Multi-objective Evolutionary-Fuzzy for Vessel Tortuosity Characterisation

The tortuosity characterisation of vascular networks in digital retinal fundus images plays important roles in biomedicine for the diagnosis and early detection of different human illness such as diseases of the artery and vein vessels of the retina, hypertension, and varying forms of retinopathies. Although literature findings have revealed that varying techniques have been proposed, studies have shown that there are needs for further investigation to improve the performance of automated vascular network tortuosity characterisation. This paper investigates the suitability of multi-objective evolutionary-fuzzy classification approach for the tortuosity characterisation of the vascular networks utilising the extracted geometric features of the vascular networks. The method proposed in this study seems promising as the performance accuracy rates of 88.57%, 90%, 95%, and 100% are obtained for varying training sample sizes.