Skip to Main Content
Previous work has demonstrated that fMRI activations in individual anatomic regions can be used as a basis for classifying subjects for schizophrenia. Here we demonstrate that this classification, which is approximately 75-83% accurate for the hippocampal formation, can be improved by incorporating hippocampal volume information, which alone classifies at chance levels of accuracy. The best combined classifier is about 87% accurate and uses only right hippocampal activation and volume data. These accuracy figures should be considered approximate due to the limited size of our test dataset (N=23, 15 patients and 8 controls).