arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2403.16776eess.IVcs.CVcs.LG
收藏

Diff-Def:用于条件图谱的扩散生成形变场

Diff-Def: Diffusion-Generated Deformation Fields for Conditional Atlases

School of Computation, Information and Technology and the School of Medicine and Health, TUM Klinikum, Technical University of Munich(计算信息学院和医学健康学院,慕尼黑技术大学克林克医院,慕尼黑技术大学);Munich Center for Machine Learning (MCML), Munich, Germany(慕尼黑机器学习中心(MCML),德国慕尼黑);Department of Computing, Imperial College London, UK(计算学院,伦敦帝国学院,英国);FAU Erlangen-Nürnberg, Germany(埃尔兰根-纽伦堡大学,德国)

Sophie Starck, Vasiliki Sideri-Lampretsa, Bernhard Kainz, Martin J. Menten, Tamara T. Mueller, Daniel Rueckert

AI总结:

本文提出Diff-Def,利用潜在扩散模型生成形变场而非直接合成图像,将一般人群图谱转化为特定亚群条件图谱,在UK Biobank脑MR图像上优于现有方法,确保结构完整性和高解剖保真度。

摘要:

解剖图谱广泛用于群体研究和分析。条件图谱针对由特定条件(如人口统计学特征或病理)定义的特定亚群,并允许研究细粒度的解剖差异,例如与衰老或疾病相关的形态变化。现有方法要么使用基于配准的方法,这些方法通常难以处理大的解剖变异;要么使用生成对抗网络,这些网络由于可能遭受训练不稳定性而难以训练。我们不直接以强度形式生成图谱,而是提出使用潜在扩散模型来生成形变场,该形变场将一般群体图谱转换为代表特定亚群的图谱。我们的方法通过生成该形变场并使用图像邻域对其进行正则化,确保了结构完整性,增强了可解释性,并避免了直接图像合成中可能出现的幻觉。我们使用来自英国生物银行的脑部MR图像,将我们的方法与几种最先进的图谱生成方法进行了比较。我们的方法生成了具有平滑变换和高解剖保真度的高度逼真图谱,优于现有基线。我们通过全面评估证明了这些图谱的质量,包括解剖准确性、感知相似性的定量指标,以及显示生成图谱一致性和真实性的定性分析。

英文摘要:

Anatomical atlases are widely used for population studies and analysis. Conditional atlases target a specific sub-population defined via certain conditions, such as demographics or pathologies, and allow for the investigation of fine-grained anatomical differences like morphological changes associated with ageing or disease. Existing approaches use either registration-based methods that are often unable to handle large anatomical variations or generative adversarial models, which are challenging to train since they can suffer from training instabilities. Instead of generating atlases directly in as intensities, we propose using latent diffusion models to generate deformation fields, which transform a general population atlas into one representing a specific sub-population. Our approach ensures structural integrity, enhances interpretability and avoids hallucinations that may arise during direct image synthesis by generating this deformation field and regularising it using a neighbourhood of images. We compare our method to several state-of-the-art atlas generation methods using brain MR images from the UK Biobank. Our method generates highly realistic atlases with smooth transformations and high anatomical fidelity, outperforming existing baselines. We demonstrate the quality of these atlases through comprehensive evaluations, including quantitative metrics for anatomical accuracy, perceptual similarity, and qualitative analyses displaying the consistency and realism of the generated atlases.

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。

↑