Hybrid UNet Architecture based on Residual Learning of Fundus Images for Retinal Vessel Segmentation

Abstract

This paper deals with the new segmentation techniques for retinal blood vessels on fundus images. This technique aims at extracting thin vessels to reduce the intensity difference between thick and thin vessels. This paper proposes the modified UNet model by incorporating ResNet blocks into it which includes structured prediction. In this work we generate the visualization of blood vessels from retinal fundus image for two loss functions namely cross entropy loss and Dice loss where the network classifies several pixels simultaneously. The results shows higher accuracy by considering a much more expressive UNet algorithm and outperforms the past algorithms for Retinal Vessel Segmentation. The benefits of this approach will be demonstrated empirically.

VG-DropDNet a Robust Architecture for Blood Vessels Segmentation on Retinal Image

“…The MCC on the VG-DropDNet architecture is higher than the results of Zhuo et al [74] and Orlando et al [73]. Although the MCC is lower than the results of Nagdeote and Prabhu [77], and Guo et al [75]., the VG-DropDNet architecture on DRIVE is able to have a good balance for blood vessels and nonblood vessel (background) segmentation on retinal images. It is indicated by the results of the MCC, G-mean, and F1-score which are all close to 1 on DRIVE.…”

“…In table V, the highest MCC and F1-score for DRIVE are obtained by Nagdeote and Prabhu [77]. However, the highest G-mean is obtained by the VG-DropDNet architecture.…”