Search papers, labs, and topics across Lattice.
This paper addresses the challenge of domain shift in magnetic resonance imaging (MRI) caused by varying magnetic field strengths, which affects image quality and anatomical visibility. The authors introduce a 3D unpaired cross-field MRI translation framework that utilizes field-conditioned content-style pretraining to disentangle anatomical content from field-dependent contrast characteristics, enabling effective translation across different field strengths. Experimental results show that the proposed method successfully adapts to multiple challenge settings while maintaining anatomical fidelity in synthesized MRI volumes.
Unpaired MRI translation can effectively preserve anatomical structures across varying magnetic field strengths, a significant leap in addressing domain shift challenges in medical imaging.
Magnetic field strength is a major source of domain shift in magnetic resonance imaging (MRI), affecting signal-to-noise ratio, tissue contrast, spatial detail, and the visibility of anatomical boundaries. The MRIxFields 2026 challenge investigates this problem through cross-field MRI translation across acquisitions at 0.1T, 1.5T, 3T, 5T, and 7T. Its three tasks, Any-to-7T, 0.1T-to-High, and Any-to-Any synthesis, require the generation of target-field image characteristics while preserving subject-specific anatomy. This problem is particularly challenging because paired acquisitions of the same subject across multiple field strengths are rarely available for training. We propose a 3D unpaired cross-field MRI translation framework based on field-conditioned content-style pretraining. The proposed framework first learns controllable field-to-field translation across all available field strengths by disentangling anatomical content from field-dependent contrast characteristics. The pretrained backbone is then adapted to task-specific target domains. Our model comprises a 3D content encoder, a 3D style encoder, a field-conditioned style generator, an AdaIN-modulated decoder, and a multi-field discriminator. Adversarial learning encourages realistic target-field appearance, while cycle-consistency, identity, content, style, and diversity constraints promote anatomical fidelity and controllable translation. We evaluate the proposed method on MRIxFields data spanning five field strengths and three MRI modalities. Experiments on paired test data demonstrate that the framework can adapt to the three challenge settings while preserving three-dimensional anatomical structure in the synthesized volumes. The implementation code is publicly available at https://github.com/Idea89560041/3D-MRI-Field-Translation.
Tsinghua AI