Transformation Invariant Control of Voxel-Wise False Discovery Rate.

ABSTRACT: Multiple testing for statistical maps remains a critical and challenging problem in brain mapping. Since the false discovery rate (FDR) criterion was introduced to the neuroimaging community a decade ago, many variations have been proposed, mainly to enhance detection power. However, a fundamental geometrical property known as transformation invariance has not been adequately addressed, especially for the voxel-wise FDR. Correction of multiple testing applied after spatial transformation is not necessarily equivalent to transformation applied after correction in the original space. Without the invariance property, assigning different testing spaces will yield different results. We find that normalized residuals of linear models with Gaussian noises are uniformly distributed on a unit high-

SUBMITTER: Li J 

PROVIDER: S-EPMC5052119 | biostudies-literature | 2016 Oct

REPOSITORIES: biostudies-literature