Deep Learning-based Quantification of Anterior Segment OCT Parameters

To investigate the effect of laser peripheral iridotomy (LPI) in subgroups of angle closure eyes based on anterior segment optical coherence tomography (AS OCT)-derived parameters. Angle closure (primary angle closure [PAC] or PAC glaucoma [PACG]) eyes were imaged using AS OCT before and 2 weeks after LPI. A hierarchical cluster analysis was performed using AS OCT parameters obtained before LPI, such as anterior chamber depth (ACD), anterior chamber width (ACW), iris cross-sectional area (IA), angle opening distance and iris thickness at 750 μm from the scleral spur (AOD750, IT750), iris curvature (IC), lens vault (LV), and anterior chamber area (ACA) to subclassify angle closure eyes. After LPI, parameters were compared to evaluate whether the outcome of LPI differed among the subgroups determined by cluster analysis. Cluster analysis generated two distinct clusters showing significantly different anatomical features. Cluster 1 (61 eyes) had lower ACD (P<0.001), higher LV (P=0.008), lower AOD750 (P<0.001), and lower ACW (P<0.001) than cluster 2 (27 eyes). The proportional change in AOD750 after LPI was significantly greater in cluster 1 than in cluster 2 (116.4±117.3% and 46.4±45.7%, respectively, P=0.007). The outcomes of LPI differed between angle closure subgroups with different anatomical characteristics. Our results provide evidence that angle closure patients can be grouped according to different anatomical anterior segment features and that the pathogenic mechanism of angle closure may differ among subgroups, suggesting that the treatment and follow-up plan should be customized according to subgroup features.

AimTo 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.MethodsPatients...

To compare the accuracy in measurement of the anterior chamber (AC) angle by anterior segment optical coherence tomography (AS-OCT) and ultrasound biomicroscopy (UBM) in European patients with suspected primary angle closure (PACS), primary angle closure (PAC), or primary angle-closure glaucoma (PACG). Cross-sectional study. In all, 55 eyes of 33 consecutive patients presenting with PACS, PAC, or PACG were examined with AS-OCT, followed by UBM. The trabecular-iris angle (TIA) was measured in all four quadrants. The angle-opening distance (AOD) was measured at 500 microm from the scleral spur. The Bland-Altman method was used for assessing agreement between the two methods. The mean (+/-SD) superior TIA was 19.3+/-15.8 degrees in AS-OCT and 15.7+/-15.0 degrees in UBM (P=0.50) and inferior TIA was 17.9+/-12.9 degrees (AS-OCT) and 16.7+/-14.1 degrees (UBM) (P=0.71). The superior AOD(500) was 0.17+/-0.16 mm in UBM and 0.21+/-0.16 mm in AS-OCT (P=0.06). Bland-Altman analysis showed a mean SD of+/-9.4 degrees for superior and inferior TIA and a mean SD of +/-0.10 mm for superior and inferior AOD(500). This comparative study shows that AS-OCT measurements are significantly correlated with UBM measurements but show poor agreement with each other. We do not believe that AS-OCT can replace UBM for the quantitative assessment of the AC angle.

PurposeTo measure changes in anterior chamber structure before and after exercise in healthy individuals using anterior segment optical coherence tomography (ASOCT).MethodsThirty‐two healthy young individuals performed jogging for 20 min. Eye blinking rate was recorded during rest and exercise. The anterior chamber angle (ACA), angle opening distance at 500 μm from the scleral spur (AOD500), trabecular‐iris space area at 500 μm from the scleral spur (TISA500), iris concavity (IC), iris concavity ratio (CR), iris thickness at 750 μm from the scleral spur (IT750), anterior chamber depth (ACD), anterior chamber width (ACW), pupil diameter (PD), intraocular pressure (IOP), blood pressure (BP) and heart rate (HR) were recorded before and after exercise. Anterior chamber angle (ACA), AOD500, TISA500, IC, IT750, ACD, ACW and PD were measured with ASOCT.ResultsCompared with rest, the blinking rate during exercise did not change significantly (13.04 ± 5.80 versus 13.52 ± 5.87 blinks/min, p = 0.645). The average IOP (15.4 ± 2.4 versus 12.4 ± 2.1 mmHg), ACA (35.96 ± 11.35 versus 40.25 ± 12.64 degrees), AOD500 (0.800 ± 0.348 versus 0.942 ± 0.387 mm), TISA500 (0.308 ± 0.155 versus 0.374 ± 0.193 mm2), IC (−0.078 ± 0.148 versus −0.153 ± 0.159 mm) and CR (−0.027 ± 0.050 versus −0.054 ± 0.056) changed significantly (all p < 0.001), while the average IT750 (0.463 ± 0.084 versus 0.465 ± 0.086 mm; p = 0.492), ACD (3.171 ± 0.229 versus 3.175 ± 0.238 mm; p = 0.543) and ACW (11.768 ± 0.377 versus 11.755 ± 0.378 mm; p = 0.122) showed no significant change after exercise.ConclusionThe blinking rate did not change significantly during exercise, while ACA, AOD500 and TISA500 increased after exercise. Exercise also induced or increased IC. These changes in anterior chamber structure were only associated with exercise, but not with the postexercise change in PD or IOP.

