Scale-Cascaded Diffusion Models for Super-Resolution in Medical Imaging

TL;DR

This paper introduces a multiscale diffusion model framework for medical image super-resolution, leveraging Laplacian pyramids and separate priors to enhance quality and efficiency across MRI data.

Contribution

It proposes a novel multiscale diffusion approach using Laplacian pyramids, improving super-resolution quality and inference speed in medical imaging.

Findings

Enhanced perceptual quality over baseline methods

Reduced inference time with smaller coarse-scale networks

Unified multiscale reconstruction and diffusion priors

Abstract

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.