AI & ML Breaks Assumption

Exposes a major flaw in medical super-resolution research where models trained on downsampled data fail to recover actual lost structures in real low-resolution scans.

March 25, 2026

Original Paper

VoDaSuRe: A Large-Scale Dataset Revealing Domain Shift in Volumetric Super-Resolution

August Leander Høeg, Sophia Wiinberg Bardenfleth, Hans Martin Kjer, Tim Bjørn Dyrby, Vedrana Andersen Dahl, Anders Bjorholm Dahl

arXiv · 2603.23153

The Takeaway

The paper challenges the validity of current SOTA benchmarks in volumetric super-resolution. It provides the VoDaSuRe dataset of paired real scans, forcing a shift in how practitioners evaluate and train models for clinical applications.

From the abstract

Recent advances in volumetric super-resolution (SR) have demonstrated strong performance in medical and scientific imaging, with transformer- and CNN-based approaches achieving impressive results even at extreme scaling factors. In this work, we show that much of this performance stems from training on downsampled data rather than real low-resolution scans. This reliance on downsampling is partly driven by the scarcity of paired high- and low-resolution 3D datasets. To address this, we introduce