Survey on Automatic Detection of Glaucoma through Deep Learning Using Retinal Fundus Images

Glaucoma - a disease which causes damage to our eye's optic nerve and subsequently blinds the vision. This occurs due to increased intraocular pressure (IOP) which causes the damage of optic nerve axons at the back of the eye, with eventual deterioration of vision. Presently, many works have been done towards automatic glaucoma detection using Fundus Images (FI) by extracting structural features. Structural features can be extracted from optic nerve head (ONH) analysis, cup to disc ratio (CDR) and Inferior, Superior, Nasal, Temporal (ISNT) rule in Fundus Image for glaucoma assessment.This survey presents various techniques for the early detection of glaucoma. Among the various techniques, retinal image-based detection plays a major role as it comes under non-invasive methods of detection. Here, a review and study were conducted for the different techniques of glaucoma detection using retinal fundus images. The objective of this survey is to obtain a technique which automatically analyze the retinal images of the eye with high efficiency and accuracy

Accuracy of the ISNT rule and its variants for differentiating glaucomatous from normal eyes in a population-based study

AimsTo evaluate the accuracy of the ISNT rule (I=inferior, S=superior, N=nasal, T=temporal) and its variants with neuroretinal rim width and retinal nerve fibre layer (RNFL) thickness measurements differentiating normal from glaucomatous eyes.MethodsThe diagnosis accuracy of the ISNT rule and its variants was evaluated in a population-based study. Neuroretinal rim widths were measured on monoscopic optic disc photographs with an image-processing program. RNFL thickness measurements were obtained with spectral-domain optical coherence tomography (SD-OCT).ResultsIn this study including 940 normal subjects and 93 patients with glaucoma, the sensitivity of the ISNT rule with optic disc photographs was 94.1% (95% CI 90.2 to 98.1), whereas its specificity was 49.2% (46.9 to 51.6). When using the IST rule, the sensitivity decreased to 69.9% (62.1 to 77.6) with a higher specificity, 87.0% (85.3 to 88.6). All the diagnosis indicators were somewhat lower for the different rules using RNFL thickness: the sensitivity of the ISNT rule was 79.4% (72.6 to 86.2) and its specificity was 34.1% (31.9 to 36.4). With the IST rule, the sensitivity decreased to 50.0% (41.6 to 58.4) while the specificity increased to 64.9% (62.7 to 67.2).ConclusionsThe ISNT and IST rules applied to neuroretinal rim width measurement by optic disc photographs are useful and simple tools for differentiating normal from glaucomatous eyes. The translation of these rules to RNFL thickness by SD-OCT is of limited value.

ISNT rule satisfaction in Korean non-glaucomatous subjects

Purpose: To evaluate the efficacy of the ISNT rule in normal eyes of Koreans. Methods: We retrospectively reviewed medical records of 890 subjects with fundus photographs and evaluated the presence of the cup and ISNT rule satisfaction. If the ISNT rule was violated, the quadrants in which the neuroretinal rim was thinnest and thickest, respectively, were evaluated. Results: Among 890 eyes, 84.7% showed the cup. The subjects without the cup were significantly younger and more hyperopic. Among 754 eyes with the cup, 53.5% showed ISNT rule satisfaction. In 351 eyes violating the ISNT rule, the most common quadrant showing the thickest neuroretinal rim was the inferior (65.5%), and the thinnest, the temporal (98.3%). Conclusion: The cup was absent in 15% of the assessed eyes. Only about half of eyes with the cup showed ISNT rule satisfaction. Even in eyes violating ISNT rule, the inferior was the most common quadrant with the thickest neuroretinal rim and the temporal, the thinnest.

Computer-Aided Glaucoma Diagnosis Using Stochastic Watershed Transformation on Single Fundus Images

Glaucoma is a chronic eye disease and one of the major causes of permanent blindness. Since it does not show initial symptoms, early diagnosis is important to limit its progression. This paper presents an automatic optic nerve characterization algorithm for glaucoma diagnosis based only on retinal fundus images. For optic cup segmentation, we used a new method based on the stochastic watershed transformation applied on the YIQ colour space to extract clinical indicators such as the Cup/Disc ratio, the area Cup/Disc ratio and the ISNT rule. Afterwards, an assessment between normal and glaucomatous fundus images is performed. The proposed algorithm was evaluated on 6 different (private and public) databases containing 723 images (377 normal and 346 glaucomatous images) which achieved a specificity and sensitivity of 0.674 and 0.675, respectively. Moreover, an F-score of 0.770 was obtained when evaluating this method on the publicly available database Drishti-GS1. A comparison of the proposed work with other state-of-the-art methods demonstrates the robustness of the proposed algorithm; because it was tested using images from different databases with high variability, which is a common issue in this area. Additional comparisons with existing works for cup segmentation, that use the publicly available database Drishti-GS1, are also presented in this paper.