Automated expert-level scleral spur detection and quantitative biometric analysis on the ANTERION anterior segment OCT system.

AIM: To perform an independent validation of deep learning (DL) algorithms for automated scleral spur detection and measurement of scleral spur-based biometric parameters in anterior segment optical coherence tomography (AS-OCT) images. METHODS: Patients receiving routine eye care underwent AS-OCT imaging using the ANTERION OCT system (Heidelberg Engineering, Heidelberg, Germany). Scleral spur locations were marked by three human graders (reference, expert and novice) and predicted using DL algorithms developed by Heidelberg Engineering that prioritise a false positive rate <4% (FPR4) or true positive rate >95% (TPR95). Performance of human graders and DL algorithms were evaluated based on agreement of scleral spur locations and biometric measurements with the reference grader. RESULTS: 1308 AS-OCT images were obtained from 117 participants. Median differences in scleral spur locations from reference locations were significantly smaller (p<0.001) for the FPR4 (52.6±48.6 µm) and TPR95 (55.5±50.6 µm) algorithms compared with the expert (61.1±65.7 µm) and novice (79.4±74.9 µm) graders. Intergrader reproducibility of biometric measurements was excellent overall for all four (intraclass correlation coefficient range 0.918-0.997). Intergrader reproducibility of the expert grader (0.567-0.965) and DL algorithms (0.746-0.979) exceeded that of the novice grader (0.146-0.929) for images with narrow angles defined by OCT measurement of angle opening distance 500 µm anterior to the scleral spur (AOD500)<150 µm. CONCLUSIONS: DL algorithms on the ANTERION approximate expert-level measurement of scleral spur-based biometric parameters in an independent patient population. These algorithms could enhance clinical utility of AS-OCT imaging, especially for evaluating patients with angle closure and performing intraocular lens calculations.

Classification of Angle Closure Severity by Hierarchical Cluster Analysis of Ocular Biometrics in the Dark and Light.

Purpose: The purpose of this study was to investigate the classification of angle closure eyes based on hierarchical cluster analysis of ocular biometrics measured in the dark and light using anterior segment optical coherence tomography (AS-OCT). Methods: Participants of the Chinese American Eye Study received complete eye examinations to identify primary angle closure suspects (PACS) and primary angle closure without/with glaucoma (PAC/G). AS-OCT was performed in the dark and light. Biometric parameters describing the angle, iris, lens, and anterior chamber were analyzed. Hierarchical clustering was performed using Ward's method. Post hoc logistic regression models were developed to identify biometric predictors of angle closure staging. Results: Analysis of 159 eyes with PACS (N = 120) or PAC/G (N = 39) produced 2 clusters in the dark and light. In both analyses, cluster 1 (N = 132 in the dark and N = 126 in the light) was characterized by smaller angle opening distance (AOD)750 and trabecular iris space area (TISA)750, greater iris curvature (IC), and greater lens vault (LV; P < 0.001) than cluster 2. The proportion of PAC/PACG to PACS eyes was significantly higher in cluster 1 than 2 in the light (36:90 and 3:30, respectively; P = 0.02), but not the dark (36:96 and 3:24, respectively; P = 0.08). On multivariable regression analyses, smaller TISA750 (odds ratio [OR] = 0.84 per 0.01 mm2) and AOD750 (OR = 0.93 per 0.01 mm) in the light and smaller TISA750 (OR = 0.86 per 0.01 mm2) in the dark conferred higher risk of PAC/G (P ≤ 0.02). Conclusions: Unsupervised cluster analysis of ocular biometrics can classify angle closure eyes by severity. Static biometrics measured in the light and dark are both predictive of PAC/G. Translational Relevance: Clustering of biometrics measured in the light could provide an alternative source of information to risk-stratify angle closure eyes for more severe disease.

Role of Static and Dynamic Ocular Biometrics Measured in the Dark and Light as Risk Factors for Angle Closure Progression.

PURPOSE: To assess the role of static and dynamic ocular biometric parameters measured in the dark and light for predicting progression of primary angle closure suspect (PACS) to primary angle closure (PAC). DESIGN: Retrospective cohort study using prospective randomized controlled trial data from untreated, control eyes. METHODS: Zhongshan Angle Closure Prevention Trial subjects underwent anterior segment optical coherence tomography (AS-OCT) imaging in the dark and light. Static biometric parameters were measured, consisting of angle, iris, lens, and anterior chamber parameters. Dynamic change parameters were calculated by subtracting light measurements from dark measurements. Cox proportional hazards regression models were developed to assess risk factors for PACD progression. RESULTS: A total of 861 eyes of 861 participants were analyzed (36 progressors). On univariable analysis, TISA500 measurements in the light and dark were associated with progression (P < .001), whereas dynamic change parameters were not (P ≥ .08). In the primary multivariable model, older age (hazard ratio [HR] = 1.09 per year), higher intraocular pressure (IOP) (HR = 1.13 per mm Hg), and smaller TISA500 in the light (HR = 1.28 per 0.01 mm2) were significantly associated with greater risk of progression (P ≤ .04). Dark TISA500 had similar significance (HR = 1.28, P = .002) when replacing light TISA500. Risk of progression was more predictive among eyes in the lowest quartile of light TISA500 measurements (HR = 4.56, P < .001) compared to dark measurements (HR = 2.89, P = .003). CONCLUSION: Static parameters measured in the light are as predictive, and possibly more so, of angle closure progression as those measured in the dark. Ocular biometrics measured under light and dark conditions may provide additional information for risk-stratifying patients for angle closure progression.

