Deep Learning Estimation of 10-2 and 24-2 Visual Field Metrics Based on Thickness Maps from Macula OCT.

PURPOSE: To develop deep learning (DL) systems estimating visual function from macula-centered spectral-domain (SD) OCT images. DESIGN: Evaluation of a diagnostic technology. PARTICIPANTS: A total of 2408 10-2 visual field (VF) SD OCT pairs and 2999 24-2 VF SD OCT pairs collected from 645 healthy and glaucoma subjects (1222 eyes). METHODS: Deep learning models were trained on thickness maps from Spectralis macula SD OCT to estimate 10-2 and 24-2 VF mean deviation (MD) and pattern standard deviation (PSD). Individual and combined DL models were trained using thickness data from 6 layers (retinal nerve fiber layer [RNFL], ganglion cell layer [GCL], inner plexiform layer [IPL], ganglion cell-IPL [GCIPL], ganglion cell complex [GCC] and retina). Linear regression of mean layer thicknesses were used for comparison. MAIN OUTCOME MEASURES: Deep learning models were evaluated using R2 and mean absolute error (MAE) compared with 10-2 and 24-2 VF measurements. RESULTS: Combined DL models estimating 10-2 achieved R2 of 0.82 (95% confidence interval [CI], 0.68-0.89) for MD and 0.69 (95% CI, 0.55-0.81) for PSD and MAEs of 1.9 dB (95% CI, 1.6-2.4 dB) for MD and 1.5 dB (95% CI, 1.2-1.9 dB) for PSD. This was significantly better than mean thickness estimates for 10-2 MD (0.61 [95% CI, 0.47-0.71] and 3.0 dB [95% CI, 2.5-3.5 dB]) and 10-2 PSD (0.46 [95% CI, 0.31-0.60] and 2.3 dB [95% CI, 1.8-2.7 dB]). Combined DL models estimating 24-2 achieved R2 of 0.79 (95% CI, 0.72-0.84) for MD and 0.68 (95% CI, 0.53-0.79) for PSD and MAEs of 2.1 dB (95% CI, 1.8-2.5 dB) for MD and 1.5 dB (95% CI, 1.3-1.9 dB) for PSD. This was significantly better than mean thickness estimates for 24-2 MD (0.41 [95% CI, 0.26-0.57] and 3.4 dB [95% CI, 2.7-4.5 dB]) and 24-2 PSD (0.38 [95% CI, 0.20-0.57] and 2.4 dB [95% CI, 2.0-2.8 dB]). The GCIPL (R2 = 0.79) and GCC (R2 = 0.75) had the highest performance estimating 10-2 and 24-2 MD, respectively. CONCLUSIONS: Deep learning models improved estimates of functional loss from SD OCT imaging. Accurate estimates can help clinicians to individualize VF testing to patients.

OCT Angiography Artifacts in Glaucoma.

PURPOSE: To determine the prevalence of different types of artifacts seen in OCT angiography (OCTA) images of healthy and glaucoma eyes and evaluate the characteristics associated with poor-quality images. DESIGN: Retrospective study. PARTICIPANTS: A total of 649 eyes of 368 healthy, glaucoma suspect, and glaucoma patients. METHODS: Angiovue (Optovue Inc) high-density (HD) and non-HD optic nerve head and macula OCTA images of participants were evaluated by 4 expert reviewers for the presence of different artifacts, including eye movement, defocus, shadow, decentration, segmentation error, blink, and Z offset in the superficial vascular layer. Each OCTA scan was designated to have good or poor quality based on the presence of artifacts. The association of demographic and ocular characteristics with the likelihood of obtaining poor-quality OCTA images was evaluated. MAIN OUTCOME MEASURES: The prevalence of OCTA artifacts and the factors associated with increased likelihood of capturing poor-quality OCTA images. RESULTS: A total of 5263 OCTA images were evaluated. Overall, 33.9% of the OCTA images had poor quality. The majority of images with acceptable quality scores (QS ≥ 4) had no artifacts (76.6%). Other images had 1 (13.6%) or 2 or more artifacts (9.8%). Older age (P < 0.001), male gender (P = 0.045), worse visual field mean deviation (P < 0.001), absence of eye tracking (P < 0.001), and macular scan area (P < 0.001) were associated with a higher likelihood of obtaining poor-quality images. In images with acceptable QS, the commercially available quality measures including QS and signal strength index had the area under the receiver operating characteristic curves of 0.65 (95% confidence interval [CI], 0.62-0.69) and 0.70 (95% CI, 0.68-0.73) to detect good-quality images, respectively. CONCLUSIONS: OCTA artifacts associated with poor-quality images are frequent, and their prevalence is affected by ocular and patient characteristics. One should not rely solely on the quantitative assessments that are provided automatically by OCTA instruments. A systematic scan review should be conducted to ensure appropriate interpretation of OCTA images. Given the high prevalence of poor-quality OCTA images, the images should be reacquired whenever an apparent and correctable artifact is present on a captured image.

