Visual C++ Implementation of Sinogram-based Adaptive Iterative Reconstruction for Sparse View X-Ray CT

D. Trinca,Y. Zhong,Y. Wang,T. Mamyrbayev,E. Libin

DOI: https://doi.org/10.48550/arXiv.1608.06248

2016-08-23

Abstract:With the availability of more powerful computing processors, iterative reconstruction algorithms have recently been successfully implemented as an approach to achieving significant dose reduction in X-ray CT. In this report, we describe our proposal of an adaptive iterative reconstruction algorithm for X-ray CT, that is shown to provide results comparable to those obtained by proprietary algorithms, both in terms of reconstruction accuracy and execution time. Implementation code in the C language is provided, along with example of user interface.

Medical Physics

What problem does this paper attempt to address?

This paper aims to solve the problem of reducing radiation dose in X - ray computed tomography (CT). Specifically, the authors developed an adaptive iterative reconstruction algorithm based on sparse views to achieve reconstruction accuracy and execution time comparable to existing proprietary algorithms while providing a significant dose - reduction effect. Through this method, the radiation dose received by patients in clinical examinations can be reduced while maintaining image quality.

### Main contributions
1. **Dose reduction**: This algorithm can achieve approximately 50% dose reduction while ensuring image quality, which is of great significance for improving patient safety and reducing radiation risks.
2. **Free and open - source**: Unlike existing proprietary algorithms, this algorithm is completely open to the scientific research community and can be freely used and further optimized.
3. **High efficiency**: By using parallel computing and multi - core processors, the execution time of this algorithm is comparable to that of existing fast algorithms (such as SAFIRE), making it suitable for practical clinical applications.

### Technical details
- **Initialization**: The algorithm first initializes the reconstruction matrix with an initial solution. This step includes calculating the segment lengths and corresponding density values on each X - ray beam path.
- **Iterative process**: In each iteration, the algorithm updates the density matrix according to the current reconstruction result and sinogram data. This process is achieved by calculating the contribution values on each X - ray beam path and gradually approaching the final reconstruction result.
- **Optimization**: To improve the execution efficiency of the algorithm, the authors proposed several optimization methods, including parallel computing, removing zero - value terms, and using different initial solutions.

### Result verification
The paper verified the effectiveness of the algorithm through experiments on the Shepp - Logan phantom. The results showed that the adaptive iterative reconstruction algorithm using 198 projection angles can achieve the same image quality as the filtered back - projection algorithm (FBP) using 360 projection angles, thus achieving approximately 45% dose reduction. In addition, the execution time of this algorithm on a desktop computer is approximately 1.75 seconds, demonstrating its feasibility in practical applications.

### Conclusion
This study proposed an efficient adaptive iterative reconstruction algorithm, which can not only significantly reduce the radiation dose in X - ray CT, but also provide image quality and execution time comparable to existing proprietary algorithms. Through the open - source approach, this algorithm provides new possibilities for scientific research and industrial applications, especially in non - destructive testing and other fields.