An AI model to estimate visual acuity based solely on cross-sectional OCT imaging of various diseases

Abstract

Purpose To develop an artificial intelligence (AI) model for estimating best-corrected visual acuity (BCVA) using horizontal and vertical optical coherence tomography (OCT) scans of various retinal diseases and examine factors associated with its accuracy. Methods OCT images and associated BCVA measurements from 2,700 OCT images (accrued from 2004 to 2018 with an Atlantis, Triton; Topcon, Tokyo, Japan) of 756 eyes of 469 patients and their BCVA were retrospectively analysed. For each eye, one horizontal and one vertical OCT scan in cross-line mode were used. The GoogLeNet architecture was implemented. The coefficient of determination (R2), root mean square error (RMSE) and mean absolute error (MAE) were computed to evaluate the performance of the trained network. Results R2, RMSE, and MAE were 0.512, 0.350, and 0.321, respectively. R2 was higher in phakic eyes than in pseudophakic eyes. Multivariable regression analysis showed that a higher R2 was significantly associated with better BCVA (p < 0.001) and a higher standard deviation of BCVA (p < 0.001). However, the performance was worse in an external validation, with R2 of 0.19. R2 values for retinal vein occlusion and age-related macular degeneration were 0.961 and 0.373 in the internal validation but 0.20 and 0.22 in the external validation. Conclusion Although underspecification appears to be a fundamental problem to be addressed in AI models for predicting visual acuity, the present results suggest that AI models might have potential for estimating BCVA from OCT in AMD and RVO. Further research is needed to improve the utility of BCVA estimation for these diseases.

Novel Artificial Intelligence-Based Assessment of Imaging Biomarkers in Full-Thickness Macular Holes: Preliminary Data from a Pivotal Trial

“…The parameters that were evaluated in previous studies, after conducting a literature review using Pubmed and Google Scholar with the key words "MH, AI and OCT", did not include IRF, SRF, ELM and EZ interruption, and HRF. Instead, they considered BCVA and hole dimensions to create a predictive model for postoperative anatomical and functional outcomes [19][20][21][22][23].…”

Prediction of Visual Acuity in Pathologic Myopia with Myopic Choroidal Neovascularization Treated with Anti-Vascular Endothelial Growth Factor Using a Deep Neural Network Based on Optical Coherence Tomography Images

“…Research on the prediction of VA in patients with mCNV is scarcely found [ 27 ]. According to a recent study by Inoda et al [ 27 ], prediction of VA in mCNV was inferior to that in neovascular age-related macular degeneration (nAMD) or retinal vein occlusion, with their estimated and actual VA corresponding to an R 2 value of 0.36.…”

“…Research on the prediction of VA in patients with mCNV is scarcely found [ 27 ]. According to a recent study by Inoda et al [ 27 ], prediction of VA in mCNV was inferior to that in neovascular age-related macular degeneration (nAMD) or retinal vein occlusion, with their estimated and actual VA corresponding to an R 2 value of 0.36. However, their study aimed to predict VA using cross-sectional OCT images at a single time point rather than predicting post-injection VA with preinjection OCT images.…”

“…In our study, using 25 volume scan images at 1, 2, and 3 years improved the VA-predictive power compared with using only H/V images. Previous deep learning studies using OCT-based convolutional neural network (CNN) models predicted VA using fewer than five H/V OCT images [ 27 , 29 ]. This implies that training with more OCT images per eye, as in our study, enhances the model’s predictive performance, which requires validation in future studies using OCT-based deep learning models for VA prediction.…”