METHODS: The morphological features of the disc that is characteristic of glaucoma are clearly seen in the fundus images. However, manual inspection of the acquired fundus images may be prone to inter-observer variation. Therefore, a computer-aided detection (CAD) system is proposed to make an accurate, reliable and fast diagnosis of glaucoma based on the optic nerve features of fundus imaging. In this paper, we reviewed existing techniques to automatically diagnose glaucoma.
RESULTS: The use of CAD is very effective in the diagnosis of glaucoma and can assist the clinicians to alleviate their workload significantly. We have also discussed the advantages of employing state-of-art techniques, including deep learning (DL), when developing the automated system. The DL methods are effective in glaucoma diagnosis.
CONCLUSIONS: Novel DL algorithms with big data availability are required to develop a reliable CAD system. Such techniques can be employed to diagnose other eye diseases accurately.
Methods: A total of 3843 participants (7,020 healthy eyes) were enrolled from the Singapore Epidemiology of Eye Diseases (SEED) study, a population-based study composing of three major ethnic groups-Malay, Indian, and Chinese-in Singapore. Ocular examinations were performed, and spectral-domain optical coherence tomography (SD-OCT) was used to measure circumpapillary RNFL thickness. We selected 35 independent glaucoma-associated genetic loci for analysis. An linear regression model was conducted to determine the association of these variants with circumpapillary RNFL, assuming an additive genetic model. We conducted association analysis in each of the three ethnic groups, followed by a meta-analysis of them.
Results: The mean age of the included participants was 59.4 ± 8.9 years, and the mean RFNL thickesss is 92.3 ± 11.2 µm. In the meta-analyses, of the 35 glacuoma loci, we found that only SIX6 was significantly associated with reduction in global RNFL thickness (rs33912345; β = -1.116 um per risk allele, P = 1.64E-05), and the effect size was larger in the inferior RNFL quadrant (β = -2.015 µm, P = 2.9E-6), and superior RNFL quadrant (β = -1.646 µm, P = 6.54E-5). The SIX6 association were consistently observed across all three ethnic groups. Other than RNFL, we also found several genetic varaints associated with vertical cuo-to-disc ratio (ATOH7, CDKN2B-AS1, and TGFBR3-CDC7), rim area (SIX6 and CDKN2B-AS1), and disc area (SIX6, ATOH7, and TGFBR3-CDC7). The association of SIX6 rs33912345 with NRFL thickness remained similar after further adjusting for disc area and 3 other disc parameter associated SNPs (ATOH7, CDKN2B-AS1, and TGFBR3-CDC7).
Conclusions: Of the 35 glaucoma identified risk loci, only SIX6 is significantly and independently associated with thinner RNFL. Our study further supports the involvement of SIX6 with RNFL thickness and pathogensis of glaucoma.