L-Seg: An end-to-end unified framework for multi-lesion segmentation of fundus images

doi:10.1016/j.neucom.2019.04.019

CORC > 计算技术研究所 > 中国科学院计算技术研究所 > 中国科学院计算技术研究所期刊论文 > 英文

	L-Seg: An end-to-end unified framework for multi-lesion segmentation of fundus images
	Wang, Kai 1,2; Guo, Song 1,5; Li, Tao 1,5; Kang, Hong 1,4,5; Li, Ning 1,5; Zhang, Yujun 3
刊名	NEUROCOMPUTING
	2019-07-15
卷号	349 页码:52-63
关键词	Multi-lesion segmentation Fundus image Diabetic retinopathy Class-imbalance
ISSN号	0925-2312
DOI	10.1016/j.neucom.2019.04.019
英文摘要	Diabetic retinopathy and diabetic macular edema are the two leading causes for blindness in working-age people, and the quantitative and qualitative diagnosis of these two diseases usually depends on the presence and areas of lesions in fundus images. The main related lesions include soft exudates, hard exudates, microaneurysms, and haemorrhages. However, segmentation of these four kinds of lesions is difficult due to their uncertainty in size, contrast, and high interclass similarity. Therefore, we aim to design a multi-lesion segmentation model. We have designed the first small object segmentation network (L-Seg) that can segment the four kinds of lesions simultaneously. Taking into account that small lesion regions could not response at high level of network, we propose a multi-scale feature fusion method to handle this problem. In addition, when considering the cases of both class-imbalance and loss-imbalance problems, we propose a multi-channel bin loss. We have evaluated L-Seg on three fundus datasets including two publicly available datasets - IDRiD and e-ophtha and one private dataset - DDR. Extensive experiments have demonstrated that L-Seg achieves better performance in small lesion segmentation than other deep learning models and traditional methods. Specially, the mAUC score of L-Seg is over 16.8%, 1.51% and 3.11% higher than that of DeepLab v3 + on IDRiD, e-ophtha and DDR datasets, respectively. Moreover, our framework shows competitive performance compared with top-3 teams in IDRiD challenge. (C) 2019 Elsevier B. V. All rights reserved.
资助项目	National Natural Science Foundation[61872200] ; National Key Research and Development Program of China[2018YFB1003405] ; National Key Research and Development Program of China[2016YFC0400709] ; Natural Science Foundation of Tianjin[18YFYZCG00060]
WOS研究方向	Computer Science
语种	英语
出版者	ELSEVIER SCIENCE BV
WOS记录号	WOS:000467536900006
内容类型	期刊论文
源URL	[http://119.78.100.204/handle/2XEOYT63/4251]
专题	中国科学院计算技术研究所期刊论文_英文
通讯作者	Wang, Kai
作者单位	1.Nankai Univ, Coll Comp Sci, Tianjin, Peoples R China 2.Key Lab Med Data Anal & Stat Res Tianjin, Tianjin, Peoples R China 3.Chinese Acad, Inst Comp Technol, Beijing, Peoples R China 4.Beijing Shanggong Med Technol Co Ltd, Beijing, Peoples R China 5.Nankai Univ, Tianjin Key Lab Network & Data Secur Technol, Tianjin, Peoples R China
推荐引用方式 GB/T 7714	Wang, Kai,Guo, Song,Li, Tao,et al. L-Seg: An end-to-end unified framework for multi-lesion segmentation of fundus images[J]. NEUROCOMPUTING,2019,349:52-63.
APA	Wang, Kai,Guo, Song,Li, Tao,Kang, Hong,Li, Ning,&Zhang, Yujun.(2019).L-Seg: An end-to-end unified framework for multi-lesion segmentation of fundus images.NEUROCOMPUTING,349,52-63.
MLA	Wang, Kai,et al."L-Seg: An end-to-end unified framework for multi-lesion segmentation of fundus images".NEUROCOMPUTING 349(2019):52-63.