Physics-Guided Diffusion Priors for Multi-Slice Reconstruction in Scientific Imaging
This addresses the computational and memory challenges in accelerating acquisition for medical and scientific imaging, though it appears incremental as a hybrid approach.
The paper tackles the problem of multi-slice reconstruction from limited measurement data in scientific imaging by integrating partitioned diffusion priors with physics-based constraints, resulting in reduced memory usage per GPU while preserving high reconstruction quality and outperforming baselines in MRI and 4D-STEM modalities.
Accurate multi-slice reconstruction from limited measurement data is crucial to speed up the acquisition process in medical and scientific imaging. However, it remains challenging due to the ill-posed nature of the problem and the high computational and memory demands. We propose a framework that addresses these challenges by integrating partitioned diffusion priors with physics-based constraints. By doing so, we substantially reduce memory usage per GPU while preserving high reconstruction quality, outperforming both physics-only and full multi-slice reconstruction baselines for different modalities, namely Magnetic Resonance Imaging (MRI) and four-dimensional Scanning Transmission Electron Microscopy (4D-STEM). Additionally, we show that the proposed method improves in-distribution accuracy as well as strong generalization to out-of-distribution datasets.