Related Experiment Video
Updated: Jul 12, 2025

Assessing Early Stage Open-Angle Glaucoma in Patients by Isolated-Check Visual Evoked Potential
Published on: May 25, 2020
Deep learning visual field global index prediction with optical coherence tomography parameters in glaucoma patients
Dongbock Kim1, Sat Byul Seo1, Seong Joon Park1
1Department of Mathematics Education, School of Education, Kyungnam University, 7 Kyugnamdaehak‑ro, Masanhappo‑gu, Changwon, Gyeongsangnam-do, 51767, Republic of Korea.
A deep-learning model accurately predicts visual field indexes (MD, PSD, VFI) from optical coherence tomography (OCT) scans. This AI tool aids glaucoma diagnosis and monitoring, especially when visual field testing is impractical.
Area of Science:
- Ophthalmology
- Artificial Intelligence
- Medical Imaging
Background:
- Glaucoma diagnosis and monitoring rely on visual field (VF) testing and optical coherence tomography (OCT).
- Predicting key VF global indexes (mean deviation, pattern standard deviation, visual field index) from OCT parameters can streamline patient management.
- Current VF testing is time-consuming and subjective, necessitating alternative diagnostic approaches.
Purpose of the Study:
- To develop and evaluate a deep-learning model for predicting visual field (VF) global indexes from optical coherence tomography (OCT) parameters.
- To assess the accuracy of the deep neural network (DNN) in estimating mean deviation (MD), pattern standard deviation (PSD), and visual field index (VFI).
Main Methods:
- A deep neural network (DNN) algorithm was trained using OCT parameters, including Bruch's Membrane Opening-Minimum Rim Width (BMO-MRW) and retinal nerve fiber layer (RNFL) thickness.
- The study included a diverse cohort of eyes with glaucoma suspects and various types of glaucoma (NTG, PACG, PEXG, POAG).
- Model performance was evaluated using mean absolute error (MAE) and Pearson's correlation coefficients.
Main Results:
- The DNN model demonstrated strong predictive capabilities for VF indexes: MAE ranged from 1.9-2.9 dB for MD, 1.6-2.0 dB for PSD, and 5.0-7.0% for VFI.
- Pearson's correlation coefficients indicated good agreement between predicted and actual VF values, ranging from 0.76-0.85 for MD, 0.74-0.82 for PSD, and 0.70-0.81 for VFI.
- The model achieved high accuracy in predicting VF global indexes across different glaucoma classifications.
Conclusions:
- The developed deep-learning model shows significant potential for aiding in the diagnosis and follow-up management of glaucoma.
- This AI-driven approach offers a valuable alternative for predicting visual field status, particularly when traditional VF testing is not feasible.
- The model's ability to predict VF indexes from OCT data may enhance the efficiency and accessibility of glaucoma care.
Related Concept Videos
Glaucoma: Overview
Open Angle Glaucoma: Treatment
Drugs such as carbonic anhydrase inhibitors, α2- and...
Angle Closure Glaucoma: Treatment