Ganglion Cell Complex Thickness and Macular Vessel Density Loss in Primary Open-Angle Glaucoma.

PURPOSE: To characterize the change rate of ganglion cell complex (GCC) thickness and macular vessel density in healthy, preperimetric glaucoma and primary open-angle glaucoma (POAG) eyes. DESIGN: Prospective, longitudinal study. PARTICIPANTS: One hundred thirty-nine eyes (23 healthy eyes, 36 preperimetric glaucoma eyes, and 80 POAG eyes) of 94 patients who had at least 3 visits were included from the Diagnostic Innovations in Glaucoma Study. The mean follow-up was 2.0 years for healthy eyes, 2.6 years for preperimetric glaucoma eyes, and 2.6 years for POAG eyes. METHODS: OCT angiography (OCTA)-based vessel density and OCT-based structural thickness of the same 3×3-mm2 GCC scan slab were evaluated. The dynamic range-based normalized rates of vessel density and thickness change were calculated and compared within each diagnostic group. The association between the rates of thickness and vessel density change and potential factors were evaluated. MAIN OUTCOME MEASURES: The rates of GCC thinning and macular vessel density loss. RESULTS: Significant rates of GCC thinning and macular vessel density decrease were detectable in all diagnostic groups (all P < 0.05). In healthy eyes and preperimetric glaucoma eyes, the normalized rates of GCC thinning and macular vessel density decrease were comparable (all P > 0.1). In contrast, the normalized rate (mean, 95% confidence interval) of macular vessel density decrease in the POAG eyes (-7.12 [-8.36, -5.88]%/year) was significantly faster than GCC thinning (-2.13 [-3.35, -0.90]%/year; P < 0.001). In the POAG group, more than two thirds of the eyes showed faster macular vessel density decrease than GCC thinning; faster macular vessel density decrease rate was associated significantly with worse glaucoma severity (P = 0.037). The association between GCC thinning rate and glaucoma severity was not significant (P = 0.586). Intraocular pressure during follow-up significantly affected the rate of GCC thinning in all groups (all P < 0.05) but showed no association with the rate of macular vessel density decrease. CONCLUSIONS: Both GCC thinning and macular vessel density decrease were detectable over time in all diagnostic groups. In POAG eyes, macular vessel density decrease was faster than GCC thinning and was associated with severity of disease. Macular vessel density is useful for evaluating glaucoma progression, particularly in more advanced disease.

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

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 R2 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 = 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 R2 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 (R2 = 0.60) and superior nasal (R2 = 0.67) sectors, moderate accuracy in inferior (R2 = 0.26) and superior (R2 = 0.35) sectors, and lower accuracy in the central (R2 = 0.15) and temporal (R2 = 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.

Measurement Floors and Dynamic Ranges of OCT and OCT Angiography in Glaucoma.

PURPOSE: To determine if OCT angiography (OCTA)-derived vessel density measurements can extend the available dynamic range for detecting glaucoma compared with spectral-domain (SD) OCT-derived thickness measurements. DESIGN: Observational, cross-sectional study. PARTICIPANTS: A total of 509 eyes from 38 healthy participants, 63 glaucoma suspects, and 193 glaucoma patients enrolled in the Diagnostic Innovations in Glaucoma Study. METHODS: Relative vessel density and tissue thickness measurement floors of perifoveal vessel density (pfVD), circumpapillary capillary density (cpCD), circumpapillary retinal nerve fiber (cpRNFL) thickness, ganglion cell complex (GCC) thickness, and visual field (VF) mean deviation (MD) were investigated and compared with a previously reported linear change point model (CPM) and locally weighted scatterplot smoothing curves. MAIN OUTCOME MEASURES: Estimated vessel density and tissue thickness measurement floors and corresponding dynamic ranges. RESULTS: Visual field MD ranged from -30.1 to 2.8 decibels (dB). No measurement floor was found for pfVD, which continued to decrease constantly until very advanced disease. A true floor (i.e., slope of approximately 0 after observed CPM change point) was detected for cpRNFL thickness only. The post-CPM estimated floors were 49.5±2.6 µm for cpRNFL thickness, 70.7±1.0 µm for GCC thickness, and 31.2±1.1% for cpCD. Perifoveal vessel density reached the post-CPM estimated floor later in the disease (VF MD, -25.8±3.8 dB) than cpCD (VF MD, -19.3±2.4 dB), cpRNFL thickness (VF MD, -17.5±3.3 dB), and GCC thickness (VF MD, -13.9±1.8 dB; P < 0.001). The number of available measurement steps from normal values to the CPM estimated floor was greatest for cpRNFL thickness (8.9), followed by GCC thickness (7.4), cpCD (4.5), and pfVD (3.8). CONCLUSIONS: In late-stage glaucoma, particularly when VF MD is worse than -14 dB, OCTA-measured pfVD is a promising tool for monitoring progression because it does not have a detectable measurement floor. However, the number of steps within the dynamic range of a parameter also needs to be considered. Although thickness parameters reached the floor earlier than OCTA-measured pfVD, there are more such steps with thickness than OCTA parameters.

