Unsupervised Self-Prior Embedding Neural Representation for Iterative Sparse-View CT Reconstruction

2025-02-08

AI Summary

In sparse-view CT (SVCT) reconstruction, unsupervised methods suffer from noise sensitivity and insufficient exploitation of image priors, limiting their performance. To address this, we propose Spener—a novel unsupervised iterative implicit neural representation (INR) framework. Its core innovation is the “self-prior embedding” mechanism: intermediate reconstructions from the current iteration are processed via local feature extraction and dynamically injected into subsequent INR optimization steps, thereby constructing a data-driven, domain-specific image prior from imperfect reconstructions. Spener requires no paired ground truth or labels. On in-domain test data, it achieves state-of-the-art (SOTA) performance comparable to supervised methods; on out-of-domain data and noisy sinograms, it significantly outperforms supervised approaches, demonstrating superior generalization and robustness.

Technology Category

• Computer Vision: Representation Learning for Vision
• Machine Learning: Unsupervised & Self-Supervised Learning
• Reasoning under Uncertainty: Uncertainty Representations

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: Graph embeddings and representation learning for Web-related graphs
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Emerging unsupervised implicit neural representation (INR) methods, such as NeRP, NeAT, and SCOPE, have shown great potential to address sparse-view computed tomography (SVCT) inverse problems. Although these INR-based methods perform well in relatively dense SVCT reconstructions, they struggle to achieve comparable performance to supervised methods in sparser SVCT scenarios. They are prone to being affected by noise, limiting their applicability in real clinical settings. Additionally, current methods have not fully explored the use of image domain priors for solving SVCsT inverse problems. In this work, we demonstrate that imperfect reconstruction results can provide effective image domain priors for INRs to enhance performance. To leverage this, we introduce Self-prior embedding neural representation (Spener), a novel unsupervised method for SVCT reconstruction that integrates iterative reconstruction algorithms. During each iteration, Spener extracts local image prior features from the previous iteration and embeds them to constrain the solution space. Experimental results on multiple CT datasets show that our unsupervised Spener method achieves performance comparable to supervised state-of-the-art (SOTA) methods on in-domain data while outperforming them on out-of-domain datasets. Moreover, Spener significantly improves the performance of INR-based methods in handling SVCT with noisy sinograms. Our code is available at https://github.com/MeijiTian/Spener.

Problem

• Enhance sparse-view CT reconstruction accuracy
• Improve noise robustness in CT imaging
• Leverage image priors for unsupervised learning

Innovation

• Unsupervised Self-Prior Embedding
• Iterative Reconstruction Algorithms
• Image Domain Prior Features