Coronary Vessel Wall Magnetic Resonance Imaging Based On Sparse Sampling

Coronary artery disease(CAD)generally refers to the accumulation of plaque in the inside layers of the arteries.The plaque will slowly narrow the flow of blood through the vessel,and the myocardium it supplies will not get enough blood.CAD is becoming one of the major causes of death in China.Magnetic resonance imaging(MRI)is one of the most powerful imaging technologies,which is widely applied in both clinical diagnosis and research settings.MRI is a non-ionizing radiation imaging modality with excellent soft tissue contrast.MRI has important features that it provides multi-parameters,multi-orientation and high resolution imaging.The research on the application of MRI in coronary vessel wall imaging is active recent years.However,as the coronary arteries are rather small,high-resolution and large-field-of-view(FOV)MR imaging is required to examine the coronary vessel wall.In addition,cardiac motion and respiratory movement limit the acquisition window during the cardiac cycle,which can lead to a low sampling efficiency and extended scan duration.Long MR scan causes the patient discomfort and may alter the patient's heart or respiration pattern,resulting in a low successful rate of coronary vessel wall MR imaging.To address these issues,a new technique concerning pulse sequence and image reconstruction on coronary vessel wall MR imaging is proposed in this work.In respect of pulse sequence,recent research on three-dimensional(3D)coronary dark-blood interleaved with gray-blood(cDIG)MR imaging has shown potential on the imaging of coronary vessel wall.However,this technique suffers motion sensitivity and long scanning time.In this work,we applied a stack-of-stars sampling pattern with tiny golden angle radial trajectory instead of the 3D Cartesian sampling trajectory,so as to improve the robustness to motion artifact and increase the sparsity of the k-space data.In order to reduce the scanning time,a straight method is to reduce the amount of sampling data while accurately reconstruct the vessel wall.Based on the theory of compressed sensing,a novel block-weighted total variation(BWTV)regularization is presented in this work.The BWTV regularization is an extension of the conventional total variation(TV).In BWTV,the reconstructed image is divided into two blocks according to the position of the vessel wall.Different penalty weights are then given to these blocks.Therefore the walls can be protected from over-smoothing,and in the same time the influence of high-level noise data points can be alleviated.In order to further improve the reconstruction quality,parallel magnetic resonance imaging(pMRI)and multi-contrast imaging can be combined in the image reconstruction modality.Based on the combination of compressed sensing and JSENSE,we utilize BWTV regularization to exploit the sparsity of coil images instead of the output image,to exploit more priori information of pMRI.Besides,we use the coil sensitivities between the dark-blood and gray-blood images as sharable information during the reconstruction.Therefore the k-space data from multi-contrast images can be jointly utilized in the same linear system while only one additional unknown parameter is added.The reconstruction quality can be highly improved with the proposed method in consequence.In this work,simulation experiment and in vivo experiments have been conducted to verify the performance of the proposed method.The experiment results indicated that the image is more robust to motion artifact and the data is sparser by applying the proposed imaging sequence.The coronary vessel wall is clearer and sharper in the reconstructed image while the influence of noise is alleviated by the usage of the proposed BWTV regularization.And with the sharing of the coil sensitivity between dark-blood and gray-blood images,the reconstruction quality of the vessel wall is further improved.In summary,the proposed technique has a great potential in improving coronary vessel wall MR imaging.