# WFM: 3D Wavelet Flow Matching for Ultrafast Multi-Modal MRI Synthesis

arXiv: https://arxiv.org/abs/2604.21146 · HTML: https://mlsanity.com/paper/2604.21146/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- Placed in Medical imaging and segmentation, one of 14 directions in Diffusion models.
- Builds on 17 of its 22 references that are in the map.
- First posted on arXiv 22 Apr 2026.

## Placed in

- Diffusion models / [Medical imaging and segmentation](https://mlsanity.com/diffusion/medical-imaging-and-segmentation/)

## Builds on (most cited first, 17 of 17)

- [Denoising Diffusion Implicit Models](https://mlsanity.com/paper/2010.02502/) (2020)
- [Image-to-Image Translation with Conditional Adversarial Networks](https://mlsanity.com/paper/1611.07004/) (2016)
- [Consistency Models](https://mlsanity.com/paper/2303.01469/) (2023)
- [The RSNA-ASNR-MICCAI BraTS 2021 Benchmark on Brain Tumor Segmentation and Radiogenomic Classification](https://mlsanity.com/paper/2107.02314/) (2021)
- [BBDM: Image-to-image Translation with Brownian Bridge Diffusion Models](https://mlsanity.com/paper/2205.07680/) (2022)
- [I2SB: Image-to-Image Schr"odinger Bridge](https://mlsanity.com/paper/2302.05872/) (2023)
- [Unsupervised Medical Image Translation with Adversarial Diffusion Models](https://mlsanity.com/paper/2207.08208/) (2022)
- [ResViT: Residual vision transformers for multi-modal medical image synthesis](https://mlsanity.com/paper/2106.16031/) (2021)
- [Diffusion Schr"odinger Bridge Matching](https://mlsanity.com/paper/2303.16852/) (2023)
- [Wavelet Diffusion Models are fast and scalable Image Generators](https://mlsanity.com/paper/2211.16152/) (2022)
- [Adaptive Latent Diffusion Model for 3D Medical Image to Image Translation: Multi-modal Magnetic Resonance Imaging Study](https://mlsanity.com/paper/2311.00265/) (2023)
- [WDM: 3D Wavelet Diffusion Models for High-Resolution Medical Image Synthesis](https://mlsanity.com/paper/2402.19043/) (2024)
- [LBM: Latent Bridge Matching for Fast Image-to-Image Translation](https://mlsanity.com/paper/2503.07535/) (2025)
- [A Survey on Training Challenges in Generative Adversarial Networks for Biomedical Image Analysis](https://mlsanity.com/paper/2201.07646/) (2022)
- [cWDM: Conditional Wavelet Diffusion Models for Cross-Modality 3D Medical Image Synthesis](https://mlsanity.com/paper/2411.17203/) (2024)
- [HACA3: A Unified Approach for Multi-site MR Image Harmonization](https://mlsanity.com/paper/2212.06065/) (2022)
- [Towards General Text-guided Image Synthesis for Customized Multimodal Brain MRI Generation](https://mlsanity.com/paper/2409.16818/) (2024)

## Built on by (most cited first, 4 of 4)

- [Dense Temporal Contrast Synthesis via Conditioned Latent Transport](https://mlsanity.com/paper/2607.29394/) (2026)
- [VoxStruct3D: Structure-Leading Flow Matching for Voxel-Space 3D MRI Synthesis](https://mlsanity.com/paper/2608.04557/) (2026)
- [MRIComp4Flow: Compression of 3D Brain MRI for Training Multi-Modal Generative Models](https://mlsanity.com/paper/2608.10291/) (2026)
- [Latent Space Is Not Flat: Rethinking Latent Structure for 3D Medical Image Synthesis](https://mlsanity.com/paper/2609.32794/) (2026)

## Abstract (arXiv)

Diffusion models have achieved remarkable quality in multi-modal MRI synthesis, but their computational cost (hundreds of sampling steps and separate models per modality) limits clinical deployment. We observe that this inefficiency stems from an unnecessary starting point: diffusion begins from pure noise, discarding the structural information already present in available MRI sequences. We propose WFM (Wavelet Flow Matching), which instead learns a direct flow from an informed prior, the mean of conditioning modalities in wavelet space, to the target distribution. Because the source and target share underlying anatomy and differ primarily in contrast, this formulation enables accurate synthesis in just 1-2 integration steps. A single 82M-parameter model with class conditioning synthesizes all four BraTS modalities (T1, T1c, T2, FLAIR), replacing four separate diffusion models totaling 326M parameters. On BraTS 2024, WFM achieves 26.8 dB PSNR and 0.94 SSIM, within 1-2 dB of diffusion baselines, while running 250-1000x faster (0.16-0.64s vs. 160s per volume). This speed-quality trade-off makes real-time MRI synthesis practical for clinical workflows. Code is available at https://github.com/yalcintur/WFM.