Learning Active Contour Models for Medical Image Segmentation

Chen, Xu, Williams, Bryan M ORCID: 0000-0001-5930-287X, Vallabhaneni, Srinivasa R, Czanner, Gabriela, Williams, Rachel ORCID: 0000-0002-1954-0256 and Zheng, Yalin ORCID: 0000-0002-7873-0922 (2019) Learning Active Contour Models for Medical Image Segmentation. In: 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2019-6-15 - 2019-6-20, Long Beach, CA, USA.

This is the latest version of this item.

Access the full-text of this item by clicking on the Open Access link.

Text
Chen_Learning_Active_Contour_Models_for_Medical_Image_Segmentation_CVPR_2019_paper.pdf - Author Accepted Manuscript
Available under License : See the attached licence file.
Download (1MB)

Official URL: https://ieeexplore.ieee.org/document/8953484

Abstract

Image segmentation is an important step in medical image processing and has been widely studied and developed for refinement of clinical analysis and applications. New models based on deep learning have improved results but are restricted to pixel-wise fitting of the segmentation map. Our aim was to tackle this limitation by developing a new model based on deep learning which takes into account the area inside as well as outside the region of interest as well as the size of boundaries during learning. Specifically, we propose a new loss function which incorporates area and size information and integrates this into a dense deep learning model. We evaluated our approach on a dataset of more than 2,000 cardiac MRI scans. Our results show that the proposed loss function outperforms other mainstream loss function Cross-entropy on two common segmentation networks. Our loss function is robust while using different hyperparameter lambda.

Item Type:	Conference or Workshop Item (Unspecified)
Uncontrolled Keywords:	Biomedical MRI
Depositing User:	Symplectic Admin
Date Deposited:	22 Jun 2020 08:29
Last Modified:	18 Jan 2023 23:49
DOI:	10.1109/cvpr.2019.01190
Open Access URL:	http://openaccess.thecvf.com/content_CVPR_2019/pap...
Related URLs:	Author Publisher
URI:	https://livrepository.liverpool.ac.uk/id/eprint/3089844