To compare the abilities of current commercially available versions of 3 optical imaging techniques: scanning laser polarimetry with variable corneal compensation (GDx VCC), confocal scanning laser ophthalmoscopy (HRT II [Heidelberg Retina Tomograph]), and optical coherence tomography (Stratus OCT) to discriminate between healthy eyes and eyes with glaucomatous visual field loss. Methods: We included 107 patients with glaucomatous visual field loss and 76 healthy subjects of a similar age. All individuals underwent imaging with a GDx VCC, HRT II, and fast retinal nerve fiber layer scan with the Stratus OCT as well as visual field testing within a 6-month period. Receiver operating characteristic curves and sensitivities at fixed specificities (80% and 95%) were calculated for parameters reported as continuous variables. Diagnostic categorization (outside normal limits, borderline, or within normal limits) provided by each instrument after comparison with its respective normative database was also evaluated, and likelihood ratios were reported. Agreement on categorization between methods (weighted) was assessed. Results: After the exclusion of subjects with unacceptable images, the final study sample included 141 eyes of
The analysis and interpretation of rates of SAP and SD-OCT change, as indicators of the velocity of neural damage in glaucoma, should take into account the severity of the disease.
The ability of deep learning architectures to identify glaucomatous optic neuropathy (GON) in fundus photographs was evaluated. A large database of fundus photographs (n = 14,822) from a racially and ethnically diverse group of individuals (over 33% of African descent) was evaluated by expert reviewers and classified as GON or healthy. Several deep learning architectures and the impact of transfer learning were evaluated. The best performing model achieved an overall area under receiver operating characteristic (AUC) of 0.91 in distinguishing GON eyes from healthy eyes. It also achieved an AUC of 0.97 for identifying GON eyes with moderate-to-severe functional loss and 0.89 for GON eyes with mild functional loss. A sensitivity of 88% at a set 95% specificity was achieved in detecting moderate-to-severe GON. In all cases, transfer improved performance and reduced training time. Model visualizations indicate that these deep learning models relied on, in part, anatomical features in the inferior and superior regions of the optic disc, areas commonly used by clinicians to diagnose GON. The results suggest that deep learning-based assessment of fundus images could be useful in clinical decision support systems and in the automation of large-scale glaucoma detection and screening programs.
To quantitatively assess and compare the thickness of the retinal nerve fiber layer (RNFL) in ocular hypertensive eyes with normal and glaucomatous eyes using the Optical Coherence Tomograph (OCT 2000, software version A4X1; Humphrey Instruments, San Leandro, Calif). Methods:The mean RNFL thickness of ocular hypertensive (n = 28) eyes was compared with age-matched normal (n = 30) and glaucomatous (n = 29) eyes. Subject eyes were classified into diagnostic groups based on intraocular pressure, stereoscopic disc photographs, and standard automated perimetry. Three circular scans were obtained for each eye at a diameter of 3.4 mm around the optic disc. In each eye, average RNFL thickness measurements were obtained in temporal, superior, nasal, and inferior quadrants. A single index of average RNFL thickness throughout 360°also was obtained.Results: Mean (95% confidence interval) RNFL was significantly thinner in ocular hypertensive eyes than in normal eyes, 72.8 µm (66.4-78.1 µm) and 85.8 µm (80.2-91.7 µm), respectively. More specifically, RNFL was significantly thinner in ocular hypertensive eyes than in normal eyes in the inferior quadrant, 84.8 µm (75.6-94.0 µm) vs 107.6 µm (99.3-115.9 µm); and in the nasal quadrant, 44.1 µm (37.5-51.7 µm) vs 61.8 µm (53.0-65.6 µm). Retinal nerve fiber layer was significantly thinner in glaucomatous eyes than in ocular hypertensive and normal eyes throughout 360°and in all quadrants. Conclusion:These findings suggest that quantitative differences in RNFL thickness exist between age-matched ocular hypertensive, normal, and glaucomatous eyes.
