用于医学图像分割的去噪监督
Supervision by Denoising for Medical Image Segmentation
AI总结：
针对医学图像标注稀缺且变异性大的问题，提出半监督框架“去噪监督”（SUD），利用模型自身去噪输出作为软标签，在时空去噪框架下交替去噪与权重更新，显著提升了3D脑部重建与2D皮质分割的重建效果。
摘要：
基于学习的图像重建模型（例如基于U-Net的模型）若要保证良好的泛化能力，需要大量带标签的图像。然而，在某些成像领域中，由于获取成本高昂，具有像素或体素级标签精度的带标签数据十分稀缺。在医学成像等领域，这一问题进一步加剧，因为不存在单一的基准真值标签，导致标签中存在大量的重复变异性。因此，通过同时从带标签和无标签样本中学习（称为半监督学习）来训练重建网络以获得更好的泛化能力，是一个具有实际和理论意义的问题。然而，传统的用于图像重建的半监督学习方法通常需要针对特定的成像问题手工设计一个可微的正则化器，这可能极其耗时。在这项工作中，我们提出了“去噪监督”（SUD），这是一个使我们能够利用模型自身的去噪输出作为软标签来监督重建模型的框架。SUD在时空去噪框架下统一了随机平均和空间去噪技术，并在半监督的优化框架中交替进行去噪和模型权重更新步骤。作为示例应用，我们将SUD应用于生物医学成像中的两个问题——解剖学脑部重建（3D）和皮质分割（2D）——以证明与仅监督和随机平均基线相比，图像重建获得了显著改善。
英文摘要：
Learning-based image reconstruction models, such as those based on the U-Net, require a large set of labeled images if good generalization is to be guaranteed. In some imaging domains, however, labeled data with pixel- or voxel-level label accuracy are scarce due to the cost of acquiring them. This problem is exacerbated further in domains like medical imaging, where there is no single ground truth label, resulting in large amounts of repeat variability in the labels. Therefore, training reconstruction networks to generalize better by learning from both labeled and unlabeled examples (called semi-supervised learning) is problem of practical and theoretical interest. However, traditional semi-supervised learning methods for image reconstruction often necessitate handcrafting a differentiable regularizer specific to some given imaging problem, which can be extremely time-consuming. In this work, we propose "supervision by denoising" (SUD), a framework that enables us to supervise reconstruction models using their own denoised output as soft labels. SUD unifies stochastic averaging and spatial denoising techniques under a spatio-temporal denoising framework and alternates denoising and model weight update steps in an optimization framework for semi-supervision. As example applications, we apply SUD to two problems arising from biomedical imaging -- anatomical brain reconstruction (3D) and cortical parcellation (2D) -- to demonstrate a significant improvement in the image reconstructions over supervised-only and stochastic averaging baselines.
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。
