A Novel Approach To Predict Glaucomatous Impairment in the Central 10° Visual Field, Excluding the Effect of Cataract.
Tomita Ryo, Asaoka Ryo, Hirasawa Kazunori, Fujino Yuri, Omura Tetsuro, Inatomi Tsutomu, Obana Akira, Nishiguchi Koji M, Tanito Masaki
AI Summary
A study accurately predicted central 10° glaucomatous visual field loss after cataract removal using a machine learning model (RFM) with visual acuity and visual field data, but adding OCT data didn't improve accuracy.
Abstract
Purpose
Our previous study predicted genuine glaucomatous visual field (VF) impairment in the central 10° VF, excluding the effect of cataract, using visual acuity (VA) and global indexes of VF more accurately than pattern deviation (PD). This study aimed to improve the accuracy by using pointwise total deviation (TD) values with the machine-learning method of random forest model (RFM) and to investigate whether incorporating optical coherence tomography-measured ganglion cell-inner plexiform layer (GCIPL) thickness is useful.
Methods
This retrospective study included 89 eyes with open-angle glaucoma that underwent successful cataract surgery (with or without iStent implantation or ab interno trabeculotomy). Postoperative TD in each of the 68 VF points was predicted using preoperative (1) PD, (2) VA and VF with a linear regression model (LM), and (3) VA and VF with RFM, and averaged as predicted mean TD (mTDpost). Further prediction was made by incorporating the preoperative GCIPL into the best model.
Results
The mean absolute error (MAE) between the actual and predicted mTDpost with RFM (1.25 ± 1.03 dB) was significantly smaller than that with PD (3.20 ± 4.06 dB, p < 0.01) and LM (1.42 ± 1.06 dB, p < 0.05). The MAEs with the model incorporating GCIPL into RFM (1.24 ± 1.04 dB) and RFM were not significantly different.
Conclusions
Accurate prediction of genuine glaucomatous VF impairment was achieved using pointwise TD with RFM. No merit was observed by incorporating the GCIPL into this model.
Translational relevance: This pointwise RFM could clinically reduce cataract effect on VF.
MeSH Terms
Shields Classification
Key Concepts5
The mean absolute error (MAE) between actual and predicted mean total deviation (mTDpost) with the random forest model (RFM) was 1.25 ± 1.03 dB in 89 eyes with open-angle glaucoma that underwent successful cataract surgery.
The mean absolute error (MAE) for predicting mean total deviation (mTDpost) with the random forest model (RFM) (1.25 ± 1.03 dB) was significantly smaller than with pattern deviation (PD) (3.20 ± 4.06 dB, p < 0.01) in 89 eyes with open-angle glaucoma that underwent successful cataract surgery.
The mean absolute error (MAE) for predicting mean total deviation (mTDpost) with the random forest model (RFM) (1.25 ± 1.03 dB) was significantly smaller than with a linear regression model (LM) (1.42 ± 1.06 dB, p < 0.05) in 89 eyes with open-angle glaucoma that underwent successful cataract surgery.
Incorporating preoperative ganglion cell-inner plexiform layer (GCIPL) thickness into the random forest model (RFM) for predicting mean total deviation (mTDpost) did not significantly improve accuracy (MAE with GCIPL: 1.24 ± 1.04 dB vs. RFM alone: 1.25 ± 1.03 dB) in 89 eyes with open-angle glaucoma that underwent successful cataract surgery.
A retrospective study included 89 eyes with open-angle glaucoma that underwent successful cataract surgery (with or without iStent implantation or ab interno trabeculotomy) to predict postoperative total deviation in the central 10° visual field.
Related Articles5
Structure-function analysis for macular surgery in patients with coexisting glaucoma.
Case SeriesLongitudinal variability outcomes of frontloaded visual field testing.
Longitudinal StudyEyecatcher 3.0 - Validating the Use of "Smart Glasses" as a Low-Cost, Portable Method of Assessing Visual Fields.
Clinical TrialStandard automated perimetry for glaucoma and diseases of the retina and visual pathways: Current and future perspectives.
ReviewPrediction of visual field progression with serial optic disc photographs using deep learning.
Cohort StudyIs this article assigned to the wrong chapter(s)? Let us know.