Comparative Analysis for Blood Vessel Segmentation based on CNN Models

Abstract

Retinal fundus images are images of color representing the inner surface of the human eye; they provide the anatomical structure of retinal blood vessels. Early diagnosis of eye-related diseases is crucial in order to take precautionary protocols to prevent major vision loss. In this study, a Machine learning-based approach for blood vessel segmentation is pro-posed. To this end, two different supervised Machine learning algorithms were implemented to analyze their performance and efficiency on blood vessel segmentation. These two algorithms are based on U-net modeling and ResNet50. A comparative analysis between the developed models and the state-of-the-art was conducted to determine a suitable solution for accurate blood vessel segmentation.