AI-based clinical assessment of optic nerve head robustness superseding biomechanical testing.

BACKGROUND/AIMS: To use artificial intelligence (AI) to: (1) exploit biomechanical knowledge of the optic nerve head (ONH) from a relatively large population; (2) assess ONH robustness (ie, sensitivity of the ONH to changes in intraocular pressure (IOP)) from a single optical coherence tomography (OCT) volume scan of the ONH without the need for biomechanical testing and (3) identify what critical three-dimensional (3D) structural features dictate ONH robustness. METHODS: 316 subjects had their ONHs imaged with OCT before and after acute IOP elevation through ophthalmo-dynamometry. IOP-induced lamina cribrosa (LC) deformations were then mapped in 3D and used to classify ONHs. Those with an average effective LC strain superior to 4% were considered fragile, while those with a strain inferior to 4% robust. Learning from these data, we compared three AI algorithms to predict ONH robustness strictly from a baseline (undeformed) OCT volume: (1) a random forest classifier; (2) an autoencoder and (3) a dynamic graph convolutional neural network (DGCNN). The latter algorithm also allowed us to identify what critical 3D structural features make a given ONH robust. RESULTS: All three methods were able to predict ONH robustness from a single OCT volume scan alone and without the need to perform biomechanical testing. The DGCNN (area under the curve (AUC): 0.76±0.08) outperformed the autoencoder (AUC: 0.72±0.09) and the random forest classifier (AUC: 0.69±0.05). Interestingly, to assess ONH robustness, the DGCNN mainly used information from the scleral canal and the LC insertion sites. CONCLUSIONS: We propose an AI-driven approach that can assess the robustness of a given ONH solely from a single OCT volume scan of the ONH, and without the need to perform biomechanical testing. Longitudinal studies should establish whether ONH robustness could help us identify fast visual field loss progressors. PRECIS: Using geometric deep learning, we can assess optic nerve head robustness (ie, sensitivity to a change in IOP) from a standard OCT scan that might help to identify fast visual field loss progressors.

Adduction induces large optic nerve head deformations in subjects with normal-tension glaucoma.

PURPOSE: To assess intraocular pressure (IOP)-induced and gaze-induced optic nerve head (ONH) strains in subjects with high-tension glaucoma (HTG) and normal-tension glaucoma (NTG). DESIGN: Clinic-based cross-sectional study. METHODS: The ONH from one eye of 228 subjects (114 subjects with HTG (pre-treatment IOP≥21 mm Hg) and 114 with NTG (pre-treatment IOP<21 mm Hg)) was imaged with optical coherence tomography (OCT) under the following conditions: (1) OCT primary gaze, (2) 20° adduction from OCT primary gaze, (3) 20° abduction from OCT primary gaze and (4) OCT primary gaze with acute IOP elevation (to approximately 33 mm Hg). We then performed digital volume correlation analysis to quantify IOP-induced and gaze-induced ONH tissue deformations and strains. RESULTS: Across all subjects, adduction generated high effective strain (4.4%±2.3%) in the LC tissue with no significant difference (p>0.05) with those induced by IOP elevation (4.5%±2.4%); while abduction generated significantly lower (p=0.01) effective strain (3.1%±1.9%). The lamina cribrosa (LC) of HTG subjects exhibited significantly higher effective strain than those of NTG subjects under IOP elevation (HTG: 4.6%±1.7% vs NTG: 4.1%±1.5%, p<0.05). Conversely, the LC of NTG subjects exhibited significantly higher effective strain than those of HTG subjects under adduction (NTG: 4.9%±1.9% vs HTG: 4.0%±1.4%, p<0.05). CONCLUSION: We found that NTG subjects experienced higher strains due to adduction than HTG subjects, while HTG subjects experienced higher strain due to IOP elevation than NTG subjects-and that these differences were most pronounced in the LC tissue.

Deep Learning Classification of Angle Closure based on Anterior Segment OCT.

