Research On Regularization Reconstruction Algorithm For PET Sparse Detection System

Positron emission tomography is widely recognized as the most sensitive non-invasive technique for studying biophysiology,metabolism,and molecular synthesis pathways in vivo.It can quantitatively calculate the metabolic level of various organs in the living body by detecting the distribution of the radioactive tracer in the living body.With the gradual maturity of clinical PET/CT technology,it is crucial for the early detection of clinical diseases,disease staging and efficacy evaluation.Long-axis PET is a trend in the development of current PET systems.An important advantage of the long-axis is to increase system sensitivity and reduce detector scanning time.Improving the axial FOV by sparseing the detector module is an important research direction.This paper mainly studies the image reconstruction effect of the sparse detection system of the regularization algorithm.The main work includes the following contents:The DKL-fTV algorithm has pretty good reconstruction performance for sparse detection system and can effectively reduce image degradation caused by sparse data.In this part of work.The algorithm principle is studied and the code is implemented to verify the correctness of the reconstruction algorithm programming.On this basis,the main parameters in the reconstruction are analyzed,including the influence of parameters on the reconstructed image,the estimation method of the parameter value and the influence of the parameters on the reconstruction convergence speed.Aiming at the slow convergence speed of DKL-fTV algorithm,the DKL-fTV algorithm improved by preconditon technology is deduced and realized.The convergence speed improvement effect of the algorithm is compared.The effect of precondition DKL-fTV algorithm on real data reconstruction is tested.Further study on the performance of regularized reconstruction algorithms under the sparse detector structure.In this paper,a variety of regularization reconstruction algorithms are implemented,and the detection data of different sparsity are reconstructed.The reconstruction results are compared and found that the DKL-fTV algorithm has the best effect of sparse data reconstruction,and the image quality better than other regularization reconstruction algorithms.Furthermore,the DKL-fTV sparse reconstruction performance is studied,and the DKL-fTV algorithm using anisotropic TV is proposed.The performance of the two DKL-fTV sparse reconstruction algorithms is poor in the simulation data and the actual data: when the sparsity is low Both DKL-fTV can reconstruct a satisfactory image;when the sparsity is very high,the anisotropic TV algorithm has better reconstruction performance.Through the reconstruction experiment of sparse detection system,it is verified that the sparse detector structure design has certain feasibility.PET sparse reconstruction has certain practical meaning.