DRACO: Differentiable Reconstruction for Arbitrary CBCT Orbits

TL;DR

A novel method for reconstructing cone beam computed tomography images for arbitrary orbits, addressing the computational and memory challenges associated with traditional iterative reconstruction algorithms by employing a differentiable shift-variant filtered backprojection neural network, optimized for arbitrary trajectories.

Abstract

This paper introduces a novel method for reconstructing cone beam computed tomography (CBCT) images for arbitrary orbits using a differentiable shift-variant filtered backprojection (FBP) neural network. Traditional CBCT reconstruction methods for arbitrary orbits, like iterative reconstruction algorithms, are computationally expensive and memory-intensive. The proposed method addresses these challenges by employing a shift-variant FBP algorithm optimized for arbitrary trajectories through a deep learning approach that adapts to a specific orbit geometry. This approach overcomes the limitations of existing techniques by integrating known operators into the learning model, minimizing the number of parameters, and improving the interpretability of the model. The proposed method is a significant advancement in interventional medical imaging, particularly for robotic C-arm CT systems, enabling faster and more accurate CBCT reconstructions with customized orbits. Especially this method can also be used for the analytical reconstruction of non-continuous orbits like circular plus arc. The experimental results demonstrate that the proposed method significantly accelerates the reconstruction process compared to conventional iterative algorithms. It achieves comparable or superior image quality, as evidenced by metrics such as the mean squared error (MSE), the peak signal-to-noise ratio (PSNR), and the structural similarity index measure (SSIM). The validation experiments show that the method can handle data from different trajectories, demonstrating its flexibility and robustness across different scan geometries. Our method demonstrates a significant improvement, particularly for the sinusoidal trajectory, achieving a 38.6% reduction in MSE, a 7.7% increase in PSNR, and a 5.0% improvement in SSIM. Furthermore, the computation time for reconstruction was reduced by more than 97%.

Figures (17)

• Figure 1: Detector coordinate system with origin $O_\lambda$ at the centre of the detector. The vectors $l_x^\lambda$ and $l_y^\lambda$ lying in the detector plane, while $l_z^\lambda$ as the unit vector pointing towards the source
• Figure 2: Grangeat’s intermediate function part of the neural network: The input of the neural network is cone-beam projections, and the output is grangeat’s intermediate function.
• Figure 3: Filtering part of the neural network: The input of the neural network is grangeat’s intermediate function, and the output is filtered cone beam projections.
• Figure 4: Backprojection part for the neural network: Backproject the filtered cone-beam projections into 3D volume
• Figure 5: Grangeat's intermediate function part, filtering part, and backprojection part are combined to form the differentiable shift-variant FBP neural network architecture.