PROJECT SUMMARY/ABSTRACT Parkinson's disease (PD) is a neurodegenerative disease that leads to the abnormalities of patients' movement and other functions. The current clinical diagnosis of PD cannot accomplish the desired accuracy. Imaging biomarkers of PD has shown great promising in improving the diagnosis accuracy. Our co-investigators developed a high spatial resolution diffusion MRI, which can improve the spatial resolution substantially and diminish geometric distortion. Pioneer studies conducted by our co-investigators identified changes in left side of substantia nigra of right-handed patients with PD. However, the potential of the data has not been fully realized due to the lack of appropriate statistical methods for this new type of data. This proposal aims to develop new statistical and computational tools to identify imaging biomarkers via parameters in the continuous time random walk (CTRW) model using the high resolution MRI data. Existing statistical methods for the CTRW model did not take the advantage of the high resolution data. The proposed statistical methods will perform high dimensional inference to integrate the information from a large number of pixels in the MRI data to achieve the power that cannot be attained by conventional low dimensional methods. This proposal has two specific objectives (1 ): develop high dimensional statistical inference methods for the CTRW model using high spatial resolution diffusion MRI; (2): integrate patients' clinical characteristics, such as disease duration, and neurological test scores and relevant biological variables such as age and sex, with imaging biomarkers in improving the diagnosis accuracy for PD. The developed statistics methods will be applied to diffusion MRI data sets of PD patients collected by co-investigators and their collaborators. User friendly software and computational tools will be made available for public use.