Macular and Optic Nerve Head Vessel Density and Progressive Retinal Nerve Fiber Layer Loss in Glaucoma.

PURPOSE: To investigate prospectively the relationship between macular and peripapillary vessel density and progressive retinal nerve fiber layer (RNFL) loss in patients with mild to moderate primary open-angle glaucoma. DESIGN: Prospective, observational study. PARTICIPANTS: One hundred thirty-two eyes of 83 patients with glaucoma followed up for at least 2 years (average: 27.3±3.36 months). METHODS: Measurements of macular whole image vessel density (m-wiVD) and optic nerve head whole image vessel density (onh-wiVD) were acquired at baseline using OCT angiography. RNFL, minimum rim width (MRW), and ganglion cell plus inner plexiform layer (GCIPL) thickness were obtained semiannually using spectral-domain OCT. Random-effects models were used to investigate the relationship between baseline vessel density parameters and rates of RNFL loss after adjusting for the following confounding factors: baseline visual field mean deviation, MRW, GCIPL thickness, central corneal thickness (CCT), and mean intraocular pressure during follow-up and disc hemorrhage, with or without including baseline RNFL. MAIN OUTCOME MEASURES: Effects of m-wiVD and onh-wiVD on rates of RNFL loss over time. RESULTS: Average baseline RNFL thickness was 79.5±14.8 µm, which declined with a mean slope of -1.07 µm/year (95% confidence interval, -1.28 to -0.85). In the univariate model, including only a predictive factor and time and their interaction, each 1% lower m-wiVD and onh-wiVD was associated with a 0.11-µm/year (P < 0.001) and 0.06-µm/year (P = 0.031) faster rate of RNFL decline, respectively. A similar relationship between low m-wiVD and onh-wiVD and faster rates of RNFL loss was found using different multivariate models. The association between vessel density measurements and rate of RNFL loss was weak (r2 = 0.125 and r2 = 0.033 for m-wiVD and onh-wiVD, respectively). Average CCT also was a predictor for faster RNFL decline in both the univariate (0.11 µm/year; P < 0.001) and multivariate models. CONCLUSIONS: Lower baseline macular and optic nerve head (ONH) vessel density are associated with a faster rate of RNFL progression in mild to moderate glaucoma. Assessment of ONH and macular vessel density may add significant information to the evaluation of the risk of glaucoma progression and prediction of rates of disease worsening.

Rate and Pattern of Rim Area Loss in Healthy and Progressing Glaucoma Eyes.