PURPOSE: To assess the performance and generalizability of a convolutional neural network (CNN) model for objective and high-throughput identification of primary angle-closure disease (PACD) as well as PACD stage differentiation on anterior segment swept-source OCT (AS-OCT). DESIGN: Cross-sectional. PARTICIPANTS: Patients from 3 different eye centers across China and Singapore were recruited for this study. Eight hundred forty-one eyes from the 2 Chinese centers were divided into 170 control eyes, 488 PACS, and 183 PAC + PACG eyes. An additional 300 eyes were recruited from Singapore National Eye Center as a testing data set, divided into 100 control eyes, 100 PACS, and 100 PAC + PACG eyes. METHODS: Each participant underwent standardized ophthalmic examination and was classified by the presiding physician as either control, primary angle-closure suspect (PACS), primary angle closure (PAC), or primary angle-closure glaucoma (PACG). Deep Learning model was used to train 3 different CNN classifiers: classifier 1 aimed to separate control versus PACS versus PAC + PACG; classifier 2 aimed to separate control versus PACD; and classifier 3 aimed to separate PACS versus PAC + PACG. All classifiers were evaluated on independent validation sets from the same region, China and further tested using data from a different country, Singapore. MAIN OUTCOME MEASURES: Area under receiver operator characteristic curve (AUC), precision, and recall. RESULTS: Classifier 1 achieved an AUC of 0.96 on validation set from the same region, but dropped to an AUC of 0.84 on test set from a different country. Classifier 2 achieved the most generalizable performance with an AUC of 0.96 on validation set and AUC of 0.95 on test set. Classifier 3 showed the poorest performance, with an AUC of 0.83 and 0.64 on test and validation data sets, respectively. CONCLUSIONS: Convolutional neural network classifiers can effectively distinguish PACD from controls on AS-OCT with good generalizability across different patient cohorts. However, their performance is moderate when trying to distinguish PACS versus PAC + PACG. FINANCIAL DISCLOSURES: The authors have no proprietary or commercial interest in any materials discussed in this article.

Investigating the determinants of iridolenticular contact area: a novel parameter for angle closure.

BACKGROUND/AIMS: To identify ocular determinants of iridolenticular contact area (ILCA), a recently introduced swept-source optical coherence tomography (SSOCT) derived parameter, and assess the association between ILCA and angle closure. METHODS: In this population-based cross-sectional study, right eyes of 464 subjects underwent SSOCT (SS-1000, CASIA, Tomey Corporation, Nagoya, Japan) imaging in the dark. Eight out of 128 cross-sectional images (evenly spaced 22.5° apart) were selected for analysis. Matlab (Matworks, Massachusetts, USA) was used to measure ILCA, defined as the circumferential extent of contact area between the pigmented iris epithelium and anterior lens surface. Gonioscopic angle closure (GAC) was defined as non-visibility of the posterior trabecular meshwork in two or more angle quadrants. RESULTS: The mean age of subjects was 62±6.6 years, with the majority being female (65.5%). 143/464 subjects (28.6%) had GAC. In multivariable linear regression analysis, ILCA was significantly associated with anterior chamber width (ß=1.03, p=0.003), pupillary diameter (ß=-1.9, p<0.001) and iris curvature (ß=-17.35, p<0.001). ILCA was smaller in eyes with GAC compared with those with open angles (4.28±1.6 mm2 vs 6.02±2.71 mm2, p<0.001). ILCA was independently associated with GAC (ß=-0.03, p<0.001), iridotrabecular contact index (ß=-6.82, p<0.001) or angle opening distance (ß=0.02, p<0.001) after adjusting for covariates. The diagnostic performance of ILCA for detecting GAC was acceptable (AUC=0.69). CONCLUSIONS: ILCA is a significant predictor of angle closure independent of other biometric factors and may reflect unique anatomical information associated with pupillary block. ILCA represents a novel biometric risk factor in eyes with angle closure.

Three-Dimensional Structural Phenotype of the Optic Nerve Head as a Function of Glaucoma Severity.

