Parallel Magnetic Resonance Imaging Based On Partial K-Space Data

Magnetic resonance imaging (MRI) is a human body imaging technology, which is based on the principle of nuclear magnetic resonance (NMR) and can be used to produce high quality images with the functional, structural and lesion's information of an organism. Nowadays, it has been widely used in clinical diagnosis. However, conventional MRI is a time consuming technique for its slow sampling speed limited by Nyquist sampling theorem. So, it can't be used on some special applications such as real-time cardiac imaging. How to accelerate the MRI speed is a hot and difficult topic in this area. To address this issue, after the systematic study of the existing fast MRI methods, the parallel MRI (pMRI) based on the partial K-space data is investigated in this thesis. The main works are described as follows:Firstly, the pMRI based on the partial Fourier (PF_pMRI) method is investigated to improve the MRI speed on the condition of maintaining the reconstruction quality. Based on the advantages of PF and pMRI respectively, this paper combines the PF with pMRI methods together to further accelerate the MRI speed. Both 2D and 3D experimental results show that the PF_pMRI method can not only provide better reconstruction than pMRI method, but also can overcome the disadvantage of low SNR in high reduction factor for the pMRI method.Secondly, the performance and the applicable condition of the two methods PF_GRAPPA and 2D GRAPPA were quantitatively analyzed in 3D imaging, which could provide a useful guide for their clinic applications. The analytical results for g factor map, noise standard deviation and artifact power of a series of 3D images show that the PF_GRAPPA method has obvious advantages over 2D GRAPPA. The PF_GRAPPA can be used to realize thin and thick slab 3D imaging while the 2D GRAPPA method is only suitable for thick slab due to its reconstruction sensitivity to the thickness change.Finally, the method of pMRI based on compressed sensing (CS_pMRI) was studied. According to make good use of the signal sparsity and space sensitivity information, the MR data can be further down sampling in random from the skipped sampled pMRI data, then the reconstruction can be obtained by TVl1_l2 iteration algorithm and parallel reconstruction sequentially. Experimental results show that the CS_pMRI method can accelerate the MRI speed and maintain the reconstruction quality.