Meeting Banner
Abstract #2368

Diagnosis of Schizophrenia using CBF Measures as a Classification Feature A FBIRN Phase 3 Multisite ASL Study at 3T

David Shin 1 , Burak Ozyurt 1 , Jerod Rasmussen 2 , Juan Bustillo 3 , Theodorus Van Erp 2 , Jatin Vaidya 4 , Daniel Mathalon 5 , Bryon Mueller 6 , James Voyvodic 7 , Douglas Greve 8 , Judith Ford 5 , Gary Glover 9 , Gregory Brown 1 , Steven Potkin 2 , and Thomas Liu 1

1 University of California, San Diego, La Jolla, CA, United States, 2 University of California, Irvine, Irvine, CA, United States, 3 University of New Mexico, Albuquerque, NM, United States, 4 University of Iowa, Iowa City, IA, United States, 5 University of California, San Francisco, San Francisco, CA, United States, 6 University of Minnesota, Twin Cities, Minneapolis, MN, United States, 7 Duke University, Durham, NC, United States, 8 Harvard Medical School, Massachusetts General Hospital, Charlestown, MA, United States, 9 Stanford University, Stanford, CA, United States

Resting state CBF maps (n=234) collected from the multisite FBIRN Phase 3 Study ( were used as learning features in a support vector machines (SVM) classification technique to evaluate its performance in differentiating Schizophrenic patients (SCZ) from healthy controls (CNT). For feature extraction, clusters of voxels with group differences (SCZ vs. CNT) were first identified (t-test, p<0.01) and the mean CBF values across these clusters were used as the training data. The sensitivity, specificity, and accuracy for the leave-one-out cross validation were found to be 75.9%, 75.5%, and 75.7%, respectively. The results suggest that the CBF map acquired from a 5-minute ASL scan combined with SVM may be a useful complimentary tool for diagnosis of SCZ.

This abstract and the presentation materials are available to members only; a login is required.

Join Here