发表机构
Imperial College London; The Royal Marsden Hospital; Imperial College Healthcare NHS Trust（帝国理工学院; 皇家马斯登医院; 帝国理工医疗国民保健信托基金）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
提出潜空间到潜空间的流匹配方法，利用图像和标签的编码表示进行医学体积随机分割，在放疗计划勾画和多器官分割中效率提升达14倍且保持临床性能。
AI 中文摘要
医学图像分割中由观察者间变异性引起的不确定性在制定治疗计划中起着重要作用。该领域的研究因大规模医学数据集缺乏多重标注而受到抑制，尤其是体积数据，其还面临额外的缩放和计算复杂性挑战。流匹配已成为生成建模的强大框架，并已被证明在处理图像的潜表示时能保持强劲性能。在这项工作中，我们提出了一种潜空间到潜空间流技术，通过图像和标签空间的编码表示对医学体积进行随机分割。我们在两个具有挑战性的应用上评估了我们的方法，涵盖放射治疗计划中的勾画不确定性和多器官结构分割，与全分辨率模型相比，效率提升高达14倍，同时保持临床相关性能。
英文摘要
Uncertainty arising from inter-observer variability in medical image segmentation plays an important role in developing treatment plans. Research in this area is inhibited by the lack of multiple annotations for large-scale medical datasets, especially for volumetric data, which suffers from additional scaling and computational complexity challenges. Flow matching has emerged as a powerful framework for generative modelling and has also been demonstrated to maintain strong performance when working with latent representations of images. In this work, we introduce a latent-to-latent flow technique for stochastic segmentation of medical volumes via encoded representations of both the image and label space. We evaluate our method on two challenging applications covering delineation uncertainty for radiotherapy planning and multiple organ structure segmentation, improving efficiency up to 14x compared with full resolution models while maintaining clinically relevant performance.
CommentsAccepted at MICCAI 2026
