We are testing a spatial filtering procedure to harmonize the noise across diffusion MRI datasets from many different sites and scanner vendors in a cohort of patients with different forms of Parkinsonism. The rationale is to create an algorithm that distinguishes between the types of Parkinsonism across many different diffusion MRI sequences from different vendors. If it can differentiate between normal and atypical parkinsonism, the program could help decrease the number of misdiagnoses. Using the data from a 2019 study by Dr. Vaillancourt, “Development and validation of the automated imaging differentiation in parkinsonism (AID-P): a multicentre machine learning study,” about 1,000 participants will be processed with different full-width half maximum spatial filters. We will test no filter, 2mm, 4mm, and 6mm spatial filters on the data. This includes data from 8 cohorts internationally (including University of Florida, Penn State Hershey Medical Center, Medical University Innsbruck, Northwestern University, University of Michigan, Parkinson’s Progression Marker’s Initiative, and 4 Repeat Tauopathy Neuroimaging Initiative) totaling the use of 17 different MRI scanners. Introduction of a non-invasive biomarker can help improve rates and classifications of diagnoses in future research projects and eventually for physicians in a clinical setting. If physicians are able to upload the MRI imaging to the program, no matter the resolution, the impact could be felt worldwide.