基于期望最大化的伪标注算法在有限注释分割中的应用【英文版】

英文标题：Expectation Maximization Pseudo Labelling for Segmentation with Limited Annotations中文摘要：该研究探讨了伪标记在半监督医学图像分割中的应用，提出了 Bayesian Pseudo Labels 这一伪标记的全面推广方法，并展示了该方法在三维肺血管、MRI 脑肿瘤和三维 MRI 脑肿瘤二元分割方面的应用。英文摘要：We study pseudo labelling and its generalisation for semi-supervisedsegmentation of medical images. Pseudo labelling has achieved great empiricalsuccesses in semi-supervised learning, by utilising raw inferences onunlabelled data as pseudo labels for self-training. In our paper, we build aconne