Retinal blood vessel segmentation using AI

Retinal microvascular is a dependable marker of abnormalities in vessel morphology, that have been linked to a variety of clinical disorders, both in ocular and metastatic disease. However, accurate vessel segmentation, which would be intricate- and time-intensive, is required for objective and statistical evaluation of the retinal blood vessels. In terms of segmenting retinal vessels, artificial intelligence (AI) has shown a significant amount of promise. In this study, the fundus images retinal blood vessel is segmented using deep learning methods. The data set required for this study is collected from the Kaggle website and pre-processed using various techniques to make it compatible with the deep learning models. The pre-processed images are then segmented using deep learning models such as LadderNet and UNet. The efficiency of the deep learning models are validated using performance metrics such as Intersection of Union (IoU), accuracy and F1 score. This study shows an accuracy of 0.98% using the UNet deep learning model and it is deemed to be an efficient model than the pre-existing models.