Research On Low-dose CT Image Restoration Based On Sparse Rspresentation And Dictionary Learning

With the continuous development of computer technology,CT imaging technology has been widely used in clinical disease diagnosis and treatment,has become the first choice for the diagnosis of brain disease.However,the amount of X-ray dose used in CT imaging scans increases,increasing the risk of disease induced by subjects,while low-dose CT can reduce the risk of disease.However,low dose also causes a decline in image quality,so it is of great significance to study the recovery of low dose CT images.Because sparse representation and dictionary learning are applied to the problem of image denoising and recovery due to its excellent characteristics,it is of great significance and value to apply this method to low-dose CT image problem.In order to solve the problem of low dose CT image degradation,this paper has done the following work:First,the low-dose CT image restoration studies based on the sparse representation of the dictionary learning(MOD,K-SVD,OLM,FDL-PG)show that the FDL-PG algorithm is superior in visual perception and objectivity to other algorithms Good adaptability,convergence speed,but there are still some noise and lack of information and other issues.Then,in view of the above problems,this paper proposes two improved low-dose CT image recovery methods based on sparse representation and dictionary learning.One method is to perform PCA for low-dose phantom and clinical brain CT images,and to perform dictionary training(FDL-PG)and denoising with reduced dimension data.The method(FDL-PG-PCA)enhances the denoising performance,but there are still a few details that are missing.Another method is to first low-dose phantom and clinical brain CT images of BM3 D filter processing,and then filter the data for dictionary training(FDL-PG)and denoising,the method(FDL-PG-BM3D)on the details The information remains well.Experimental results of phantom and clinical CT images show that the two methods have higher denoising performance and further suppress the noise.These two methods in clinical use,is expected to ensure that the doctor diagnosed accurately while reducing the patient receiving X-ray radiation dose.