Low-dose CT Imaging Via Noise Reduction In Sinogram

Computed tomography is to obtain an image of the cross section of an object.The quality of CT reconstruction image is closely related to the dose of X ray radiation,the higher the dose,the better the image quality,but the excessive X ray radiation can cause serious harm to human beings.Therefore,in order to ensure the image quality while reducing the dose of X ray radiation,the development of CT technology has become an important goal and has important scientific research value and clinical value.(1)For the noise of projection data,total generalized variation minimization method is proposed about projection data.First of all,the nonlinear Anscombe transform is used to transform the projection data satisfying the Poisson distribution into the approximate Gaussian distribution.Then,the transformed data are denoised by the total generalized variational regularization model.Finally,the Anscombe transform is used to implement the traditional filtered backprojection CT reconstruction.The numerical results show that the proposed method has better performance in suppressing noise and strip artifacts.(2)For the Poisson noise of gray images,weighted kernel norm minimization method is proposed for the Poisson noise images denoising.First of all,the image containing Poisson noise is transformed into Gaussian type data by Anscombe transform.Then,the Gaussian noise is denoised by the optimization algorithm.Finally,the denoised image is obtained by inverse transform.The experimental results show that the weighted kernel norm method is superior to the Poisson noise image.(3)For the low dose CT image reconstruction in spatial domain,a semi analytical method is proposed by this paper.In this method,the total variation regularization model is used,the problem of solving the spatial domain is transformed into the frequency domain,and the convergence of the algorithm is analyzed.The experimental results show the effectiveness of the proposed method.