Importance: The 3-dimensional (3-D) structural phenotype of glaucoma as a function of severity was thoroughly described and analyzed, enhancing understanding of its intricate pathology beyond current clinical knowledge. Objective: To describe the 3-D structural differences in both connective and neural tissues of the optic nerve head (ONH) between different glaucoma stages using traditional and artificial intelligence-driven approaches. Design, Setting, and Participants: This cross-sectional, clinic-based study recruited 541 Chinese individuals receiving standard clinical care at Singapore National Eye Centre, Singapore, and 112 White participants of a prospective observational study at Vilnius University Hospital Santaros Klinikos, Vilnius, Lithuania. The study was conducted from May 2022 to January 2023. All participants had their ONH imaged using spectral-domain optical coherence tomography and had their visual field assessed by standard automated perimetry. Main Outcomes and Measures: (1) Clinician-defined 3-D structural parameters of the ONH and (2) 3-D structural landmarks identified by geometric deep learning that differentiated ONHs among 4 groups: no glaucoma, mild glaucoma (mean deviation [MD], ≥-6.00 dB), moderate glaucoma (MD, -6.01 to -12.00 dB), and advanced glaucoma (MD, <-12.00 dB). Results: Study participants included 213 individuals without glaucoma (mean age, 63.4 years; 95% CI, 62.5-64.3 years; 126 females [59.2%]; 213 Chinese [100%] and 0 White individuals), 204 with mild glaucoma (mean age, 66.9 years; 95% CI, 66.0-67.8 years; 91 females [44.6%]; 178 Chinese [87.3%] and 26 White [12.7%] individuals), 118 with moderate glaucoma (mean age, 68.1 years; 95% CI, 66.8-69.4 years; 49 females [41.5%]; 97 Chinese [82.2%] and 21 White [17.8%] individuals), and 118 with advanced glaucoma (mean age, 68.5 years; 95% CI, 67.1-69.9 years; 43 females [36.4%]; 53 Chinese [44.9%] and 65 White [55.1%] individuals). The majority of ONH structural differences occurred in the early glaucoma stage, followed by a plateau effect in the later stages. Using a deep neural network, 3-D ONH structural differences were found to be present in both neural and connective tissues. Specifically, a mean of 57.4% (95% CI, 54.9%-59.9%, for no to mild glaucoma), 38.7% (95% CI, 36.9%-40.5%, for mild to moderate glaucoma), and 53.1 (95% CI, 50.8%-55.4%, for moderate to advanced glaucoma) of ONH landmarks that showed major structural differences were located in neural tissues with the remaining located in connective tissues. Conclusions and Relevance: This study uncovered complex 3-D structural differences of the ONH in both neural and connective tissues as a function of glaucoma severity. Future longitudinal studies should seek to establish a connection between specific 3-D ONH structural changes and fast visual field deterioration and aim to improve the early detection of patients with rapid visual field loss in routine clinical care.

How Myopia and Glaucoma Influence the Biomechanical Susceptibility of the Optic Nerve Head.

Purpose: The purpose of this study was to assess optic nerve head (ONH) deformations following acute intraocular pressure (IOP) elevations and horizontal eye movements in control eyes, highly myopic (HM) eyes, HM eyes with glaucoma (HMG), and eyes with pathologic myopia (PM) alone or PM with staphyloma (PM + S). Methods: We studied 282 eyes, comprising of 99 controls (between +2.75 and -2.75 diopters), 51 HM (< -5 diopters), 35 HMG, 21 PM, and 75 PM + S eyes. For each eye, we imaged the ONH using spectral-domain optical coherence tomography (OCT) under the following conditions: (1) primary gaze, (2) 20 degrees adduction, (3) 20 degrees abduction, and (4) primary gaze with acute IOP elevation (to ∼35 mm Hg) achieved through ophthalmodynamometry. We then computed IOP- and gaze-induced ONH displacements and effective strains. Effective strains were compared across groups. Results: Under IOP elevation, we found that HM eyes exhibited significantly lower strains (3.9 ± 2.4%) than PM eyes (6.9 ± 5.0%, P < 0.001), HMG eyes (4.7 ± 1.8%, P = 0.04), and PM + S eyes (7.0 ± 5.2%, P < 0.001). Under adduction, we found that HM eyes exhibited significantly lower strains (4.8% ± 2.7%) than PM + S eyes (6.0 ± 3.1%, P = 0.02). We also found that eyes with higher axial length were associated with higher strains. Conclusions: Our study revealed that eyes with HMG experienced significantly greater strains under IOP compared to eyes with HM. Furthermore, eyes with PM + S had the highest strains on the ONH of all groups.

Changes in Anterior Segment Parameters After Laser Peripheral Iridotomy in Caucasian Eyes With Different Primary Angle Closure Mechanisms.

