Accelerated Optimization of Implicit Neural Representations for CT Reconstruction

2025-04-18

AI Summary

In low-dose or sparse-view CT reconstruction, implicit neural representations (INRs) suffer from slow gradient-based optimization convergence—requiring thousands of iterations—leading to poor computational efficiency. To address this, we propose a dual-path acceleration framework: first, a condition-number-aware hybrid loss combining ℓ₂ fidelity and total variation regularization to mitigate Hessian ill-conditioning; second, embedding INR reconstruction within the alternating direction method of multipliers (ADMM) framework to decouple data fidelity and regularization subproblems for efficient alternating optimization. Evaluated on a synthetic breast CT phantom under sparse-view settings, our method reduces required iterations by 3–5×, significantly accelerating reconstruction while preserving high fidelity and structural consistency. To the best of our knowledge, this is the first work synergistically integrating ADMM with condition-number-optimized loss for INR-based CT reconstruction.

Technology Category

• Search and Optimization: Non-convex Optimization
• Computer Vision: Learning & Optimization for CV
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• User Modeling, Personalization and Recommendation: User modeling for targeted and personalized online advertising

Abstract

Inspired by their success in solving challenging inverse problems in computer vision, implicit neural representations (INRs) have been recently proposed for reconstruction in low-dose/sparse-view X-ray computed tomography (CT). An INR represents a CT image as a small-scale neural network that takes spatial coordinates as inputs and outputs attenuation values. Fitting an INR to sinogram data is similar to classical model-based iterative reconstruction methods. However, training INRs with losses and gradient-based algorithms can be prohibitively slow, taking many thousands of iterations to converge. This paper investigates strategies to accelerate the optimization of INRs for CT reconstruction. In particular, we propose two approaches: (1) using a modified loss function with improved conditioning, and (2) an algorithm based on the alternating direction method of multipliers. We illustrate that both of these approaches significantly accelerate INR-based reconstruction of a synthetic breast CT phantom in a sparse-view setting.

Problem

• Accelerate optimization of implicit neural representations for CT
• Improve slow convergence of INRs in CT reconstruction
• Enhance sparse-view CT reconstruction using faster INR training

Innovation

• Modified loss function for better conditioning
• ADMM algorithm to accelerate optimization
• Implicit neural representations for CT reconstruction