Fungal hyphae growing into anterior chamber from cornea

PurposeTo assess the interdevice agreement between swept-source Fourier-domain and time-domain anterior segment optical coherence tomography (AS-OCT).MethodsFifty-three eyes from 41 subjects underwent CASIA2 and Visante OCT imaging. One hundred eighty–degree axis images were measured with the built-in two-dimensional analysis software for the swept-source Fourier-domain AS-OCT (CASIA2) and a customized program for the time-domain AS-OCT (Visante OCT). In both devices, we examined the angle opening distance (AOD), trabecular iris space area (TISA), angle recess area (ARA), anterior chamber depth (ACD), anterior chamber width (ACW), and lens vault (LV). Bland-Altman plots and intraclass correlation (ICC) were performed. Orthogonal linear regression assessed any proportional bias.ResultsICC showed strong correlation for LV (0.925) and ACD (0.992) and moderate agreement for ACW (0.801). ICC suggested good agreement for all angle parameters (0.771–0.878) except temporal AOD500 (0.743) and ARA750 (nasal 0.481; temporal 0.481). There was a proportional bias in nasal ARA750 (slope 2.44, 95% confidence interval [CI]: 1.95–3.18), temporal ARA750 (slope 2.57, 95% CI: 2.04–3.40), and nasal TISA500 (slope 1.30, 95% CI: 1.12–1.54). Bland-Altman plots demonstrated in all measured parameters a minimal mean difference between the two devices (−0.089 to 0.063); however, evidence of constant bias was found in nasal AOD250, nasal AOD500, nasal AOD750, nasal ARA750, temporal AOD500, temporal AOD750, temporal ARA750, and ACD. Among the parameters with constant biases, CASIA2 tends to give the larger numbers.ConclusionsBoth devices had generally good agreement. However, there were proportional and constant biases in most angle parameters. Thus, it is not recommended that values be used interchangeably.

To assess the reproducibility and agreement of measurement values obtained from the Tomey CASIA2 and Heidelberg Spectralis OCT2 anterior segment optical coherence tomographic devices. Twenty eyes from 10 subjects ranging from age 28 to 45 years with no history of eye conditions or intraocular surgery were included. Two scans were obtained with each device in a standardized dark room environment after a period of dark adaptation. One anterior segment optical coherence tomography image along the horizontal (temporal nasal) meridian was analyzed per eye and per scan. Lens vault, pupil diameter, anterior chamber width, angle opening distance, trabecular iris space area, and scleral spur angle were measured using manufacturer-provided image analysis programs. Intraclass correlation coefficient (ICC) values, coefficients of variation, and Bland-Altman plots were computed to assess the intradevice correlation and interdevice agreement of measurement values. There was excellent intradevice reproducibility of measurement values for both the CASIA (ICC range, 0.86 to 0.99) and Spectralis (ICC range, 0.79 to 1.00). There was also excellent interdevice correlation of measurement values (ICC range, 0.78 to 0.93) for all parameters except anterior chamber width (ICC 0.20). Linear regression models and Bland-Altman plots showed that this relationship was strongest when measurement values were small. There is excellent intradevice reproducibility and good interdevice agreement of anterior segment parameter measurement values for the CASIA2 and Spectralis OCT2. However, the measurements obtained with each device should not be considered interchangeable.