Racial and Ethnic Differences in the Roles of Myopia and Ocular Biometrics as Risk Factors for Primary Open-Angle Glaucoma.

Purpose: Assess how the roles of refractive error (RE) and ocular biometrics as risk factors for primary open-angle glaucoma (POAG) differ by race and ethnicity. Methods: Data from the Los Angeles Latino Eye Study (LALES) and the Chinese American Eye Study (CHES), two population-based epidemiological studies, were retrospectively analyzed. Multivariable logistic regression and interaction term analyses were performed to assess relationships between POAG and its risk factors, including RE and axial length (AL), and to assess effect modification by race/ethnicity. Results: Analysis included 7601 phakic participants of LALES (47.3%) and CHES (52.7%) with age ≥ 50 years. Mean age was 60.6 ± 8.3 years; 60.9% were female. The prevalence and unadjusted risk of POAG were higher in LALES than CHES (6.0% and 4.0%, respectively; odds ratio [OR] = 1.55; P < 0.001). In the multivariable analysis, significant risk factors for POAG included Latino ethnicity (OR = 2.25; P < 0.001), refractive myopia (OR = 1.54 for mild, OR = 2.47 for moderate, OR = 3.94 for high compared to non-myopes; P ≤ 0.003), and longer AL (OR = 1.37 per mm; P < 0.001). AL (standardized regression coefficient [SRC] = 0.3) was 2.7-fold more strongly associated with POAG than high myopia status (SRC = 0.11). There was no modifying effect by race/ethnicity on the association between RE (per diopter) or AL (per millimeter) and POAG (P = 0.49). Conclusions: Although the POAG risk conferred by myopic RE and longer AL is similar between Latino and Chinese Americans, the difference in POAG prevalence between the two groups is narrowed by higher myopia prevalence among Chinese Americans. Racial/ethnic populations with higher myopia incidence may become disproportionately affected by POAG in the context of the global myopia epidemic.

Biometric Risk Factors for Angle Closure Progression After Laser Peripheral Iridotomy.

Importance: Laser peripheral iridotomy (LPI) is the most common primary treatment for primary angle closure disease (PACD). However, there are sparse data guiding the longitudinal care of PAC suspect (PACS) eyes after LPI. Objective: To elucidate the anatomic effects of LPI that are associated with a protective outcome against progression from PACS to PAC and acute angle closure (AAC) and to identify biometric factors that predict progression after LPI. Design, Setting, and Participants: This was a retrospective analysis of data from the Zhongshan Angle Closure Prevention (ZAP) trial, a study of mainland Chinese people aged 50 to 70 years with bilateral PACS who received LPI in 1 randomly selected eye. Gonioscopy and anterior-segment optical coherence tomography (AS-OCT) imaging were performed 2 weeks after LPI. Progression was defined as the development of PAC or an acute angle closure (AAC) attack. Cohort A included a random mix of treated and untreated eyes, and cohort B included only eyes treated with LPI. Univariable and multivariable Cox regression models were developed to assess biometric risk factors for progression in cohorts A and B. Data were analyzed from January 4 to December 22, 2022. Main Outcome and Measure: Six-year progression to PAC or AAC. Results: Cohort A included 878 eyes from 878 participants (mean [SD] age, 58.9 [5.0] years; 726 female [82.7%]) of whom 44 experienced progressive disease. In a multivariable analysis, treatment (hazard ratio [HR], 0.67; 95% CI, 0.34-1.33; P = .25) was no longer associated with progression after adjusting for age and trabecular iris space area at 500 µm (TISA at 500 µm) at the 2-week visit. Cohort B included 869 treated eyes from 869 participants (mean [SD] age, 58.9 [5.0] years; 717 female [82.5%]) of whom 19 experienced progressive disease. In multivariable analysis, TISA at 500 µm (HR, 1.33 per 0.01 mm2 smaller; 95% CI, 1.12-1.56; P = .001) and cumulative gonioscopy score (HR, 1.25 per grade smaller; 95% CI, 1.03-1.52; P = .02) at the 2-week visit were associated with progression. Persistent angle narrowing on AS-OCT (TISA at 500 µm ≤0.05 mm2; HR, 9.41; 95% CI, 3.39-26.08; P <.001) or gonioscopy (cumulative score ≤6; HR, 2.80; 95% CI, 1.13-6.93; P =.04) conferred higher risk of progression. Conclusions and Relevance: Study results suggest that persistent angle narrowing detected by AS-OCT or cumulative gonioscopy score was predictive of disease progression in PACS eyes after LPI. These findings suggest that AS-OCT and gonioscopy may be performed to identify patients at high risk of developing angle closure who may benefit from closer monitoring despite patent LPI.

