Abstract
In volume-to-volume translations in medical images, existing models often struggle to capture the inherent volumetric distribution using 3D voxelspace representations, due to high computational dataset demands. We present Score-Fusion, a novel volumetric translation model that effectively learns 3D representations by ensembling perpendicularly trained 2D diffusion models in score function space. By carefully initializing our model to start with an average of 2D models as in existing models, we reduce 3D training to a fine-tuning process, mitigating computational and data demands. Furthermore, we explicitly design the 3D model’s hierarchical layers to learn ensembles of 2D features, further enhancing efficiency and performance. Moreover, Score-Fusion naturally extends to multi-modality settings by fusing diffusion models conditioned on different inputs for flexible, accurate integration. We demonstrate that 3D representation is essential for better performance in downstream recognition tasks, such as tumor segmentation, where most segmentation models are based on 3D representation. Extensive experiments demonstrate that Score-Fusion achieves superior accuracy and volumetric fidelity in 3D medical image super-resolution and modality translation. Additionally, we extend Score-Fusion to video superresolution by integrating 2D diffusion models on time-space slices with a spatial-temporal video diffusion backbone, highlighting its potential for general-purpose volume translation and providing broader insight into learning-based approaches for score function fusion.
| Original language | English (US) |
|---|---|
| Pages (from-to) | 79868-79902 |
| Number of pages | 35 |
| Journal | Proceedings of Machine Learning Research |
| Volume | 267 |
| State | Published - 2025 |
| Event | 42nd International Conference on Machine Learning, ICML 2025 - Vancouver, Canada Duration: Jul 13 2025 → Jul 19 2025 |
ASJC Scopus subject areas
- Software
- Control and Systems Engineering
- Statistics and Probability
- Artificial Intelligence
Fingerprint
Dive into the research topics of 'Introducing 3D Representation for Dense Volume-to-Volume Translation via Score Fusion'. Together they form a unique fingerprint.Cite this
- APA
- Standard
- Harvard
- Vancouver
- Author
- BIBTEX
- RIS