Multimodal synthesis of MRI and tabular data with diffusion in a joint latent space via cross-attention

Daniel Mensing Daniel Mensing, Jan Kapar, Jochen G. Hirsch, Matthias Günther, Horst Hahn, Marvin N. Wright · 2026

Read Multimodal synthesis of MRI and tabular data with diffusion in a joint latent space via cross-attention by Daniel Mensing Daniel Mensing, Jan Kapar, Jochen G. Hirsch, Matthias Günther, Horst Hahn, Marvin N. Wright (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-06699.txt.

Description

Daniel Mensing and five co-authors make synthetic patients by modeling MRI scans and clinical table data together in one diffusion model. Separate tools can produce mismatched pairs, like a scan that contradicts the listed body measures. Their model fuses both data types with an encoder-decoder into a shared latent space linked by cross-attention, then runs diffusion there with separate decoders.

Trained on over ten thousand German National Cohort people, it produces plausible scans whose body makeup matches the generated traits. Image quality scores are strong, and its tables rival standard generators. The authors present it as a first proof of concept toward digital twins in healthcare.

Catalog record

CREATORDaniel Mensing Daniel Mensing, Jan Kapar, Jochen G. Hirsch, Matthias Günther, Horst Hahn, Marvin N. Wright
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE17 KB
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.06699

Browse the full library · About the Tower of Babel