RBGNet: Reliable Boundary-Guided Segmentation of Choroidal Neovascularization

doi:10.1007/978-3-031-43901-8_16

科研成果详情

题名	RBGNet: Reliable Boundary-Guided Segmentation of Choroidal Neovascularization
作者	Chen, Tao 1,2; Zhao, Yitian 1; Mou, Lei 1; Zhang, Dan 3; Xu, Xiayu 4; Liu, Mengting 5; Fu, Huazhu 6; Zhang, Jiong 1
会议录名称	SPRINGER INTERNATIONAL PUBLISHING AG 影响因子和分区
语种	英语
原始文献类型	Proceedings Paper ; Conference Paper
会议名称	26th International Conference on Medical Image Computing and Computer-Assisted Intervention (MICCAI)
会议日期	OCT 08-12, 2023
会议地点	Vancouver, CANADA
关键词	CNV OCTA Transformer Uncertainty Multi-task
其他关键词	ANGIOGRAPHY
摘要	Choroidal neovascularization (CNV) is a leading cause of visual impairment in retinal diseases. Optical coherence tomography angiography (OCTA) enables non-invasive CNV visualization with micrometerscale resolution, aiding precise extraction and analysis. Nevertheless, the irregular shape patterns, variable scales, and blurred lesion boundaries of CNVs present challenges for their precise segmentation in OCTA images. In this study, we propose a Reliable Boundary-Guided choroidal neovascularization segmentation Network (RBGNet) to address these issues. Specifically, our RBGNet comprises a dual-stream encoder and a multi-task decoder. The encoder consists of a convolutional neural network (CNN) stream and a transformer stream. The transformer captures global context and establishes long-range dependencies, compensating for the limitations of the CNN. The decoder is designed with multiple tasks to address specific challenges. Reliable boundary guidance is achieved by evaluating the uncertainty of each pixel label, By assigning it as a weight to regions with highly unstable boundaries, the network's ability to learn precise boundary locations can be improved, ultimately leading to more accurate segmentation results. The prediction results are also used to adaptively adjust the weighting factors between losses to guide the network's learning process. Our experimental results demonstrate that RBGNet outperforms existing methods, achieving a Dice score of 90.42% for CNV region segmentation and 90.25% for CNV vessel segmentation. https://github.com/iMED-Lab/RBGnet-Pytorch.git.
资助项目	Ningbo Natural Science Foundation[2022J143];A*STAR[A20H4b0141];National Science Foundation Program of China[62103398,62272444];Zhejiang Provincial Natural Science Foundation of China[LQ23F010002,LR22F020008,LZ23F010002]
出版地	CHAM
出版者	Springer Science and Business Media Deutschland GmbH
ISSN	0302-9743
EISSN	1611-3349
卷号	14223 LNCS
页码	163-172
DOI	10.1007/978-3-031-43901-8_16
页数	10
URL	查看原文
WOS类目	Computer Science, Software Engineering ; Computer Science, Theory & Methods ; Radiology, Nuclear Medicine & Medical Imaging
WOS研究方向	Computer Science ; Radiology, Nuclear Medicine & Medical Imaging
WOS记录号	WOS:001109630700016
收录类别	CPCI ; CPCI-S ; EI ; SCOPUS
发表日期	2023
EI入藏号	20234314955270
EI主题词	Optical tomography
EI分类号	461.6 Medicine and Pharmacology ; 716.1 Information Theory and Signal Processing ; 723.2 Data Processing and Image Processing ; 741.3 Optical Devices and Systems
通讯作者地址	[Zhao, Yitian;Zhang, Jiong]Chinese Acad Sci, Ningbo Inst Mat Technol & Engn, Inst Biomed Engn, Ningbo, Peoples R China.
Scopus记录号	2-s2.0-85174699895
Scopus学科分类	Theoretical Computer Science;Computer Science (all)
引用统计
文献类型	会议论文
条目标识符	https://kms.wmu.edu.cn/handle/3ETUA0LF/210066
专题	其他_温州医科大学慈溪生物医药研究院
通讯作者	Zhao, Yitian; Zhang, Jiong
作者单位	1.Chinese Acad Sci, Ningbo Inst Mat Technol & Engn, Inst Biomed Engn, Ningbo, Peoples R China; 2.Wenzhou Med Univ, Cixi Biomed Res Inst, Ningbo, Peoples R China; 3.Ningbo Univ Technol, Sch Cyber Sci & Engn, Ningbo, Peoples R China; 4.Xi An Jiao Tong Univ, Sch Life Sci & Technol, Xian, Peoples R China; 5.Sun Yat Sen Univ, Sch Biomed Engn, Shenzhen, Peoples R China; 6.ASTAR, Inst High Performance Comp, Singapore, Singapore
第一作者单位	其他_温州医科大学慈溪生物医药研究院
推荐引用方式 GB/T 7714	Chen, Tao,Zhao, Yitian,Mou, Lei,et al. RBGNet: Reliable Boundary-Guided Segmentation of Choroidal Neovascularization[C]. CHAM:Springer Science and Business Media Deutschland GmbH,2023:163-172.