PURPOSE: To characterize the rate and pattern of age-related and glaucomatous neuroretinal rim area changes in subjects of African and European descent. DESIGN: Prospective longitudinal study. PARTICIPANTS: Two hundred ninety-six eyes of 157 healthy subjects (88 patients of African descent and 69 of European descent) and 73 progressing glaucoma eyes of 67 subjects (24 patients of African descent and 43 of European descent) from the Diagnostic Innovations in Glaucoma Study and the African Descent and Glaucoma Evaluation Study were included. METHODS: Global and sectoral rim areas were measured using confocal laser scanning ophthalmoscopy. Masked stereophotograph review determined progression of glaucomatous optic disc damage. The rates of absolute rim area loss and percentage rim area loss in healthy and progressing glaucomatous eyes were compared using multivariate, nested, mixed-effects models. MAIN OUTCOME MEASURES: Rate of rim area loss over time. RESULTS: The median follow-up time was 5.0 years (interquartile range, 2.0-7.4 years) for healthy eyes and 8.3 years (interquartile range, 7.5-9.9 years) for progressing glaucoma eyes. The mean rate of global rim area loss was significantly faster in progressing glaucomatous eyes compared with healthy eyes for both rim area loss (-10.2×10(-3) vs. -2.8×10(-3) mm(2)/year, respectively; P < 0.001) and percentage rim area loss (-1.1% vs. -0.2%/year, respectively; P < 0.001), but considerable overlap existed between the 2 groups. Sixty-three percent of progressing glaucoma eyes had a rate of change faster than the fifth quantile of healthy eyes. For both healthy and progressing eyes, the pattern of rim area loss and percentage rim area loss were similar, tending to be fastest in the superior temporal and inferior temporal sectors. The rate of change was similar in progressing eyes of patients of African or European descent. CONCLUSIONS: Compared with healthy eyes, the mean rate of global rim area loss was 3.7 times faster and the mean rate of global percentage rim area loss was 5.4 times faster in progressing glaucoma eyes. A reference database of healthy eyes can be used to help clinicians distinguish age-related rim area loss from rim area loss resulting from glaucoma.

African Descent and Glaucoma Evaluation Study (ADAGES): Racial Differences in Optic Disc Hemorrhage and Beta-Zone Parapapillary Atrophy.

PURPOSE: To investigate the differences in the frequency of optic disc hemorrhage (DH) and prevalence of beta-zone parapapillary atrophy (ßPPA) between individuals of African descent (AD) and European descent (ED). DESIGN: Prospective, multicenter, observational cohort. PARTICIPANTS: A total of 1950 eyes of 1172 participants of the African Descent and Glaucoma Evaluation Study (ADAGES). METHODS: Stereoscopic disc photographs of subjects with and without glaucomatous optic neuropathy (GON) followed during the first 13 years of the ADAGES underwent masked review searching for DH and ßPPA. A total of 928 eyes (non-GON, 581; GON, 347) of 551 AD patients (non-GON, 334; GON, 217) and 1022 eyes (non-GON, 568; GON, 454) of 611 ED patients (non-GON, 334; GON, 277) were included. We compared the number of eyes with detected DH at any time during follow-up and eyes with ßPPA between the AD and ED groups. The analyses were then adjusted for clinical parameters using multivariable logistic regression. MAIN OUTCOME MEASURES: Differences in frequency of DH and prevalence of ßPPA. RESULTS: A total of 9395 stereoscopic disc photographs were reviewed. More ED eyes experience DH than AD eyes (49/1022 [4.8%] vs. 10/928 eyes [1.1%], respectively; P < 0.001), whereas ßPPA had higher prevalence in AD eyes (675 eyes [72%] vs. 659 eyes [64%]; P < 0.001). In the final multivariable model, after controlling for confounders, AD eyes were less likely to have at least 1 detected DH than ED eyes (odds ratio [OR], 0.21; 95% CI, 0.10-0.45; P < 0.001) but were more likely to have ßPPA than ED eyes (OR, 1.55; 95% CI, 1.12-2.14; P = 0.008). CONCLUSIONS: Subjects of ED are at higher risk for developing DH compared with AD subjects, whereas AD subjects have greater prevalence of ßPPA. These findings suggest that there are structural differences within the optic nerve complex between these groups. Further research is needed to determine whether racial differences in the frequency of DH and prevalence of ßPPA affect the likelihood of glaucomatous progression.

Detecting glaucomatous change in visual fields: Analysis with an optimization framework.

Detecting glaucomatous progression is an important aspect of glaucoma management. The assessment of longitudinal series of visual fields, measured using Standard Automated Perimetry (SAP), is considered the reference standard for this effort. We seek efficient techniques for determining progression from longitudinal visual fields by formulating the problem as an optimization framework, learned from a population of glaucoma data. The longitudinal data from each patient's eye were used in a convex optimization framework to find a vector that is representative of the progression direction of the sample population, as a whole. Post-hoc analysis of longitudinal visual fields across the derived vector led to optimal progression (change) detection. The proposed method was compared to recently described progression detection methods and to linear regression of instrument-defined global indices, and showed slightly higher sensitivities at the highest specificities than other methods (a clinically desirable result). The proposed approach is simpler, faster, and more efficient for detecting glaucomatous changes, compared to our previously proposed machine learning-based methods, although it provides somewhat less information. This approach has potential application in glaucoma clinics for patient monitoring and in research centers for classification of study participants.

