Boundary-Aware Network for Kidney Tumor Segmentation

Shishuai Hu, Jianpeng Zhang, Yong Xia

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

14 Scopus citations

Abstract

Segmentation of the kidney and kidney tumors using computed tomography (CT) is a crucial step in related surgical procedures. Although many deep learning models have been constructed to solve this problem, most of them ignore the boundary information. In this paper, we propose a boundary-aware network (BA-Net) for kidney and kidney tumor segmentation. This model consists of a shared 3D encoder, a 3D boundary decoder, and a 3D segmentation decoder. In contrast to existing boundary-involved methods, we first introduce the skip connections from the boundary decoder to the segmentation decoder, incorporating the boundary prior as the attention that indicates the error-prone regions into the segmentation process, and then define the consistency loss to push both decoders towards producing the same result. Besides, we also use the strategies of multi-scale input and deep supervision to extract hierarchical structural information, which can alleviate the issues caused by variable tumor sizes. We evaluated the proposed BA-Net on the kidney tumor segmentation challenge (KiTS19) dataset. The results suggest that the boundary decoder and consistency loss used in our model are effective and the BA-Net is able to produce relatively accurate segmentation of the kidney and kidney tumors.

Original languageEnglish
Title of host publicationMachine Learning in Medical Imaging - 11th International Workshop, MLMI 2020, Held in Conjunction with MICCAI 2020, Proceedings
EditorsMingxia Liu, Chunfeng Lian, Pingkun Yan, Xiaohuan Cao
PublisherSpringer Science and Business Media Deutschland GmbH
Pages189-198
Number of pages10
ISBN (Print)9783030598600
DOIs
StatePublished - 2020
Event11th International Workshop on Machine Learning in Medical Imaging, MLMI 2020, held in conjunction with the 23rd International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2020 - Lima, Peru
Duration: 4 Oct 20204 Oct 2020

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume12436 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference11th International Workshop on Machine Learning in Medical Imaging, MLMI 2020, held in conjunction with the 23rd International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2020
Country/TerritoryPeru
CityLima
Period4/10/204/10/20

Keywords

  • Boundary detection
  • Computed tomography
  • Deep learning
  • Kidney and kidney tumor segmentation

Fingerprint

Dive into the research topics of 'Boundary-Aware Network for Kidney Tumor Segmentation'. Together they form a unique fingerprint.

Cite this