Refractive Error and Anterior Chamber Depth as Risk Factors in Primary Angle Closure Disease: The Chinese American Eye Study.

PRCIS: The risk of primary angle closure disease (PACD) rises rapidly with greater hyperopia while remaining relatively low for all degrees of myopia. Refractive error (RE) is useful for angle closure risk stratification in the absence of biometric data. PURPOSE: To assess the role of RE and anterior chamber depth (ACD) as risk factors in PACD. METHODS: Chinese American Eye Study participants received complete eye examinations including refraction, gonioscopy, amplitude-scan biometry, and anterior segment ocular coherence tomography imaging. PACD included primary angle closure suspect (≥3 quadrants of angle closure on gonioscopy) and primary angle closure/primary angle closure glaucoma (peripheral anterior synechiae or intraocular pressure >21 mm Hg). Logistic regression models were developed to assess associations between PACD and RE and/or ACD adjusted for sex and age. Locally weighted scatterplot smoothing curves were plotted to assess continuous relationships between variables. RESULTS: Three thousand nine hundred seventy eyes (3403 open angle and 567 PACD) were included. The risk of PACD increased with greater hyperopia [odds ratio (OR) = 1.41 per diopter (D); P < 0.001] and shallower ACD (OR = 1.75 per 0.1 mm; P < 0.001). Hyperopia (≥ + 0.5 D; OR = 5.03) and emmetropia (-0.5 D to +0.5 D; OR = 2.78) conferred a significantly higher risk of PACD compared with myopia (≤0.5 D). ACD (standardized regression coefficient = -0.54) was a 2.5-fold stronger predictor of PACD risk compared with RE (standardized regression coefficient = 0.22) when both variables were included in one multivariable model. The sensitivity and specificity of a 2.6 mm ACD cutoff for PACD were 77.5% and 83.2% and of a +2.0 D RE cutoff were 22.3% and 89.1%. CONCLUSION: The risk of PACD rises rapidly with greater hyperopia while remaining relatively low for all degrees of myopia. Although RE is a weaker predictor of PACD than ACD, it remains a useful metric to identify patients who would benefit from gonioscopy in the absence of biometric data.

Generalisability and performance of an OCT-based deep learning classifier for community-based and hospital-based detection of gonioscopic angle closure.

PURPOSE: To assess the generalisability and performance of a deep learning classifier for automated detection of gonioscopic angle closure in anterior segment optical coherence tomography (AS-OCT) images. METHODS: A convolutional neural network (CNN) model developed using data from the Chinese American Eye Study (CHES) was used to detect gonioscopic angle closure in AS-OCT images with reference gonioscopy grades provided by trained ophthalmologists. Independent test data were derived from the population-based CHES, a community-based clinic in Singapore, and a hospital-based clinic at the University of Southern California (USC). Classifier performance was evaluated with receiver operating characteristic curve and area under the receiver operating characteristic curve (AUC) metrics. Interexaminer agreement between the classifier and two human examiners at USC was calculated using Cohen's kappa coefficients. RESULTS: The classifier was tested using 640 images (311 open and 329 closed) from 127 Chinese Americans, 10 165 images (9595 open and 570 closed) from 1318 predominantly Chinese Singaporeans and 300 images (234 open and 66 closed) from 40 multiethnic USC patients. The classifier achieved similar performance in the CHES (AUC=0.917), Singapore (AUC=0.894) and USC (AUC=0.922) cohorts. Standardising the distribution of gonioscopy grades across cohorts produced similar AUC metrics (range 0.890-0.932). The agreement between the CNN classifier and two human examiners (Ò =0.700 and 0.704) approximated interexaminer agreement (Ò =0.693) in the USC cohort. CONCLUSION: An OCT-based deep learning classifier demonstrated consistent performance detecting gonioscopic angle closure across three independent patient populations. This automated method could aid ophthalmologists in the assessment of angle status in diverse patient populations.

Anterior segment biometric measurements explain misclassifications by a deep learning classifier for detecting gonioscopic angle closure.