Evaluation of progressive neuroretinal rim loss as a surrogate end point for development of visual field loss in glaucoma.

PURPOSE: To evaluate the validity of using progressive loss of neuroretinal rim area as a surrogate end point for the development of visual field loss in glaucoma. DESIGN: Prospective, observational cohort study. PARTICIPANTS: The study group included 492 eyes of 328 patients classified with suspected glaucoma at the baseline visit. These eyes had an average of 7.4±2.8 confocal scanning laser ophthalmoscopy (CSLO) images during a mean follow-up time of 6.6±1.6 years. METHODS: Rim area measurements were acquired with CSLO during follow-up. The visual field end point was considered the development of 3 consecutive abnormal visual fields on standard automated perimetry. Strong predictive ability and large proportion of treatment effect (PTE) explained are requisites for a suitable surrogate end point. A joint longitudinal survival model was used to evaluate the ability of rates of rim area loss in predicting visual field development, adjusting for confounding variables (baseline age, race, and corneal thickness and follow-up measurements of intraocular pressure [IOP] and pattern standard deviation). The PTE was calculated by comparing the effect of IOP on the risk of development of visual field loss when incorporating rim area loss in the same model with the effect of IOP in the model excluding rim area measurements. MAIN OUTCOME MEASURES: Predictive strength was measured by survival-adapted R(2) and PTE. RESULTS: Sixty-two of 492 eyes (13%) developed visual field loss during follow-up. The mean rate of rim area change in eyes that developed visual field loss was -0.011 mm(2)/year versus -0.003 mm(2)/year in eyes that did not (P<0.001). In the multivariable model, each 0.01 mm(2)/year faster rate of rim area loss was associated with a 2.94 higher risk of visual field loss (hazard ratio, 2.94; 95% confidence interval, 1.38-6.23; P = 0.005). R(2) values were 62% and 81% for univariable and multivariable models, respectively. The PTE was 65%. CONCLUSIONS: Progressive rim area loss was highly predictive of the development of visual field loss in glaucoma and explained a significant PTE on the clinically relevant outcome. These findings suggest that rim area measurements may be suitable surrogate end points in glaucoma clinical trials.

Wide-Field Optical Coherence Tomography Imaging Improves Rate of Change Detection in Progressing Glaucomatous Eyes Compared With Standard-Field Imaging.

Purpose: To compare rates of retinal nerve fiber layer change over time in healthy, eyes with nonprogressing glaucoma and eyes with progressing glaucoma using single wide-field (SWF) and optic nerve head (ONH) cube scan optical coherence tomography (OCT) images. Methods: Forty-five eyes of 25 healthy individuals and 263 eyes of 161 glaucoma patients from the Diagnostic Innovations in Glaucoma Study were included. All eyes underwent 24-2 visual field testing and OCT (Spectralis SD-OCT) ONH and macular imaging. SWF images (up to 43° × 28°) were created by stitching together ONH cube scans centered on the optic disc and macular cube scans centered on the fovea. Visual field progression was defined as guided progression analysis likely progression and/or a significant (P < 0.01) mean deviation slope of less than -1.0 dB/year. Mixed effects models were used to compare rates of change. Highly myopic eyes were included. Results: Thirty glaucomatous eyes were classified as progressing. In eyes with glaucoma, mean global rate of change was -1.22 µm/year (P < 0.001) using SWF images and -0.83 µm/year (P = 0.003) using ONH cube scans. Rate of change was significantly greater in eyes with progressing glaucoma compared with eyes with nonprogressing glaucoma (-1.51 µm/year vs. -1.24 µm/year; P = 0.002) using SWF images and was similar using ONH cube scans (P = 0.27). Conclusions: In this cohort that includes eyes with and without high axial myopia, the mean rate of retinal nerve fiber layer thinning measured using SWF images was faster in eyes with progressing glaucoma than in eyes with nonprogressing glaucoma. Wide-field OCT images including the ONH and macula can be effective for monitoring glaucomatous progression in patients with and without high myopia.

Diagnostic Accuracy of Optic Nerve Head and Macula OCT Parameters for Detecting Glaucoma in Eyes With and Without High Axial Myopia.

