Abstract
Magnetic Resonance Imaging (MRI) varies significantly between scanners, protocols, and magnetic field strengths (0.1T to 7T), limiting cross-site generalisation and robust multi-centre analysis. We propose a Latent Conditional Rectified Flow Model (LCRFM) for controllable MRI synthesis across modalities (T1w, T2w, and FLAIR) and field strengths. The method operates in a pretrained latent space and is conditioned on modality and field-strength embeddings, combined with anatomically standardised representations and tissue priors, enabling learning without paired supervision. We evaluate the approach on the MRIxFields2026 benchmark for ultra-high-field synthesis, low-field enhancement, and field-to-field translation. The results demonstrate that the proposed framework is capable of synthesising MRI across a range of modalities and magnetic field strengths while maintaining anatomically consistent outputs. Overall, the framework provides a unified approach for controllable MRI synthesis under heterogeneous imaging conditions using a single model without task-specific training.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/MRIxFields2026_008.pdf
SharedIt Link: Not yet available
SpringerLink (DOI): Not yet available
Supplementary Material: https://papers.miccai.org/miccai-2026-sat/supp/MRIxFields2026_008_supp.pdf
Link to Open Review
Open Review Page: https://openreview.net/forum?id=NIE1LN5Ttk
BibTex
@InProceedings{CarAgu_Latent_MICCAISAT2026,
author = { Cartaya Lathulerie, Agustin AND Casamitjana, Adrià AND Oliver, Arnau AND Lladó, Xavier},
title = { { Latent Conditional Rectified Flow for MRI Synthesis Across Modalities and Field Strengths in the MRIxFields2026 Challenge } },
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026 Workshops and Challenges},
year = {2026},
publisher = {Springer Nature Switzerland},
volume = {LNCS 17274},
month = {pending},
page = {pending}
}