PRCIS: Subgrouping of angle closure mechanisms based on the swept-source optical coherence tomography images may help to identify the predominant underlying anatomic mechanism, evaluate personal treatment, and improve the better outcomes. PURPOSE: The purpose of this study was to evaluate changes in anterior segment parameters in Caucasian eyes with different angle closure mechanisms before and after laser peripheral iridotomy (LPI). METHODS: Sixty-six subjects underwent swept-source optical coherence tomography (CASIA, Tomey Corporation) angle imaging in the dark before and 7 days after LPI. On the basis of the baseline swept-source optical coherence tomography images, the eyes were categorized into 4 angle closure mechanisms, namely pupillary block (PB), plateau iris configuration (PIC), thick peripheral iris (TPI), and large lens vault (LLV). Sixteen out of 128 cross-sectional images (11.25 degrees apart) per volume scan were selected for analysis. We used a generalized estimating equation to compare quantitative parameters among angle closure mechanisms and between before and after LPI after adjusting the intereye correlation. RESULTS: The mean age of subjects was 67.7±9.2 years, with the majority being female (82.2%). One hundred twenty-nine eyes (67 primary angle closure suspects, 34 primary angle closure, and 28 primary angle closure glaucoma) were categorized into PB (n=71, 55%), PIC (n=40, 31%), TPI (n=14, 10.9%), and LLV (n=4, 3.1%). Anterior chamber depth was the shallowest in the LLV, followed by TPI, PB, and PIC group at baseline. Widening of the angle and reduction of the iris curvature (IC) due to LPI were observed in all groups (all P <0.01). When compared to the PB group, the LPI-induced angle widening in the TPI group was significantly less even though the iris curvature reduction in the TPI group was greater (all P <0.05). CONCLUSIONS: In patients with angle closure, anterior segment morphology and LPI-induced angle widening were different among the various angle closure mechanisms.

Intravitreal Neuroglobin Mitigates Primate Experimental Glaucomatous Structural Damage in Association with Reduced Optic Nerve Microglial and Complement 3-Astrocyte Activation.

Current management of glaucomatous optic neuropathy is limited to intraocular pressure control. Neuroglobin (Ngb) is an endogenous neuroprotectant expressed in neurons and astrocytes. We recently showed that exogenous intravitreal Ngb reduced inflammatory cytokines and microglial activation in a rodent model of hypoxia. We thus hypothesised that IVT-Ngb may also be neuroprotective in experimental glaucoma (EG) by mitigating optic nerve (ON) astrogliosis and microgliosis as well as structural damage. In this study using a microbead-induced model of EG in six Cynomolgus primates, optical coherence imaging showed that Ngb-treated EG eyes had significantly less thinning of the peripapillary minimum rim width, retinal nerve fibre layer thickness, and ON head cupping than untreated EG eyes. Immunohistochemistry confirmed that ON astrocytes overexpressed Ngb following Ngb treatment. A reduction in complement 3 and cleaved-caspase 3 activated microglia and astrocytes was also noted. Our findings in higher-order primates recapitulate the effects of neuroprotection by Ngb treatment in rodent EG studies and suggest that Ngb may be a potential candidate for glaucoma neuroprotection in humans.

Medical Application of Geometric Deep Learning for the Diagnosis of Glaucoma.

Purpose: (1) To assess the performance of geometric deep learning in diagnosing glaucoma from a single optical coherence tomography (OCT) scan of the optic nerve head and (2) to compare its performance to that obtained with a three-dimensional (3D) convolutional neural network (CNN), and with a gold-standard parameter, namely, the retinal nerve fiber layer (RNFL) thickness. Methods: Scans of the optic nerve head were acquired with OCT for 477 glaucoma and 2296 nonglaucoma subjects. All volumes were automatically segmented using deep learning to identify seven major neural and connective tissues. Each optic nerve head was then represented as a 3D point cloud with approximately 1000 points. Geometric deep learning (PointNet) was then used to provide a glaucoma diagnosis from a single 3D point cloud. The performance of our approach (reported using the area under the curve [AUC]) was compared with that obtained with a 3D CNN, and with the RNFL thickness. Results: PointNet was able to provide a robust glaucoma diagnosis solely from a 3D point cloud (AUC = 0.95 ± 0.01).The performance of PointNet was superior to that obtained with a 3D CNN (AUC = 0.87 ± 0.02 [raw OCT images] and 0.91 ± 0.02 [segmented OCT images]) and with that obtained from RNFL thickness alone (AUC = 0.80 ± 0.03). Conclusions: We provide a proof of principle for the application of geometric deep learning in glaucoma. Our technique requires significantly less information as input to perform better than a 3D CNN, and with an AUC superior to that obtained from RNFL thickness. Translational Relevance: Geometric deep learning may help us to improve and simplify diagnosis and prognosis applications in glaucoma.

