Retinal vessel segmentation using dense U-net with multiscale inputs

Kejuan Yue; Beiji Zou; Zailiang Chen; Qing Liu

doi:10.1117/1.JMI.6.3.034004

Retinal vessel segmentation using dense U-net with multiscale inputs

J Med Imaging (Bellingham). 2019 Jul;6(3):034004. doi: 10.1117/1.JMI.6.3.034004. Epub 2019 Sep 27.

Authors

Kejuan Yue^{1

2

3}, Beiji Zou^{1

2}, Zailiang Chen^{1

2}, Qing Liu^{1

2}

Affiliations

¹ Central South University, School of Computer Science and Engineering, Changsha, China.
² Central South University, Engineering Research Center of Machine Vision and Intelligent Medicine, Changsha, China.
³ Hunan First Normal University, School of Information Science and Engineering, Changsha, China.

Abstract

A color fundus image is an image of the inner wall of the eyeball taken with a fundus camera. Doctors can observe retinal vessel changes in the image, and these changes can be used to diagnose many serious diseases such as atherosclerosis, glaucoma, and age-related macular degeneration. Automated segmentation of retinal vessels can facilitate more efficient diagnosis of these diseases. We propose an improved U-net architecture to segment retinal vessels. Multiscale input layer and dense block are introduced into the conventional U-net, so that the network can make use of richer spatial context information. The proposed method is evaluated on the public dataset DRIVE, achieving 0.8199 in sensitivity and 0.9561 in accuracy. Especially for thin blood vessels, which are difficult to detect because of their low contrast with the background pixels, the segmentation results have been improved.

Keywords: U-net; dense block; multiscale; retinal vessel segmentation.