Quantitative relaxation parameters of subcortical nuclei based on SyMRI can help identify different motor subtypes of Parkinson's disease

Motivation: Pakinson subtypes show significant heterogeneity, yet underlying pathological mechanisms remain unclear. Goal(s): To assess differences in subcortical nuclei parameters via SyMRI across PD motor subtypes and develop an interpretable model for subtype differentiation. Approach: Quantitative parameters of basal ganglia nuclei were extracted and compared among PD subtypes and HC, followed by predictive modeling and SHAP-based interpretability analysis. Results: PIGD subtype showed more extensive SyMRI parameter alterations, especially in T1 and T2, than the TD subtype. AUC of the SVM model was 0.799 in the training set and 0.739 in the validation set. SHAP analysis helps explain the internal decision-making process of the model. Impact: SyMRI can detect microscopic brain damage across PD subtypes, with specific relaxation parameters offering potential as neuroimaging biomarkers to distinguish difference subtype of PD.