Related Experiment Video
Updated: Dec 11, 2025
07:45
Assessing Binocular Central Visual Field and Binocular Eye Movements in a Dichoptic Viewing Condition
Published on: July 21, 2020
4.8K
Estimating Global Visual Field Indices in Glaucoma by Combining Macula and Optic Disc OCT Scans Using 3-Dimensional
Hsin-Hao Yu1, Stefan R Maetschke1, Bhavna J Antony1
1IBM Research Australia, Melbourne, Victoria, Australia.
Ophthalmology. Glaucoma
|August 23, 2020
Summary
Deep neural networks estimate visual field indices from OCT scans. Combining macula (MAC) and optic nerve head (ONH) data improves accuracy in glaucoma patients, suggesting distinct disease progression in these regions.
Area of Science:
- Ophthalmology
- Medical Imaging
- Artificial Intelligence
Background:
- Glaucoma diagnosis and monitoring rely on visual field tests and optical coherence tomography (OCT) imaging.
- Estimating global visual field indices from OCT scans could offer a more integrated diagnostic approach.
- Understanding the distinct contributions of the macula (MAC) and optic nerve head (ONH) to visual field deficits is crucial.
Purpose of the Study:
- To assess the accuracy of deep neural networks in estimating visual field global indices from OCT scans.
- To quantify the specific contributions of the macula (MAC) and optic nerve head (ONH) to these estimations.
- To evaluate the combined impact of MAC and ONH data on estimation accuracy.
Main Methods:
- A cohort of 10,370 eyes from healthy individuals, glaucoma suspects, and glaucoma patients was studied.
- Three-dimensional convolutional neural networks were trained using OCT scans (centered on MAC, ONH, or both) to predict visual field indices.
- Performance was evaluated using Spearman's rank correlation, Pearson's correlation coefficient, and absolute errors for visual field index (VFI) and mean deviation (MD).
Main Results:
- The combined MAC + ONH approach yielded a Spearman's correlation of 0.76 and Pearson's correlation of 0.87 for VFI and MD.
- Median absolute errors were 2.7 for VFI and 1.57 decibels (dB) for MD.
- Estimates using only MAC or ONH were less accurate; accuracy varied with OCT signal strength and glaucoma severity.
Conclusions:
- Integrating MAC and ONH information from OCT scans significantly improves the accuracy of estimating global visual field indices, particularly in advanced glaucoma.
- This suggests that structural changes in the macula and optic nerve head follow different temporal patterns throughout glaucoma progression.
- Deep learning models utilizing combined OCT data show promise for enhanced glaucoma assessment.
