considered by most experts to represent the state-of-the-art analysis for ictal SPECT (O’Brien et al, 1998). Statistical Parametric Mapping (SPM) is an increasingly accepted method for analysis of functional imaging data, including SPECT. ISAS (Ictal-interictal SPECT Analysis using SPM) is an SPM-Based technique for ictal-interictal SPECT analysis (McNally et al, 2005). ISAS utilizes a group of 14 normal scan pairs to identify and assess the significance of perfusion changes between a patient s paired interictal and ictal SPECT scans. We sought to compare the impact of these two analysis techniques with regard to the conclusiveness of localization and the inter-rater reliability.Methods: In 57 epilepsy surgery patients an ictal scan was compared with interictal scan using both techniques. SISCOM was performed using Analyze 4.0 (O’Brien et al, 1998), as well as ISAS as previously outlined (McNally et al, 2005, also see http://spect.yale.edu). Three blinded reviewers (LP, BO, KL) rated both sets of output images with regard to the binary measure; “1”-study is conclusively localizing, or “0”-study is inconclusive. Reliability analysis was performed yielding the average interclass correlation coefficient between the three reviewers for each technique (SISCOM and ISAS). The proportion of conclusive studies from each technique was tabulated and the significance of the difference calculated.Results: The average interclass correlation coefficient between the three reviewers for SISCOM was 0.6641, while that for ISAS was 0.8811. The lower limit 95% CI for ISAS was 0.8155 while the upper 95% CI for SISCOM was 0.7911, demonstrating the difference in inter-rater reliability as significant. The proportion of conclusively localizing studies from SISCOM was 53 out of 171 (all 3 reviewers), while the yield from ISAS was 88 out of 171. The significance of this difference has a calculated value of p<0.000121.Conclusions: These results demonstrate that ISAS produces a significantly higher proportion of conclusively localizing results. The inter-rater reliability is significantly higher with ISAS than with SISCOM. These results also suggest that ISAS may produce a higher sensitivity and specificity results with regard to localization.