Improving Quality In Low-dose Myocardial Perfusion CT Imaging Based On Redundancy Information And Tensor Total Generalized Variation

Coronary artery disease(CAD)is caused by coronary artery atherosclerosis which induces stenosis and occlusion,and leads to myocardial ischemia.Recently,the incidence of CAD is on a rising tendency,the early and accurate examination of CAD is significant for clinical diagnosis,treatment and prognosis.Myocardial perfusion computed tomography(MPCT),as a non-invasive functional imaging technique for CAD,has received more and more attention.With the mass intravenous injection of iodine contrast agent,MPCT can acquire time density curve(TDC)of every pixel in region of interest(ROI)by scanning a single slice of myocardium in a cine mode.TDC reflects myocardial density changes caused by iodine contrast agent when it pass through myocardium.So we can combine different mathematical models and TDC to calculate myocardial perfusion hemodynamic parameters including myocardial blood flow(MBF),myocardial blood volume(MBV)and mean transmit time(MTT).Myocardial perfusion hemodynamic parameters reflect perfusion changes of myocardium so that we can locate and grade myocardial ischemia.However,compared with conventional computed tomography(CT),MPCT increase the X-ray radiation dose for patients due to its continuous dynamic scan protocols.Excessive radiation exposure would induce cancer,leukemia and the other genetic disease.At present,lowering kVp,milliampere-seconds(mAs)or decreasing the number of X-ray projections per rotation are always utilized to reduce the radiation dose in the clinic.But with the reduction of radiation dose,the quality of MPCT images are degraded by noise induced artifact,which hampers the diagnostic accuracy.Therefore,how to reduce the radiation dose without degrade MPCT images quality has become one of important research topics in the CT field.Recently,various investigations for the low-dose MPCT imaging has been explored.There are two categories about projection data researches.The first one is to restore the ideal integral projection data from acquired noisy data by building the projection restoration model based on the noise distribution characteristics in low-dose MPCT imaging.The second one is accurately modeling the statistical properties of low-dose projection data and utilizing iterative reconstruction algorithm to obtain high quality MPCT images directly in projection domain.Among reconstruction algorithms,statistical iterative reconstruction algorithm is about to describe CT system by building an accurate model based on photon statistics,which usually consist of data fidelity term and regularization term in the objective function.The data fidelity term can model the statistical measurement,which is the base and precondition for high quality reconstruction.The regularization term introduces prior knowledge and penalize the solution,which has important significance for high precision image reconstruction.In MPCT imaging,perfusion information changes continuously by time and anatomical structure basically remain the same.MPCT images have redundant structural information within frames and highly similarity between adjacent frames.Therefore,we can reconstruct high quality images from low-dose projection data by taking advantage of intrinsic properties and prior of MPCT image.Based on the information above,the main work in this paper can be summarized as follow:(1)Redundancy information induced image reconstruction for low-dose MPCT is proposed.Except the perfusion information changes caused by iodine contrast agent,adjacent frames in MPCT sequence images have high structural similarity and the noise obedient to independent random distribution.Therefore,to reduce the random noise within frames,an average image is generated from adjacent frames of the current frame image and introduced as reference image into nonlocal means(NLM)framework.Combining average-value image induced NLM(aviNLM)with total variation(TV),a penalized weighted least-squares(PWLS)model incorporating aviNLM and TV based hybrid constraints is utilized to reconstruct MPCT images,which is referred to as PWLS-aviNLM-TV for simplicity.The PWLS-aviNLM-TV algorithm can effectively improve the quality of low-dose MPCT images by taking into account the similarity between adjacent frames and redundancy information within frames.(2)A low-dose MPCT images reconstruction algorithm based on tensor total generalized variation(TTGV)is proposed.Considering the spatial correlation of the myocardial structure and the temporal continuity of contrast agent intake during the perfusion,we incorporate TTGV regularization into the PWLS framework,which is referred to as PWLS-TTGV for simplicity.The proposed algorithm can remove staircase artifact and remain edge structure information by TTGV regularization which takes full advantage of the higher-order derivatives.The Chambolle-Pock(CP)algorithm is utilized to optimize the objective function.