Motivation: Susceptibility map-weighted imaging (SMwI) aids Parkinson's disease (PD) diagnosis by detecting abnormalities in nigrosome 1 (N1) of the substantia nigra, but its accurate interpretation remains challenging. Goal(s): We assessed the effect of deep learning-based AI software on the diagnostic performance of PD Approach: Four radiologists retrospectively read SMwIs of PD patients and normal cohorts with and without AI software, which evaluated PD risk based on N1 abnormalities. Diagnostic performance metrics for the radiologists were calculated. Results: With AI, radiologists' performance enhanced across all experience levels, with significant increases in specificity, positive predictive value, and accuracy in three readers, and AUROC in two readers. Impact: Deep learning-based AI improves the diagnostic accuracy of nigrosome 1 abnormalities on Susceptibility map-weighted imaging (SMwI) for Parkinson's disease across all radiological experience levels, which may extend its role in clinical decision-making for PD diagnosis.
Park et al. (Tue,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: