Research On Low Dose Dynamic CT Myocardial Perfusion Based On Motion Correction

CT myocardial perfusion imaging is an effective method for the diagnosis and treatment of coronary heart disease.However,during CT myocardial perfusion imaging(CT-MPI),some spatial displacement caused by the motion of the human heart will affect the quality of the reconstructed image.Moreover,CT-MPI needs continuous enhancement scannings the region of interest,which greatly increases the radiation dose received by the patient and causes certain adverse effects on the patients.For these two problems,many domestic and foreign researchers have conducted various preliminary researches on low-dose CT myocardial perfusion imaging.Tao et al.firstly applied the statistical iterative reconstruction technique to myocardial perfusion CT image reconstruction,and effectively reduced the streak artifacts induced by image noise.Bian et al.developed a motion-adaptive sparse prior model under penalty weighted least squares framework to improve the image quality of low-dose myocardial perfusion CT.The common solution is to introduce motion correction into the reconstruction process to reduce the influence of motion,make full use of prospective ECG gating techniques,optimize scanning protocols,optimize KVp and mAs,and adopt advanced image processing and reconstruction techniques to reduce patients radiation dose received.This work is also focusing on the above two problems.In this work,a registration algorithm based on structure image representation and optical flow model is proposed for motion correction in CT-MPI.The motion correction technique and the tensor total variation regularization are introduced into the framework of penalty weighted least squares(PWLS).A new reconstruction model has been built into the energy function of CT image reconstruction.The quantitative and qualitative analysis on the results of the XCAT phantom simulation data and the preclinical pig data were performed,the results show the effectiveness of this method.The detailed work is described as followsing:(1)A registration algorithm based on structural image representation and optical flow model is proposed.First,we use Shannon entropy to calculate the entropy value of the specified size field around all pixels in the reference image and the floating image.And then,the calculated entropy value is taken as the gray value of the corresponding point in the entropy image to generate the corresponding two entropy images.Finally,the two entropy images are registered using the optical flow field model registration algorithm.The experimental results show that the proposed method has higher registration accuracy than the optical flow model registration algorithm and the residual complexity registration algorithm.(2)A reconstruction algorithm based on motion compensation and tensor total variation regularization(MCTTV)is proposed.This algorithm considers the gradient and the spatial-temporal structure sparse characteristics of the difference map of CT perfusion images of adjacent frames after correction and introduces the process of heartbeat correction into the penalty weighted least squares algorithm to improve the image quality of CT-MPI.We refer to this method as PWLS-MCTTV.Finally,we use the steepest descent optimization algorithm to optimize the objective function.Experimental results show that the proposed method can effectively improve the quality of reconstructed images compared with the filtered back projection reconstruction algorithm(FBP)and the PWLS reconstruction algorithm based on tensor total variation regularization(PWLS-TTV).