When are Diffusion Priors Helpful in Sparse Reconstruction? A Study with Sparse-view CT

2025-02-04

AI Summary

This work systematically evaluates diffusion models as priors for sparse-view CT reconstruction—specifically, low-dose chest-wall CT fat quantification—assessing their applicability and limitations. We compare denoising diffusion probabilistic models (DDPMs) against sparse and Tikhonov regularization across projection counts ranging from 10 to 128, using PSNR, SSIM, and downstream fat-quantification accuracy as evaluation metrics. Our key finding is the first empirical identification of a robustness inflection point: diffusion priors significantly outperform classical methods under extremely sparse conditions (≈10–15 views), yet exhibit limited fine-detail recovery; performance saturates rapidly thereafter and falls below that of Tikhonov or sparse regularization at moderate-to-high data regimes. This challenges the prevailing assumption that “less data favors diffusion models” in clinical CT reconstruction and provides critical evidence for the cautious, context-aware deployment of diffusion priors in high-stakes medical imaging.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Calibration & Uncertainty Quantification
• Reasoning under Uncertainty: Relational Probabilistic Models

Application Category

• User Modeling, Personalization and Recommendation: User privacy protection in personalized systems
• Web Mining and Content Analysis: Content-based information diffusion
• Security and Privacy: Large-scale security measurements

Abstract

Diffusion models demonstrate state-of-the-art performance on image generation, and are gaining traction for sparse medical image reconstruction tasks. However, compared to classical reconstruction algorithms relying on simple analytical priors, diffusion models have the dangerous property of producing realistic looking results emph{even when incorrect}, particularly with few observations. We investigate the utility of diffusion models as priors for image reconstruction by varying the number of observations and comparing their performance to classical priors (sparse and Tikhonov regularization) using pixel-based, structural, and downstream metrics. We make comparisons on low-dose chest wall computed tomography (CT) for fat mass quantification. First, we find that classical priors are superior to diffusion priors when the number of projections is ``sufficient''. Second, we find that diffusion priors can capture a large amount of detail with very few observations, significantly outperforming classical priors. However, they fall short of capturing all details, even with many observations. Finally, we find that the performance of diffusion priors plateau after extremely few ($approx$10-15) projections. Ultimately, our work highlights potential issues with diffusion-based sparse reconstruction and underscores the importance of further investigation, particularly in high-stakes clinical settings.

Problem

• Evaluate diffusion priors in sparse CT reconstruction.
• Compare diffusion and classical priors' performance metrics.
• Assess diffusion priors' limitations in clinical settings.

Innovation

• Diffusion models for image reconstruction
• Comparison with classical priors
• Performance analysis in sparse CT