Statistical Inferences for Complex Dependence of Multimodal Imaging Data

Statistical analysis of multimodal imaging data is a challenging task, since the involves high-dimensionality, strong spatial correlations and complex structures. In this article, we propose rigorous statistical testing procedures for making inferences on dependence data. Motivated by multi-task fMRI in Human Connectome Project (HCP) study, particularly address three hypothesis problems: (a) independence among modalities over brain regions, (b) between regions within modalities, (c) across different modalities. Considering general form all tests, develop global procedure multiple controlling false discovery rate. We study theoretical properties proposed tests computationally efficient distributed algorithm. The methods theory are relevant many problems structure components high-dimensional random vectors with arbitrary also illustrate our via extensive simulations five task contrast maps HCP study. Supplementary materials article available online.