arXiv 2404.089512024-04-13cs.CVcs.LG

混合域半监督医学图像分割中中间域的构建与探索

Constructing and Exploring Intermediate Domains in Mixed Domain Semi-supervised Medical Image Segmentation

Nanjing University（南京大学）；Southeast University（东南大学）；Shandong Women’s University（山东女子学院）；State Key Laboratory for Novel Software Technology（现代软件技术国家重点实验室）；National Institute of Healthcare Data Science（国家医疗数据科学研究院）

Qinghe Ma, Jian Zhang, Lei Qi, Qian Yu, Yinghuan Shi, Yang Gao

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

针对医学图像分割中标注有限与域偏移共存的问题，提出混合域半监督分割场景MiDSS，通过UCP构建中间域、SymGD策略与TP-RAM方法提升性能，在Prostate数据集上Dice系数较现有最优方法提升13.57%。

摘要：

标注有限与域偏移都是医学图像分割领域普遍存在的挑战。传统的半监督分割方法与无监督域适应方法分别单独解决其中一个问题。然而，标注有限与域偏移共存的情况十分常见，这促使我们提出一种全新且具有挑战性的场景：混合域半监督医学图像分割（Mixed Domain Semi-supervised medical image Segmentation，MiDSS）。在该场景下，我们处理来自多个医疗中心的数据，其中仅单个域有有限的标注，而多个域都存在大量无标注数据。我们发现，解决该问题的关键在于当标注数据存在域偏移时，如何为无标注数据生成可靠的伪标签。为解决这一问题，我们采用图像间的统一复制粘贴（Unified Copy-Paste，UCP）来构建中间域，促进知识从有标注数据的域向无标注数据的域迁移。为充分利用中间域内的信息，我们提出了对称引导训练策略（symmetric Guidance training strategy，SymGD），该策略通过融合中间样本的伪标签，额外为无标注数据提供直接引导。随后，我们提出了训练过程感知随机幅度混合（Training Process aware Random Amplitude MixUp，TP-RAM），以逐步将风格转换组件融入中间样本。在三个公开数据集上的实验结果表明，与现有最优方法相比，我们的方法在Prostate数据集上的Dice系数取得了13.57%的显著提升。我们的代码可在https://github.com/MQinghe/MiDSS 获取。

英文摘要：

Both limited annotation and domain shift are prevalent challenges in medical image segmentation. Traditional semi-supervised segmentation and unsupervised domain adaptation methods address one of these issues separately. However, the coexistence of limited annotation and domain shift is quite common, which motivates us to introduce a novel and challenging scenario: Mixed Domain Semi-supervised medical image Segmentation (MiDSS). In this scenario, we handle data from multiple medical centers, with limited annotations available for a single domain and a large amount of unlabeled data from multiple domains. We found that the key to solving the problem lies in how to generate reliable pseudo labels for the unlabeled data in the presence of domain shift with labeled data. To tackle this issue, we employ Unified Copy-Paste (UCP) between images to construct intermediate domains, facilitating the knowledge transfer from the domain of labeled data to the domains of unlabeled data. To fully utilize the information within the intermediate domain, we propose a symmetric Guidance training strategy (SymGD), which additionally offers direct guidance to unlabeled data by merging pseudo labels from intermediate samples. Subsequently, we introduce a Training Process aware Random Amplitude MixUp (TP-RAM) to progressively incorporate style-transition components into intermediate samples. Compared with existing state-of-the-art approaches, our method achieves a notable 13.57% improvement in Dice score on Prostate dataset, as demonstrated on three public datasets. Our code is available at https://github.com/MQinghe/MiDSS .

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。