PURPOSE: To characterize structural differences and assess the diagnostic accuracy of optic nerve head (ONH) and macula optical coherence tomography (OCT) parameters to detect glaucoma in eyes with and without high axial myopia. DESIGN: Cross-sectional study. METHODS: Three hundred sixty-eight glaucoma and 411 healthy eyes with no axial myopia, 393 glaucoma and 271 healthy eyes with mild axial myopia and 124 glaucoma and 85 healthy eyes with high axial myopia were included. Global and sectoral peripapillary retinal nerve fiber layer thickness (pRNFLT), Bruch's membrane opening minimum rim width (BMO-MRW), ganglion cell inner plexiform layer thickness (GCIPLT), and macula RNFLT (mRNFLT) were compared and the diagnostic accuracy for glaucoma detection was evaluated using the adjusted area under the receiver operating characteristic curve (AUC). RESULTS: Diagnostic accuracy for ONH and macula parameters to detect glaucoma was generally high and differed by myopia group. For ONH parameters the diagnostic accuracy was highest for global (AUC = 0.95) and inferotemporal (AUC = 0.91) pRNFLT for high myopes and global BMO-MRW for nonmyopes (AUC = 1.0) and mild myopes (AUC = 0.97). For macula parameters, the diagnostic accuracy was higher in high myopes with 6 of the 11 GCIPLT global/sectors having adjusted AUCs > 0.90 compared to nonhigh myopes with no AUCs > 0.90. In all myopia groups, mRNFLT had lower AUCs than GCIPLT. CONCLUSIONS: The diagnostic accuracy for pRNFL and GCIPL was high for high axial myopic eyes and shows promise for glaucoma detection in high myopes. Further analysis is needed to determine whether the high diagnostic accuracy can be confirmed in other populations.

Central visual field damage in glaucoma eyes with choroidal microvasculature dropout with and without high axial myopia.

PURPOSE: To characterise the relationship between a deep-layer microvasculature dropout (MvD) and central visual field (VF) damage in primary open-angle glaucoma (POAG) patients with and without high axial myopia. DESIGN: Cross-sectional study. METHODS: Seventy-one eyes (49 patients) with high axial myopia and POAG and 125 non-highly myopic POAG eyes (97 patients) were enrolled. Presence, area and angular circumference of juxtapapillary MvD were evaluated on optical coherence tomography angiography B-scans and en-face choroidal images. RESULTS: Juxtapapillary MvD was detected more often in the highly myopic POAG eyes (43 eyes, 86%) than in the non-highly myopic eyes (73 eyes, 61.9%; p=0.002). In eyes with MvD, MvD area and angular circumference (95% CI) were significantly larger in the highly myopic eyes compared with the non-highly myopic eyes (area: (0.69 (0.40, 0.98) mm2 vs 0.31 (0.19, 0.42) mm2, p=0.011) and (angular circumference: 84.3 (62.9, 105.8) vs 74.5 (58.3, 90.9) degrees, p<0.001), respectively. 24-2 VF mean deviation (MD) was significantly worse in eyes with MvD compared with eyes without MvD in both groups (p<0.001). After adjusting for 24-2 MD VF, central VF defects were more frequently found in eyes with MvD compared with eyes without MvD (82.7% vs 60.9%, p<0.001). In multivariable analysis, higher intraocular pressure, worse 24-2 VF MD, longer axial length and greater MvD area and angular circumference were associated with worse 10-2 VF MD. CONCLUSIONS: MvD was more prevalent and larger in POAG eyes with high myopia than in non-highly myopic POAG eyes. In both groups, eyes with MvD showed worse glaucoma severity and more central VF defects.

Deep Learning Identifies High-Quality Fundus Photographs and Increases Accuracy in Automated Primary Open Angle Glaucoma Detection.

Purpose: To develop and evaluate a deep learning (DL) model to assess fundus photograph quality, and quantitatively measure its impact on automated POAG detection in independent study populations. Methods: Image quality ground truth was determined by manual review of 2815 fundus photographs of healthy and POAG eyes from the Diagnostic Innovations in Glaucoma Study and African Descent and Glaucoma Evaluation Study (DIGS/ADAGES), as well as 11,350 from the Ocular Hypertension Treatment Study (OHTS). Human experts assessed a photograph as high quality if of sufficient quality to determine POAG status and poor quality if not. A DL quality model was trained on photographs from DIGS/ADAGES and tested on OHTS. The effect of DL quality assessment on DL POAG detection was measured using area under the receiver operating characteristic (AUROC). Results: The DL quality model yielded an AUROC of 0.97 for differentiating between high- and low-quality photographs; qualitative human review affirmed high model performance. Diagnostic accuracy of the DL POAG model was significantly greater (P < 0.001) in good (AUROC, 0.87; 95% CI, 0.80-0.92) compared with poor quality photographs (AUROC, 0.77; 95% CI, 0.67-0.88). Conclusions: The DL quality model was able to accurately assess fundus photograph quality. Using automated quality assessment to filter out low-quality photographs increased the accuracy of a DL POAG detection model. Translational Relevance: Incorporating DL quality assessment into automated review of fundus photographs can help to decrease the burden of manual review and improve accuracy for automated DL POAG detection.

