To demonstrate the effect of glaucoma on the optical attenuation coefficient of the retinal nerve fiber layer (RNFL) in Spectral Domain Optical Coherence Tomography (SD-OCT) images.
We analyzed ...images of the peripapillary areas in 10 healthy and 30 glaucomatous eyes (mild, moderate, and advanced glaucoma, 10 eyes each), scanned with the Spectralis OCT (Heidelberg Engineering GmbH, Dossenheim, Germany). To calculate the RNFL attenuation coefficient (μ(att)), determined by the scattering properties of the RNFL, we used a model that normalized the reflectivity of the RNFL by the retinal pigment epithelium. The analysis was performed at four preset locations at 1.3 and 1.7 mm from the center of the optic nerve head (ONH) (i.e., temporally, superiorly, nasally, and inferiorly) and on averages per eye. To assess the structure-function relationship, we correlated the μ(att) to the mean deviation (MD) in standard automated perimetry.
The μ(att) of the RNFL decreased up to 40% with increasing disease severity, on average as well as in each location around the ONH (Jonckheere-Terpstra test, P < 0.019 in all tests). The μ(att) of the RNFL depended significantly on the location around the ONH in all eyes (Kruskal-Wallis test, P < 0.014) and was lowest nasally from the ONH. The μ(att) correlated significantly with the MD in SAP (R(2) = 0.337).
The measurements clearly demonstrated that the μ(att) of the RNFL decreased with increasing disease severity. The RNFL attenuation coefficient may serve as a new method to quantify glaucoma in SD-OCT images.
The early detection of glaucoma is essential in preventing visual impairment. Artificial intelligence (AI) can be used to analyze color fundus photographs (CFPs) in a cost-effective manner, making ...glaucoma screening more accessible. While AI models for glaucoma screening from CFPs have shown promising results in laboratory settings, their performance decreases significantly in real-world scenarios due to the presence of out-of-distribution and low-quality images. To address this issue, we propose the Artificial Intelligence for Robust Glaucoma Screening (AIROGS) challenge. This challenge includes a large dataset of around 113,000 images from about 60,000 patients and 500 different screening centers, and encourages the development of algorithms that are robust to ungradable and unexpected input data. We evaluated solutions from 14 teams in this paper and found that the best teams performed similarly to a set of 20 expert ophthalmologists and optometrists. The highest-scoring team achieved an area under the receiver operating characteristic curve of 0.99 (95% CI: 0.98-0.99) for detecting ungradable images on-the-fly. Additionally, many of the algorithms showed robust performance when tested on three other publicly available datasets. These results demonstrate the feasibility of robust AI-enabled glaucoma screening.
To determine the diagnostic accuracy of the GDx VCC in the diagnosis of glaucoma.
Prospective, comparative, observational, clinic-based case series.
One eye each of 77 healthy subjects and 162 ...patients with primary open-angle glaucoma of Caucasian racial origin. Healthy subjects had normal visual fields (VFs), healthy-looking optic discs, and intraocular pressures of ≤21 mmHg in both eyes. Glaucoma patients had a reproducible glaucomatous VF defect and a glaucomatous appearance of the optic disc in at least one eye.
All subjects were measured with the GDx VCC with an automated variable corneal compensator. We constructed receiver operating characteristic (ROC) curves for all available parameters. Subsequently, we calculated sensitivity, specificity, and multilevel likelihood ratios for the best discriminating parameter in the entire group. In addition, we calculated sensitivity and specificity in patients with mild, moderate, and severe glaucomatous damage separately.
Software-derived parameters TSNIT (temporal, superior, nasal, inferior, temporal) Average, Superior Average, Inferior Average, TSNIT Std. Dev. (standard deviation), and Nerve Fiber Indicator (NFI).
The areas under the ROC curve for TSNIT Average, Superior Average, Inferior Average, TSNIT Std. Dev., and NFI were 0.93, 0.94, 0.90, 0.92, and 0.98, respectively. For the best discriminating parameter NFI, the sensitivity and specificity with a cutoff point of ≥40 were 89.0% and 95.9%, respectively. The multilevel likelihood ratios for glaucoma were 0.07 at NFI values of <35, 1.30 at values between 35 and 44, and 61.50 at values of ≥44. At the cutoff level of ≥40, the sensitivities of the NFI for correctly identifying glaucoma patients with mild, moderate, and severe damage were 83.8%, 92.9%, and 90.1%, respectively.
The GDx VCC allowed easy, rapid, and accurate discrimination between healthy and glaucomatous eyes. The NFI was the best discriminating parameter. The GDx VCC seems to fulfill criteria for a glaucoma screening device.
To investigate the relationship between retinal light sensitivity measured with standard automated perimetry (SAP) and retardation of the peripapillary retinal nerve fiber layer (RNFL) measured with ...the GDx VCC (Laser Diagnostic Technologies, Inc., San Diego, CA).