BACKGROUND/AIMS: To identify biometric parameters that explain misclassifications by a deep learning classifier for detecting gonioscopic angle closure in anterior segment optical coherence tomography (AS-OCT) images. METHODS: Chinese American Eye Study (CHES) participants underwent gonioscopy and AS-OCT of each angle quadrant. A subset of CHES AS-OCT images were analysed using a deep learning classifier to detect positive angle closure based on manual gonioscopy by a reference human examiner. Parameter measurements were compared between four prediction classes: true positives (TPs), true negatives (TNs), false positives (FPs) and false negatives (FN). Logistic regression models were developed to differentiate between true and false predictions. Performance was assessed using area under the receiver operating curve (AUC) and classifier accuracy metrics. RESULTS: 584 images from 127 participants were analysed, yielding 271 TPs, 224 TNs, 77 FPs and 12 FNs. Parameter measurements differed (p<0.001) between prediction classes among anterior segment parameters, including iris curvature (IC) and lens vault (LV), and angle parameters, including angle opening distance (AOD). FP resembled TP more than FN and TN in terms of anterior segment parameters (steeper IC and higher LV), but resembled TN more than TP and FN in terms of angle parameters (wider AOD). Models for detecting FP (AUC=0.752) and FN (AUC=0.838) improved classifier accuracy from 84.8% to 89.0%. CONCLUSIONS: Misclassifications by an OCT-based deep learning classifier for detecting gonioscopic angle closure are explained by disagreement between anterior segment and angle parameters. This finding could be used to improve classifier performance and highlights differences between gonioscopic and AS-OCT definitions of angle closure.

Age-Related Changes in Dynamic Iris Behavior Assessed Using a Programmable Closed-Loop Iris Control System.

Purpose: The purpose of this study was to develop and test a programmable closed-loop system for tracking, modulating, and assessing dynamic iris behavior, including in the mid-dilated position. Methods: A programmable closed-loop iris control system was developed by customizing an ANTERION OCT device (Heidelberg Engineering, Heidelberg, Germany). Custom software was developed to store camera and optical coherence tomography (OCT) images, track pupillary diameter (PD), control a light-emitting diode (LED), and modulate ambient lighting to maintain the iris in a dilated, constricted, or mid-dilated position in real-time. Study participants underwent 3 consecutive 65-second scan sessions. Dynamic iris behavior in the form of peak constriction velocity (PCV) and mid-dilated iris activity (MDIA) were calculated and analyzed offline. Results: Among 58 participants, 56 (96.6%) were eligible for analysis based on achieving and maintaining mean PD within ±10% of the calculated mid-dilated PD. Mean participant age was 49.8 ± 18.9 years. Mean PCV was 3.92 ± 0.83 mm/s, and mean MDIA was 0.37 ± 0.15 mm. The mean difference between the calculated and achieved mid-dilated PD was 0.166 ± 0.192 mm. There were significant negative correlations between PCV and age (slope = -0.022, P < 0.001) and MDIA and age (slope = -0.004, P < 0.001). Success rates were lower (69.0%) but relationships between dynamic iris behavior and age were similar based on achieving and maintaining mean PD within ±5% of the calculated mid-dilated PD. Conclusions: A programmable closed-loop iris control system can modulate dynamic iris behavior and maintain the iris in a mid-dilated position. Pupillary constriction velocity and iris activity in the mid-dilated position decrease with age. Translational Relevance: This system can be applied to study dynamic disease processes involving the iris and establish novel biometric measures that could serve as risk factors for acute and chronic primary angle closure glaucoma (PACG).

Ocular Biometric Risk Factors for Progression of Primary Angle Closure Disease: The Zhongshan Angle Closure Prevention Trial.

PURPOSE: To assess baseline ocular biometric risk factors for progression from primary angle closure suspect (PACS) to primary angle closure (PAC) or acute angle closure (AAC). DESIGN: Prospective, observational study. PARTICIPANTS: Six hundred forty-three mainland Chinese with untreated PACS. METHODS: Participants underwent baseline clinical examinations, including gonioscopy, anterior segment OCT (AS-OCT) imaging, and A-scan ultrasound biometry as part of the Zhongshan Angle Closure Prevention (ZAP) Trial. Primary angle closure suspect was defined as an inability to visualize pigmented trabecular meshwork in 2 or more quadrants based on static gonioscopy. Primary angle closure was defined as development of intraocular pressure above 24 mmHg or peripheral anterior synechiae. Progression was defined as development of PAC or an AAC attack. Multivariable logistic regression models were developed to assess biometric risk factors for progression. MAIN OUTCOME MEASURES: Six-year progression from PACS to PAC or AAC. RESULTS: Six hundred forty-three untreated eyes (609 nonprogressors, 34 progressors) of 643 participants were analyzed. In a multivariable model with continuous parameters, narrower horizontal angle opening distance of 500 µm from the scleral spur (AOD500; odds ratio [OR], 1.10 per 0.01-mm decrease; P = 0.03), flatter horizontal iris curvature (IC; OR, 1.96 per 0.1-mm decrease; P = 0.01), and older age (OR, 1.11 per 1-year increase; P = 0.01) at baseline were associated significantly with progression (area under the receiver operating characteristic curve [AUC], 0.73). Smaller cumulative gonioscopy score was not associated with progression (OR, 1.03 per 1-modified Shaffer grade decrease; P = 0.85) when replacing horizontal AOD500 in the multivariable model. In a separate multivariable model with categorical parameters, participants in the lowest quartile of horizontal AOD500 (OR, 3.10; P = 0.002) and IC (OR, 2.48; P = 0.014) measurements and 59 years of age or older (OR, 2.68; P = 0.01) at baseline showed higher odds of progression (AUC, 0.72). CONCLUSIONS: Ocular biometric measurements can help to risk-stratify patients with early angle closure for more severe disease. Anterior segment OCT measurements of biometric parameters describing the angle and iris are predictive of progression from PACS to PAC or AAC, whereas gonioscopy grades are not.