Repeatability and reproducibility of anterior ocular biometric measurements with 2-dimensional and 3-dimensional optical coherence tomography

BackgroundFirst reported study to assess the effect of diurnal variation on anterior chamber angle measurements, as well as, to re-test the effects of lighting and angle-of-incidence variation on anterior chamber angle (ACA) measurements acquired by time-domain anterior segment optical coherence tomography (AS-OCT).MethodsA total of 30 eyes from 15 healthy, normal subjects underwent anterior chamber imaging using a Visante time-domain AS-OCT according to an IRB-approved protocol. For each eye, the inferior angle was imaged twice in the morning (8 am – 10 am) and then again in the afternoon (3 pm – 5 pm), under light meter-controlled conditions with ambient room lighting ‘ON’ and lights ‘OFF’, and at 5° angle of incidence increments. The ACA metrics measured for each eye were: angle opening distance (AOD, measured 500 and 750 μm anterior from scleral spur), the trabecular-iris-space area (TISA, measured 500 and 750 μm anterior from scleral spur), and scleral spur angle. Measurements were performed by masked, certified Reading Center graders using the Visante’s Internal Measurement Tool. Differences in measurements between morning and afternoon, lighting variations, and angle of incidence were compared.ResultsMean age of the participants was 31.2 years (range 23–58). Anterior chamber angle metrics did not differ significantly from morning to afternoon imaging, or when the angle of incidence was offset by 5° in either direction away from the inferior angle 6 o’clock position. (p-value 0.13-0.93). Angle metrics at the inferior corneal limbus, 6 o’clock position (IC270), with room lighting ‘OFF’, showed a significant decrease (p < 0.05) compared to room lighting ‘ON’.ConclusionsThere does not appear to be significant diurnal variation in AS-OCT parameters in normal individuals, but lighting conditions need to be strictly controlled since variation in lighting led to significant variability in AS-OCT parameters. No changes in ACA parameters were noted by varying the angle-of-incidence, which gives confidence in being able to perform longitudinal studies in approximately the same area (plus/minus 5° of original scan location).

PurposeThe 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).MethodsParticipants 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.ResultsAnalysis 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).ConclusionsUnsupervised 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 RelevanceClustering of biometrics measured in the light could provide an alternative source of information to risk-stratify angle closure eyes for more severe disease.

To assess the ability of swept-source (SS) optical coherence tomography (OCT) of the anterior segment (AS) to measure anterior chamber (AC) inflammation (both flare and cells) objectively. To compare OCT-derived inflammatory indices with standard techniques. Prospective evaluation of a diagnostic test. Patients diagnosed with anterior uveitis (active or inactive) and controls. Participants underwent an AC inflammation evaluation including: clinical cell and flare grading and laser flare photometry (LFP). Uveitis patients were divided into active or inactive uveitis status according to clinical grading. Anterior segment SS-OCT scans were obtained for each participant. Tomographic images were analyzed to count the AC cells, and to calculate to absolute measurements of aqueous signal intensity. The absolute values were compared with the signal measured by the scan outside the eye, generating an optical density ratio (aqueous-to-air relative intensity [ARI] index). Correlations between OCT-derived AC inflammatory indexes and LFP, clinical grading, participant category (active or inactive uveitis, control), age, gender, and central corneal thickness (CCT) were assessed. Correlation between OCT-derived AC inflammatory indexes (ARI index and AC cells on OCT) and standard clinical techniques (LFP, clinical cell grading). Two hundred thirty-seven eyes (70 active uveitis, 97 inactive uveitis, and 70 controls) were included. Anterior chamber cells count on OCT did not differ between inactive uveitis and controls, but was significantly higher in active uveitis compared to the other categories (both P < 0.0001). All groups had different LFP (all P < 0.0001). Active uveitis had significantly higher ARI index compared with inactive uveitis and controls (both P < 0.0001). Interobserver agreement (intraclass correlation coefficient) for ARI index was 0.78. The ARI index correlated positively with age (P= 0.043) and negatively with CCT (P= 0.006). The ARI index correlated with LFP in the active uveitis group (P < 0.0001), but not in the others. Anterior chamber cells on OCT increased among all cell clinical grades (P < 0.0001). The ARI index increased among all flare clinical grades (P < 0.005). Anterior segment SS-OCT could be used for a comprehensive assessment of AC inflammation, providing objective measurements of inflammatory cells and aqueous flare.