Forty-seven healthy subjects and 101 patients with glaucoma were examined with SAP and with the commercially available scanning laser polarimeter GDx VCC, with automated individualized compensation of anterior segment birefringence. Individual visual field test points and peripapillary RNFL retardation measurements were grouped into six corresponding sectors. The correlation between perimetry and GDx VCC measurements was determined, and the relationship between RNFL retardation and perimetry, expressed both in the standard decibel scale and in an unlogged scale, was described with linear regression analysis.
A statistically significant correlation was found in most sectors between perimetry and GDx VCC measurements in patients with glaucoma, but not in healthy subjects. A linear relationship was found between the unlogged sensitivities and GDx VCC measurements for the superotemporal and inferotemporal sectors. In the decibel scale, this relationship was curvilinear.
GDx VCC measurements of the peripapillary RNFL relate well with functional loss in glaucoma. Based on the observed relationships between function and structure, patients with mild to moderate visual field loss in glaucoma may be better monitored with the GDx VCC and patients who have severe loss, with SAP.
Patient satisfaction with glaucoma treatment has been poorly studied to date. Because glaucoma is a chronic condition in which the therapeutic response is dependent on adherence to treatment, patient ...acceptability is an important factor in achieving satisfactory outcomes. This multicenter, international (Belgium, the Netherlands, and Spain), epidemiological convenience sample survey among patients commencing treatment with preservative-free latanoprost collected data on patient satisfaction with particular regard to tolerability. A total of 1,541 patients were recruited who were predominantly elderly (74% were over 60 years of age) and female (61%). Most of the patients had previously received preserved topical glaucoma medication (69%), 6.7% had previously received preservative-free medication, whereas 24% had not previously been treated for glaucoma. The great majority of patients (>95%) were satisfied with the preservative-free latanoprost treatment. Among the patients who had previously received preserved medication, 73% of patients found preservative-free latanoprost to be better tolerated and 89% found it at least as easy to use as their prior treatment. Patient satisfaction (determined by a 0-100 mm visual analog scale) was improved by 47% on a switch from preserved treatment to preservative-free latanoprost. Intraocular pressure was similar in patients who had previously received preserved (18.3 mmHg), preservative-free (17.8 mmHg) glaucoma medication or who were naïve to treatment (20.3 mmHg). Preservative-free latanoprost provided effective reduction of intraocular pressure with better tolerability and patient satisfaction than preserved glaucoma medication. This tolerability profile can be expected to improve adherence to treatment in glaucoma patients.
Corneal endothelium images obtained by in vivo specular microscopy provide important information to assess the health status of the cornea. Estimation of clinical parameters, such as cell density, ...polymegethism, and pleomorphism, requires accurate cell segmentation. State-of-the-art techniques to automatically segment the endothelium are error-prone when applied to images with low contrast and/or large variation in cell size. Here, we propose an automatic method to segment the endothelium. Starting with an oversegmented image comprised of superpixels obtained from a stochastic watershed segmentation, the proposed method uses intensity and shape information of the superpixels to identify and merge those that constitute a cell, using support vector machines. We evaluated the automatic segmentation on a data set of in vivo specular microscopy images (Topcon SP-1P), obtaining 95.8% correctly merged cells and 2.0% undersegmented cells. We also evaluated the parameter estimation against the results of the vendor's built-in software, obtaining a statistically significant better precision in all parameters and a similar or better accuracy. The parameter estimation was also evaluated on three other data sets from different imaging modalities (confocal microscopy, phase-contrast microscopy, and fluorescence confocal microscopy) and tissue types ( ex vivo corneal endothelium and retinal pigment epithelium). In comparison with the estimates of the data sets' authors, we achieved statistically significant better accuracy and precision in all parameters except pleomorphism, where a similar accuracy and precision were obtained.
To determine the diagnostic accuracy of judging optic disc photographs for glaucoma by ophthalmologists.
Evaluation of diagnostic test and technology.
A total of 243 of 875 invited ophthalmologists ...in 11 European countries.
We determined how well each participant classified 40 healthy eyes and 48 glaucomatous eyes with varying severity of the disease on stereoscopic slides. Duplicate slides were provided for determining intraobserver agreement. All eyes were also imaged with the GDx with variable corneal compensation (GDx-VCC) (Carl Zeiss Meditec AG, Jena, Germany) and the Heidelberg Retina Tomograph (HRT) I (Heidelberg Engineering GmbH, Heidelberg, Germany). Diagnostic accuracies of clinicians were compared with those of the best machine classifiers.
Accuracy of classification, expressed as sensitivity, specificity, and overall accuracy. Intraobserver agreement (kappa).
The overall diagnostic accuracy of ophthalmologists was 80.5% (standard deviation SD, 6.8; range, 61.4%-94.3%). The machine classifiers outperformed most observers in diagnostic accuracy; the GDx-VCC nerve fiber indicator and the HRT's best classifier correctly classified 93.2% and 89.8% of eyes, respectively. The intraobserver agreement (kappa) varied between -0.13 and 1.0 and was on average good (0.7).
