Comparison of PET [15O]water studies with 6-minute and 10-minute interscan intervals: single-subject and group analyses.

TLDR: Analysis of results suggests that with a small increase in the number of scans and the use of a 6-minute interscan interval, a comparable level of statistical significance may be achieved for single-subject experiments that use a local variance estimate, with an overall shortening of the study duration.

Abstract: The authors recently showed that [15O]water PET data obtained with a short interscan interval (6 minutes) produced similar results whether or not the residual background from the previous scan is subtracted. The purpose of the present study was to compare scans obtained during motor activation using a short (6-minute) interscan interval protocol with those obtained with a standard (10-minute) protocol in the same scanning session. Single-subject and group analyses were performed using Worsley's method, which uses a pooled variance estimate and statistical parametric mapping with a local variance estimate. High consistency in both the activation maps, i.e., the number of activated motor brain structures and the Talairach coordinates of peak intensities of the activated regions, was obtained in the 6- and 10-minute studies in both single-subject and group analyses. However, in comparison to the 6-minute studies, a larger cluster size of activated brain regions and an approximately 20% higher peak activation in these regions were observed in the 10-minute studies with the same number of replicates. Analysis of these results suggests that using a 6-minute interval with an increased number of replications, i.e., without changing the subject's total study duration, should produce comparable statistical power to that of the 10-minute interval for group analysis and increased statistical power for single-subject analyses that use a local variance estimate because of increased degrees of freedom. Alternatively, with a small increase in the number of scans and the use of a 6-minute interscan interval, a comparable level of statistical significance may be achieved for single-subject experiments that use a local variance estimate, with an overall shortening of the study duration.