Utilizing combined quantitative multiparametric MRI as potential biomarkers for improved early-stage Parkinson's disease diagnosis

Motivation: Improving early Parkinson's disease diagnosis using advanced MRI techniques and machine learning to detect subtle neuroanatomical and microstructural changes. Goal(s): To develop a machine learning model using quantitative MRI features to accurately diagnose early-stage Parkinson's disease. Approach: This study used QSM and DKI features, analyzed with an SVM model, to differentiate early-stage Parkinson's disease from healthy controls and advanced-stage PD. Results: The SVM model demonstrated moderate accuracy in detecting early-stage PD (accuracy: 0.78, AUC: 0.90) and high accuracy for advanced-stage PD (accuracy: 0.97, AUC: 0.97), with DKI's kurtosis feature crucial for classification. Impact: This study demonstrates the potential of combining QSM and DKI features to improve early-stage Parkinson's disease diagnosis, offering clinicians a non-invasive tool for detection. It paves the way for future research into MRI-based biomarkers for disease progression monitoring.