Predicting the Glaucomatous Central 10-Degree Visual Field From Optical Coherence Tomography Using Deep Learning and...

Area of Science:

• Ophthalmology
• Medical Imaging
• Artificial Intelligence

Background:

• Glaucoma is a leading cause of irreversible blindness worldwide.
• Accurate prediction of visual field (VF) defects is crucial for glaucoma management.
• Optical coherence tomography (OCT) provides detailed structural measurements of retinal layers.

Purpose of the Study:

• To predict the central 10° visual field (VF) in glaucoma patients using optical coherence tomography (OCT) measurements.
• To evaluate the efficacy of deep learning and tensor regression models for VF prediction.

Main Methods:

• A cross-sectional study involving 304 glaucoma patients and 43 normal subjects.
• Two convolutional neural network (CNN) models were developed: CNN-PR and CNN-TR (using tensor regression).
• Prediction accuracy was assessed using root mean squared error (RMSE) and compared against multiple linear regression (MLR) and support vector regression (SVR).

Main Results:

• The CNN-TR model achieved the lowest average RMSE (6.32 ± 3.76 dB), outperforming CNN-PR (6.76 ± 3.86 dB), SVR (7.18 ± 3.87 dB), and MLR (8.56 ± 3.69 dB).
• The absolute mean prediction error for the whole VF using CNN-TR was 2.72 ± 2.60 dB.
• The CNN-TR model demonstrated significantly higher prediction accuracy (P < .05).

Conclusions:

• Deep learning combined with tensor regression can effectively predict Humphrey 10-2 visual fields from OCT-measured retinal layer thicknesses.
• This approach holds potential for non-invasive glaucoma monitoring and diagnosis.