Retinal Vessel Segmentation Algorithm Based on Attention Mechanism

The current retinal vessel segmentation algorithm suffers from the problems of poor microvascular feature capture and low segmentation efficiency. To address these problems, this paper proposes a retinal vessel segmentation algorithm based on the attention mechanism. The algorithm uses LadderNet as the base network and adds a channel improve the extraction capability of retinal microvascular features, thus enhancing the segmentation quality and efficiency of blood vessels. Experimental validation on the publicly available DRIVE dataset yielded accuracy, sensitivity and specificity as high as 95.61%, 78.55% and 98.10% respectively, and the experimental results show that the algorithm in this paper can segment retinal blood vessels effectively.