SCOUT: Fast Spectral CT Imaging in Ultra LOw-data Regimes via PseUdo-label GeneraTion

2026-02-28

AI Summary

This work addresses the challenge of image quality degradation in ultra-low-dose CT imaging caused by noise and detector ring artifacts, particularly in scenarios lacking external training data or annotated projection measurements. The authors propose a novel self-supervised reconstruction method that requires neither pretraining nor external supervision. By leveraging intrinsic structural properties of raw projection data—specifically, spatial non-local similarity and conjugate symmetry in the projection domain—the method constructs a lightweight self-supervised framework to generate pseudo-3D labels for high-quality, efficient reconstruction. The approach significantly suppresses ring artifacts, enhances fine-detail recovery, and maintains high-fidelity imaging even under extremely low-dose conditions. Furthermore, it achieves fast reconstruction without reliance on large annotated datasets or iterative optimization, thereby establishing a new paradigm for CT reconstruction based on endogenous data priors.

Technology Category

• Machine Learning: Unsupervised & Self-Supervised Learning
• Computer Vision: Learning & Optimization for CV
• Search and Optimization: Non-convex Optimization

Application Category

• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Noise and artifacts during computed tomography (CT) scans are a fundamental challenge affecting disease diagnosis. However, current methods either involve excessively long reconstruction times or rely on data-driven models for optimization, failing to adequately consider the valuable information inherent in the data itself, especially medical 3D data. This work proposes a reconstruction method under ultra-low raw data conditions, requiring no external data and avoiding lengthy pre-training processes. By leveraging spatial nonlocal similarity and the conjugate properties of the projection domain to generate pseudo-3D data for self-supervised training, high-fidelity results can be achieved in a very short time. Extensive experiments demonstrate that this method not only mitigates detector-induced ring artifacts but also exhibits unprecedented capabilities in detail recovery. This method provides a new paradigm for research using unlabeled raw projection data. Code is available at https://github.com/yqx7150/SCOUT.

Problem

• spectral CT
• low-data regime
• noise and artifacts
• ring artifacts
• detail recovery

Innovation

• spectral CT
• ultra-low-data reconstruction
• pseudo-label generation
• self-supervised learning
• ring artifact reduction