Regression is all you need for medical image translation
Authors:
Sebastian Rassmann,
David Kügler,
Christian Ewert,
Martin Reuter
Abstract:
The acquisition of information-rich images within a limited time budget is crucial in medical imaging. Medical image translation (MIT) can help enhance and supplement existing datasets by generating synthetic images from acquired data. While Generative Adversarial Nets (GANs) and Diffusion Models (DMs) have achieved remarkable success in natural image generation, their benefits - creativity and im…
▽ More
The acquisition of information-rich images within a limited time budget is crucial in medical imaging. Medical image translation (MIT) can help enhance and supplement existing datasets by generating synthetic images from acquired data. While Generative Adversarial Nets (GANs) and Diffusion Models (DMs) have achieved remarkable success in natural image generation, their benefits - creativity and image realism - do not necessarily transfer to medical applications where highly accurate anatomical information is required. In fact, the imitation of acquisition noise or content hallucination hinder clinical utility. Here, we introduce YODA (You Only Denoise once - or Average), a novel 2.5D diffusion-based framework for volumetric MIT. YODA unites diffusion and regression paradigms to produce realistic or noise-free outputs. Furthermore, we propose Expectation-Approximation (ExpA) DM sampling, which draws inspiration from MRI signal averaging. ExpA-sampling suppresses generated noise and, thus, eliminates noise from biasing the evaluation of image quality. Through extensive experiments on four diverse multi-modal datasets - comprising multi-contrast brain MRI and pelvic MRI-CT - we show that diffusion and regression sampling yield similar results in practice. As such, the computational overhead of diffusion sampling does not provide systematic benefits in medical information translation. Building on these insights, we demonstrate that YODA outperforms several state-of-the-art GAN and DM methods. Notably, YODA-generated images are shown to be interchangeable with, or even superior to, physical acquisitions for several downstream tasks. Our findings challenge the presumed advantages of DMs in MIT and pave the way for the practical application of MIT in medical imaging.
△ Less
Submitted 6 May, 2025; v1 submitted 4 May, 2025;
originally announced May 2025.
Learning Anatomical Segmentations for Tractography from Diffusion MRI
Authors:
Christian Ewert,
David Kügler,
Anastasia Yendiki,
Martin Reuter
Abstract:
Deep learning approaches for diffusion MRI have so far focused primarily on voxel-based segmentation of lesions or white-matter fiber tracts. A drawback of representing tracts as volumetric labels, rather than sets of streamlines, is that it precludes point-wise analyses of microstructural or geometric features along a tract. Traditional tractography pipelines, which do allow such analyses, can be…
▽ More
Deep learning approaches for diffusion MRI have so far focused primarily on voxel-based segmentation of lesions or white-matter fiber tracts. A drawback of representing tracts as volumetric labels, rather than sets of streamlines, is that it precludes point-wise analyses of microstructural or geometric features along a tract. Traditional tractography pipelines, which do allow such analyses, can benefit from detailed whole-brain segmentations to guide tract reconstruction. Here, we introduce fast, deep learning-based segmentation of 170 anatomical regions directly on diffusion-weighted MR images, removing the dependency of conventional segmentation methods on T 1-weighted images and slow pre-processing pipelines. Working natively in diffusion space avoids non-linear distortions and registration errors across modalities, as well as interpolation artifacts. We demonstrate consistent segmentation results between 0 .70 and 0 .87 Dice depending on the tissue type. We investigate various combinations of diffusion-derived inputs and show generalization across different numbers of gradient directions. Finally, integrating our approach to provide anatomical priors for tractography pipelines, such as TRACULA, removes hours of pre-processing time and permits processing even in the absence of high-quality T 1-weighted scans, without degrading the quality of the resulting tract estimates.
△ Less
Submitted 9 September, 2020;
originally announced September 2020.
