Differentiation of artery and vein in digital fundus photograph

In recent studies of hypertension, evaluating blood vessel change is considered as an important parameter for estimating disease progression. CT/MRI angiographies or gluteal muscle biopsy can be performed for this, however, these methods are expensive and invasive. In this paper, we performed the evaluation of the blood vessels on the digital fundus photograph of the eye. The method focuses on blood vessels near the optic disc and classifies them into arteries and veins. Using the red value of the RGB image, the optic disc can be segmented. And the center value and the radius of the optic disc were acquired. Subsequently, vessels between the twice and thrice radius of the optic disc were detected and analyzed. Using intensity profile of the circular area, vessel candidates were detected, and through a series of filters, vessels were selected. Veins and arteries showed distinct patterns in the intensity profile. Using those patterns, differentiation of retinal vein and artery was performed, and the result was compared with the reading of a retina specialist.