Exploring the Feasibility of Zero-Shot Super-Resolution in Preclinical Imaging
- 20 September 2025
- book chapter
- Published by Springer Nature
- Vol. 15961, 186-196
Abstract
Preclinical imaging studies are vital to the research, development, and evaluation of new medical therapies. Images acquired during these studies often have high in-plane resolution but low through-plane resolution, resulting in highly anisotropic volumes that hamper downstream volumetric analysis. Additionally, since there are no image acquisition standards across studies, training data for conventional supervised super-resolution (SR) methods is limited. In this work, we compare two SR methods that do not require additional training data. The first is ECLARE, a self-SR approach that creates its own training data from in-plane patches drawn from the anisotropic volume. The second is Biplanar Denoising diffusion null space model (DDNM) Averaging (BiDA), a proposed method leveraging two independently pre-trained denoising diffusion probabilistic models and the DDNM posterior sampling technique. We evaluate both methods first on rat data at two scale factors (\(2.5\times \) and \(5\times \)) and compare signal recovery and downstream task performance. We further evaluate these methods on a different species (mice) to measure their generalizability. Both methods experimentally resulted in good signal recovery performance, but only the images super-resolved by BiDA were accurately skullstripped downstream. Although both methods performed well on the in-domain rat data, BiDA did not fully generalize to the out-of-domain mouse data.
Keywords
This publication has 27 references indexed in Scilit:
- TS-SR3: Time-Strided Denoising Diffusion Probabilistic Model for MR Super-Resolution2024
- Slice-Consistent 3D Volumetric Brain CT-to-MRI Translation with 2D Brownian Bridge Diffusion Model2024
- Self-Supervised Super-Resolution for Anisotropic MR Images with and Without Slice Gap2023
- Deep Filter Bank Regression for Super-Resolution of Anisotropic MR Brain Images2022
- SRFlow: Learning the Super-Resolution Space with Normalizing Flow2020
- Efficient and Accurate MRI Super-Resolution Using a Generative Adversarial Network and 3D Multi-level Densely Connected Network2018
- Self Super-Resolution for Magnetic Resonance Images2016
- Small Animal Imaging2011
- Magnetic resonance imaging and neurosphere therapy of stroke in ratAnnals of Neurology, 2003
- The synthesis of pulse sequences yielding arbitrary magnetization vectorsMagnetic Resonance in Medicine, 1989