The Development Of An Image Moment-based System For Low-dose CT Reconstruction

Since CT(X-ray Computerized Tomography,CT)can provide enough anatomical information of human organs and internal structures,it has become ones of the efficient tools for clinical diagnosis and therapy.However,a full C T scan will introduce ionizing radiation damage to patients,which will bring about some diseases,such as cancer,especially when patients are exposed to the high dose X-rays.Although reducing C T projections can decrease the dose of X-rays and relieve ionizing radiation damage to patients,it will degrade the reconstructed C T images at the same time.That would ultimately affect the clinical diagnosis.To enhance the quality of 2D CT reconstructed images when some of parallel projections are missing,this thesis proposes a limited-view 2D CT reconstruction method based on discrete Krawtchouk moments.This method uses the properties of discrete Krawtchouk moments and 2D Radon transform to establish the relationship between discrete Krawtchouk moments and 2D Radon transform.In the next step,this relationship is used to obtain discrete Krawtchouk moments from the known projections,and then the unknown projections can be estimated from the obtained discrete Krawtchouk moments.In this way,the missing CT projections can be completed and the quality of reconstructed images will be improved.The experimental results show that the proposed method can improve the quality of reconstructed 2D CT images when missing some of parallel projections.Besides,for the purpose of improving the quality of reconstructed 3D CT images when the cone-beam projections in some views are missing,this thesis develops a limited-view cone-beam CT reconstruction method based on 3D geometric moments.At first,it transforms the obtained cone-beam CT projections to known 3D Radon data.Then,the relationship between 3D geometric moments and 3D Radon transform is established,which will be used to calculate the 3D geometric moments from known 3D Radon data.After that,the unknown 3D Radon data can be estimated from the obtained 3D geometric moments.Finally,the 3D CT images are reconstructed by 3D inverse Radon transform directly.From the experimental results,it can be seen that even though cone-beam projections in some views are missing,the proposed method also can reconstruct good CT images.