Ocular Biometric Determinants of Dark-to-Light Change in Angle Width: The Chinese American Eye Study.

PURPOSE: To assess ocular biometric determinants of dark-to-light change in anterior chamber angle width and identify dynamic risk factors in primary angle closure disease (PACD). DESIGN: Population-based cross-sectional study. METHODS: Chinese American Eye Study (CHES) participants underwent anterior segment optical coherence tomography imaging in the dark and light. Static dark and light biometric parameters, including angle opening distance, 750 µm (AOD750), anterior chamber width (ACW), lens vault (LV), and pupillary diameter (PD) were measured, and dynamic dark-to-light changes were calculated. Contributions by static and dynamic parameters to dark-to-light changes in AOD750 were assessed using multivariable linear regression models with standardized regression coefficients (SRCs) and semipartial correlation coefficients squared (SPCC2). PACD was defined as ≥3 quadrants of gonioscopic angle closure. RESULTS: The analysis included 1011 participants. All biometric parameters differed between dark and light (P < .05). On multivariable regression analysis, change in ACW (SRC = -0.35, SPCC2 = 0.081) and PD (SRC = -0.46, SPCC2 = 0.072) were the strongest determinants of dark-to-light change in AOD750 (overall R2 = 0.40). Dark-to-light increase in AOD750 was less in eyes with than without PACD (0.081 mm and 0.111 mm, respectively; P < .001). ACW increased in eyes with PACD and decreased in eyes without PACD from dark to light (P < .025), whereas change in PD was similar (P = .28). CONCLUSIONS: Beneficial angle widening effects of transitioning from dark to light are attenuated in eyes with PACD, which appears related to aberrant dark-to-light change in ACW. These findings highlight the importance of assessing the angle in both dark and light to identify potential dynamic mechanisms of angle closure.

Glaucoma Expert-Level Detection of Angle Closure in Goniophotographs With Convolutional Neural Networks: The Chinese American Eye Study.

PURPOSE: To compare the performance of a novel convolutional neural network (CNN) classifier and human graders in detecting angle closure in EyeCam (Clarity Medical Systems, Pleasanton, California, USA) goniophotographs. DESIGN: Retrospective cross-sectional study. METHODS: Subjects from the Chinese American Eye Study underwent EyeCam goniophotography in 4 angle quadrants. A CNN classifier based on the ResNet-50 architecture was trained to detect angle closure, defined as inability to visualize the pigmented trabecular meshwork, using reference labels by a single experienced glaucoma specialist. The performance of the CNN classifier was assessed using an independent test dataset and reference labels by the single glaucoma specialist or a panel of 3 glaucoma specialists. This performance was compared to that of 9 human graders with a range of clinical experience. Outcome measures included area under the receiver operating characteristic curve (AUC) metrics and Cohen kappa coefficients in the binary classification of open or closed angle. RESULTS: The CNN classifier was developed using 29,706 open and 2,929 closed angle images. The independent test dataset was composed of 600 open and 400 closed angle images. The CNN classifier achieved excellent performance based on single-grader (AUC = 0.969) and consensus (AUC = 0.952) labels. The agreement between the CNN classifier and consensus labels (κ = 0.746) surpassed that of all non-reference human graders (κ = 0.578-0.702). Human grader agreement with consensus labels improved with clinical experience (P = 0.03). CONCLUSION: A CNN classifier can effectively detect angle closure in goniophotographs with performance comparable to that of an experienced glaucoma specialist. This provides an automated method to support remote detection of patients at risk for primary angle closure glaucoma.

Anatomic Changes and Predictors of Angle Widening after Laser Peripheral Iridotomy: The Zhongshan Angle Closure Prevention Trial.

