IEEE Trans Med Imaging. 2023 May;42(5):1388-1400. doi: 10.1109/TMI.2022.3228254. Epub 2023 May 2.
Well-annotated medical datasets enable deep neural networks (DNNs) to gain strong power in extracting lesion-related features. Building such large and well-designed medical datasets is costly due to the need for high-level expertise. Model pre-training based on ImageNet is a common practice to gain better generalization when the data amount is limited. However, it suffers from the domain gap between natural and medical images. In this work, we pre-train DNNs on ultrasound (US) domains instead of ImageNet to reduce the domain gap in medical US applications. To learn US image representations based on unlabeled US videos, we propose a novel meta-learning-based contrastive learning method, namely Meta Ultrasound Contrastive Learning (Meta-USCL). To tackle the key challenge of obtaining semantically consistent sample pairs for contrastive learning, we present a positive pair generation module along with an automatic sample weighting module based on meta-learning. Experimental results on multiple computer-aided diagnosis (CAD) problems, including pneumonia detection, breast cancer classification, and breast tumor segmentation, show that the proposed self-supervised method reaches state-of-the-art (SOTA). The codes are available at https://github.com/Schuture/Meta-USCL.
标注良好的医学数据集使深度神经网络 (DNN) 能够在提取与病变相关的特征方面具有强大的能力。由于需要高水平的专业知识，因此构建此类大型且设计良好的医学数据集成本很高。在数据量有限的情况下，基于 ImageNet 的模型预训练是提高泛化能力的常见做法。然而，它存在自然图像和医学图像之间的领域差距。在这项工作中，我们在超声 (US) 领域对 DNN 进行预训练，以减少医学 US 应用中的领域差距。为了基于未标记的 US 视频学习 US 图像表示，我们提出了一种基于元学习的新对比学习方法，即元超声对比学习 (Meta-USCL)。为了解决对比学习中获得语义一致样本对的关键挑战，我们提出了一个正样本生成模块以及一个基于元学习的自动样本加权模块。在多个计算机辅助诊断 (CAD) 问题上的实验结果，包括肺炎检测、乳腺癌分类和乳腺肿瘤分割，表明所提出的自监督方法达到了最先进的水平。代码可在 https://github.com/Schuture/Meta-USCL 上获得。
