Temporal Neural Cellular Automata: Application to modeling of contrast enhancement in breast MRI

Daniel M. Lang1,2, Richard Osuala3, Veronika Spieker1,2, Karim Lekadir3, Rickmer Braren4, Julia A. Schnabel1,2
1Helmholtz Munich, 2Technical University Munich, 3Universitat de Barcelona, 4Klinikum Rechts der Isar
arXiv Code

TeNCA is able to continuously model the evolution of contrast enhancement solely based on pre-contrast acquisitions.

Abstract

Synthetic contrast enhancement offers fast image acquisition and eliminates the need for intravenous injection of contrast agent. This is particularly beneficial for breast imaging, where long acquisition times and high cost are significantly limiting the applicability of magnetic resonance imaging (MRI) as a widespread screening modality.

Recent studies have demonstrated the feasibility of synthetic contrast generation. However, current state-of-the-art (SOTA) methods lack sufficient measures for consistent temporal evolution. Neural cellular automata (NCA) offer a robust and lightweight architecture to model evolving patterns between neighboring cells or pixels.

In this work we introduce TeNCA (Temporal Neural Cellular Automata), which extends and further refines NCAs to effectively model temporally sparse, non-uniformly sampled imaging data. To achieve this, we advance the training strategy by enabling adaptive loss computation and define the iterative nature of the method to resemble a physical progression in time. This conditions the model to learn a physiologically plausible evolution of contrast enhancement. We rigorously train and test TeNCA on a diverse breast MRI dataset and demonstrate its effectiveness, surpassing the performance of existing methods in generation of images that align with ground truth post-contrast sequences.

Method overview

At each step, our neural cellular automata backbone transitions the images gradually to reflect the next time point. During training, intermediate states are conditioned at all time points with a given ground truth DCE-MRI available.

Dynamic modeling of contrast enhancement

Notably, TeNCA and CC-Net are able to generate post-contrast sequences from an early stage, whereas ground truth DCE generated post-contrast images and U-Net predictions become only available at later time points.

Related Links

Towards learning contrast kinetics with multi-condition latent diffusion models introduces CC-Net, a latent diffusion model conditioning on acquisition time and supplementary imaging information using a ControlNet.

Virtual Dynamic Contrast Enhanced Breast MRI Using 2D U-Net Architectures develops two U-Net based architectures to model contrast enhancement.

Simulating Dynamic Tumor Contrast Enhancement in Breast MRI using Conditional Generative Adversarial Networks uses a generative adversarial network based on pix2pixHD.

Code for computation of the Fréchet Radiomics Distance (FRD) can be found here.

Data used in this study can be downloaded from the MAMA-MIA website and from TCIA.

BibTeX

@article{lang2025temporal,
        title={Temporal Neural Cellular Automata: Application to modeling of contrast enhancement in breast MRI}, 
        author={Daniel M. Lang and Richard Osuala and Veronika Spieker and Karim Lekadir and Rickmer Braren and Julia A. Schnabel},
        year={2025},
        eprint={2506.18720},
        archivePrefix={arXiv},
        primaryClass={eess.IV},
        url={https://arxiv.org/abs/2506.18720},
}