Scale-Cascaded Diffusion Models for Super-Resolution in Medical Imaging
This work addresses super-resolution in medical imaging, offering a method that is incremental by unifying multiscale reconstruction with diffusion priors.
The paper tackles the problem of medical image super-resolution by proposing a scale-cascaded diffusion model that decomposes images into Laplacian pyramid scales and trains separate diffusion priors for each frequency band, resulting in improved perceptual quality and reduced inference time on brain, knee, and prostate MRI data.
Diffusion models have been increasingly used as strong generative priors for solving inverse problems such as super-resolution in medical imaging. However, these approaches typically utilize a diffusion prior trained at a single scale, ignoring the hierarchical scale structure of image data. In this work, we propose to decompose images into Laplacian pyramid scales and train separate diffusion priors for each frequency band. We then develop an algorithm to perform super-resolution that utilizes these priors to progressively refine reconstructions across different scales. Evaluated on brain, knee, and prostate MRI data, our approach both improves perceptual quality over baselines and reduces inference time through smaller coarse-scale networks. Our framework unifies multiscale reconstruction and diffusion priors for medical image super-resolution.