Semi-supervised lung tumor segmentation based on multi-scale consistency and regional reliability perception
10.3969/j.issn.1005-202X.2024.09.004
- VernacularTitle:基于多尺度区域可靠性感知的半监督肺肿瘤分割
- Author:
Weipeng LIU
1
;
Yedong QI
;
Jian LI
;
Haixing XU
Author Information
1. 河北工业大学人工智能与数据科学学院,天津 300130;河北工业大学高端装备智能感知与先进控制研究所,天津 300130
- Keywords:
semi-supervised learning;
medical image segmentation;
lung tumor;
reliability perception;
multi-scale consistency
- From:
Chinese Journal of Medical Physics
2024;41(9):1078-1085
- CountryChina
- Language:Chinese
-
Abstract:
A semi-supervised learning method based on multi-scale consistency and regional reliability perception is proposed to combine unlabeled data with a small amount of labeled data to achieve high-performance lung tumor segmentation tasks.A multi-scale consistency mean teacher framework is used to construct a multi-scale consistency loss and constrain the outputs in the mean teacher network to be consistent across multiple scales,so that the model learns richer consistency knowledge.In addition,a regional reliability perception scheme is adopted to make the knowledge exchange between consistency learning more efficient,enabling the model to learn more valid and reliable knowledge from unlabeled data.The evaluation on the lung tumor dataset in the Medical Segmentation Decathlon shows superior performance of the proposed method over current state-of-the-art semi-supervised learning methods,validating its effectiveness.