PURPOSE: To assess anatomic changes after laser peripheral iridotomy (LPI) and predictors of angle widening based on anterior segment (AS) OCT and angle opening based on gonioscopy. DESIGN: Prospective observational study. PARTICIPANTS: Primary angle-closure suspects (PACSs) 50 to 70 years of age. METHODS: Participants of the Zhongshan Angle Closure Prevention (ZAP) Trial underwent gonioscopy and AS-OCT imaging at baseline and 2 weeks after LPI. Primary angle-closure suspect was defined as the inability to visualize pigmented trabecular meshwork in 2 or more quadrants on static gonioscopy. Laser peripheral iridotomy was performed on 1 eye per patient in superior (between 11 and 1 o'clock) or temporal or nasal locations (at or below 10:30 or 1:30 o'clock). Biometric parameters in horizontal and vertical AS-OCT scans were measured and averaged. Linear and logistic regression modeling were performed to determine predictors of angle widening, defined as change in mean angle opening distance measured at 750 µm from the scleral spur (AOD750); poor angle widening, defined as the lowest quintile of change in mean AOD750; and poor angle opening, defined as residual PACS after LPI based on gonioscopy. MAIN OUTCOME MEASURES: Anatomic changes and predictors of angle widening and opening after LPI. RESULTS: Four hundred fifty-four patients were included in the analysis. Two hundred nineteen underwent superior LPI and 235 underwent temporal or nasal LPI. Significant changes were found among most biometric parameters (P < 0.006) after LPI, including greater AOD750 (P < 0.001). One hundred twenty eyes (26.4%) showed residual PACS after LPI. In multivariate regression analysis, superior LPI location (P = 0.004), smaller AOD750 (P < 0.001), and greater iris curvature (P < 0.001), were predictive of greater angle widening. Temporal or nasal LPI locations (odds ratio [OR], 2.60, P < 0.001) was predictive of poor angle widening. Smaller mean gonioscopy grade (OR, 0.34, 1-grade increment) was predictive of poor angle opening. CONCLUSIONS: Superior LPI location results in significantly greater angle widening compared with temporal or nasal locations in a Chinese population with PACS. This supports consideration of superior LPI locations to optimize anatomic changes after LPI.

Differences in Ocular Biometric Measurements among Subtypes of Primary Angle Closure Disease: The Chinese American Eye Study.

PURPOSE: To assess differences in ocular biometric measurements between primary angle closure suspect (PACS) eyes and primary angle closure (PAC) and primary angle closure glaucoma (PACG) eyes. DESIGN: Cross-sectional study. PARTICIPANTS: Patients with primary angle closure disease (PACD) were identified from the Chinese American Eye Study, a population-based study in Los Angeles, California. METHODS: Patients previously underwent complete ocular examinations including gonioscopy and anterior segment (AS)-OCT imaging with the Tomey CASIA SS-1000 (Tomey Corporation). Four AS-OCT images were analyzed per eye. Averaged and sectoral measurements of biometric parameters, including angle recess area (ARA), trabecular iris space area (TISA), iris area, iris curvature, lens vault, anterior chamber depth, and anterior chamber area, were compared between early PACD (PACS) and late PACD (PAC and PACG) groups. Machine learning classifiers that attempt to differentiate between early and late PACD eyes were developed by applying different regression algorithms to a training dataset of sectoral parameter measurements. Classifier performance was assessed using an independent test dataset. MAIN OUTCOME MEASURES: Averaged and sectoral measurements of biometric parameters. RESULTS: Two hundred ninety-eight eyes (231 PACS, 67 PAC or PACG) of 298 patients were analyzed. No difference was found in averaged biometric measurements between the 2 groups before (P > 0.09) or after (P > 0.14) adjusting for age and gender. Differences (P < 0.04) between the 2 groups were found for 11 sectoral parameter measurements, including ARA and TISA. The performance of machine learning classifiers developed using sectoral parameter measurements was poor on the independent test dataset for all regression algorithms (area under the receiver operating characteristic curve, 0.529-0.628). CONCLUSIONS: Differences in biometric measurements between subtypes of PACD eyes were small in a population-based cohort of Chinese Americans. The poor performance of classifiers based on these measurements highlights potential challenges of developing quantitative methods to detect late PACD.

Intradevice Repeatability and Interdevice Agreement of Ocular Biometric Measurements: A Comparison of Two Swept-Source Anterior Segment OCT Devices.

Purpose: To assess the repeatability and agreement of ocular biometric parameters measured using the Tomey CASIA SS-1000 and Heidelberg ANTERION anterior segment optical coherence tomography (AS-OCT) devices. Methods: Both eyes of subjects 18 years of age or older were scanned three times with the CASIA and ANTERION under standardized dark lighting. One AS-OCT image along the horizontal (temporal-nasal) meridian was analyzed per eye and per scan. Pupillary diameter (PD) was within 15% for all pairwise comparisons. Anterior chamber depth, lens vault, anterior chamber width, angle opening distance, trabecular iris space area, and scleral spur angle (SSA500) were measured using manufacturer-provided image analysis software. Intraclass correlation (ICC), Wilcoxon signed-rank, and Bland-Altman analyses were performed to assess intradevice repeatability and interdevice agreement of measurements. Results: Thirty-two eyes of 21 subjects were analyzed. There was excellent agreement (ICC >0.98) and no significant difference (P > 0.05) in PD across all comparisons. Intradevice measurement repeatability was excellent for both the CASIA (ICC range 0.93-0.99) and ANTERION (ICC range 0.97-0.99). Interdevice measurement agreement was also excellent (ICC range 0.85-0.96). Measurements within and between devices were similar (P > 0.06) for all parameters except SSA500 (P = 0.03). Linear regression and Bland-Altman plots showed the relationship was consistent across the entire range of measurements. Conclusions: Intradevice measurement repeatability is excellent for the CASIA and ANTERION. Interdevice measurement agreement between the two devices exceeds metrics reported by previous comparison studies. Translational Relevance: Modern swept-source AS-OCT devices produce highly repeatable measurements of ocular biometric parameters that are nearly interchangeable across devices.

