Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

3D Photon Counting CT Image Super-Resolution Using Conditional Diffusion Model

Chuang Niu, Christopher Wiedeman, Mengzhou Li, Jonathan S Maltz, Ge Wang

TL;DR

The paper tackles the challenge of improving PCCT resolution by employing a conditional denoising diffusion probabilistic model (DDPM) framework trained on CatSim-simulated LR-HR CT pairs. To manage high dimensionality, it decomposes 3D super-resolution into two 2D networks for in-plane and through-plane SR and uses a joint inference strategy to enforce 3D consistency, enabling efficient training and inference. Results show that the conditional DDPMs recover high-frequency details more effectively than baseline references, with 2D joint inference mitigating artifacts and providing sharper textures in both planar directions. This realistic-simulation plus diffusion-based approach demonstrates potential to enhance PCCT resolution in a clinically relevant, computationally tractable manner.

Abstract

This study aims to improve photon counting CT (PCCT) image resolution using denoising diffusion probabilistic models (DDPM). Although DDPMs have shown superior performance when applied to various computer vision tasks, their effectiveness has yet to be translated to high dimensional CT super-resolution. To train DDPMs in a conditional sampling manner, we first leverage CatSim to simulate realistic lower resolution PCCT images from high-resolution CT scans. Since maximizing DDPM performance is time-consuming for both inference and training, especially on high-dimensional PCCT data, we explore both 2D and 3D networks for conditional DDPM and apply methods to accelerate training. In particular, we decompose the 3D task into efficient 2D DDPMs and design a joint 2D inference in the reverse diffusion process that synergizes 2D results of all three dimensions to make the final 3D prediction. Experimental results show that our DDPM achieves improved results versus baseline reference models in recovering high-frequency structures, suggesting that a framework based on realistic simulation and DDPM shows promise for improving PCCT resolution.

3D Photon Counting CT Image Super-Resolution Using Conditional Diffusion Model

TL;DR

The paper tackles the challenge of improving PCCT resolution by employing a conditional denoising diffusion probabilistic model (DDPM) framework trained on CatSim-simulated LR-HR CT pairs. To manage high dimensionality, it decomposes 3D super-resolution into two 2D networks for in-plane and through-plane SR and uses a joint inference strategy to enforce 3D consistency, enabling efficient training and inference. Results show that the conditional DDPMs recover high-frequency details more effectively than baseline references, with 2D joint inference mitigating artifacts and providing sharper textures in both planar directions. This realistic-simulation plus diffusion-based approach demonstrates potential to enhance PCCT resolution in a clinically relevant, computationally tractable manner.

Abstract

This study aims to improve photon counting CT (PCCT) image resolution using denoising diffusion probabilistic models (DDPM). Although DDPMs have shown superior performance when applied to various computer vision tasks, their effectiveness has yet to be translated to high dimensional CT super-resolution. To train DDPMs in a conditional sampling manner, we first leverage CatSim to simulate realistic lower resolution PCCT images from high-resolution CT scans. Since maximizing DDPM performance is time-consuming for both inference and training, especially on high-dimensional PCCT data, we explore both 2D and 3D networks for conditional DDPM and apply methods to accelerate training. In particular, we decompose the 3D task into efficient 2D DDPMs and design a joint 2D inference in the reverse diffusion process that synergizes 2D results of all three dimensions to make the final 3D prediction. Experimental results show that our DDPM achieves improved results versus baseline reference models in recovering high-frequency structures, suggesting that a framework based on realistic simulation and DDPM shows promise for improving PCCT resolution.
Paper Structure (8 sections, 9 equations, 5 figures)

This paper contains 8 sections, 9 equations, 5 figures.

Table of Contents

  1. Introduction
  2. Methods
  3. Data and Simulation
  4. Conditional DDPM
  5. 3D Super-Resolution with Joint 2D Inference
  6. Implementation Details
  7. Experiments and Results
  8. Discussion and conclusion

Figures (5)

  • Figure 1: In-plane (XY) results.
  • Figure 2: Through-plane ($z$--$x$) results.
  • Figure 3: Application to CT images of the human temporal bone.
  • Figure 4: MTF Comparison (In-Plane). The frequency axis has been suppressed for proprietary reasons.
  • Figure 5: MTF Comparison (Through-Plane). The frequency axis has been suppressed for proprietary reasons.