The overall goal of this proposal is to identify a robust magnetic resonance imaging (MRI) marker for posttraumatic stress disorder (PTSD), resulting in improved diagnosis of this condition and assessment of treatment interventions. Currently, an accurate diagnosis of posttraumatic stress disorder (PTSD) is extremely difficult because the assessment is exclusively based on symptoms and often confounding brain conditions with similar symptoms, such as traumatic brain injury are present. Multimodal MRI studies of PTSD, including structural, functional, perfusion, and diffusion imaging, have revealed multiple brain alterations in PTSD that hold promise as potential markers for this condition. So far, however, all studies of PTSD analyzed the multimodal MRI measures separately, ignoring relationships between the measures that might carry important diagnostic information. Similarly, regional relationships between brain alterations that could reflect the anatomical footprint of PTSD have not been exploited, especially not for multimodal MRI. Moreover, the ability to identify individuals with PTSD based on multimodal MRI has not been established. This proposal aims at resolving all these issues. Our aims are threefold: First, we will develop a novel multivariate statistical framework that maximally exploits information from multimodal brain MRI in PTSD by jointly analyzing structural, perfusion and diffusion MRI data as well as their spatial relationships. Second, we will develop a novel approach in image analysis to associate MRI data from individuals with the chance that they have PTSD. Third, we will test the new methods first on a group of clinically well-characterized veterans with (N=20) and without (N=20) PTSD, who participated in a pilot MRI study at 4Tesla before validating the methods on another cohort of veterans with (N=20) and without (N=20) PTSD, who will be recruited as part of a funded biomarker study by Dr. Marmar. We predict that the new analysis algorithms for multimodal MRI will provide superior diagnostic ability for PTSD relative to currently used methods and also will facilitate much more robust classifications of individuals with PTSD. Furthermore, since the methods developed in this project are general, they also will be useful for MRI studies of other brain disorders, including traumatic brain injury, Alzheimer’s disease and Parkinson’s disease.