Deep Neural Network for Scleral Spur Detection in Anterior Segment OCT Images: The Chinese American Eye Study.

Purpose: To develop a deep neural network that detects the scleral spur in anterior segment optical coherence tomography (AS-OCT) images. Methods: Participants in the Chinese American Eye Study, a population-based study in Los Angeles, California, underwent complete ocular examinations, including AS-OCT imaging with the Tomey CASIA SS-1000. One human expert grader provided reference labels of scleral spur locations in all images. A convolutional neural network (CNN)-based on the ResNet-18 architecture was developed to detect the scleral spur in each image. Performance of the CNN model was assessed by calculating prediction errors, defined as the difference between the Cartesian coordinates of reference and CNN-predicted scleral spur locations. Prediction errors were compared with intragrader variability in detecting scleral spur locations by the reference grader. Results: The CNN was developed using a training dataset of 17,704 images and tested using an independent dataset of 921 images. The mean absolute prediction errors of the CNN model were 49.27 ± 42.07 µm for X-coordinates and 47.73 ± 39.70 µm for Y-coordinates. The mean absolute intragrader variability was 52.31 ± 47.75 µm for X-coordinates and 45.88 ± 45.06 µm for Y-coordinates. Distributions of prediction errors for the CNN and intragrader variability for the reference grader were similar for X-coordinates (P = 0.609) and Y-coordinates (P = 0.378). The mean absolute prediction error of the CNN was 73.08 ± 52.06 µm and the mean absolute intragrader variability was 73.92 ± 60.72 µm. Conclusions: A deep neural network can detect the scleral spur on AS-OCT images with performance similar to that of a human expert grader. Translational Relevance: Deep learning methods that automate scleral spur detection can facilitate qualitative and quantitative assessments of AS-OCT images.

Effect of Angle Narrowing on Sectoral Variation of Anterior Chamber Angle Width: The Chinese American Eye Study.

PURPOSE: To characterize the relationship between mean and sectoral variation of anterior chamber angle (ACA) width using anterior segment optical coherence tomography (AS-OCT). METHODS: Subjects aged 50 years or older were identified from the Chinese American Eye Study (CHES), a population-based epidemiological study in Los Angeles, CA. Each subject underwent a complete ocular examination including gonioscopy and AS-OCT imaging. Primary angle closure disease (PACD) was defined as inability to visualize pigmented trabecular meshwork in 3 or more quadrants. Four AS-OCT images from one eye per subject were analyzed and parameters describing ACA width were measured at 500 and 750 µm from the scleral spur: angle opening distance (AOD), trabecular iris space area (TISA), and scleral spur angle (SSA). The relationship between mean and sectoral variation of ACA width was assessed using locally-weighted scatterplot smoothing (LOWESS) regression and change-point analyses and Spearman correlation coefficients. RESULTS: 674 eyes (337 with PACD, 337 without PACD) from 674 subjects were analyzed. Overall, sectoral variation of ACA width decreased as mean ACA width decreased. This relationship was divided into two phases based on the change-point analysis. Sectoral variation of ACA width was strongly and significantly correlated (P < 0.001) with mean ACA width with below parameter-specific change points for most parameters: AOD500 (r = 0.599), AOD750 (r = 0.246), TISA500 (r = 0.734), TISA750 (r = 0.664), SSA500 (r = 0.661), SSA750 (r = 0.394). Correlations were weaker but still significant (P < 0.004) above these change points for most parameters: AOD500 (r = 0.321), AOD750 (r = 0.550), TISA500 (r = 0.122), TISA750 (r = 0.275), SSA500 (r = -0.036), SSA750 (r = 0.313). Correlations to the left and right of the change points strengthened when sectoral variation of ACA width was adjusted for mean ACA width. CONCLUSIONS: Correlations between mean and sectoral variation of ACA width strengthen as the severity of angle narrowing worsens. This relationship likely reflects anatomical changes related to chronic angle closure and may be relevant for refining current definitions and management of PACD.

Ocular Biometric Determinants of Anterior Chamber Angle Width in Chinese Americans: The Chinese American Eye Study.

