RSR-NF: Neural Field Regularization by Static Restoration Priors for Dynamic Imaging

2025-03-13

AI Summary

In dynamic CT reconstruction, severe ill-posedness arises from single-view acquisition and extremely low sampling rates. To address this, we propose a neural field-based reconstruction method that operates without dynamic ground-truth labels. Our approach integrates a static-image pre-trained deep restoration operator as a structural prior into a time-varying neural field representation, and solves the inverse problem within the Regularization-by-Denoising (RED) framework via the Alternating Direction Method of Multipliers (ADMM). This work constitutes the first successful transfer of static priors to variational dynamic neural field reconstruction, eliminating reliance on strong temporal constraints or dynamic annotation data. Experimental results demonstrate substantial improvements in structural fidelity and motion consistency over baseline methods—including purely temporal-regularized neural fields, low-rank + RED, and deep image priors—under realistic sparse-view dynamic CT settings.

Technology Category

• Computer Vision: Other Foundations of Computer Vision
• Machine Learning: Matrix & Tensor Methods
• Intelligent Robots: State Estimation

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

Dynamic imaging involves the reconstruction of a spatio-temporal object at all times using its undersampled measurements. In particular, in dynamic computed tomography (dCT), only a single projection at one view angle is available at a time, making the inverse problem very challenging. Moreover, ground-truth dynamic data is usually either unavailable or too scarce to be used for supervised learning techniques. To tackle this problem, we propose RSR-NF, which uses a neural field (NF) to represent the dynamic object and, using the Regularization-by-Denoising (RED) framework, incorporates an additional static deep spatial prior into a variational formulation via a learned restoration operator. We use an ADMM-based algorithm with variable splitting to efficiently optimize the variational objective. We compare RSR-NF to three alternatives: NF with only temporal regularization; a recent method combining a partially-separable low-rank representation with RED using a denoiser pretrained on static data; and a deep-image prior-based model. The first comparison demonstrates the reconstruction improvements achieved by combining the NF representation with static restoration priors, whereas the other two demonstrate the improvement over state-of-the art techniques for dCT.

Problem

• Reconstructs dynamic objects from undersampled measurements in dynamic imaging.
• Addresses lack of ground-truth data for supervised learning in dynamic CT.
• Improves reconstruction by combining neural fields with static restoration priors.

Innovation

• Neural field represents dynamic imaging objects.
• Static restoration priors enhance variational formulation.
• ADMM algorithm optimizes dynamic imaging reconstruction.