Fig. 2.

Percent change in predicted “Significant” area by number of bootstrapped iterations. Within this dataset, and using a cutoff of 95 and 90 % voxel-wise confidence interval, area predicted to be “above threshold” stabilized markedly over the first 100 iterations. Approximately 350 iterations were required to stabilize predictions to an arbitrarily defined baseline with change of <+/−5 %. Arrow marks approximate lower limit of iterations among “left out” subjects