1887

Abstract

Retinal examination is a diagnostic pillar in ophthalmology for the early detection of eye diseases such as retinopathy (hemorrhage/exudate) and glaucoma (optic disc abnormality). Large population-based studies have shown that retinal vessel morphological quantification may be useful in predicting the development and progression of hypertension, diabetes and cardiovascular diseases (CVD). The rationale for this is that the retinal microcirculatory bed shares similar anatomical and physiological characteristics with the coronary and cerebrovascular circulations. Therefore, quantification of retinal vessel metrics may reveal insights into the development of coronary artery and cerebrovascular disease. Indeed, retinal blood vessel analysis has gained increasing interest in recent years for predicting the development of hypertension, coronary heart disease, stroke and type II diabetes. A number of software algorithms have been used to quantify retinal vessel morphology, but they require manual input and are time- and labor-intensive. Fully automated image analysis allows objective and accurate assessment of retinal vessel parameters. VITO has developed and released IFLEXIS, software for semi-automated retinal vessel analysis, which includes algorithms to determine blood vessel widths, vessel branching and vessel network complexity, integrated in a user-friendly workflow. During the Belgian Economic Mission to Qatar in March 2015, WCM-Q, VITO and Qatar Biobank signed a Memorandum of Understanding to explore the potential of retinal analysis for the early detection of retinal vessel abnormalities in subjects attending the QBB. Fundus images from 774 people attending the Qatar Biobank contained images from healthy controls, subjects with Impaired Glucose Tolerance (IGT), Hypertension (HT) and/or Type 2 diabetes (T2DM). Here, we report the utility of IFLEXIS in a subset of 597 fundus images obtained from 574 persons which could be analyzed. The following parameters were quantified: (i) FD (fractal dimension, using the box counting method), FFD (Fourier fractal dimension) and lacunarity of the vessel network, and (ii) CRAE, CRVE (central artery/ vein equivalent) and AVR (artery-to-vein ratio). Analysis revealed the following mean (range) for this population: 1.39 (1.32 – 1.53) for FD, 2.75 (1.77 – 2.98) for FFD, 1.00 (0.94 – 1.08) for lacunarity, 154.73 (93.79 – 195.67) for CRAE, 233.26 (159.67 – 307.94) for CRVE and 0.67 (0.48 – 0.93) for AVR. Statistical tests for the retinal parameters reveal interesting differences between the studied disease groups. When comparing HT with control subjects, statistically significant differences were found for the CRAE (150.05 ± 2.17 versus 157.88 ± 1.91, p=1.02×10-6) and the CRVE (231.74 ± 3.60 versus 238.28 ± 2.83, p=0.031). Comparing subjects with HT and T2DM with controls also revealed significant differences in CRAE (145.15 ± 3.62 versus 157.88 ± 1.91, p=4.66×10-8) and CRVE (218.47 ± 5.16 versus 238.28 ± 2.83, p=3.31×10-9). This study showed the feasibility of retinal vessel analysis of standard fundus images from the QBB using IFLEXIS. Despite the fact that the fundus images were originally not taken for this purpose, IFLEXIS software can extract novel retinal blood vessel dimensions. As blood vessel analysis has been shown to be relevant for chronic disease prediction, we propose to utilize the baseline assessment to augment the Qatar Biobank database to predict incident disease in follow-up studies. Single retinal features already appear to differentiate subjects with hypertension and/or diabetes compared to controls. We will undertake quantification of a larger image data set and will perform further data analysis to refine both the diagnostic and/or prognostic use of retinal metric analysis in the QBB population. To this end, we will combine retinal vessel metrics with demographics such as age, duration of disease and other metabolic parameters and study the association with whole genome sequence patterns in this population.

Loading

Article metrics loading...

/content/papers/10.5339/qfarc.2018.HBPD893
2018-03-15
2020-09-26
Loading full text...

Full text loading...

http://instance.metastore.ingenta.com/content/papers/10.5339/qfarc.2018.HBPD893
Loading
This is a required field
Please enter a valid email address
Approval was a Success
Invalid data
An Error Occurred
Approval was partially successful, following selected items could not be processed due to error