Abstract

Abstract Algebraic reconstruction is more suitable for image reconstruction than analytic reconstruction as its high contrast and precision for incomplete projecting data. The calculation of projection coefficients is the key step of algebraic reconstruction, which has crucial effect on the reconstructed quality and speed. In this paper, an efficient method to calculate projection coefficients is proposed for cone-beam computed tomography (CBCT), referring to the structure characteristics of graphics processing unit (GPU) and intersection law of ray and grids. Firstly, an array is used to store the projection coefficient value under different intersections of grids and ray. Then, another array is designed to correct it in another direction. Finally, the projection coefficients can be solved by combining the two arrays. The proposed method has the advantage of less calculation amount and less program branches, and is suitable for parallel acceleration based on compute unified device architecture (CUDA). The calculated speed can be accelerated significantly. Experimental results show that the proposed algorithm is 10 times faster than Siddon algorithm, and almost 30 times faster in GPU.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call