| Magnetic resonance imaging(MRI)is widely used in medical practice as one of non-invasive imaging methods,this method is relatively safer than other medical imaging meth-ods(such as CT).However,MRI collects data for a long time and can’t deal with imaging time-related object.Laterly,people proposed parallel MRI,named as PMRI(parallel mag-netic resonance imaging).PMRI samples K-space data simultaneously through multiple coils,so the data acquisition time is reduced.In PMRI,each coil collects incomplete K-space data,so the sampling data of each coil can’t recover the image independently.PMRI reconstruction methods include image domain methods(such as SENSE),K-space domain methods(such as GRAPPA)and a combination of both.Because the image domain method can effectively use the prior information of the reconstructed image,the method has attracted extensive attention and research.Owning to the characteristic of PMRI and Fourier,in K-space,high-frequency area’s cofficient mainly reflect image noise,the low frequency area’s mainly reflects the contour of image.We propose adaptive PMRI reconstruction models on the foundation of data characteristic.In the data fitting terms of the model,high and low frequency areas use different weight factors as fitting coefficients.In the low frequency area,the weight factor is large,in the high frequency area,the weight factor is small.We also derive the corresponding primal dual algorithm to solve the model.Sensitivity map plays a crucial role for the image domain method.The accuracy of the sensitivity map affects the qualities of the reconstructed image.The sensitivity map has different estimation methods,but the estimated sensitivity map usually con-tains errors,which affects the qualities of the reconstructed image.We proposes PMRI reconstruction model which is based on sensitivity map regularization,making sensitivity map be corrected during the reconstructed image.We also derive proximal alternating minimization algorithm to solve the model and analyze the convergence of the algorithm.Finally,we experient by different PMR.I data.The experimental results show that the reconstructed images of adaptive PMRI model proposed in the dissertation has higher qualities,and reconstructed images of the PMRI model which is based on sensitivity map regularization can make sensitivity map be corrected and improve the reconstructed images’ qualities further. |
