Diffusion Active Learning: Towards Data-Driven Experimental Design in Computed Tomography

2025-04-04

AI Summary

To address the ill-posed inverse problem of low-dose, few-view CT reconstruction, this paper proposes an active learning framework integrating generative diffusion models with data-driven sequential experimental design. Methodologically, it introduces an unconditional diffusion model as a structured prior within a closed-loop active learning pipeline; reconstruction uncertainty is quantified via diffusion posterior sampling, enabling adaptive selection of the most informative projection angles—thereby jointly optimizing measurement strategy and reconstruction quality. The framework comprises three stages: diffusion model pretraining, uncertainty-driven active query selection, and iterative co-updating of reconstruction and acquisition. Evaluated on multiple real-world CT datasets, the method achieves equivalent X-ray dose reductions of 30–50% while improving PSNR by 2.1–3.8 dB, significantly outperforming conventional sparse-angle reconstruction and state-of-the-art active learning approaches.

Technology Category

• Machine Learning: Active Learning
• Constraint Satisfaction and Optimization: Constraint Learning and Acquisition
• Reasoning under Uncertainty: Stochastic Optimization

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Web Mining and Content Analysis: Content-based information diffusion

Abstract

We introduce Diffusion Active Learning, a novel approach that combines generative diffusion modeling with data-driven sequential experimental design to adaptively acquire data for inverse problems. Although broadly applicable, we focus on scientific computed tomography (CT) for experimental validation, where structured prior datasets are available, and reducing data requirements directly translates to shorter measurement times and lower X-ray doses. We first pre-train an unconditional diffusion model on domain-specific CT reconstructions. The diffusion model acts as a learned prior that is data-dependent and captures the structure of the underlying data distribution, which is then used in two ways: It drives the active learning process and also improves the quality of the reconstructions. During the active learning loop, we employ a variant of diffusion posterior sampling to generate conditional data samples from the posterior distribution, ensuring consistency with the current measurements. Using these samples, we quantify the uncertainty in the current estimate to select the most informative next measurement. Our results show substantial reductions in data acquisition requirements, corresponding to lower X-ray doses, while simultaneously improving image reconstruction quality across multiple real-world tomography datasets.

Problem

• Reducing data acquisition in CT scans using diffusion models
• Improving image reconstruction quality with active learning
• Lowering X-ray doses through adaptive experimental design

Innovation

• Combines generative diffusion with active learning
• Uses diffusion model as data-dependent prior
• Employs diffusion posterior sampling for uncertainty