IMPORTANCE A deep learning system (DLS) that could automatically detect glaucomatous optic neuropathy (GON) with high sensitivity and specificity could expedite screening for GON.OBJECTIVE To establish a DLS for detection of GON using retinal fundus images and glaucoma diagnosis with convoluted neural networks (GD-CNN) that has the ability to be generalized across populations. DESIGN, SETTING, AND PARTICIPANTSIn this cross-sectional study, a DLS for the classification of GON was developed for automated classification of GON using retinal fundus images obtained from the Chinese Glaucoma Study Alliance, the Handan Eye Study, and online databases. The researchers selected 241 032 images were selected as the training data set. The images were entered into the databases on June 9, 2009, obtained on July 11, 2018, and analyses were performed on December 15, 2018. The generalization of the DLS was tested in several validation data sets, which allowed assessment of the DLS in a clinical setting without exclusions, testing against variable image quality based on fundus photographs obtained from websites, evaluation in a population-based study that reflects a natural distribution of patients with glaucoma within the cohort and an additive data set that has a diverse ethnic distribution. An online learning system was established to transfer the trained and validated DLS to generalize the results with fundus images from new sources. To better understand the DLS decision-making process, a prediction visualization test was performed that identified regions of the fundus images utilized by the DLS for diagnosis. EXPOSURES Use of a deep learning system. MAIN OUTCOMES AND MEASURES Area under the receiver operating characteristics curve (AUC), sensitivity and specificity for DLS with reference to professional graders. RESULTS From a total of 274 413 fundus images initially obtained from CGSA, 269 601 images passed initial image quality review and were graded for GON. A total of 241 032 images (definite GON 29 865 [12.4%], probable GON 11 046 [4.6%], unlikely GON 200 121 [83%]) from 68 013 patients were selected using random sampling to train the GD-CNN model. Validation and evaluation of the GD-CNN model was assessed using the remaining 28 569 images from CGSA. The AUC of the GD-CNN model in primary local validation data sets was 0.996 (95% CI, 0.995-0.998), with sensitivity of 96.2% and specificity of 97.7%. The most common reason for both false-negative and false-positive grading by GD-CNN (51 of 119 [46.3%] and 191 of 588 [32.3%]) and manual grading (50 of 113 [44.2%] and 183 of 538 [34.0%]) was pathologic or high myopia.CONCLUSIONS AND RELEVANCE Application of GD-CNN to fundus images from different settings and varying image quality demonstrated a high sensitivity, specificity, and generalizability for detecting GON. These findings suggest that automated DLS could enhance current screening programs in a cost-effective and time-efficient manner.
PURPOSE-To evaluate RTVue spectral-domain optical coherence tomography (OCT) (Optovue Inc, Fremont, California, USA) reproducibility and to assess agreement with Stratus time-domain OCT (Carl Zeiss Meditec, Dublin, California, USA) measurements. DESIGN-Observational clinical study.METHODS-Scans were obtained from both eyes of all participants 3 times using the RTVue nerve head map 4-mm diameter protocol and once using Stratus OCT within the same session. RTVue reproducibility and agreement with Stratus OCT were evaluated for retinal nerve fiber layer (RNFL) and optic disc measurements.RESULTS-Thirty healthy participants (60 eyes) and 38 glaucoma patients (76 eyes) were included in the study. RTVue reproducibility was good in both healthy participants and patients. For average RNFL thickness, the intraclass correlation coefficients in healthy eyes and patient eyes were 0.97 whereas for rim area they were 0.97 and 0.96, respectively. The correlation between RTVue and Stratus measurements generally was good, especially for average RNFL thickness (healthy eyes and patient eyes, r 2 = 0.82 and 0.86, respectively) and rim volume (healthy eyes and patient eyes, r 2 = 0.78 and 0.76, respectively). Bland-Altman plots showed good agreement between the instruments, with better agreement for average RNFL thickness (95% limits of agreement in healthy eyes and patient eyes, −8.6 to 12 µm and −5.6 to −14.8 µm, respectively) than optic disc parameters. Cup-to-disc ratio 95% limits of agreement in healthy eyes and patient eyes were −0.3 to 0.4 and −0.2 to 0.3, respectively. Optic disc measurements with RTVue were smaller than those with Stratus OCT (eg, disc area was on average 0.4 mm 2 smaller and rim area was 0.3 mm 2 smaller with RTVue).CONCLUSIONS-Reproducibility of RTVue RNFL and optic disc measurements was excellent in both groups. The level of agreement between RTVue and Stratus measurements suggests that RTVue has the potential to detect glaucomatous structural changes.Optical Coherence Tomography (Oct) is a noninvasive retinal nerve fiber layer (RNFL) and optic disc imaging method that provides micrometer-scale resolution.1 -4 OCT technology has changed considerably in recent years with the incorporation of spectral-domain (SD) imaging that offers significant advantages over the traditional time-domain (TD) OCT techniques. Unlike TD-OCT, SD-OCT uses a stationary reference mirror, and the OCT signal is acquired using a spectrometer as a detector.6 , 7 SD technology currently is capable of an acquisition speed of up to 29,000 A scans per second.8 In addition, SD-OCT offers a higher resolution than TD-OCT2 and can provide a significant reduction in motion artifacts and an increased signal-to-noise ratio compared with TD-OCT.9 , 10 Until very recently, ophthalmic applications of OCT technology were performed exclusively using TD-OCT (Stratus OCT; Carl Zeiss Meditec, Dublin, California, USA). The recently introduced RTVue (Optovue Inc, Fremont, California, USA) is one of several ultra high-speed, high-resolution OCT ...
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.