| Computed tomography (CT) has been applied in the clinical diagnosis for some decades. Due to its superiority of the high resolution, the high sensitivity and the multi levels, CT is playing a more important clinical value than the traditional X ray examination. Cone Beam CT (Cone Beam Computed Tomography, CBCT) is a new technology which is applied to the clinical diagnosis at the end of the 20th century.It uses the X rays which are low energy and cone beam to revolve around the interested parts of patients with flat panel detector. It takes shorter reconstruction time and low radiation dose.The CBCT can display the diseased tissue structures and the normal tissue structures from the perspective of the three dimensional image in order to avoid the inherent defects of the two dimensional image.Such as the overlap of the organization structures in the image. So the CBCT has been used in clinical diagnosis widely.The reconstruction of image for CBCT system has been a hot topic in the field of the radiation. In the clinical practice,on the one hand,the technician need to increase the radiation dose and the scan range in order to increase the diagnostic accuracy of the patients. Because a clear image can help the doctors to find the small lesions.So it can decrease the occurrence of misjudgment. On the other hand,the technician need to reduce the radiation dose and the range of scanning.Because the radiation dose is bad for human’s reproductive system, nervous system and immune system. High doses of radiation will give rise to the cardiovascular disease, the diabetes, the cancer mutations, the abortion and infertility of pregnant women, the fetal malformation, and can directly affect the body tissue and bone growth of minors. Also it will bring about the decline of vision, memory and liver hematopoietic function.The severe case can lead to the retinal detachment. So you must to balance the relationship between the high radiation dose and the low radiation dose at the same time. In order to improve the quality of the CBCT image and decline the radiation dose in patient’s body,the method can be divided into the hardware technology,the scanning technology and the reconstructed technology.The hardware technology contain the ways of how to set the parameters.The scanning technology contain the ways of scanning and the time of scanning.The reconstructed technology contain the ways how to improve the reconstruction algorithm.In the paper,it is talking about the reconstruction technology of CBCT.The algorithm of the image reconstruction for CBCT system are divided into the analytical reconstruction algorithm and the iterative reconstruction algorithm.The analytical reconstruction algorithm is mainly consist of the FDK algorithm which is based on the filter back projection (FBP) algorithm.The FBP algorithm is used in the reconstruction of the parallel beam and the fan beam.The main process of the FBP algorithm is that.First of all, the projection data is transformed by the one dimensional fourier. After that, the result is multiplied by the transfer function of the slope filter. Finally, the result is transformed to the image by using the one dimensional fourier inversion. In the fact,the FDK algorithm is the three dimensional extension of the two dimensional FBP algorithm.It is used in the cone beam. The main process of the FDK algorithm is that. The first step, cosine function is used to weight the projection data.The second step, the projection data is filtered by the one dimensional filter in the horizontal line and in the different angle.The third step, along the direction of X ray,the projection data is transformed to the three dimensional image by the analytical reconstruction algorithm.Its advantage is that it can reconstruct image simply and fastly. When the projection data is sanning by the full range,the analytical reconstruction algorithm can reconstruct the images with the high quality.But if the projection data are sampled sparsely, the quality of the reconstruction image significantly decreased by using the analytical reconstruction algorithm.In the CBCT scanning, the scan field of view(SFOV) is determined by the X ray source and the flat panel detector array.Because the cost of the flat panel detector array is higher,at the same time, in order to reduce the radiation dose of patient, so the size of the detector is limited. Because the traditional scanning geometry structure is limited.So sometimes the detector can not completely cover the area of scanning. When there is a obesity patient and the small size of detector or the patient can not be placed accurately, the X rays beams are not sufficient to cover the patient’s scanning area, which is meaning beyond the SFOV. So the part of the area which is not covered by the X rays beams will not have the projection date.If it use the traditional analytical reconstruction algorithm directly to reconstruct,the reconstruction images will appear the truncation artifacts which is manifested as the bright ring artifact and the bright speck artifact.The truncation artifacts will decline the reconstruction image’s quality significantly and give rise to the reconstruction’s error.It will affect the doctor to check the interesting areas. The bright ring artifact mainly because the detector’s width is too short, which will result in the missing of both sides in the projection data. There is a bright ring around the SFOV in the reconstruction of image because the data deviation. To solve the problem of the truncation, Ohnesorge and other people proposed a image projection algorithm which name is the symmetric mirroring algorithm.The algorithm’s implementation process includes that.The first step, to find the value which is two times as compared with the boundary value of the truncated data. The second step, to use the symmetric mirroring algorithm to add the deviated values.The third step, to modify the data by the sine or cosine function.The fourth step, to reconstruct the image by the FDK algorithm.This algorithm can remove the bright ring artifact, but the position of the value which is two times as compared with the boundary value of the truncated data is uncertain. So the algorithm failed to remove the bright ring artifact in some reconstruction images. The bright speck artifact mainly because