EUVIMEDOpdag EUVIMED
Crossref

Deep self‐cleansing for medical image segmentation with noisy labels

Jiahua Dong, Yue Zhang, Qiuli Wang, Ruofeng Tong, Shihong Ying, Shaolin Gong, Xuanpu Zhang, Lanfen Lin, Yen-Wei Chen, Shaohua Kevin Zhou

Medical Physics · 2025

DOI: 10.1002/mp.70007 ↗

At a glance

Journal article in Medical Physics, 2025. Written by Jiahua Dong and 9 co-authors. Institution: College of Computer Science and Technology, Zhejiang University, Hangzhou, China. Major MEDLINE topics: Image Processing, Computer-Assisted, Deep Learning, Signal-To-Noise Ratio. Cited 2 times so far (Crossref); 47 references. Funded by: National Natural Science Foundation of China, National Key Research and Development Program of China, Natural Science Foundation of Zhejiang Province.

AbstractSubjectsBibliografiske oplysningerCiteRelated studies

Abstract (original language)

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.

Subjects

MeSH terms (MEDLINE)

Image Processing, Computer-AssistedDeep LearningSignal-To-Noise RatioHumansMagnetic Resonance ImagingTomography, X-Ray ComputedHeartLiver Neoplasms

Author keywords

label self‐cleansingmedical image segmentationnoisy labels

Authors and institutions

  • College of Computer Science and Technology, Zhejiang University, Hangzhou, China.
  • Center for Medical Imaging, Robotics, Analytic Computing & Learning (MIRACLE), Suzhou Institute for Advanced Research, University of Science and Technology of China, Suzhou, Hangzhou, China.
  • National Key Laboratory of Intelligent Collaborative Computing, University of Electronic Science and Technology of China, Chengdu, China.
  • 7T Magnetic Resonance Translational Medicine Research Center, Department of Radiology, Southwest Hospital, Army Medical University (Third Military Medical University), Chongqing, China.
  • Department of Radiology, The First Affiliated Hospital, College of Medicine, Zhejiang University, Hangzhou, China.
  • College of Information Science and Engineering, Ritsumeikan University, Shiga, Japan.

Funding

  • National Natural Science Foundation of China
  • National Key Research and Development Program of China
  • Natural Science Foundation of Zhejiang Province
  • Suzhou Basic Research Program
  • Open Fund Project of Guangdong Acedemy of Medical Science

Cite

APA

Dong, J., Zhang, Y., Wang, Q., Tong, R., Ying, S., Gong, S., Zhang, X., Lin, L., Chen, Y. W., & Zhou, S. K. (2025). Deep self‐cleansing for medical image segmentation with noisy labels. Medical Physics. https://doi.org/10.1002/mp.70007

Vancouver

Dong J, Zhang Y, Wang Q, Tong R, Ying S, Gong S, et al. Deep self‐cleansing for medical image segmentation with noisy labels. Medical Physics. 2025. doi:10.1002/mp.70007

BibTeX
@article{Dong2025,
  author = {Dong, Jiahua and Zhang, Yue and Wang, Qiuli and Tong, Ruofeng and Ying, Shihong and Gong, Shaolin and Zhang, Xuanpu and Lin, Lanfen and Chen, Yen-Wei and Zhou, Shaohua Kevin},
  title = {Deep self‐cleansing for medical image segmentation with noisy labels},
  journal = {Medical Physics},
  year = {2025},
  doi = {10.1002/mp.70007}
}
Start for free →