Geometric Deep Learning to Identify the Critical 3D Structural Features of the Optic Nerve Head for Glaucoma Diagnosis.

PURPOSE: To compare the performance of 2 relatively recent geometric deep learning techniques in diagnosing glaucoma from a single optical coherence tomographic (OCT) scan of the optic nerve head (ONH); and to identify the 3-dimensional (3D) structural features of the ONH that are critical for the diagnosis of glaucoma. DESIGN: Comparison and evaluation of deep learning diagnostic algorithms. METHODS: In this study, we included a total of 2247 nonglaucoma and 2259 glaucoma scans from 1725 participants. All participants had their ONHs imaged in 3D with Spectralis OCT. All OCT scans were automatically segmented using deep learning to identify major neural and connective tissues. Each ONH was then represented as a 3D point cloud. We used PointNet and dynamic graph convolutional neural network (DGCNN) to diagnose glaucoma from such 3D ONH point clouds and to identify the critical 3D structural features of the ONH for glaucoma diagnosis. RESULTS: Both the DGCNN (area under the curve [AUC]: 0.97±0.01) and PointNet (AUC: 0.95±0.02) were able to accurately detect glaucoma from 3D ONH point clouds. The critical points (ie, critical structural features of the ONH) formed an hourglass pattern, with most of them located within the neuroretinal rim in the inferior and superior quadrant of the ONH. CONCLUSIONS: The diagnostic accuracy of both geometric deep learning approaches was excellent. Moreover, we were able to identify the critical 3D structural features of the ONH for glaucoma diagnosis that tremendously improved the transparency and interpretability of our method. Consequently, our approach may have strong potential to be used in clinical applications for the diagnosis and prognosis of a wide range of ophthalmic disorders.

Angle closure extent, anterior segment dimensions and intraocular pressure.

AIM: To investigate the association between the extent of iridotrabecular contact and other quantitative anterior segment dimensions measured by swept-source optical coherence tomography (SS-OCT; CASIA SS-1000, Tomey, Nagoya, Japan) with intraocular pressure (IOP). METHODS: Cross-sectional study. All subjects who were ≥50 years with no history of glaucoma, ocular surgery or trauma, underwent SS-OCT imaging (eight equally spaced radial scans), Goldman applanation tonometry and gonioscopy on the same day. We measured iridotrabecular contact (ITC) index and area, total volume of trabeculo-iris space area and angle opening distance at 500 and 750 from the scleral spur (TISA 500 and 750, AOD 500 and 750, respectively), anterior chamber depth (ACD), volume, area and width, pupil diameter, lens vault and iris volume.Their relationship with IOP (dependent variable) was assessed by locally weighted scatterplot smoothing (Lowess) regression with change-point analysis and generalised additive models adjusted for confounders. RESULTS: 2027 right eyes of mostly Chinese Singaporeans (90%) were analysed. ITC index above a threshold of ~60% (95% CI 34% to 92%) was significantly associated with higher IOP. Independent of the extent of ITC, ACD was also significantly associated with higher IOP below a threshold of 2.5 mm (95% CI 2.33 mm to 2.71 mm). Greater ITC index and shallower ACD had a joint association with IOP. A model including ACD and ITC index was more predictive of IOP than a model considering these variables separately, particularly for women with gonioscopically closed angles (R2 52.7%, p<0.05). CONCLUSIONS: The extent of angle closure and the ACD below a certain threshold had a significant joint association with IOP. These parameters, as biometrical surrogates of mechanical obstruction of the aqueous outflow, may jointly contribute to elevated IOP, particularly in women with gonioscopic angle closure.

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.

Differing Associations between Optic Nerve Head Strains and Visual Field Loss in Patients with Normal- and High-Tension Glaucoma.

