Design Of C-arm X-ray Machine Host Computer Control Software And Research On 3D Reconstruction Algorithm

The cone-beam CT has become a hotspot in the current research because of its high radiation utilization efficiency and isotropic reconstruction results.C-arm X-ray machine is a common cone-beam CT medical equipment,whose three-dimensional reconstruction technology is being widely studied.The research of this thesis relies on the related scientific and technological research projects,In the project,the design of the control software of the C-arm X-ray host machine is mainly completed.In the Qt development environment,the functions of image acquisition,motion control and three-dimensional photographing are realized by using the C++ programming language.And the key technologies of three-dimensional reconstruction of CT image denoising,FDK algorithm and iterative algorithm are studied.The main contents of this paper are as follows:(1)The C-arm X-ray machine control program is achievesd.Aiming at the software function requirement and hardware structure of this machine,the overall architecture design of control software is completed.The software module is divided into the exposure and acquisition module,the motion control module,the image processing module and the three-dimensional drawing module upon the function and theses modules were designed.And these functions are achieved on the Qt platform.(2)The nonlocal mean denoising algorithm is improved.For the problem of noise at low dose,the nonlocal mean denoising algorithm is used to denoise the CT image.The existing problems of the algorithm are analyzed and the algorithm is improved: First,the exponential kernel function of the algorithm is improved using the cosine kernel function;Second,a pre-criterion method based on structural similarity is proposed to select the weighted pixel block to be calculated.Aiming at the problem that the algorithm operates a long time,a parallel acceleration scheme based on CUDA is designed and implemented.After the improvement the image quality is improved and the use of CUDA acceleration can greatly shorten the calculation time.(3)A short scan trajectory FDK reconstruction algorithm CUDA acceleration scheme is designed.The FDK reconstruction algorithm for short scan is studied.The influence of scanning trajectory on FDK algorithm is analyzed.Combined with the existing research,this thesis analyzes and implements FDK algorithm that applies the short scan trajectory.In order to meet the requirements of real time,combined with the calculation characteristics of each step of the algorithm,the CUDA acceleration scheme is designed.(4)The ASD-POCS three-dimensional reconstruction algorithm based on compression perceptual theory is improved.The commonly used iterative reconstruction algorithm is analyzed and implemented.In order to obtain better reconstruction results in the condition of sparse angle,the ASD-POCS algorithm is analyzed.The experimental results show that the method can obtain better image results than the traditional reconstruction algorithm.However,the algorithm has the problems of a lot of iterations and slow convergence.For the problem,a two-step improvement is proposed in this paper: First,an iterative termination condition of TV minimization iterative process based on the result pixel difference ratio is proposed.Second,the algorithm is improved by combining the fast iteration shrink threshold algorithm.The improved algorithm converges faster and requires fewer iterations to obtain good reconstruction results.The improved algorithm can obtain more accurate reconstruction results than the original algorithm when the projection sparseness becomes larger and the projection range becomes smaller.