This thesis focuses on the development of automated methods for artery-vein (AV) segmentation in fundus images, an essential step in enabling early and non-invasive detection of systemic and neurodegenerative diseases. The retina offers a unique window into neuronal and vascular health, as its vascularization closely reflects changes in the brain and cardiovascular system. Therefore, accurate
AV segmentation plays a key role in extracting vascular biomarkers with diagnostic and prognostic value. Despite its importance, AV segmentation remains a technically complex task. Arteries and veins have very similar visual and structural characteristics, leading to ambiguity even among expert annotators. This difficulty is compounded by the scarcity of high-quality, pixel-level annotated datasets needed to train reliable models. The goal of this thesis is to address these challenges by developing automated methods for reliable, reproducible, and scalable AV segmentation to extract vascular measurements from fundus images.