The Study For Computed Tomography Based On Iterative Algorithm And Compressed Sensing

Medical Imaging Technology has been an important means of diagnosis of the disease, but also people first engineering methods used for medical diagnosis.Just three decades, computed tomography has made great progress. CT reconstruction algorithm for conventional projection data requirements are very strictly, need to complete the projection angle and lower sampling interval. In order to improve signal to noise ratio of data collection, in practice often need to increase the radiation dose. Radiation dose for X-ray has strict requirements in clinical, but rather the reconstruction of sparse projection data quality is often not good enough. This study attempts to solve the sparse projection data and projection from the point of view is incomplete, uneven redevelopment projection data are high quality reconstructed image.This study fund by PKU College of engineering - School of Basic Medical joint Molecular Medical Imaging Laboratory related subject. Based on the previous research, using filtered back projection (FBP) algorithm, Algebraic Reconstruction Technique (ART) and TV + ART (TV: Total Variation) algorithm to reconstruct the experimental data which acquisition from the multi-modal molecular imaging system from the following aspects:1. Fast two-dimensional ray traversal algorithm for pixel. The line traverses pixels for general intersection problem of slow process of calculation, a straight line traversing the intersection pixels and fast algorithm. In the algorithm, a distance parameter to determine the ray through the pixel index and calculate the length of the ray through the pixel. Incremental calculation of the distance parameter, operation efficiency is very high. ART algorithm is also applied the algorithm to calculate weights matrix and the projection process, and using the general line of pixels traversing the intersection algorithm ART algorithm, greatly shorten the reconstruction time.2. Reconstruction of real data by TV+ART algorithm. In this paper, on the Shepp-logan model and simulation results of self-built model based on the PET, CT real data reconstruction of projection data. The results show that, TV + ART algorithm for edge information well, regular shape reconstruction with a reconstruction of the fault cross-section of high precision, fast convergence characteristics of the edge, compared with the conventional reconstruction algorithm for tomography with sparse projection data for CT reconstruction can be better The reconstruction quality and reconstruction accuracy, TV + ART PET image reconstruction algorithm is applied to obtain the ideal reconstruction quality.