Retinal vascular segmentation based on depth-separable convolution and attention mechanisms

Xiaopeng Liu, Dongxu Gao, Congyi Zhang, Hongwei Gao, Zhaojie Ju

Research output: Chapter in Book/Report/Conference proceedingConference contribution

3 Downloads (Pure)

Abstract

Retinal vascular segmentation is an important research direction in the field of medical image processing, its main purpose is to automatically segment the vascular area from the fundus image, and provide doctors with more accurate diagnosis results and treatment plans. In recent years, with the continuous development of deep learning technology, retinal vascular segmentation algorithm based on deep learning has gradually become a research hotspot. In this paper, the retinal vascular segmentation algorithm based on deep learning is mainly improved, and the retinal vascular segmentation algorithm based on IPN-V2 is improved, in an attempt to make new explorations.

The retinal vascular segmentation algorithm based on IPN-V2 provides global information, but requires a large amount of image data and label information, the image size is different, and most importantly, the accuracy of the model for the segmentation of the original image is not enough. Therefore, this paper improves the retinal vascular segmentation algorithm based on IPN-V2, introduces the attention mechanism, and constructs a retinal vascular segmentation model based on ASR-IPN-V2, which enables the model to extract more image details from the original image through the depth-separable convolution and convolutional block attention mechanisms.

Experiments show that the retinal vascular segmentation model based on ASR-IPN-V2 greatly improves the efficiency of retinal vascular segmentation.
Original languageEnglish
Title of host publicationIntelligent Robotics and Applications 16th International Conference, ICIRA 2023, Hangzhou, China, July 5–7, 2023, Proceedings, Part III
EditorsHuayong Yang, Honghai Liu, Jun Zou, Zhouping Yin, Lianqing Liu, Geng Yang, Xiaoping Ouyang, Zhiyong Wang
PublisherSpringer
Pages145-160
Number of pages16
ISBN (Electronic)9789819964895
ISBN (Print)9789819964888
DOIs
Publication statusPublished - 11 Oct 2023
EventInternational Conference on Intelligent Robotics and Applications - Hangzhou, China
Duration: 5 Jul 20237 Jul 2023

Publication series

NameLecture Notes in Computer Science
PublisherSpringer
Volume14269
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

ConferenceInternational Conference on Intelligent Robotics and Applications
Country/TerritoryChina
CityHangzhou
Period5/07/237/07/23

Keywords

  • IPN-V2 model
  • segmentation of retinal blood vessels
  • attention mechanism

Fingerprint

Dive into the research topics of 'Retinal vascular segmentation based on depth-separable convolution and attention mechanisms'. Together they form a unique fingerprint.

Cite this