Related Experiment Video
Updated: May 5, 2026
09:33
An Acute Retinal Model for Evaluating Blood Retinal Barrier Breach and Potential Drugs for Treatment
Published on: September 13, 2016
6.6K
Interpretable Machine Learning Predictions of Bruch's Membrane Opening-Minimum Rim Width Using Retinal Nerve Fiber
Sat Byul Seo1, Hyun-Kyung Cho2,3
1Department of Mathematics Education, Kyungnam University, 7 Kyungnamdaehak-ro, Changwon-si 51767, Geongsangnam-do, Republic of Korea.
Bioengineering (Basel, Switzerland)
|March 28, 2025
Summary
This study predicts Bruch's membrane opening-minimum rim width (BMO-MRW) using optical coherence tomography (OCT) and visual field (VF) data. Machine learning accurately estimated BMO-MRW, aiding glaucoma diagnosis when OCT is unavailable.
Area of Science:
- Ophthalmology
- Medical Imaging
- Machine Learning
Background:
- Bruch's membrane opening-minimum rim width (BMO-MRW) is a novel glaucoma parameter.
- Conventional optical coherence tomography (OCT) and visual field (VF) tests provide structural and functional data.
- Predicting BMO-MRW without direct measurement is clinically valuable.
Purpose of the Study:
- To develop and validate a machine learning model for predicting BMO-MRW.
- To integrate retinal nerve fibre layer (RNFL) thickness and VF global indexes (MD, PSD, VFI) for BMO-MRW prediction.
- To assess the contribution of different parameters to BMO-MRW prediction.
Main Methods:
- Development of an interpretable machine learning model.
- Integration of OCT-derived RNFL thickness and VF global indexes.
- Utilizing SHAP (SHapley Additive exPlanations) for parameter analysis.
Main Results:
- The model achieved high predictive accuracy for BMO-MRW, with R² values of 0.68 (inferotemporal), 0.67 (global), and 0.64 (superotemporal).
- RNFL parameters were significant predictors of BMO-MRW.
- Age and PSD were identified as critical contributing factors.
Conclusions:
- Machine learning models integrating structural and functional data can accurately predict BMO-MRW.
- This approach offers valuable clinical insights for glaucoma management, especially when BMO-MRW is not directly measurable.
- The model shows potential for enhancing glaucoma diagnosis and disease progression monitoring.