To quantitatively analyze the angle anatomy in eyes with a Boston type 1 keratoprosthesis (KPro) using anterior segment optical coherence tomography (AS-OCT) and to assess the diagnostic ability of AS-OCT in KPro-associated glaucoma. AS-OCT (RTVue) images from KPro eyes with and without glaucoma were reviewed. The angle opening distance at 500 μm from the scleral spur (AOD500), trabecular-iris angle at 500 μm from the scleral spur (TIA500), and trabecular-iris surface area at 500 μm from the scleral spur (TISA500) were measured by 2 observers masked to the diagnosis. The measurements for each visible quadrant were compared between KPro eyes with and without glaucoma. Twenty-two eyes with glaucoma and 17 eyes without glaucoma from 39 patients with KPro were included. Of the 4 quadrants imaged, the temporal angle was the most visible (79.5%) and angle measurements of the temporal quadrant were the only ones that differentiated the 2 groups: the mean AOD500, TIA500, and TISA500 were significantly lower in KPro eyes with glaucoma than without glaucoma (388.2 ± 234.4 μm vs. 624.5 ± 310.5 μm, P = 0.02; 26.1 ± 14.0 degrees vs. 39.1 ± 17.1 degrees, P = 0.03; and 0.15 ± 0.09 mm vs. 0.23 ± 0.12 mm, P = 0.03; respectively). The highest area under the receiver operating characteristic curve for detecting glaucoma was 0.75 for temporal TIA500 (95% confidence interval 0.57-0.94, P = 0.02) with 50% specificity at 80% of sensitivity and a cutoff value of 37 degrees. The temporal angle was the most visible on AS-OCT in eyes with a KPro. Significant narrowing of the temporal angle detected on AS-OCT was associated with glaucoma in these eyes.

Cataract Extraction and the Iris Configuration

PurposeTo develop generative adversarial networks (GANs) that synthesize realistic anterior segment optical coherence tomography (AS-OCT) images and evaluate deep learning (DL) models that are trained on real and synthetic datasets for detecting angle closure.MethodsThe GAN architecture was adopted and trained on the dataset with AS-OCT images collected from the Joint Shantou International Eye Center of Shantou University and the Chinese University of Hong Kong, synthesizing open- and closed-angle AS-OCT images. A visual Turing test with two glaucoma specialists was performed to assess the image quality of real and synthetic images. DL models, trained on either real or synthetic datasets, were developed. Using the clinicians’ grading of the AS-OCT images as the reference standard, we compared the diagnostic performance of open-angle vs. closed-angle detection of DL models and the AS-OCT parameter, defined as a trabecular-iris space area 750 µm anterior to the scleral spur (TISA750), in a small independent validation dataset.ResultsThe GAN training included 28,643 AS-OCT anterior chamber angle (ACA) images. The real and synthetic datasets for DL model training have an equal distribution of open- and closed-angle images (all with 10,000 images each). The independent validation dataset included 238 open-angle and 243 closed-angle AS-OCT ACA images. The image quality of real versus synthetic AS-OCT images was similar, as assessed by the two glaucoma specialists, except for the scleral spur visibility. For the independent validation dataset, both DL models achieved higher areas under the curve compared with TISA750. Two DL models had areas under the curve of 0.97 (95% confidence interval, 0.96–0.99) and 0.94 (95% confidence interval, 0.92–0.96).ConclusionsThe GAN synthetic AS-OCT images appeared to be of good quality, according to the glaucoma specialists. The DL models, trained on all-synthetic AS-OCT images, can achieve high diagnostic performance.Translational RelevanceThe GANs can generate realistic AS-OCT images, which can also be used to train DL models.

To assess the distribution of iris structural measurements among American Caucasians, American Chinese and mainland Chinese aged 40years and over, in order to determine associated factors of increased risk for angle closure in Chinese. Prospective multicentre cross-sectional study. Four gender- and age-matched cohorts: Caucasians, American Chinese, Southern Chinese from Guangzhou and Northern Chinese from Beijing were enrolled in this study. Anterior segment optical coherence tomography was utilized to image the anterior chamber. Customized software was used to calculate the related parameters from anterior segment optical coherence tomography images. Iris thickness at 750µm from the scleral spur, iris curvature, iris area, pupil diameter. Data from 121, 124, 121 and 120 subjects were available for American Caucasian, American Chinese, Southern and Northern Mainland Chinese subjects, respectively. The Chinese cohorts did not differ with regard to any of the parameters. Multiple linear regressions indicated that Caucasians had thinner iris thickness at 750µm from the scleral spur and smaller iris area than Chinese after controlling for potential confounders. Female Caucasians had thinner irides (P=0.015) and smaller iris area (P=0.001) than male Caucasians. Female Chinese tended to have more convex irides (P=0.021) than male Chinese. Pupil size was inversely associated with age (P<0.001) in both ethnicities. In dark conditions, Chinese have thicker irides and greater iris area than Caucasians independent of pupil size. Among Chinese, more convex irides in women may account for the higher rate of narrow angles identified in this segment of the Chinese population.