mlsanityData to 14 Oct 2026

Resolution-Robust 3D MRI Reconstruction with 2D Diffusion Priors: Diverse-Resolution Training Outperforms Interpolation

arXiv 2412.18584 ↗First posted 24 Dec 2024cs.CV cs.LG eess.IVcited by 1

On the map: Diffusion models → Medical imaging and segmentation

Abstract (arXiv)

Deep learning-based 3D imaging, in particular magnetic resonance imaging (MRI), is challenging because of limited availability of 3D training data. Therefore, 2D diffusion models trained on 2D slices are starting to be leveraged for 3D MRI reconstruction. However, as we show in this paper, existing methods pertain to a fixed voxel size, and performance degrades when the voxel size is varied, as it is often the case in clinical practice. In this paper, we propose and study several approaches for resolution-robust 3D MRI reconstruction with 2D diffusion priors. As a result of this investigation, we obtain a simple resolution-robust variational 3D reconstruction approach based on diffusion-guided regularization of randomly sampled 2D slices. This method provides competitive reconstruction quality compared to posterior sampling baselines. Towards resolving the sensitivity to resolution-shifts, we investigate state-of-the-art model-based approaches including Gaussian splatting, neural representations, and infinite-dimensional diffusion models, as well as a simple data-centric approach of training the diffusion model on several resolutions. Our experiments demonstrate that the model-based approaches fail to close the performance gap in 3D MRI. In contrast, the data-centric approach of training the diffusion model on various resolutions effectively provides a resolution-robust method without compromising accuracy.

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Placed in Medical imaging and segmentation, one of 14 directions in Diffusion models. Builds on 27 of its 38 references that are in the map. First posted on arXiv 24 Dec 2024.

Builds on

27 of 38 references.

Closest ancestors · 6

Same direction, newest first

  1. 2024INFusion: Diffusion Regularized Implicit Neural Representations for 2D and 3D accelerated MRI reconstruction19 Jun
  2. 2024DiffusionBlend: Learning 3D Image Prior through Position-aware Diffusion Score Blending for 3D Computed Tomography Reconstruction14 Jun
  3. 2023Improving 3D Imaging with Pre-Trained Perpendicular 2D Diffusion Models15 Mar
  4. 2022Solving 3D Inverse Problems using Pre-trained 2D Diffusion Models19 Nov
  5. 2021Solving Inverse Problems in Medical Imaging with Score-Based Generative Models15 Nov
  6. 2021Score-based diffusion models for accelerated MRI8 Oct

Other references · 21

  1. 2020Denoising Diffusion Probabilistic Modelscited by 21,076
  2. 2021Diffusion Models Beat GANs on Image Synthesiscited by 8,255
  3. 2015Deep Unsupervised Learning using Nonequilibrium Thermodynamicscited by 7,058
  4. 20233D Gaussian Splatting for Real-Time Radiance Field Renderingcited by 6,226
  5. 2021Improved Denoising Diffusion Probabilistic Modelscited by 3,761
  6. 2022DreamFusion: Text-to-3D using 2D Diffusioncited by 3,025
  7. 2020Fourier Neural Operator for Parametric Partial Differential Equationscited by 2,908
  8. 2020Fourier Features Let Networks Learn High Frequency Functions in Low Dimensional Domainscited by 2,200
  9. 2022Diffusion Posterior Sampling for General Noisy Inverse Problemscited by 1,416
  10. 2018fastMRI: An Open Dataset and Benchmarks for Accelerated MRIcited by 470
  11. 2023ReconFusion: 3D Reconstruction with Diffusion Priorscited by 318
  12. 2023A Variational Perspective on Solving Inverse Problems with Diffusion Modelscited by 222
9 more
  1. 2023Decomposed Diffusion Sampler for Accelerating Large-Scale Inverse Problemscited by 132
  2. 2021Generative Models as Distributions of Functionscited by 89
  3. 2021Measuring Robustness in Deep Learning Based Compressive Sensingcited by 53
  4. 2022Test-Time Training Can Close the Natural Distribution Shift Performance Gap in Deep Learning Based Compressed Sensingcited by 37
  5. 2023Multilevel Diffusion: Infinite Dimensional Score-Based Diffusion Models for Image Generationcited by 37
  6. 2023∞-Diff: Infinite Resolution Diffusion with Subsampled Mollified Statescited by 21
  7. 2023Robustness of Deep Learning for Accelerated MRI: Benefits of Diverse Training Datacited by 10
  8. 2023FastMRI Prostate: A Publicly Available, Biparametric MRI Dataset to Advance Machine Learning for Prostate Cancer Imagingcited by 9
  9. 2023Resolution-Invariant Image Classification based on Fourier Neural Operatorscited by 7

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026Piecewise Dynamic Diffusion Regularization for Reconstruction of Cardiac Cine MRIcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Improving 2D Diffusion Models for 3D Medical Imaging with Inter-Slice Consistent Stochasticity
  2. 2023Robust Physics-based Deep MRI Reconstruction Via Diffusion Purification
  3. 2025Conditional Denoising Diffusion Model-Based Robust MR Image Reconstruction from Highly Undersampled Data
  4. 2026Comparative Analysis of 3D Convolutional and 2.5D Slice-Conditioned U-Net Architectures for MRI Super-Resolution via Elucidated Diffusion Models
  5. 2024MRI Reconstruction with Regularized 3D Diffusion Model (R3DM)
  6. 2024X-Diffusion: Generating Detailed 3D MRI Volumes From a Single Image Using Cross-Sectional Diffusion Models
  7. 2026Scale-Cascaded Diffusion Models for Super-Resolution in Medical Imaging
  8. 2025Deep generative priors for 3D brain analysis