Tian Yu, Liu Fengbei, Pang Guansong, Chen Yuanhong, Liu Yuyuan, Verjans Johan W, Singh Rajvinder, Carneiro Gustavo
Harvard Ophthalmology AI Lab, Harvard Medical School, United States of America.
Australian Institute for Machine Learning, University of Adelaide, Australia.
Med Image Anal. 2023 Dec;90:102930. doi: 10.1016/j.media.2023.102930. Epub 2023 Aug 18.
Unsupervised anomaly detection (UAD) methods are trained with normal (or healthy) images only, but during testing, they are able to classify normal and abnormal (or disease) images. UAD is an important medical image analysis (MIA) method to be applied in disease screening problems because the training sets available for those problems usually contain only normal images. However, the exclusive reliance on normal images may result in the learning of ineffective low-dimensional image representations that are not sensitive enough to detect and segment unseen abnormal lesions of varying size, appearance, and shape. Pre-training UAD methods with self-supervised learning, based on computer vision techniques, can mitigate this challenge, but they are sub-optimal because they do not explore domain knowledge for designing the pretext tasks, and their contrastive learning losses do not try to cluster the normal training images, which may result in a sparse distribution of normal images that is ineffective for anomaly detection. In this paper, we propose a new self-supervised pre-training method for MIA UAD applications, named Pseudo Multi-class Strong Augmentation via Contrastive Learning (PMSACL). PMSACL consists of a novel optimisation method that contrasts a normal image class from multiple pseudo classes of synthesised abnormal images, with each class enforced to form a dense cluster in the feature space. In the experiments, we show that our PMSACL pre-training improves the accuracy of SOTA UAD methods on many MIA benchmarks using colonoscopy, fundus screening and Covid-19 Chest X-ray datasets.
无监督异常检测（UAD）方法仅使用正常（或健康）图像进行训练，但在测试期间，它们能够对正常图像和异常（或疾病）图像进行分类。UAD是一种重要的医学图像分析（MIA）方法，可应用于疾病筛查问题，因为针对这些问题的可用训练集通常仅包含正常图像。然而，仅依赖正常图像可能会导致学习到无效的低维图像表示，这些表示对检测和分割大小、外观和形状各异的未见异常病变不够敏感。基于计算机视觉技术，使用自监督学习对UAD方法进行预训练可以缓解这一挑战，但它们并非最优，因为它们没有探索领域知识来设计前置任务，并且它们的对比学习损失没有尝试对正常训练图像进行聚类，这可能导致正常图像的稀疏分布，对异常检测无效。在本文中，我们提出了一种用于MIA UAD应用的新的自监督预训练方法，名为通过对比学习的伪多类强增强（PMSACL）。PMSACL由一种新颖的优化方法组成，该方法将正常图像类与合成异常图像的多个伪类进行对比，每个类在特征空间中强制形成一个密集簇。在实验中，我们表明，我们的PMSACL预训练提高了许多使用结肠镜检查、眼底筛查和新冠病毒胸部X光数据集的MIA基准测试中SOTA UAD方法的准确性。
