Research Of Directly 3D Reconstruction Algorithm In Cone-beam CT

Since the birth of the first CT (computed tomography) scanner in 1972, CT has been widely used in the world. The wide uses of CT also generate great impetus for CT research. Several big changes have taken placed during the past 30 years. These changes mainly involve in two aspects- speed and image quality.The CT image is calculated from projection data by computer. So there are two steps to get image—using X-ray to measure projection data and calculating projection data to get image. There are three kinds of X-ray measure structures, parallel-beam, fan-beam and cone-beam. But the parallel-beam can not be realized directly in engineering. Compared with fan-beam, cone-beam projection system can gain more information in one projection, so it is of advantage to increasing scanning speed and image quality. But the relationship between cone-beam projection data and the cross section image is very complex. Accordingly the reconstruction algorithm and scan frame becomes more complex. In recent years, many scholars have researched the cone-beam scan technology, but the algorithms they developed are not very perfect, so the real commercial cone-beam CT is not appeared today. In order to solve the problem in cone-beam CT, Four main questions of cone-beam CT have been researched in this thesis.1. An intermediate function algorithm, which can be used in cone-beam CT, have been researched. It is a kind of exact resolution algorithm which is based on Fourier space, and at same time it is a kind of back-projection algorithm. The algorithm can be decomposed into three steps. Firstly, a partial Fourier transform is applied to the projection data and an intermediate function is got. Secondly, the same changes to the intermediate function are made and a form of inverse Fourier transform of object function is got. Finally, the image can be reconstructed by making back-projection to the new form of intermediate function. To verify the algorithm, some experiments have been done and the results are satisfied.2. Direct 3-D reconstruction technology has been researched. The using of 3-D image becomes more and more important in medical field. It can help doctor to findan exact place of focus and to make plan for operation. Three-dimension image of traditional CT is made from 2-D image by late processing. The main method of processing is an interpolating calculation. Because of no enough sampling in z axes, the image space resolution of traditional CT is low. Traditional CT also has motive artifact which is resulted from low scan speed. These restrictions make the quality of 3-D image of traditional CT not satisfied. Direct 3-D reconstruction is done by using a lot of cone-beam projection data. A 3-D image can be got by directly calculating the value of 3-D space points. The 3-D points which are gained by direct reconstruction have same resolution in three directions, so the 3-D image which is based on these points has a same nature in all direction.3. By Researching data sufficient condition of VOI reconstruction, a new scan framework is designed to adapt the algorithm. The new framework is named double cone-beam three axes scan. The idea of three axes scan is an extending of spiral scan. The sample density in z direction will decrease when the pitch increases in fast scan. Double cone-beam framework, which can increase the sample density in z direction, uses two different focuses to produce two cone-beams in different place. Two focuses turn on alternation to avoid overlap of projection data. To fulfill this work mode, a kind of double anode tube is put forward in this paper.4. A new scan method which can be used in heart scan is designed. The scan structure is a double cone-beam three axes scan. General CT can not get heart's stationary image, because the heart is a fast moving organ. Three-dimension image of heart can be got directly by combining double cone-beam three axes scan technology and direct 3-D reconstruction algorithm.In this thesis, some other cone-beam scan reconstruction algorithms are also analyzed. These algorithms can be sorted by exactness, the length of object and scan mode. Every algorithm has its' virtue and limitation. Algorithms which are discussed in this paper are FDK, Grangeat, PI and PHI etc.