Racial Differences in Diagnostic Accuracy of Retinal Nerve Fiber Layer Thickness in Primary Open-Angle Glaucoma.

Purpose: To evaluate the diagnostic accuracy of retinal nerve fiber layer thickness (RNFLT) by spectral-domain optical coherence tomography (OCT) in primary open-angle glaucoma (POAG) in eyes of African (AD) and European descent (ED). Design: Comparative diagnostic accuracy analysis by race. Participants: 379 healthy eyes (125 AD and 254 ED) and 442 glaucomatous eyes (226 AD and 216 ED) from the Diagnostic Innovations in Glaucoma Study and the African Descent and Glaucoma Evaluation Study. Methods: Spectralis (Heidelberg Engineering GmbH) and Cirrus (Carl Zeiss Meditec) OCT scans were taken within one year from each other. Main Outcome Measures: Diagnostic accuracy of RNFLT measurements. Results: Diagnostic accuracy for Spectralis-RNFLT was significantly lower in eyes of AD compared to those of ED (area under the receiver operating curve [AUROC]: 0.85 and 0.91, respectively, P=0.04). Results for Cirrus-RNFLT were similar but did not reach statistical significance (AUROC: 0.86 and 0.90 in AD and ED, respectively, P =0.33). Adjustments for age, central corneal thickness, axial length, disc area, visual field mean deviation, and intraocular pressure yielded similar results. Conclusions: OCT-RNFLT has lower diagnostic accuracy in eyes of AD compared to those of ED. This finding was generally robust across two OCT instruments and remained after adjustment for many potential confounders. Further studies are needed to explore the potential sources of this difference.

Proactive Decision Support for Glaucoma Treatment: Predicting Surgical Interventions with Clinically Available Data.

A longitudinal ophthalmic dataset was used to investigate multi-modal machine learning (ML) models incorporating patient demographics and history, clinical measurements, optical coherence tomography (OCT), and visual field (VF) testing in predicting glaucoma surgical interventions. The cohort included 369 patients who underwent glaucoma surgery and 592 patients who did not undergo surgery. The data types used for prediction included patient demographics, history of systemic conditions, medication history, ophthalmic measurements, 24-2 VF results, and thickness measurements from OCT imaging. The ML models were trained to predict surgical interventions and evaluated on independent data collected at a separate study site. The models were evaluated based on their ability to predict surgeries at varying lengths of time prior to surgical intervention. The highest performing predictions achieved an AUC of 0.93, 0.92, and 0.93 in predicting surgical intervention at 1 year, 2 years, and 3 years, respectively. The models were also able to achieve high sensitivity (0.89, 0.77, 0.86 at 1, 2, and 3 years, respectively) and specificity (0.85, 0.90, and 0.91 at 1, 2, and 3 years, respectively) at an 0.80 level of precision. The multi-modal models trained on a combination of data types predicted surgical interventions with high accuracy up to three years prior to surgery and could provide an important tool to predict the need for glaucoma intervention.

One-Vote Veto: Semi-Supervised Learning for Low-Shot Glaucoma Diagnosis.

Convolutional neural networks (CNNs) are a promising technique for automated glaucoma diagnosis from images of the fundus, and these images are routinely acquired as part of an ophthalmic exam. Nevertheless, CNNs typically require a large amount of well-labeled data for training, which may not be available in many biomedical image classification applications, especially when diseases are rare and where labeling by experts is costly. This article makes two contributions to address this issue: 1) It extends the conventional Siamese network and introduces a training method for low-shot learning when labeled data are limited and imbalanced, and 2) it introduces a novel semi-supervised learning strategy that uses additional unlabeled training data to achieve greater accuracy. Our proposed multi-task Siamese network (MTSN) can employ any backbone CNN, and we demonstrate with four backbone CNNs that its accuracy with limited training data approaches the accuracy of backbone CNNs trained with a dataset that is 50 times larger. We also introduce One-Vote Veto (OVV) self-training, a semi-supervised learning strategy that is designed specifically for MTSNs. By taking both self-predictions and contrastive predictions of the unlabeled training data into account, OVV self-training provides additional pseudo labels for fine-tuning a pre-trained MTSN. Using a large (imbalanced) dataset with 66,715 fundus photographs acquired over 15 years, extensive experimental results demonstrate the effectiveness of low-shot learning with MTSN and semi-supervised learning with OVV self-training. Three additional, smaller clinical datasets of fundus images acquired under different conditions (cameras, instruments, locations, populations) are used to demonstrate the generalizability of the proposed methods.

