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

2026-03-24

AI Summary

Existing voxel-based super-resolution methods are typically trained on synthetic low-resolution data generated via downscaling, lacking authentic high–low resolution pairs, which leads to an overestimation of their performance on real-world low-resolution scans. To address this gap, this work introduces VoDaSuRe, the first large-scale dataset of genuinely paired 3D volumetric data spanning both medical and scientific imaging domains. Leveraging this dataset, we conduct a systematic evaluation of prominent CNN- and Transformer-based super-resolution models, revealing substantial domain shift when applied to real data: these models often produce overly smoothed outputs or structural distortions and struggle to recover genuinely lost fine details. Our findings challenge the prevailing assumptions about the efficacy of current approaches and underscore the necessity for future research to advance on the basis of realistic, complex data.

Technology Category

• Computer Vision: 3D Computer Vision
• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Knowledge Representation and Reasoning: Diagnosis and Abductive Reasoning

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Web Mining and Content Analysis: Large pretrained models with web data
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

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 VoDaSuRe, a large-scale volumetric dataset containing paired high- and low-resolution scans. When training models on VoDaSuRe, we reveal a significant discrepancy: SR models trained on downsampled data produce substantially sharper predictions than those trained on real low-resolution scans, which smooth fine structures. Conversely, applying models trained on downsampled data to real scans preserves more structure but is inaccurate. Our findings suggest that current SR methods are overstated - when applied to real data, they do not recover structures lost in low-resolution scans and instead predict a smoothed average. We argue that progress in deep learning-based volumetric SR requires datasets with paired real scans of high complexity, such as VoDaSuRe. Our dataset and code are publicly available through: https://augusthoeg.github.io/VoDaSuRe/

Problem

• volumetric super-resolution
• domain shift
• real low-resolution scans
• paired 3D dataset
• downsampling bias

Innovation

• volumetric super-resolution
• domain shift
• real low-resolution data
• paired 3D dataset
• VoDaSuRe