XctDiff: Reconstruction of CT Images with Consistent Anatomical Structures from a Single Radiographic Projection Image

TL;DR

XctDiff is a novel framework that reconstructs high-quality CT images from a single radiograph by extracting 3D priors and guiding the reconstruction process, overcoming blurring issues and enabling potential applications in medical analysis.

Contribution

The paper introduces XctDiff, a new method that decomposes CT reconstruction into feature extraction and latent space guidance, achieving state-of-the-art results from a single radiograph.

Findings

Achieves state-of-the-art reconstruction performance.

Effectively overcomes blurring issues in CT images.

Enables promising applications in medical image analysis.

Abstract

In this paper, we present XctDiff, an algorithm framework for reconstructing CT from a single radiograph, which decomposes the reconstruction process into two easily controllable tasks: feature extraction and CT reconstruction. Specifically, we first design a progressive feature extraction strategy that is able to extract robust 3D priors from radiographs. Then, we use the extracted prior information to guide the CT reconstruction in the latent space. Moreover, we design a homogeneous spatial codebook to improve the reconstruction quality further. The experimental results show that our proposed method achieves state-of-the-art reconstruction performance and overcomes the blurring issue. We also apply XctDiff on self-supervised pre-training task. The effectiveness indicates that it has promising additional applications in medical image analysis. The code is available at:https://github.com/qingze-bai/XctDiff