Diagnostic accuracy of swept source optical coherence tomography classification algorithms for detection of gonioscopic angle closure

Abstract

PurposeTo evaluate the performance of swept source optical coherence tomography (SS-OCT) to detect gonioscopic angle closure using different classification algorithms.MethodsThis was a cross-sectional study of 2028 subjects without ophthalmic symptoms recruited from a community-based clinic. All subjects underwent gonioscopy and SS-OCT (Casia, Tomey Corporation, Nagoya, Japan) under dark room conditions. For each eye, 8 out of 128 frames (22.5° interval) were selected to measure anterior chamber parameters namely anterior chamber width, depth, area and volume (ACW, ACD, ACA, and ACV), lens vault (LV), iris curvature (IC), iris thickness (IT) from 750 µm and 2000 µm from the scleral spur, iris area and iris volume. Five diagnostic algorithms—stepwise logistic regression, random forest, multivariate adaptive regression splines, recursive partitioning and Naïve Bayes were evaluated for detection of gonioscopic angle closure (defined as ≥2 closed quadrants). The performance of the horizontal frame was compared with that of other meridians.ResultsData from 1988 subjects, including 143 (7.2%) with gonioscopic angle closure, were available for analysis. They were divided into two groups: training (1391, 70%) and validation (597, 30%). The best algorithm for detecting gonioscopic angle closure was stepwise logistic regression with an area under the curve of 0.91 (95% CI 0.88 to 0.93) using all parameters, and 0.88 (95% CI 0.82 to 0.93) using only ACA, LV and IC of the horizontal meridian scan.ConclusionsA stepwise logistic regression model incorporating SS-OCT measurements has a high diagnostic ability to detect gonioscopic angle closure.

Performance of anterior segment OCT-based algorithms in the opportunistic screening for primary angle-closure disease

“…Nongpiur et al established a stepwise logistic regression model with six ASOCT parameters to diagnose angle closure 95 % of the time [ 32 ]. Tan et al demonstrated that the combined parameters of ACA, LV, and IC derived from the horizontal meridian scans could achieve an AUC of 0.88 in detecting angle closure [ 33 ]. The variations in predictive efficacy between our study and others may stem from differences in patient demographics, disease severity, angle-closure mechanisms involved, and the selection of variables used for constructing the models.…”

“…Compared to IOP, which is the main parameter in the early development of screening tests for glaucoma, structural or functional abnormality diagnostic tests are now widely used [ 20 , [32] , [33] , [34] , [35] ]. In PACD, diagnostic tests based on structural changes in the anterior segment are more appropriate.…”

“…First, we utilized only one meridional scan and did not use multiple scans of various meridians to determine whether the accuracy of the algorithm could be improved. However, Tan et al showed that horizontal scanning had the best performance and adding all eight frames reduced the diagnostic accuracy to 63 % [ 33 ]. Second, the ASOCT software was semiautomatic, and the SS and angle recess positions needed to be adjusted manually, which could have resulted in subjectivity at some level [ 37 ].…”

Prediction of excessively low vault after implantable collamer lens implantation using iris morphology

“…Anterior segment imaging was performed using swept-source anterior segment optical coherence tomography (AS-OCT, CASIA2, Tomey corporation, Nagoya, Japan), as described previously ( 11 , 12 ). The AS-OCT parameters were measured by horizontal scan across the pupil center, including ACD, anterior chamber width (ACW), crystalline lens rise (CLR), iris concavity (IC), iris concavity ratio (ICR), iris thickness at 750 μm from the scleral spur (IT750) and postoperative vault.…”

Identifying and Exploring the Impact Factors for Intraocular Pressure Prediction in Myopic Children with Atropine Control Utilizing Multivariate Adaptive Regression Splines

“…Owing to the complexity of the interaction between variables in clinical data, traditional biostatistical methods are limited in capturing hidden information. This information contains complex multivariate information that ML can capture more effectively than traditional methods [16,18,19]. Moreover, various ML methods such as multivariate adaptive regression splines (MARS) can provide helpful insights regarding the captured information.…”