Abstract
Early and accurate diagnosis of Parkinson’s disease (PD) remains a clinical challenge. This study explores the potential of habitat-based radiomics as a novel approach to improve PD detection using routine clinical MRI scans. We analyzed MRI data from 308 participants (173 PD patients and 135 healthy controls) to extract detailed features from segmented habitats in the caudate nucleus and putamen. Machine learning models, trained on habitat-based radiomic features, achieved a diagnostic accuracy exceeding 94%. This superior performance, compared to traditional radiomics, highlights the ability of habitat-based radiomics to capture subtle tissue heterogeneity associated with PD. Our findings suggest that habitat-based radiomics could be a valuable tool for early and accurate PD diagnosis, enabling timely intervention and improved patient outcomes.