Deep Learning Analysis of Optic Nerve Morphology Associated with Parkinson's Disease and Drug-Induced Parkinsonism in the LIFE Study

A large Japanese population-based study of 14,280 participants used deep learning to analyze cup-to-disc ratios from fundus photographs as a potential imaging biomarker for Parkinson's disease and drug-induced Parkinsonism. Participants with the highest cup-to-disc ratios showed significantly elevated odds of Parkinson's disease, suggesting optic nerve morphology may aid in identifying neurodegenerative pathology.