The Denoising Algorithm Of Low-Dose CT Images Using Non-Local Information

The radiation hazard caused by the wide use of Computed Tomography(CT)has attracted more and more attention.It is found that high-dose Computed Tomography(LDCT)is harmful to human body.Therefore,the technology of Low-Dose Computed Tomography(LDCT)comes into being.Although low-dose CT reduces the harm to human body,reducing the radiation dose of CT scan will reduce the resolution and thus produce fringe artifacts,which will affect the diagnosis of doctors and cause misdiagnosis and medical malpractice in serious cases.Therefore,how to ensure the acquisition of high-quality images under LDCT scanning is the key to whether LDCT technology can really provide basis for doctors and patients.Improving the quality of LDCT image is becoming a hot topic in current research.In order to improve the image quality of low-dose CT,combining the non-local characteristics of image and noise,two kinds of post-processing algorithms are proposed to improve the image quality of LDCT.The details are as follows:1)We propose a fast denoising algorithm for low-dose CT images based on edge extraction and progressive non-local means.The proposed algorithm denoises the non-local mean information extracted by the edge and adds the processing of gradual filtering noise,which overcomes the problem that the traditional algorithm can not effectively eliminate the noise of low-dose CT graph while retaining its edge information.After finding the weighted 2-norm square value of all image blocks in the whole image,the weighted Euclidean distance is called each time.All the convolutions are calculated once through the Fourier transform,which solves the problem of long running time of the algorithm.2)We propose a fast low-dose CT image denoising algorithm based on multiscale non-local means.On the one hand,it improves the ability of the algorithm to recover the low-frequency information;on the other hand,the edge extraction progressive non-local filtering strategy is adopted to effectively protect the image details while denoising.In addition,using the Fourier transform,the fast algorithm is used to calculate the similarity of the image blocks to improve the running speed of the algorithm.Experiments were performed on real and simulated low-dose CT images and show that the algorithm can not only improve the quality of low-dose CT images,but also improve the algorithm by about 4 times.