PURPOSE: To study the associations between optic nerve head (ONH) strains under intraocular pressure (IOP) elevation with retinal sensitivity in patients with glaucoma. DESIGN: Clinic-based cross-sectional study. PARTICIPANTS: Two hundred twenty-nine patients with primary open-angle glaucoma (subdivided into 115 patients with high-tension glaucoma [HTG] and 114 patients with normal-tension glaucoma [NTG]). METHODS: For 1 eye of each patient, we imaged the ONH using spectral-domain OCT under the following conditions: (1) primary gaze and (2) primary gaze with acute IOP elevation (to approximately 35 mmHg) achieved through ophthalmodynamometry. A 3-dimensional strain-mapping algorithm was applied to quantify IOP-induced ONH tissue strain (i.e., deformation) in each ONH. Strains in the prelaminar tissue (PLT), the retina, the choroid, the sclera, and the lamina cribrosa (LC) were associated (using linear regression) with measures of retinal sensitivity from the 24-2 Humphrey visual field test (Carl Zeiss Meditec). This was performed globally, then locally according to a previously published regionalization scheme. MAIN OUTCOME MEASURES: Associations between ONH strains and values of retinal sensitivity from visual field testing. RESULTS: For patients with HTG, we found (1) significant negative linear associations between ONH strains and retinal sensitivity (P < 0.001; on average, a 1% increase in ONH strains corresponded to a decrease in retinal sensitivity of 1.1 decibels [dB]), (2) that high-strain regions colocalized with anatomically mapped regions of high visual field loss, and (3) that the strongest negative associations were observed in the superior region and in the PLT. In contrast, for patients with NTG, no significant associations between strains and retinal sensitivity were observed except in the superotemporal region of the LC. CONCLUSIONS: We found significant negative associations between IOP-induced ONH strains and retinal sensitivity in a relatively large glaucoma cohort. Specifically, patients with HTG who experienced higher ONH strains were more likely to exhibit lower retinal sensitivities. Interestingly, this trend in general was less pronounced in patients with NTG, which could suggest a distinct pathophysiologic relationship between the two glaucoma subtypes.

The three-dimensional structural configuration of the central retinal vessel trunk and branches as a glaucoma biomarker.

PURPOSE: To assess whether the 3-dimensional (3D) structural configuration of the central retinal vessel trunk and its branches (CRVT&B) could be used as a diagnostic marker for glaucoma. DESIGN: Retrospective, deep-learning approach diagnosis study. METHODS: We trained a deep learning network to automatically segment the CRVT&B from the B-scans of the optical coherence tomography (OCT) volume of the optic nerve head. Subsequently, 2 different approaches were used for glaucoma diagnosis using the structural configuration of the CRVT&B as extracted from the OCT volumes. In the first approach, we aimed to provide a diagnosis using only 3D convolutional neural networks and the 3D structure of the CRVT&B. For the second approach, we projected the 3D structure of the CRVT&B orthographically onto sagittal, frontal, and transverse planes to obtain 3 two-dimensional (2D) images, and then a 2D convolutional neural network was used for diagnosis. The segmentation accuracy was evaluated using the Dice coefficient, whereas the diagnostic accuracy was assessed using the area under the receiver operating characteristic curves (AUCs). The diagnostic performance of the CRVT&B was also compared with that of retinal nerve fiber layer (RNFL) thickness (calculated in the same cohorts). RESULTS: Our segmentation network was able to efficiently segment retinal blood vessels from OCT scans. On a test set, we achieved a Dice coefficient of 0.81 ± 0.07. The 3D and 2D diagnostic networks were able to differentiate glaucoma from nonglaucoma subjects with accuracies of 82.7% and 83.3%, respectively. The corresponding AUCs for the CRVT&B were 0.89 and 0.90, higher than those obtained with RNFL thickness alone (AUCs ranging from 0.74 to 0.80). CONCLUSIONS: Our work demonstrated that the diagnostic power of the CRVT&B is superior to that of a gold-standard glaucoma parameter, that is, RNFL thickness. Our work also suggested that the major retinal blood vessels form a "skeleton"-the configuration of which may be representative of major optic nerve head structural changes as typically observed with the development and progression of glaucoma.

Towards 'automated gonioscopy': a deep learning algorithm for 360° angle assessment by swept-source optical coherence tomography.

