A lightweight network guided with differential matched filtering for retinal vessel segmentation

Yubo Tan; Shi-Xuan Zhao; Kai-Fu Yang; Yong-Jie Li

doi:10.1016/j.compbiomed.2023.106924

A lightweight network guided with differential matched filtering for retinal vessel segmentation

Comput Biol Med. 2023 Jun:160:106924. doi: 10.1016/j.compbiomed.2023.106924. Epub 2023 Apr 20.

Authors

Yubo Tan¹, Shi-Xuan Zhao², Kai-Fu Yang³, Yong-Jie Li⁴

Affiliations

¹ The MOE Key Laboratory for Neuroinformation, Radiation Oncology Key Laboratory of Sichuan Province, University of Electronic Science and Technology of China, China. Electronic address: tyb@std.uestc.edu.cn.
² The MOE Key Laboratory for Neuroinformation, Radiation Oncology Key Laboratory of Sichuan Province, University of Electronic Science and Technology of China, China. Electronic address: zhaosx@std.uestc.edu.cn.
³ The MOE Key Laboratory for Neuroinformation, Radiation Oncology Key Laboratory of Sichuan Province, University of Electronic Science and Technology of China, China. Electronic address: yangkf@uestc.edu.cn.
⁴ The MOE Key Laboratory for Neuroinformation, Radiation Oncology Key Laboratory of Sichuan Province, University of Electronic Science and Technology of China, China. Electronic address: liyj@uestc.edu.cn.

PMID: 37146492
DOI: 10.1016/j.compbiomed.2023.106924

Abstract

The geometric morphology of retinal vessels reflects the state of cardiovascular health, and fundus images are important reference materials for ophthalmologists. Great progress has been made in automated vessel segmentation, but few studies have focused on thin vessel breakage and false-positives in areas with lesions or low contrast. In this work, we propose a new network, differential matched filtering guided attention UNet (DMF-AU), to address these issues, incorporating a differential matched filtering layer, feature anisotropic attention, and a multiscale consistency constrained backbone to perform thin vessel segmentation. The differential matched filtering is used for the early identification of locally linear vessels, and the resulting rough vessel map guides the backbone to learn vascular details. Feature anisotropic attention reinforces the vessel features of spatial linearity at each stage of the model. Multiscale constraints reduce the loss of vessel information while pooling within large receptive fields. In tests on multiple classical datasets, the proposed model performed well compared with other algorithms on several specially designed criteria for vessel segmentation. DMF-AU is a high-performance, lightweight vessel segmentation model. The source code is at https://github.com/tyb311/DMF-AU.

Keywords: Anisotropic attention; Fundus image; Matched filtering; Vessel segmentation.

MeSH terms

Algorithms*
Fundus Oculi
Image Processing, Computer-Assisted / methods
Retinal Vessels* / diagnostic imaging
Software