• 文献检索
  • 文档翻译
  • 深度研究
  • 学术资讯
  • Suppr Zotero 插件Zotero 插件
  • 邀请有礼
  • 套餐&价格
  • 历史记录
应用&插件
Suppr Zotero 插件Zotero 插件浏览器插件Mac 客户端Windows 客户端微信小程序
定价
高级版会员购买积分包购买API积分包
服务
文献检索文档翻译深度研究API 文档MCP 服务
关于我们
关于 Suppr公司介绍联系我们用户协议隐私条款
关注我们

Suppr 超能文献

核心技术专利:CN118964589B侵权必究
粤ICP备2023148730 号-1Suppr @ 2026

文献检索

告别复杂PubMed语法,用中文像聊天一样搜索,搜遍4000万医学文献。AI智能推荐,让科研检索更轻松。

立即免费搜索

文件翻译

保留排版,准确专业,支持PDF/Word/PPT等文件格式,支持 12+语言互译。

免费翻译文档

深度研究

AI帮你快速写综述,25分钟生成高质量综述,智能提取关键信息,辅助科研写作。

立即免费体验

ReFixMatch-LS:用于半监督皮肤病变分类的伪标签复用

ReFixMatch-LS: reusing pseudo-labels for semi-supervised skin lesion classification.

作者信息

Zhou Shaofeng, Tian Shenwei, Yu Long, Wu Weidong, Zhang Dezhi, Peng Zhen, Zhou Zhicheng

机构信息

College of Software, Xinjiang University, Urumqi, 830000, China.

Key Laboratory of Software Engineering Technology, Xinjiang University, Urumqi, 830000, China.

出版信息

Med Biol Eng Comput. 2023 May;61(5):1033-1045. doi: 10.1007/s11517-022-02743-5. Epub 2023 Jan 17.

DOI:10.1007/s11517-022-02743-5
PMID:36648692
Abstract

Recent research on semi-supervised learning (SSL) is mainly based on the method of consistency regularization, which relies on domain-specific data augmentation. Pseudo-labeling is a more general method that has no such restrictions but performs limited by noisy training. We combine both approaches and focus on generating pseudo-labels using domain-independent weak augmentation. In this article, we propose ReFixMatch-LS and apply it to the classification of medical images. First, we reduce the impact of noisy artificial labels by label smoothing and consistent regularization. Then, by recording high-confidence pseudo-labels generated from each epoch during training, we reuse the generated pseudo-labels to train the model in the subsequent epochs. ReFixMatch-LS effectively increases the number of pseudo-labels and improves the model performance. We validate the effectiveness of ReFixMatch-LS on skin lesion diagnosis in the ISIC 2018 and ISIC 2019 challenge datasets, obtaining AUCs of 91.54%, 93.68%, 94.55%, and 95.47% on the four proportions of labeled data from ISIC 2018.

摘要

近期关于半监督学习(SSL)的研究主要基于一致性正则化方法,该方法依赖于特定领域的数据增强。伪标签是一种更通用的方法,没有此类限制,但受噪声训练的影响有限。我们将这两种方法结合起来,专注于使用与领域无关的弱增强来生成伪标签。在本文中,我们提出了ReFixMatch-LS并将其应用于医学图像分类。首先,我们通过标签平滑和一致性正则化来减少有噪声的人工标签的影响。然后,通过记录训练期间每个轮次生成的高置信度伪标签,我们在后续轮次中重用生成的伪标签来训练模型。ReFixMatch-LS有效地增加了伪标签的数量并提高了模型性能。我们在ISIC 2018和ISIC 2019挑战数据集中验证了ReFixMatch-LS在皮肤病变诊断方面的有效性,在ISIC 2018的四个标注数据比例上分别获得了91.54%、93.68%、94.55%和95.47%的曲线下面积(AUC)。

