Semi-supervised medical image segmentation method based on consistency regularization(PDF)
《中国医学物理学杂志》[ISSN:1005-202X/CN:44-1351/R]
- Issue:
- 2025年第6期
- Page:
- 784-790
- Research Field:
- 医学影像物理
- Publishing date:
Info
- Title:
- Semi-supervised medical image segmentation method based on consistency regularization
- Author(s):
- XU Xinhui1; ZENG Zhiyong2; LIN Zhengyu3
- 1. College of Computer and Cyber Security, Fujian Normal University, Fuzhou 350117, China; 2. Digital Fujian Internet of ThingsLaboratory of Environmental Monitoring, Fuzhou 350117, China; 3. Interventional Department, the First Affiliated Hospital of FujianMedical University, Fuzhou 350000, China
- Keywords:
- semi-supervised medical image segmentation; uncertainty estimation; edge-preserving noise; exponentialmoving average
- PACS:
- R318
- DOI:
- DOI:10.3969/j.issn.1005-202X.2025.06.011
- Abstract:
- In response to the high cost and time consumption of medical image annotation, and issues such as the imprecisionof unlabeled data segmentation in semi-supervised medical image segmentation, loss of image edge information, and delayedparameter updates, a semi-supervised medical image segmentation method based on consistency regularization is presented.Firstly, an uncertainty measurement method based on the dual perspectives of entropy and variance is designed to assess theuncertainty of predictions for unlabeled data, jointly evaluating the uncertainty of unlabeled data from the perspectives ofentropy and variance. Then, edge-preserving noise based on the Canny operator is used to retain image edge information andimportant structures, thereby avoiding the potential blurring of organ edges that may result from the addition of random noise.Finally, a semi-supervised residual-driven segmentation method based on the mean teacher framework is developed, with aFrobenius norm regularization term in the exponential moving average scheme to enhance the performance of mean teacher.The proposed method is validated on the publicly available multi-organ segmentation benchmark dataset BTCV and braintumor segmentation dataset BraTS 2019. In the case of 40% labeled data in the BTCV dataset, Dice similarity coefficient andstandardized surface distance are 77.42% and 79.47%, respectively. In the case of 20% labeled data in the BraTS 2019dataset, the proposed method achieve a Dice similarity coefficient of 83.89%, a Jaccard coefficient of 74.21%, an averagesurface distance of 2.34 mm, and a 95% Hausdorff distance of 9.08 mm, demonstrating its superiority.
Last Update: 2025-07-01