In general, ophthalmologists classify optic disc photographs moderately well for detecting glaucoma. There is, however, large variability in diagnostic accuracy among and agreement within clinicians. Common imaging devices outperform most clinicians in classifying optic discs.
Proprietary or commercial disclosure may be found after the references.
•A segmentation method that operates on OCT-derived attenuation coefficient data.•A probabilistic framework combining image data and prior knowledge about the retina.•Simultaneous segmentation of ...layers via weakly coupled level sets.•Weak coupling exploiting the predefined order of the layers and thickness priors.•Evaluation on various types of data (scan area, glaucoma, two OCT instrument).
Display omitted
Optical coherence tomography (OCT) yields high-resolution, three-dimensional images of the retina. Reliable segmentation of the retinal layers is necessary for the extraction of clinically useful information. We present a novel segmentation method that operates on attenuation coefficients and incorporates anatomical knowledge about the retina. The attenuation coefficients are derived from in-vivo human retinal OCT data and represent an optical property of the tissue. Then, the layers in the retina are simultaneously segmented via a new flexible coupling approach that exploits the predefined order of the layers. The accuracy of the method was evaluated on 20 peripapillary scans of healthy subjects. Ten of those subjects were imaged again to evaluate the reproducibility. An additional evaluation was performed to examine the robustness of the method on a variety of data: scans of glaucoma patients, macular scans and scans by a two different OCT imaging devices. A very good agreement on all data was found between the manual segmentation performed by a medical doctor and the segmentation obtained by the automatic method. The mean absolute deviation for all interfaces in all data types varied between 1.9 and 8.5 µm (0.5–2.2 pixels). The reproducibility of the automatic method was similar to the reproducibility of the manual segmentation.
Significant visual impairment due to glaucoma is largely caused by the disease being detected too late.
To build a labeled data set for training artificial intelligence (AI) algorithms for glaucoma ...screening by fundus photography, to assess the accuracy of the graders, and to characterize the features of all eyes with referable glaucoma (RG).
Cross-sectional study.
Color fundus photographs (CFPs) of 113 893 eyes of 60 357 individuals were obtained from EyePACS, California, United States, from a population screening program for diabetic retinopathy.
Carefully selected graders (ophthalmologists and optometrists) graded the images. To qualify, they had to pass the European Optic Disc Assessment Trial optic disc assessment with ≥ 85% accuracy and 92% specificity. Of 90 candidates, 30 passed. Each image of the EyePACS set was then scored by varying random pairs of graders as “RG,” “no referable glaucoma (NRG),” or "ungradable (UG).” In case of disagreement, a glaucoma specialist made the final grading. Referable glaucoma was scored if visual field damage was expected. In case of RG, graders were instructed to mark up to 10 relevant glaucomatous features.
Qualitative features in eyes with RG.
The performance of each grader was monitored; if the sensitivity and specificity dropped below 80% and 95%, respectively (the final grade served as reference), they exited the study and their gradings were redone by other graders. In all, 20 graders qualified; their mean sensitivity and specificity (standard deviation SD) were 85.6% (5.7) and 96.1% (2.8), respectively. The 2 graders agreed in 92.45% of the images (Gwet’s AC2, expressing the inter-rater reliability, was 0.917). Of all gradings, the sensitivity and specificity (95% confidence interval) were 86.0 (85.2–86.7)% and 96.4 (96.3–96.5)%, respectively. Of all gradable eyes (n = 111 183; 97.62%) the prevalence of RG was 4.38%. The most common features of RG were the appearance of the neuroretinal rim (NRR) inferiorly and superiorly.
A large data set of CFPs was put together of sufficient quality to develop AI screening solutions for glaucoma. The most common features of RG were the appearance of the NRR inferiorly and superiorly. Disc hemorrhages were a rare feature of RG.
Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.
Primary open angle glaucoma (POAG) is a complex disease with a major genetic contribution. Its prevalence varies greatly among ethnic groups, and is up to five times more frequent in black African ...populations compared to Europeans. So far, worldwide efforts to elucidate the genetic complexity of POAG in African populations has been limited. We conducted a genome-wide association study in 1113 POAG cases and 1826 controls from Tanzanian, South African and African American study samples. Apart from confirming evidence of association at
TXNRD2
(rs16984299; OR
T
1.20;
P
= 0.003), we found that a genetic risk score combining the effects of the 15 previously reported POAG loci was significantly associated with POAG in our samples (OR 1.56; 95% CI 1.26–1.93;
P
= 4.79 × 10
−5
). By genome-wide association testing we identified a novel candidate locus, rs141186647, harboring
EXOC4
(OR
A
0.48;
P
= 3.75 × 10
−8
), a gene transcribing a component of the exocyst complex involved in vesicle transport. The low frequency and high degree of genetic heterogeneity at this region hampered validation of this finding in predominantly West-African replication sets. Our results suggest that established genetic risk factors play a role in African POAG, however, they do not explain the higher disease load. The high heterogeneity within Africans remains a challenge to identify the genetic commonalities for POAG in this ethnicity, and demands studies of extremely large size.