Deep self-cleansing for medical image segmentation with noisy labels

doi:10.1002/mp.70007

	Deep self-cleansing for medical image segmentation with noisy labels
	Dong, Jiahua 1; Zhang, Yue 2,3; Wang, Qiuli 4; Tong, Ruofeng 1; Ying, Shihong 5; Gong, Shaolin 5; Zhang, Xuanpu 5; Lin, Lanfen 1; Chen, Yen-Wei 6; Zhou, Shaohua Kevin 2,7,8,9
	2025-09-22
发表期刊	MEDICAL PHYSICS
ISSN	0094-2405
卷号	52 期号:10 页码:17
摘要	Background Medical image segmentation plays a pivotal role in medical imaging, significantly contributing to disease diagnosis and surgical planning. Traditional segmentation methods predominantly rely on supervised deep learning, where the accuracy of manually delineated labels is crucial for model performance. However, these labels often contain noise, such as missing annotations and imprecise boundaries, which can adversely affect the network's ability to accurately model target characteristics.Purpose This study aims to develop a robust segmentation framework capable of mitigating the impact of noisy labels during the training phase. The proposed framework is designed to preserve clean labels while cleansing noisy ones, thereby enhancing the overall segmentation accuracy.Methods We introduce a deep self-cleansing segmentation framework that incorporates two key modules as follows: a Gaussian Mixture Model (GMM)-based label filtering module (LFM) and a label cleansing module (LCM). The GMM-based LFM is employed to differentiate between noisy and clean labels. Subsequently, the LCM generates pseudo low-noise labels for the identified noisy samples. These pseudo-labels, along with the preserved clean labels, are then used to supervise the network training process.Results The framework was evaluated on a clinical liver tumor dataset (231 CT scans) and a public cardiac diagnosis dataset (200 MRI scans). Compared to baseline methods, our approach significantly improves segmentation performance, achieving a +7.31% boost in the B-model and a +12.36% improvement in the L-model. These results demonstrate the framework's ability to effectively suppress the interference of noisy labels and enhance segmentation accuracy. The method's capability to distinguish and cleanse noisy labels ensures more precise modeling of target structures, improving the robustness of the segmentation process.Conclusions The proposed deep self-cleansing segmentation framework offers a promising solution to the challenge of noisy labels in medical image segmentation. By integrating a GMM-based LFM and an LCM, the framework effectively preserves clean labels and generates pseudo low-noise labels, thereby improving the overall segmentation accuracy. The successful validation on both clinical and public datasets underscores the potential of this approach to enhance disease diagnosis and surgical planning in medical imaging.
关键词	label self-cleansing medical image segmentation noisy labels
DOI	10.1002/mp.70007
收录类别	SCI
语种	英语
资助项目	Open Fund Project of Guangdong Acedemy of Medical Science
WOS研究方向	Radiology, Nuclear Medicine & Medical Imaging
WOS类目	Radiology, Nuclear Medicine & Medical Imaging
WOS记录号	WOS:001576633200001
出版者	WILEY
引用统计
文献类型	期刊论文
条目标识符	http://119.78.100.204/handle/2XEOYT63/41697
专题	中国科学院计算技术研究所期刊论文_英文
通讯作者	Tong, Ruofeng; Ying, Shihong; Zhou, Shaohua Kevin
作者单位	1.Zhejiang Univ, Coll Comp Sci & Technol, Hangzhou, Peoples R China 2.Univ Sci & Technol China, Suzhou Inst Adv Res, Ctr Med Imaging Robot Analyt Comp & Learning, Hangzhou, Peoples R China 3.Univ Elect Sci & Technol China, Natl Key Lab Intelligent Collaborat Comp, Chengdu, Peoples R China 4.Third Mil Med Univ, Army Med Univ, 7T Magnet Resonance Translat Med Res Ctr, Southwest Hosp,Dept Radiol, Chongqing, Peoples R China 5.Zhejiang Univ, Affiliated Hosp 1, Coll Med, Dept Radiol, Hangzhou, Peoples R China 6.Ritsumeikan Univ, Coll Informat Sci & Engn, Kusatsu, Shiga, Japan 7.Univ Sci & Technol China, Sch Biomed Engn, Div Life Sci & Med, Hefei, Anhui, Peoples R China 8.Univ Sci & Technol China, Key Lab Precis & Intelligent Chem, Hefei, Anhui, Peoples R China 9.Chinese Acad Sci, Key Lab Intelligent Informat Proc, Inst Comp Technol, Beijing, Peoples R China
推荐引用方式 GB/T 7714	Dong, Jiahua,Zhang, Yue,Wang, Qiuli,et al. Deep self-cleansing for medical image segmentation with noisy labels[J]. MEDICAL PHYSICS,2025,52(10):17.
APA	Dong, Jiahua.,Zhang, Yue.,Wang, Qiuli.,Tong, Ruofeng.,Ying, Shihong.,...&Zhou, Shaohua Kevin.(2025).Deep self-cleansing for medical image segmentation with noisy labels.MEDICAL PHYSICS,52(10),17.
MLA	Dong, Jiahua,et al."Deep self-cleansing for medical image segmentation with noisy labels".MEDICAL PHYSICS 52.10(2025):17.