Multimodal Deep Learning Classifier for Primary Open Angle Glaucoma Diagnosis Using Wide-Field Optic Nerve Head Cube Scans in Eyes With and Without High Myopia.

PRCIS: An optical coherence tomography (OCT)-based multimodal deep learning (DL) classification model, including texture information, is introduced that outperforms single-modal models and multimodal models without texture information for glaucoma diagnosis in eyes with and without high myopia. BACKGROUND/AIMS: To evaluate the diagnostic accuracy of a multimodal DL classifier using wide OCT optic nerve head cube scans in eyes with and without axial high myopia. MATERIALS AND METHODS: Three hundred seventy-one primary open angle glaucoma (POAG) eyes and 86 healthy eyes, all without axial high myopia [axial length (AL) ≤ 26 mm] and 92 POAG eyes and 44 healthy eyes, all with axial high myopia (AL > 26 mm) were included. The multimodal DL classifier combined features of 3 individual VGG-16 models: (1) texture-based en face image, (2) retinal nerve fiber layer (RNFL) thickness map image, and (3) confocal scanning laser ophthalmoscope (cSLO) image. Age, AL, and disc area adjusted area under the receiver operating curves were used to compare model accuracy. RESULTS: Adjusted area under the receiver operating curve for the multimodal DL model was 0.91 (95% CI = 0.87, 0.95). This value was significantly higher than the values of individual models [0.83 (0.79, 0.86) for texture-based en face image; 0.84 (0.81, 0.87) for RNFL thickness map; and 0.68 (0.61, 0.74) for cSLO image; all P ≤ 0.05]. Using only highly myopic eyes, the multimodal DL model showed significantly higher diagnostic accuracy [0.89 (0.86, 0.92)] compared with texture en face image [0.83 (0.78, 0.85)], RNFL [0.85 (0.81, 0.86)] and cSLO image models [0.69 (0.63, 0.76)] (all P ≤ 0.05). CONCLUSIONS: Combining OCT-based RNFL thickness maps with texture-based en face images showed a better ability to discriminate between healthy and POAG than thickness maps alone, particularly in high axial myopic eyes.

Long-term reproducibility of optical coherence tomography angiography in healthy and stable glaucomatous eyes.

BACKGROUND/AIMS: To assess and compare long-term reproducibility of optic nerve head (ONH) and macula optical coherence tomography angiography (OCTA) vascular parameters and optical coherence tomography (OCT) thickness parameters in stable primary open-angle glaucoma (POAG), glaucoma suspect and healthy eyes. METHODS: Eighty-eight eyes (15 healthy, 38 glaucoma suspect and 35 non-progressing POAG) of 68 subjects who had at least three visits within 1-1.5 years with OCTA and OCT imaging (Angiovue; Optovue, Fremont, California, USA) on the same day were included. A series of vascular and thickness parameters were measured including macular parafoveal vessel density (pfVD), ONH circumpapillary capillary density (cpCD), macular parafoveal ganglion cell complex (pfGCC) and ONH circumpapillary retinal nerve fibre layer (cpRNFL). A random effects analysis of variance model was used to estimate intraclass correlation (ICC) coefficients and long-term variability estimates. RESULTS: ICC was lower for OCTA (pfVD 0.823 (95% CI 0.736 to 0.888) and cpCD 0.871 (0.818 to 0.912)) compared with OCT (pfGCC 0.995 (0.993 to 0.997) and cpRNFL 0.975 (0.964 to 0.984)). Within-subject test-retest SD was 1.17% and 1.22% for pfVD and cpCD, and 0.57 and 1.22 µm for pfGCC and cpRNFL. Older age and lower signal strength index were associated with decreasing long-term variability of vessel densities. CONCLUSIONS: OCTA-measured macula and ONH vascular parameters have good long-term reproducibility, supporting the use of this instrument for longitudinal analysis. OCTA long-term reproducibility is less than OCT-measured thickness reproducibility. This needs to be taken into consideration when serial OCTA images are evaluated for change. TRIAL REGISTRATION NUMBER: NCT00221897.