Deep Learning Approaches Predict Glaucomatous Visual Field Damage from OCT Optic Nerve Head En Face Images and Retinal Nerve Fiber Layer Thickness Maps

Abstract

Purpose: To develop and evaluate a deep learning system for differentiating between eyes with and without glaucomatous visual field damage (GVFD) and predicting the severity of GFVD from spectral domain OCT (SD OCT) optic nerve head images.Design: Evaluation of a diagnostic technology.Participants: A total of 9765 visual field (VF) SD OCT pairs collected from 1194 participants with and without GVFD (1909 eyes).Methods: Deep learning models were trained to use SD OCT retinal nerve fiber layer (RNFL) thickness maps, RNFL en face images, and confocal scanning laser ophthalmoscopy (CSLO) images to identify eyes with GVFD and predict quantitative VF mean deviation (MD), pattern standard deviation (PSD), and mean VF sectoral pattern deviation (PD) from SD OCT data.Main Outcome Measures: Deep learning models were compared with mean RNFL thickness for identifying GVFD using area under the curve (AUC), sensitivity, and specificity. For predicting MD, PSD, and mean sectoral PD, models were evaluated using R 2 and mean absolute error (MAE).Results: In the independent test dataset, the deep learning models based on RNFL en face images achieved an AUC of 0.88 for identifying eyes with GVFD and 0.82 for detecting mild GVFD significantly (P < 0.001) better than using mean RNFL thickness measurements (AUC ¼ 0.82 and 0.73, respectively). Deep learning models outperformed standard RNFL thickness measurements in predicting all quantitative VF metrics. In predicting MD, deep learning models based on RNFL en face images achieved an R 2 of 0.70 and MAE of 2.5 decibels (dB) compared with 0.45 and 3.7 dB for RNFL thickness measurements. In predicting mean VF sectoral PD, deep learning models achieved high accuracy in the inferior nasal (R 2 ¼ 0.60) and superior nasal (R 2 ¼ 0.67) sectors, moderate accuracy in inferior (R 2 ¼ 0.26) and superior (R 2 ¼ 0.35) sectors, and lower accuracy in the central (R 2 ¼ 0.15) and temporal (R 2 ¼ 0.12) sectors.Conclusions: Deep learning models had high accuracy in identifying eyes with GFVD and predicting the severity of functional loss from SD OCT images. Accurately predicting the severity of GFVD from SD OCT imaging can help clinicians more effectively individualize the frequency of VF testing to the individual patient.

Artificial Intelligence Mapping of Structure to Function in Glaucoma

“…Other studies relied on different types of OCT inputs to get predictions of SAP. 7,[33][34][35][36][37] Christopher et al 37 achieved best performance when using en face images to predict SAP global metrics and sectoral averages. It should be noted, however, that our main purpose was to use the CNN to develop an SF map.…”

A deep learning approach to predict visual field using optical coherence tomography

“…Although attempts to use machine learning algorithms for structure-function relationships in glaucoma are not novel [38][39][40], there have been few studies that predict visual field using OCT imaging. In a recent study similar to ours, Christopher et al [41] used a deep learning The combined OCT images, which were input into the deep learning architecture, are shown on the left column. The actual threshold values of visual field exams are shown in the (B) middle panel and the threshold values predicted by Inception V3 based deep learning architecture are shown on the (C) right panel.…”

Effects of Study Population, Labeling and Training on Glaucoma Detection Using Deep Learning Algorithms

“…Indeed, our research teams from UCSD and UTokyo recently reported the benefits of transfer learning using Imagenet in glaucoma detection from fundus photographs as well as diagnosing early stage glaucoma from optical coherence tomography (OCT) images. 5,6,25,26 In both cases, diagnostic performance of the deep learning model was significantly improved by pretraining using transfer learning. 5,6 Updating the trained model with new images for the same specialized task can also improve performance.…”