Exploring the Feasibility of Zero-Shot Super-Resolution in Preclinical Imaging.

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× and .5×) 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.