Andrew Mario Michael1,2, Stefi A. Baum2, Vince P. Clark1, Rex Jung1, Kelvin O. Lim3, Tonya White3, Beng C. Ho4, Randy L. Gollub5, Vince D. Calhoun1
1MIND Research Network, Albuquerque, NM, USA; 2Center for Imaging Science, Rochester Institute of Technology, Rochester, NY, USA; 3University of Minnesota, Minneapolis, MN, USA; 4University of Iowa, Iowa City, IA, USA; 5Psychiatry and Radiology, Massachusetts General Hospital, Charlestown, MA, USA
Data fusion approaches can help to find hidden traits in complex disorders such as schizophrenia. We examine all possible combinations of inter-voxel cross-correlations between structural and functional data and where these correlations show patient/control differences. We present efficient approaches to compute the structure-function correlations and to evaluate their statistical properties. We find that cross-correlations between gray matter concentration and functional MRI data from a sensorimotor task are significantly lower in patients. The cerebellum showed more positive correlations with functional data in controls versus patients and the cingulum showed more negative correlations in patients.