| Popis: |
Diabetes is exploding and pandemic all over the world. High glucose peaks cause vessel wall damage, especially in the microvasculature. The retina is particularly susceptible since it is the most oxygen-consuming tissue in the body. Retinal deterioration leads to blindness, and it is estimated that 4% of diabetics will go blind at some point. Microaneurysms, haemorrhages, hard and soft exudates are few biomarkers on retinal vasculature, to identify Diabetic retinopathy. The diameter or width of retinal vessels is also a key indicator of blood flow and metabolism in the retina. The goal of this research is to employ deep learning algorithms to efficiently segment, classify and assess the retinal vascular diameters of a healthy and diabetic individual and to look at how these methodologies contribute to the identification of diabetic retinopathy and may be used as a predictor of disease diagnosis. U-Net architecture for segmentation of vessels, SVM for classification of vessels as arteries and veins and VAMPIRE for the assessment of vascular diameters are proposed in this research work. |
