Fig. 8.

Distribution of the number of subjects over number of elapsed volumes until auto-calibration was successful with a threshold of one-fourth of the slice thickness on the HBN data set. This analysis showed a heavy-tailed distribution with a heavy concentration around 1, which indicated that for the majority of subjects (~ 200 subjects) on the HBN data set, the first volume was successfully chosen as the reference volume; whereas for a few subjects, that apparently moved continuously and significantly, it could take between 15 to 70 volumes before a “motion-free ” reference volume could be obtained. According to our fMRI protocol, for 99.5% of the subjects the auto-calibration time was less than 30 seconds; and only for two cases (among 251 subjects) the auto-calibration took 90–110 s. In addition to showing the efficacy of our auto-calibration method, this analysis is another evidence for the necessity of real-time motion monitoring as it indicates without real-time motion monitoring there is no guarantee that fMRI of sufficient quality is obtained for every subject in a cohort.