Fast forward: Rephrasing 3D deformable image registration through density alignment and splatting

Mattias P Heinrich, Alexander Bigalke, Lasse Hansen
Proceedings of The 8th International Conference on Medical Imaging with Deep Learning, PMLR 301:639-652, 2026.

Abstract

Unsupervised learning- and optimisation-based 3D registration has almost exclusively been approached using backward warping (interpolation) for transforming images. While this has practical advantages in particular the ease of implementation within common libraries it limits the robustness and accuracy in certain challenging scenarios. The alternative solution of forward splatting (extrapolation) is currently limited to very few applications, e.g. mesh or point cloud registration, requiring specific geometric learning architectures that are so far less efficient compared to dense 3D convolutional networks. In this work, we propose to use a straightforward forward splatting technique based on differentiable rasterisation. Contrary to prior work, we rephrase the problem of deformable image registration as a density alignment of rasterised volumes based on intermediate point cloud representations that can be automatically obtained through e.g. geometric vessel filters or surface segmentations. Our experimental validation demonstrates state-of-the-art performance over a wide range of registration tasks including intra- and inter-patient alignment of thorax and abdomen.

Cite this Paper

BibTeX
@InProceedings{pmlr-v301-heinrich26a, title = {Fast forward: Rephrasing 3D deformable image registration through density alignment and splatting}, author = {Heinrich, Mattias P and Bigalke, Alexander and Hansen, Lasse}, booktitle = {Proceedings of The 8th International Conference on Medical Imaging with Deep Learning}, pages = {639--652}, year = {2026}, editor = {Tasdizen, Tolga and Elhabian, Shireen and Summers, Ronald and Chen, Chen and Koch, Lisa and Zhuang, Yan}, volume = {301}, series = {Proceedings of Machine Learning Research}, month = {09--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v301/main/assets/heinrich26a/heinrich26a.pdf}, url = {https://proceedings.mlr.press/v301/heinrich26a.html}, abstract = {Unsupervised learning- and optimisation-based 3D registration has almost exclusively been approached using backward warping (interpolation) for transforming images. While this has practical advantages in particular the ease of implementation within common libraries it limits the robustness and accuracy in certain challenging scenarios. The alternative solution of forward splatting (extrapolation) is currently limited to very few applications, e.g. mesh or point cloud registration, requiring specific geometric learning architectures that are so far less efficient compared to dense 3D convolutional networks. In this work, we propose to use a straightforward forward splatting technique based on differentiable rasterisation. Contrary to prior work, we rephrase the problem of deformable image registration as a density alignment of rasterised volumes based on intermediate point cloud representations that can be automatically obtained through e.g. geometric vessel filters or surface segmentations. Our experimental validation demonstrates state-of-the-art performance over a wide range of registration tasks including intra- and inter-patient alignment of thorax and abdomen.} }
Endnote
%0 Conference Paper %T Fast forward: Rephrasing 3D deformable image registration through density alignment and splatting %A Mattias P Heinrich %A Alexander Bigalke %A Lasse Hansen %B Proceedings of The 8th International Conference on Medical Imaging with Deep Learning %C Proceedings of Machine Learning Research %D 2026 %E Tolga Tasdizen %E Shireen Elhabian %E Ronald Summers %E Chen Chen %E Lisa Koch %E Yan Zhuang %F pmlr-v301-heinrich26a %I PMLR %P 639--652 %U https://proceedings.mlr.press/v301/heinrich26a.html %V 301 %X Unsupervised learning- and optimisation-based 3D registration has almost exclusively been approached using backward warping (interpolation) for transforming images. While this has practical advantages in particular the ease of implementation within common libraries it limits the robustness and accuracy in certain challenging scenarios. The alternative solution of forward splatting (extrapolation) is currently limited to very few applications, e.g. mesh or point cloud registration, requiring specific geometric learning architectures that are so far less efficient compared to dense 3D convolutional networks. In this work, we propose to use a straightforward forward splatting technique based on differentiable rasterisation. Contrary to prior work, we rephrase the problem of deformable image registration as a density alignment of rasterised volumes based on intermediate point cloud representations that can be automatically obtained through e.g. geometric vessel filters or surface segmentations. Our experimental validation demonstrates state-of-the-art performance over a wide range of registration tasks including intra- and inter-patient alignment of thorax and abdomen.
APA
Heinrich, M.P., Bigalke, A. & Hansen, L.. (2026). Fast forward: Rephrasing 3D deformable image registration through density alignment and splatting. Proceedings of The 8th International Conference on Medical Imaging with Deep Learning, in Proceedings of Machine Learning Research 301:639-652 Available from https://proceedings.mlr.press/v301/heinrich26a.html.

Related Material