For statistical neuroimaging analysis of source current density, eLORETA applied a statistical nonparametric mapping (SnPM) method [57]. First, the difference in source localization of cortical oscillations between groups for each frequency band was assessed by voxel-by-voxel independent sample F-ratio-tests, based on eLORETA log-transformed current density power. In the resulting statistical three-dimensional images, cortical voxels showing significant differences were identified by a nonparametric randomization/permutation procedure (i.e., based on the Fisher's permutation method, thresholded at the 5% probability level) that compares the mean source power of each voxel and the distribution of the permuted values. By estimating the empirical probability distribution for the “maximal-statistic” under the null hypothesis, randomization and permutation tests have proven to be effective in controlling the family-wise error rate (Type I error) in neuroimaging studies [58]. eLORETA used 5000 data randomizations to determine the critical probability threshold values for the actually observed log F-ratio values with correction for multiple comparisons across all voxels and all frequencies, without the need to rely on Gaussianity. The use of SnPM applied to LORETA images has been validated in several studies [59], [60]. Details on the nonparametric randomization procedure are provided in the study of Nichols and Holmes [58].
