arXiv 2408.117332024-08-21cs.CV

通过组合性增强跨模态医学图像分割

Enhancing Cross-Modal Medical Image Segmentation through Compositionality

University of Amsterdam（阿姆斯特丹大学）；The Netherlands Cancer Institute（荷兰癌症研究所）；Utrecht University（乌得勒支大学）；Maastricht University Medical Center（马斯特里赫特大学医学中心）

Aniek Eijpe, Valentina Corbetta, Kalina Chupetlovska, Regina Beets-Tan, Wilson Silva

备注：: 11 pages, 3 figures, 2 tables. Accepted at Deep Generative Models workshop @ MICCAI 2024 (DGM4MICCAI). This is the submitted manuscript with added link to github repo, funding acknowledgements and authors' names and affiliations. No further post submission improvements or corrections were integrated. Final version not published yet

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

本文提出一种端到端跨模态医学图像分割框架，通过引入可学习的 von Mises-Fisher 核实现组合性归纳偏置与内容-风格解耦，有效提升了分割性能、可解释性并降低计算成本。

摘要：

跨模态医学图像分割面临重大挑战，因为不同成像模态产生的图像在分辨率、对比度以及解剖结构外观上存在差异。我们在跨模态分割网络中引入组合性作为归纳偏置，以提升分割性能和可解释性，同时降低复杂度。所提出的网络是一个端到端的跨模态分割框架，它使用可学习的 von Mises-Fisher 核对学习到的表示施加组合性约束。这些核促进了学习表示中的内容-风格解耦，生成的组合内容表示具有内在可解释性，并能有效解耦不同的解剖结构。实验结果表明，在多个医学数据集上分割性能得到提升且计算成本降低。此外，我们展示了所学组合特征的可解释性。代码和检查点将公开于：https://github.com/Trustworthy-AI-UU-NKI/Cross-Modal-Segmentation。

英文摘要：

Cross-modal medical image segmentation presents a significant challenge, as different imaging modalities produce images with varying resolutions, contrasts, and appearances of anatomical structures. We introduce compositionality as an inductive bias in a cross-modal segmentation network to improve segmentation performance and interpretability while reducing complexity. The proposed network is an end-to-end cross-modal segmentation framework that enforces compositionality on the learned representations using learnable von Mises-Fisher kernels. These kernels facilitate content-style disentanglement in the learned representations, resulting in compositional content representations that are inherently interpretable and effectively disentangle different anatomical structures. The experimental results demonstrate enhanced segmentation performance and reduced computational costs on multiple medical datasets. Additionally, we demonstrate the interpretability of the learned compositional features. Code and checkpoints will be publicly available at: https://github.com/Trustworthy-AI-UU-NKI/Cross-Modal-Segmentation.

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。