AIMS: To validate a deep learning (DL) algorithm (DLA) for 360° angle assessment on swept-source optical coherence tomography (SS-OCT) (CASIA SS-1000, Tomey Corporation, Nagoya, Japan). METHODS: This was a reliability analysis from a cross-sectional study. An independent test set of 39 936 SS-OCT scans from 312 phakic subjects (128 SS-OCT meridional scans per eye) was analysed. Participants above 50 years with no previous history of intraocular surgery were consecutively recruited from glaucoma clinics. Indentation gonioscopy and dark room SS-OCT were performed. Gonioscopic angle closure was defined as non-visibility of the posterior trabecular meshwork in ≥180° of the angle. For each subject, all images were analysed by a DL-based network based on the VGG-16 architecture, for gonioscopic angle-closure detection. Area under receiver operating characteristic curves (AUCs) and other diagnostic performance indicators were calculated for the DLA (index test) against gonioscopy (reference standard). RESULTS: Approximately 80% of the participants were Chinese, and more than half were women (57.4%). The prevalence of gonioscopic angle closure in this hospital-based sample was 20.2%. After analysing a total of 39 936 SS-OCT scans, the AUC of the DLA was 0.85 (95% CI:0.80 to 0.90, with sensitivity of 83% and a specificity of 87%) to classify gonioscopic angle closure with the optimal cut-off value of >35% of circumferential angle closure. CONCLUSIONS: The DLA exhibited good diagnostic performance for detection of gonioscopic angle closure on 360° SS-OCT scans in a glaucoma clinic setting. Such an algorithm, independent of the identification of the scleral spur, may be the foundation for a non-contact, fast and reproducible 'automated gonioscopy' in future.

Evaluation of the Diagnostic Performance of Swept-Source Anterior Segment Optical Coherence Tomography in Primary Angle Closure Disease.

PURPOSE: To evaluate the diagnostic performance of swept-source anterior segment optical coherence tomography (SS-OCT) in differentiating eyes with primary angle closure disease (PACD) from eyes of control subjects, as well as eyes with PAC and PAC glaucoma (PACG) from eyes with PAC suspect (PACS) disease. DESIGN: Multicenter cross-sectional study. METHODS: Chinese patients were classified into control, PACS, and PAC/PACG groups. The area under the receiving operating characteristic curve (AUC) from logistic regression models was used to evaluate discriminating ability. Sensitivity and specificity were calculated, and performance of the models was validated using an independent dataset. RESULTS: A total of 2928 SS-OCT images from 366 eyes of 260 patients were recruited to develop diagnostic models. The validation dataset included 1176 SS-OCT images from 147 eyes of 143 patients. For distinguishing PACD from control eyes, average anterior chamber depth had the highest AUC (0.94). With a cutoff of 2.2 mm for average anterior chamber depth, the sensitivity and specificity were 90.2% and 85.2% in the training set. For distinguishing PAC/PACG from PACS, a multivariate model had an AUC of 0.83, with sensitivity and specificity of 82.0% and 62.8% in the training set. The validation set confirmed the findings. CONCLUSIONS: SS-OCT of the anterior segment showed excellent diagnostic performance distinguishing PACD from normal eyes and moderate diagnostic ability distinguishing eyes with PAC/PACG from eyes with PACS. ACD alone may provide a simple and effective way to diagnose PACD from control subjects. As ACD can be obtained using other more available modalities, this has implications for the early diagnosis of PACD.

Describing the Structural Phenotype of the Glaucomatous Optic Nerve Head Using Artificial Intelligence.

PURPOSE: To develop a novel deep-learning approach that can describe the structural phenotype of the glaucomatous optic nerve head (ONH) and can be used as a robust glaucoma diagnosis tool. DESIGN: Retrospective, deep-learning approach diagnosis study. METHOD: We trained a deep-learning network to segment 3 neural-tissue and 4 connective-tissue layers of the ONH. The segmented optical coherence tomography images were then processed by a customized autoencoder network with an additional parallel branch for binary classification. The encoder part of the autoencoder reduced the segmented optical coherence tomography images into a low-dimensional latent space (LS), whereas the decoder and the classification branches reconstructed the images and classified them as glaucoma or nonglaucoma, respectively. We performed principal component analysis on the latent parameters and identified the principal components (PCs). Subsequently, the magnitude of each PC was altered in steps and reported how it impacted the morphology of the ONH. RESULTS: The image reconstruction quality and diagnostic accuracy increased with the size of the LS. With 54 parameters in the LS, the diagnostic accuracy was 92.0 ± 2.3% with a sensitivity of 90.0 ± 2.4% (at 95% specificity), and the corresponding Dice coefficient for the reconstructed images was 0.86 ± 0.04. By changing the magnitudes of PC in steps, we were able to reveal how the morphology of the ONH changes as one transitions from a "nonglaucoma" to a "glaucoma" condition. CONCLUSIONS: Our network was able to identify novel biomarkers of the ONH for glaucoma diagnosis. Specifically, the structural features identified by our algorithm were found to be related to clinical observations of glaucoma.

