Zero-Shot Brain MRI Inpainting with 2.5D Unconditional Flow Priors
Quick Answer
The proposed zero-shot brain MRI inpainting framework utilizes 2.5D unconditional flow priors to synthesize healthy tissue in pathological regions, achieving an SSIM of 0.816 and a PSNR of 22.923 on the BraTS 2026 validation set.
Quick Take
This approach resolves inter-slice discontinuities found in 2D methods while avoiding the computational costs of 3D models, making it effective for automated brain analysis applications.
Key Points
- Utilizes 2.5D unconditional flow priors for efficient brain MRI inpainting.
- Achieves SSIM of 0.816 and PSNR of 22.923 on BraTS 2026 validation set.
- Avoids the computational overhead of full 3D convolutions.
- Resolves inter-slice discontinuities present in traditional 2D methods.
- Code is available for public access to enhance research and application.
Paper Resources
📖 Reader Mode
~2 min readAbstract:Generative inpainting of brain MRI volumes is essential for synthesizing healthy tissue in pathological regions, improving the accuracy and reliability of automated downstream brain analysis applications such as image registration, brain extraction, and segmentation. However, standard 3D approaches are computationally prohibitive, while efficient 2D slice-wise methods suffer from severe inter-slice discontinuities. Furthermore, traditional models rely on conditional training, requiring task-specific learning of masked inputs. We propose a zero-shot brain MRI inpainting framework utilizing 2.5D unconditional flow priors to capture spatial context along the superior-inferior axis without the overhead of full 3D convolutions. During training, our flow matching model learns the joint distribution of adjacent axial slice triplets, modeling the manifold of healthy brain anatomy while explicitly excluding pathological regions from the loss function. At inference, the model processes the input triplets autoregressively along the depth axis. We employ the Restora-Flow solver to constrain the unconditional prior using the input mask, achieving accurate zero-shot inpainting. Evaluations show our 2.5D strategy resolves the structural discontinuities of 2D baselines, synthesizing plausible healthy tissue while maintaining volumetric consistency across the axial, sagittal, and coronal planes. As a final step, we generate and average an ensemble of multiple stochastic reconstructions to form the final prediction. Quantitative results benchmarked on the official BraTS 2026 Inpainting Challenge validation set demonstrate the effectiveness of our proposed approach, yielding an SSIM of 0.816 $\pm$ 0.112, MSE of 0.007 $\pm$ 0.005, and PSNR of 22.923 $\pm$ 4.343. Code is available at this https URL.
| Subjects: | Computer Vision and Pattern Recognition (cs.CV) |
| Cite as: | arXiv:2610.08983 [cs.CV] |
| (or arXiv:2610.08983v1 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2610.08983 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Arnela Hadzic [view email]
[v1]
Tue, 6 Oct 2026 18:44:24 UTC (3,218 KB)
— Originally published at arxiv.org
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.CV
See more →ProMoE-FL: Prototype-conditioned Mixture of Experts for Multimodal Federated Learning with Missing Modalities
ProMoE-FL introduces a Prototype-conditioned Mixture-of-Experts framework for multimodal federated learning, effectively addressing missing modalities. It outperforms existing methods on four chest X-ray datasets, demonstrating superior feature synthesis capabilities in both homogeneous and heterogeneous settings.