the detector’s height is limited or the objects’height is too long which will result in the missing data in the height of the projection data.If it use the FDK algorithm to reconstruct the missing projection data,the reconstruction images will appear incomplete or brightness difference at the top and bottom. To solve the problem of the truncation,Grass and other people proposed the hybrid reconstruction technology (HRT) algorithm. On the basis of the traditional analytical reconstruction algorithm,it uses the weighted projection to process the redundancy area of data, in order to make sure that the projection data is continuously and normalized, and guarantees the smoothness of border area. But the reconstruction algorithm is so complex and there is a other ring in the image beyond the SFOV.The iterative reconstruction algorithm is mainly divided into the algebraic reconstruction algorithm and the statistical iterative algorithm. The algebraic reconstruction algorithm is mainly divided into the iterative reconstruction algorithm (ART) and the simultaneous algebraic reconstruction technique algorithm (S ART). Its main idea is that the reconstruction image is obtained by solving the equations. The projection date which is obtained by the image is compared with the actual projection date which is obtained by detection.Then to correct the date to obtain the new image.The statistics iterative algorithm is mainly consist of the maximum likelihood expectation maximum algorithm (MLEM). Its main idea is that to use the poisson noise model as the target function to reconstruct image. When the projection data are sampled sparsely, the ART, SART and MLEM algorithm all can reconstruct image with better quality than the FBP and FDK algorithm, but they will take a long time. And the image’s quality which is reconstruct by the ART,SART and MLEM algorithm is still unable to meet the clinical requirements. In 2006, Sidky and other people proposed the regularization constraint algorithm based on total variation (TV),namely the ART-TV algorithm.The principle of the ART-TV algorithm is that. The use of the image’s sparseness as a prior information can improve the quality of image reconstruction. Under the condition of the sparse angle reconstruction have achieved good results. In 2008, Chen and other people, in order to further improve the image quality and reduce the radiation dose of patients, put forward the prior image constrained compressed sensing algorithm (PICCS). First of all, it should obtain high quality image as the prior image which is reconstructed by the same patient or other patient in the same structure and the same scanning parameters.After that,they use this prior image as constraints to reconstruct the interesting image. But the algorithm has a disadvantage. If the prior image’s geometric structure can get the good match of the reconstruction image,the algorithm can reconstruct good image with high quality. But when the prior image’s geometric structure has the position deviation with the reconstructed image, the reconstructed image will appear artifacts.This will lead to the decrease of the quality of the reconstructed image.To solve above problems, the paper made a basic research.In the first and second chapters, firstly, it reviewed the CBCT equipment composition and the basic knowledge of the CBCT image reconstruction.Then it study and implemented the analytical reconstruction algorithm and the iterative reconstruction algorithm. Final, it analyzed their advantages and disadvantages.In third chapter, first of all, in view of the bright speck artifact problem,the weighting reconstruction method is proposed. The first step, it need to calculate the time of the back projection of the FDK algorithm. The second step, add the weight distribution to the reconstruction image. The Shepp-Logan body model and the physical model are used to the image reconstruction. The result is that the bright speck artifact is disappear.The reconstruction image is more closer to the original image. And this method is more simple. After that,in view of the problem of the bright ring artifact problem, proposed the automation symmetric mirroring algorithm. The first step, to add projection data in each line by using the automation symmetric mirroring algorithm. The second step, using sine or cosine correction to correct the data. The third step, through the FDK algorithm for reconstruction image. The Shepp-Logan body model and the real data are using to the image reconstruction.The experimental results show that the algorithm can effectively eliminate the bright ring artifact caused by the transverse problem, and more close to the original image.In fouth chapter, in order to against disadvantages of PICCS method, it improved and realized a prior image constrained total variation regularization CT image reconstruction algorithm which can eliminate the effects of the position deviation and reduce the patient radiation dose and improve the SNR of the image and reduce the error. The concrete steps as follows:Firstly,the algorithm selects an image with good quality as the prior image which is similar to the to-be-reconstructed image and calculates the average pixel value of the prior image’s each homogeneous region as the priori information. Secondly,the projection data is reconstructed using the ART algorithm to obtain the intermediate image. The positivity constraint is added to it. Thirdly, the intermediate image is optimized using TV minimization combining with priori information. The above two steps cycle until the convergence criterion is met. Simulation experiments with the Shepp-Logan phantom demonstrate that the proposed algorithm is better than ART-TV and PICCS algorithm,significantry improving Signal to Noise Ratio and suppressing the noise, thus the image quality is improved.In this article, the existing problems in the analytical reconstruction algorithm and the iterative reconstruction algorithm is studied. Through the improvement and implementation of new algorithm can better solve the corresponding problems. But there also have some problems.Such as it didn’t have a large number of experiment data to verify our algorithm.it just do preliminary research. So in order to get the better result, then it need to be further research. |
PDF Full Text Request