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

PURPOSE: To evaluate the performance of swept source optical coherence tomography (SS-OCT) to detect gonioscopic angle closure using different classification algorithms. METHODS: This 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. RESULTS: Data 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. CONCLUSIONS: A stepwise logistic regression model incorporating SS-OCT measurements has a high diagnostic ability to detect gonioscopic angle closure.

Peripapillary sclera exhibits a v-shaped configuration that is more pronounced in glaucoma eyes.

AIMS: To compare the shape of the anterior surface of the peripapillary sclera (PPS) between glaucoma and healthy subjects. METHODS: 88 primary open angle glaucoma (POAG), 98 primary angle closure glaucoma (PACG) and 372 age-matched and gender-matched healthy controls were recruited in this study. The optic nerve head of one randomly selected eye of each subject was imaged with spectral domain optical coherence tomography. The shape of the PPS was measured through an angle defined between a line parallel to the nasal anterior PPS boundary and one parallel to the temporal side. A negative value indicated that the PPS followed an inverted v-shaped configuration (peak pointing towards the vitreous), whereas a positive value indicated that it followed a v-shaped configuration. RESULTS: The mean PPS angle in normal controls (4.56±5.99°) was significantly smaller than that in POAG (6.60±6.37°, p=0.011) and PACG (7.90±6.87°, p<0.001). The v-shaped PPS was significantly associated with older age (ß=1.79, p<0.001), poorer best-corrected visual acuity (ß=3.31, p=0.047), central corneal thickness (ß=-0.28, p=0.001), peripapillary choroidal thickness (ß=-0.21, p<0.001) and presence of POAG (ß=1.94, p<0.009) and PACG (ß=2.96, p<0.001). The v-shaped configuration of the PPS significantly increased by 1.46° (p=0.001) in healthy controls for every 10-year increase in age, but not in glaucoma groups. CONCLUSIONS: The v-shaped configuration of the PPS was more pronounced in glaucoma eyes than in healthy eyes. This posterior bowing of the PPS may have an impact on the biomechanical environment of the optic nerve head.

Changes in Iris Stiffness and Permeability in Primary Angle Closure Glaucoma.

Purpose: To evaluate the biomechanical properties of the iris by evaluating iris movement during pupil constriction and to compare such properties between healthy and primary angle-closure glaucoma (PACG) subjects. Methods: A total of 140 subjects were recruited for this study. In a dark room, the anterior segments of one eye per subject were scanned using anterior segment optical coherence tomography imaging during induced pupil constriction with an external white light source of 1700 lux. Using a custom segmentation code, we automatically isolated the iris segments from the AS-OCT images, which were then discretized and transformed into a three-dimensional point cloud. For each iris, a finite element (FE) mesh was constructed from the point cloud, and an inverse FE simulation was performed to match the clinically observed iris constriction in the AS-OCT images. Through this optimization process, we were able to identify the elastic modulus and permeability of each iris. Results: For all 140 subjects (95 healthy and 45 PACG of Indian/Chinese ethnicity; age 60.2 ± 8.7 for PACG subjects and 57.7 ± 10.1 for healthy subjects), the simulated deformation pattern of the iris during pupil constriction matched well with OCT images. We found that the iris stiffness was higher in PACG than in healthy controls (24.5 ± 8.4 kPa vs. 17.1 ± 6.6 kPa with 40 kPa of active stress specified in the sphincter region; P < 0.001), whereas iris permeability was lower (0.41 ± 0.2 mm2/kPa s vs. 0.55 ± 0.2 mm2/kPa s; p = 0.142). Conclusions: This study suggests that the biomechanical properties of the iris in PACG are different from those in healthy controls. An improved understanding of the biomechanical behavior of the iris may have implications for the understanding and management of angle-closure glaucoma.