An ensemble machine learning classifier for Parkinson's disease diagnosis using optical coherence tomography...

Area of Science:

• Ophthalmology
• Neurology
• Medical Imaging

Background:

• Parkinson's disease (PD) diagnosis is challenging due to a lack of early biomarkers.
• Retinal microvascular alterations detected via OCTA are emerging as potential non-invasive PD biomarkers.
• Objective, non-invasive tools are crucial for timely PD detection and intervention.

Purpose of the Study:

• To investigate the utility of OCTA-derived retinal vascular features for classifying Parkinson's disease.
• To develop and validate a machine learning model for PD detection using retinal imaging data.

Main Methods:

• Retrospective analysis of OCTA images from 53 PD patients and 39 healthy controls.
• Segmentation of superficial and deep vascular complexes to extract 22 quantitative features.
• Application of a multi-stage feature selection framework and ensemble machine learning models (XGBoost, Random Forest, KNN).

Main Results:

• An ensemble model achieved 74.28% accuracy, 90% sensitivity, 53.33% specificity, and 0.75 AUC on an independent test set.
• Key features included morphological descriptors (form factor, convexity, solidity, roundness) and vascular density (VAD, VSD).
• A user-friendly graphical interface (PDAI) was developed for clinical application.

Conclusions:

• OCTA-based retinal vascular analysis offers a promising non-invasive approach for Parkinson's disease classification and screening.
• The developed AI framework supports early PD detection, requiring further validation in larger, multi-center studies.