Image Reconstruction For Low-dose CT And Sparse-view

With the wide application of CT imaging techniques in the clinic, more and moreattention is caused by X ray radiation. CT radiation can cause certain harm to human body,and increase the incidence of diseases such as cancer incidences. Therefore we need toreduce the dose of X-ray radiation as much as possible on the premise of ensuring theimage quality. In order to reduce the radiation dose, Naidich et al first puts forward theconcept of low dose CT in1990. When the other scanning parameters are constant,imaging with lower tube current can satisfy the requirement of diagnosis, meanwhile it canreduce the radiation dose.When the CT tube current is too low, the photon number received by the detector willreduce, and the reconstruction image will have serious noise interference, such as thestreak artifact. Therefore it needs to suppress the noise in the image. There are three kindsof denoising meathods, including filtering noise on the reconstruction image, or theprojection data, and improving the reconstruction algorithm. The second kind of method is widely used because of high efficiency, reconstruction image uniformity and isotropy. Inour previous research, we found the projection data noise is non-stationary and similarGauss distributed, and proposed a kind of penalized weighted least squares method. Afterfiltering the noise on the projection data and reconstructing the image with filtered backprojection algorithm, our method can suppress the noise effectively, and maintain a betterspatial resolution.In order to reduce the X-ray radiation dose, reconstructing CT image with sparseprojection caused widespread concern in recent years. In2006Candes et al studied theproblem of restoring the signal from little linear measurement data. They proved that itcan use total variation minimization approach to reconstruct the signal accurately from asmall amount of measured data, if the signal is sparse or can be representated sparsely.Sidky et al introduced the theory to CT reconstruction with sparse projections, andpresented a total variation minimization iteration algorithm. It minimizes the projectiondata errors and the image TV alternately, and obtains a better reconstruction effect.Considering the combination of the methods of imaging with low tube current andreconstructing CT image with sparse projections, it can further exploit the advantages ofboth methods, and reduce radiation dose. Therefore, we did the research on low tubecurrent imaging and sparse view reconstruction algorithm separately in this work, andcombined them together to further reduce the radiation dose. The main work is dividedinto the following several points:(1)Denoising and restoring low-dose CT projections by PWLS methodWe analysed the low dose projection noise properties obtained by the homegrown64-slice CT based on the noise model established in previous work, and denoised theprojection by PWLS method. Real data experimental results show that this method iseffective for the low-dose projection data obtained by the homegrown64-slice CT and itcan provide a possible way to reduce the radiation dose for the homegrown multi-slice CTimaging.(2)Reconstructing CT image with sparse views from the resored low dose projection In order to further reduce the radiation dose, we combined the TV minimization withthe PWLS restoration. We first denoised and restored the low-dose CT projection data byPWLS method, and then reconstructed CT image by sparse projections based on totalvariation minimization method. In order to evaluate the effectiveness of the proposedmethod, we validated this algorithm by Shepp-Logan model simulation experiment andreal phantom data experiment. The results show that, it can realize the function of CTimage reconstruction by sparse restored low-dose CT projection.(3)Researching a Micro-CT system with low-dose and large FOVAt the same time, for the further application of the low dose CT imaging technology,we designe and realize a low-dose cone-beam Micro-CT imaging system with wide-field.It is used for imaging the tissue structure of large animal in vivo, such as rabbit and rat.The preliminary experimental results show that, this Micro-CT imaging system is able toreconstruct the tissue structure of the scanned animal, and can realize the CT imagereconstruction with sparse low-dose CT projections.