Few-View CT Reconstruction with Group-Sparsity Regularization

Peng Bao, Jiliu Zhou, Yi Zhang

Abstract

Classical total variation (TV) based iterative reconstruction algorithms assume that the signal is piecewise smooth, which causes reconstruction results to suffer from the over-smoothing effect. To address this problem, this work presents a novel computed tomography (CT) reconstruction method for the few-view problem called the group-sparsity regularization-based simultaneous algebraic reconstruction technique (GSR-SART). Group-based sparse representation, which utilizes the concept of a group as the basic unit of sparse representation instead of a patch, is introduced as the image domain prior regularization term to eliminate the over-smoothing effect. By grouping the nonlocal patches into different clusters with similarity measured by Euclidean distance, the sparsity and nonlocal similarity in a single image are simultaneously explored. The split Bregman iteration algorithm is applied to obtain the numerical scheme. Experimental results demonstrate that our method both qualitatively and quantitatively outperforms several existing reconstruction methods, including filtered back projection, expectation maximization, SART, and TV-based projections onto convex sets.

Introduction

Computed Tomography (CT) is a vital diagnostic tool, but the X-ray radiation it uses carries risks of cancer and genetic damage. To minimize this, researchers try to reduce the radiation dose by either lowering the X-ray current or decreasing the number of "views" (angles) taken during a scan. However, taking fewer views (few-view CT) leads to an "ill-posed" reconstruction problem. Standard algorithms like Filtered Back Projection (FBP) produce severe streak artifacts when data is missing. While iterative methods using Total Variation (TV) regularization can remove these streaks, they often assume the image is "piecewise smooth," which results in a "cartoon-like" over-smoothing that wipes out fine textures and critical clinical details.

This paper introduces GSR-SART, a novel CT reconstruction algorithm that allows for high-quality imaging using significantly fewer X-ray views. By grouping similar image patches together and applying group-sparsity regularization, the method eliminates streak artifacts while avoiding the blurriness seen in traditional TV methods.

Methods

The authors propose the Group-Sparsity Regularization-based Simultaneous Algebraic Reconstruction Technique (GSR-SART). The core innovation lies in how the image is "regularized" during reconstruction:

  1. SART Step: The algorithm starts with the Simultaneous Algebraic Reconstruction Technique, a standard iterative method that tries to find an image that matches the measured projection data.
  2. Group-Based Sparse Representation (GSR): Instead of processing the image pixel-by-pixel or patch-by-patch, the algorithm looks for "nonlocal similarity." It divides the image into patches, finds the most similar patches across the entire image using Euclidean distance, and clusters them into a "group."
  3. Adaptive Dictionary Learning: For each group, the algorithm uses Singular Value Decomposition (SVD) to learn a dictionary specifically tailored to those patches. This allows the model to represent the image data very efficiently (sparsely).
The process of Group-Based Sparse Representation (GSR), involving patch extraction, matching, and grouping to exploit nonlocal similarity.
  1. Split Bregman Iteration: Because the math involved in "sparsity" (specifically the $L_0$ norm) is notoriously difficult to solve, the authors employ the Split Bregman algorithm to break the complex optimization problem into smaller, manageable sub-problems.
Algorithm 1: The Split Bregman Iteration (SBI) used to solve the group-sparsity optimization problem.

By treating groups of similar patches as the basic unit, the algorithm preserves repetitive structures and sharp edges much better than methods that treat every patch in isolation.

Experimental Results

The researchers evaluated their method using three representative clinical slices (Abdominal, Pelvic, and Thoracic) sourced from the National Cancer Imaging Archive. The experiments simulated a "few-view" scenario using only 64 projection views evenly distributed over 360 degrees.

The GSR-SART method was compared against FBP, EM, SART, and TV-POCS. The quantitative improvements were substantial:

  • PSNR (Peak Signal-to-Noise Ratio): In the pelvic image test, GSR-SART achieved a PSNR of 45.74 dB, compared to 37.32 dB for TV-POCS and only 21.56 dB for FBP.
  • SSIM (Structural Similarity Index): GSR-SART consistently scored above 0.98, indicating the reconstructed image was nearly identical in structure to the "gold standard" full-dose reference.
  • Visual Quality: Qualitative results showed that while TV-POCS removed streaks, it blurred the edges of tissues. GSR-SART successfully removed the streaks while maintaining the sharpness of small blood vessels and bone structures.

Discussion and Conclusion

The major insight of this work is that group sparsity provides a much stronger prior than Total Variation. It doesn't just assume the image is smooth; it assumes the image is structured, which allows it to keep edges sharp while suppressing noise. Furthermore, the dictionary used to reconstruct the image isn't fixed; it is adaptively updated at each iteration based on the current estimate of the image.

Despite its performance, the authors acknowledge that the computational burden of GSR-SART is heavy due to the grouping and SVD processes. Future work will focus on using machine learning to adaptively determine optimal parameters and implementing the algorithm on GPUs to bring reconstruction time down to clinically acceptable levels.

Summary updated

Abstract

Classical total variation (TV) based iterative reconstruction algorithms assume that the signal is piecewise smooth, which causes reconstruction results to suffer from the over-smoothing effect. To address this problem, this work presents a novel computed tomography (CT) reconstruction method for the few-view problem called the group-sparsity regularization-based simultaneous algebraic reconstruction technique (GSR-SART). Group-based sparse representation, which utilizes the concept of a group as the basic unit of sparse representation instead of a patch, is introduced as the image domain prior regularization term to eliminate the over-smoothing effect. By grouping the nonlocal patches into different clusters with similarity measured by Euclidean distance, the sparsity and nonlocal similarity in a single image are simultaneously explored. The split Bregman iteration algorithm is applied to obtain the numerical scheme. Experimental results demonstrate that our method both qualitatively and quantitatively outperforms several existing reconstruction methods, including filtered back projection, expectation maximization, SART, and TV-based projections onto convex sets.
BibTeX
@article{bao2018few,
  author = {Peng Bao and Jiliu Zhou and Yi Zhang},
  title = {{Few-View CT Reconstruction with Group-Sparsity Regularization}},
  journal = {International Journal for Numerical Methods in Biomedical Engineering},
  year = {2018},
  doi = {10.1002/cnm.3101},
}
Figure