PURPOSE: We sought to investigate anatomic mechanisms of angle narrowing by assessing ocular biometric determinants of anterior chamber angle width. DESIGN: Population-based cross-sectional study. METHODS: Subjects ≥50 years of age from the Chinese American Eye Study underwent a comprehensive ocular examination, including anterior segment optical coherence tomography imaging and ultrasound A-scan. Independent variables, including anterior chamber depth (ACD), lens vault (LV), iris curvature (IC), anterior chamber width, lens thickness, vitreous cavity depth, and axial length, and dependent variables, including angle opening distance, were measured in 1 randomly selected eye per subject. Univariable and multivariable regression models with standardized regression coefficients (SRCs) and semipartial correlation coefficients squares (SPCC2) were used to assess relative and unique contributions by independent variables to angle width. RESULTS: Two thousand two hundred twenty-five subjects (1433 women and 834 men) were included in the analysis. All biometric parameters except lens thickness differed between men and women (age-adjusted P < .001). In model 1A (R2 = 0.66), which included ACD, lens thickness, and vitreous cavity depth, ACD (SRC = 0.64, SPCC2 = 0.19) and IC (SRC = -0.26, SPCC2 = 0.041) were the strongest determinants of angle opening distance. In model 1B (R2 = 0.58), which included LV and axial length, LV (SRC = -0.46, SPCC2 = 0.1) and IC (SRC = -0.3, SPCC2 = 0.047) were the strongest determinants of angle opening distance. Determinants of angle width were similar in separate multivariable models for men and women. CONCLUSIONS: ACD, LV, and IC are the strongest determinants of angle width in Chinese Americans. Sex-related differences in angle width are explained by differences among biometric measurements. These results provide insights into anatomic mechanisms of angle narrowing and have important implications for quantitative assessments of angle closure eyes.

Deep Learning Classifiers for Automated Detection of Gonioscopic Angle Closure Based on Anterior Segment OCT Images.

PURPOSE: To develop and test deep learning classifiers that detect gonioscopic angle closure and primary angle closure disease (PACD) based on fully automated analysis of anterior segment OCT (AS-OCT) images. METHODS: Subjects were recruited as part of the Chinese-American Eye Study (CHES), a population-based study of Chinese Americans in Los Angeles, California, USA. Each subject underwent a complete ocular examination including gonioscopy and AS-OCT imaging in each quadrant of the anterior chamber angle (ACA). Deep learning methods were used to develop 3 competing multi-class convolutional neural network (CNN) classifiers for modified Shaffer grades 0, 1, 2, 3, and 4. Binary probabilities for closed (grades 0 and 1) and open (grades 2, 3, and 4) angles were calculated by summing over the corresponding grades. Classifier performance was evaluated by 5-fold cross-validation and on an independent test dataset. Outcome measures included area under the receiver operating characteristic curve (AUC) for detecting gonioscopic angle closure and PACD, defined as either 2 or 3 quadrants of gonioscopic angle closure per eye. RESULTS: A total of 4036 AS-OCT images with corresponding gonioscopy grades (1943 open, 2093 closed) were obtained from 791 CHES subjects. Three competing CNN classifiers were developed with a cross-validation dataset of 3396 images (1632 open, 1764 closed) from 664 subjects. The remaining 640 images (311 open, 329 closed) from 127 subjects were segregated into a test dataset. The best-performing classifier was developed by applying transfer learning to the ResNet-18 architecture. For detecting gonioscopic angle closure, this classifier achieved an AUC of 0.933 (95% confidence interval, 0.925-0.941) on the cross-validation dataset and 0.928 on the test dataset. For detecting PACD based on 2- and 3-quadrant definitions, the ResNet-18 classifier achieved AUCs of 0.964 and 0.952, respectively, on the test dataset. CONCLUSION: Deep learning classifiers effectively detect gonioscopic angle closure and PACD based on automated analysis of AS-OCT images. These methods could be used to automate clinical evaluations of the ACA and improve access to eye care in high-risk populations.

Differences in Anterior Chamber Angle Assessments Between Gonioscopy, EyeCam, and Anterior Segment OCT: The Chinese American Eye Study.

PURPOSE: To quantify interquadrant differences in anterior chamber angle (ACA) configuration assessed on gonioscopy, EyeCam, and anterior segment optical coherence tomography (AS-OCT) in a cohort of Chinese Americans. METHODS: Subjects aged 50 years or older were recruited from the Chinese American Eye Study (CHES), a population-based epidemiologic study in Los Angeles, CA. Each subject underwent a complete ocular exam, including gonioscopy, EyeCam, and AS-OCT, under dark ambient lighting. Gonioscopy and AS-OCT imaging and EyeCam image grading were performed by trained ophthalmologists. RESULTS: Seven hundred nine eyes from 709 subjects were analyzed. Less anatomic variation among the quadrants was detected on gonioscopy and EyeCam compared with AS-OCT (P < 0.05). The mean gonioscopy grade, EyeCam grade, and AS-OCT measurement for each quadrant varied by up to 10.3%, 6.4%, and 46.2% of the superior quadrant value, respectively. There were significant interquadrant differences (P < 0.05) among mean AOD750 measurements when grouping by quadrant and gonioscopy or EyeCam grade. Mean AOD750 measurements were smallest for the superior quadrant by between 14.3% and 38.1% and 17.4% and 37.9% on gonioscopy and EyeCam, respectively, compared with other quadrants. CONCLUSIONS: Gonioscopy and EyeCam significantly underrepresent anatomic variations of the ACA compared with AS-OCT. Gonioscopy or EyeCam grades from different quadrants do not appear to be comparable or interchangeable, which supports reconsideration of current definitions and methods used to diagnose and manage primary angle closure disease. TRANSLATIONAL RELEVANCE: AS-OCT imaging raises concerns about current clinical definitions and methods that rely gonioscopy or EyeCam to assess the ACA.