Abstract

Image reconstruction for dynamic inverse problems with highly undersampled data poses a major challenge: not accounting for the dynamics of the process leads to a non-realistic motion with no time regularity. Variational approaches that penalize time derivatives or introduce motion model regularizers have been proposed to relate subsequent frames and improve image quality using grid-based discretization. Neural fields offer an alternative parametrization of the desired spatiotemporal quantity with a deep neural network, a lightweight, continuous, and biased towards smoothness representation. The inductive bias has been exploited to enforce time regularity for dynamic inverse problems resulting in neural fields optimized by minimizing a data-fidelity term only. In this paper we investigate and show the benefits of introducing explicit PDE-based motion regularizers, namely, the optical flow equation, in 2D+time computed tomography for the optimization of neural fields. We also compare neural fields against a grid-based solver and show that the former outperforms the latter.
Original languageEnglish
JournalJournal of Mathematical Imaging and Vision
Publication statusAcceptance date - 14 Jun 2025

Acknowledgements

The authors gratefully acknowledge the University of Bath’s Research Computing Group (doi.org/10.15125/b6cd-s854) for their support in this work.

Funding

PA is supported by a scholarship from the EPSRC Centre for Doctoral Training in Statistical Applied Mathematics at Bath (SAMBa), under the project EP/S022945/1. MJE acknowledges support from the EPSRC (EP/S026045/1, EP/T026693/1, EP/V026259/1, EP/Y037286/1) and the European Union Horizon 2020 research and innovation programme under the Marie Skodowska-Curie grant agreement REMODEL.

FundersFunder number
Engineering and Physical Sciences Research CouncilEP/S022945/1 , EP/S026045/1

Keywords

  • eess.IV
  • cs.CV

Fingerprint

Dive into the research topics of 'Enhancing Dynamic CT Image Reconstruction with Neural Fields Through Explicit Motion Regularizers'. Together they form a unique fingerprint.

Cite this