相似文献

1
ReFixMatch-LS: reusing pseudo-labels for semi-supervised skin lesion classification.ReFixMatch-LS:用于半监督皮肤病变分类的伪标签复用
Med Biol Eng Comput. 2023 May;61(5):1033-1045. doi: 10.1007/s11517-022-02743-5. Epub 2023 Jan 17.
2
FaxMatch: Multi-Curriculum Pseudo-Labeling for semi-supervised medical image classification.FaxMatch:用于半监督医学图像分类的多课程伪标签
Med Phys. 2023 May;50(5):3210-3222. doi: 10.1002/mp.16312. Epub 2023 Feb 21.
3
CPSS: Fusing consistency regularization and pseudo-labeling techniques for semi-supervised deep cardiovascular disease detection using all unlabeled electrocardiograms.CPSS:利用所有未标记的心电图进行半监督深度心血管疾病检测的一致性正则化和伪标记技术融合。
Comput Methods Programs Biomed. 2024 Sep;254:108315. doi: 10.1016/j.cmpb.2024.108315. Epub 2024 Jul 4.
4
Semantic contrast with uncertainty-aware pseudo label for lumbar semi-supervised classification.基于具有不确定性感知的伪标签的语义对比进行腰椎半监督分类。
Comput Biol Med. 2024 Aug;178:108754. doi: 10.1016/j.compbiomed.2024.108754. Epub 2024 Jun 15.
5
STAMP: A Self-training Student-Teacher Augmentation-Driven Meta Pseudo-Labeling Framework for 3D Cardiac MRI Image Segmentation.STAMP:一种用于3D心脏MRI图像分割的基于自我训练的师生增强驱动的元伪标签框架。
Med Image Underst Anal. 2022;13413:371-386. doi: 10.1007/978-3-031-12053-4_28. Epub 2022 Jul 25.
6
FMixCutMatch for semi-supervised deep learning.FMixCutMatch 用于半监督深度学习。
Neural Netw. 2021 Jan;133:166-176. doi: 10.1016/j.neunet.2020.10.018. Epub 2020 Nov 10.
7
Mutual learning with reliable pseudo label for semi-supervised medical image segmentation.用于半监督医学图像分割的基于可靠伪标签的相互学习
Med Image Anal. 2024 May;94:103111. doi: 10.1016/j.media.2024.103111. Epub 2024 Feb 21.
8
Deep Source Semi-Supervised Transfer Learning (DS3TL) for Cross-Subject EEG Classification.深度源半监督迁移学习 (DS3TL) 在跨被试 EEG 分类中的应用。
IEEE Trans Biomed Eng. 2024 Apr;71(4):1308-1318. doi: 10.1109/TBME.2023.3333327. Epub 2024 Mar 20.
9
Deep virtual adversarial self-training with consistency regularization for semi-supervised medical image classification.深度对偶对抗自训练与一致性正则化在半监督医学图像分类中的应用。
Med Image Anal. 2021 May;70:102010. doi: 10.1016/j.media.2021.102010. Epub 2021 Feb 22.
10
Improving Skin Lesion Segmentation with Self-Training.通过自我训练改进皮肤病变分割
Cancers (Basel). 2024 Mar 11;16(6):1120. doi: 10.3390/cancers16061120.

引用本文的文献

1
Advancements and challenges of artificial intelligence in dermatology: a review of applications and perspectives in China.人工智能在皮肤病学中的进展与挑战:中国应用与前景综述
Front Digit Health. 2025 Aug 13;7:1544520. doi: 10.3389/fdgth.2025.1544520. eCollection 2025.
2
Dual-branch Transformer for semi-supervised medical image segmentation.双分支Transformer 用于半监督医学图像分割。
J Appl Clin Med Phys. 2024 Oct;25(10):e14483. doi: 10.1002/acm2.14483. Epub 2024 Aug 12.
3
Deep learning with noisy labels in medical prediction problems: a scoping review.
深度学习中带噪标签在医学预测问题中的应用:范围综述。
J Am Med Inform Assoc. 2024 Jun 20;31(7):1596-1607. doi: 10.1093/jamia/ocae108.
4
A lightweight bladder tumor segmentation method based on attention mechanism.基于注意力机制的轻量级膀胱肿瘤分割方法。
Med Biol Eng Comput. 2024 May;62(5):1519-1534. doi: 10.1007/s11517-024-03018-x. Epub 2024 Feb 2.