Memory-Efficient 3D Denoising Diffusion Models for Medical Image Processing

Florentin Bieder, Julia Wolleb, Alicia Durrer, Robin Sandkuehler, Philippe C. Cattin
Medical Imaging with Deep Learning, PMLR 227:552-567, 2024.

Abstract

Denoising diffusion models have recently achieved state-of-the-art performance in many image-generation tasks. They do, however, require a large amount of computational resources. This limits their application to medical tasks, where we often deal with large 3D volumes, like high-resolution three-dimensional data. In this work, we present a number of different ways to reduce the resource consumption for 3D diffusion models and apply them to a dataset of 3D images. The main contribution of this paper is the memory-efficient patch-based diffusion model PatchDDM, which can be applied to the total volume during inference while the training is performed only on patches. Without limiting the application of the proposed diffusion model for image generation, we evaluate the method on the tumor segmentation task of the BraTS2020 dataset and demonstrate that we can generate meaningful three-dimensional segmentations.

Cite this Paper

BibTeX
@InProceedings{pmlr-v227-bieder24a, title = {Memory-Efficient 3D Denoising Diffusion Models for Medical Image Processing}, author = {Bieder, Florentin and Wolleb, Julia and Durrer, Alicia and Sandkuehler, Robin and Cattin, Philippe C.}, booktitle = {Medical Imaging with Deep Learning}, pages = {552--567}, year = {2024}, editor = {Oguz, Ipek and Noble, Jack and Li, Xiaoxiao and Styner, Martin and Baumgartner, Christian and Rusu, Mirabela and Heinmann, Tobias and Kontos, Despina and Landman, Bennett and Dawant, Benoit}, volume = {227}, series = {Proceedings of Machine Learning Research}, month = {10--12 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v227/bieder24a/bieder24a.pdf}, url = {https://proceedings.mlr.press/v227/bieder24a.html}, abstract = {Denoising diffusion models have recently achieved state-of-the-art performance in many image-generation tasks. They do, however, require a large amount of computational resources. This limits their application to medical tasks, where we often deal with large 3D volumes, like high-resolution three-dimensional data. In this work, we present a number of different ways to reduce the resource consumption for 3D diffusion models and apply them to a dataset of 3D images. The main contribution of this paper is the memory-efficient patch-based diffusion model PatchDDM, which can be applied to the total volume during inference while the training is performed only on patches. Without limiting the application of the proposed diffusion model for image generation, we evaluate the method on the tumor segmentation task of the BraTS2020 dataset and demonstrate that we can generate meaningful three-dimensional segmentations.} }
Endnote
%0 Conference Paper %T Memory-Efficient 3D Denoising Diffusion Models for Medical Image Processing %A Florentin Bieder %A Julia Wolleb %A Alicia Durrer %A Robin Sandkuehler %A Philippe C. Cattin %B Medical Imaging with Deep Learning %C Proceedings of Machine Learning Research %D 2024 %E Ipek Oguz %E Jack Noble %E Xiaoxiao Li %E Martin Styner %E Christian Baumgartner %E Mirabela Rusu %E Tobias Heinmann %E Despina Kontos %E Bennett Landman %E Benoit Dawant %F pmlr-v227-bieder24a %I PMLR %P 552--567 %U https://proceedings.mlr.press/v227/bieder24a.html %V 227 %X Denoising diffusion models have recently achieved state-of-the-art performance in many image-generation tasks. They do, however, require a large amount of computational resources. This limits their application to medical tasks, where we often deal with large 3D volumes, like high-resolution three-dimensional data. In this work, we present a number of different ways to reduce the resource consumption for 3D diffusion models and apply them to a dataset of 3D images. The main contribution of this paper is the memory-efficient patch-based diffusion model PatchDDM, which can be applied to the total volume during inference while the training is performed only on patches. Without limiting the application of the proposed diffusion model for image generation, we evaluate the method on the tumor segmentation task of the BraTS2020 dataset and demonstrate that we can generate meaningful three-dimensional segmentations.
APA
Bieder, F., Wolleb, J., Durrer, A., Sandkuehler, R. & Cattin, P.C.. (2024). Memory-Efficient 3D Denoising Diffusion Models for Medical Image Processing. Medical Imaging with Deep Learning, in Proceedings of Machine Learning Research 227:552-567 Available from https://proceedings.mlr.press/v227/bieder24a.html.

Related Material