Research On Dynamic MRI Algorithm Based On Artificial Sparsity

Temporal resolution and spatial resolution are very important for the clinical diagnosis of dynamic magnetic resonance imaging,contributing to a number of rapid imaging methods.As an extension of k-t broad-use linear acquisition speed-up technique(k-t BLAST),k-t principal component analysis(k-t PCA)is a classic algorithm with high temporal resolution and high spatial resolution,which has been widely applied to the dynamic fast imaging with wide bandwidth,such as free-breathing myocardial perfusion and cerebral perfusion.It uses the principal component analysis technique to extract the temporal and spatial correlation as the temporal basic function,and it fundamentally changes the inverse problem of signal reconstruction and improves the temporal fidelity of the reconstructed image.However,as the acceleration factor increases,the reconstruction result of k-t PCA will generate increasing noise and intolerable residual aliasing artifacts,which greatly affect its applications in fast imaging.In recent years,reconstructions of sparse data has been applied in many areas of dynamic parallel imaging.Compared to the reconstruction of original data,better image quality can be obtained by applying the reconstruction algorithm directly to sparse data.Therefore,in view of the above problems of k-t PCA,we carry out the research on dynamic MRI algorithm based on artificial sparsity,proposing a sparse k-t PCA reconstruction technique,which aims to further improve the reconstruction performance of k-t PCA by combining the artificial sparsity.By removing the original reconstruction error from spare reproduction,it can realize the improvement of image temporal resolution and signal-to-noise ratio,and can obtain higher image quality in the high-speed accelerated reconstruction.In addition,based on the existing residual k-space sparsity method,we propose the residual k-space k-t PCA algorithm as a comparison.In this paper,we reconstructed the original k-t PCA and the two k-t PCA improved algorithms by using the simulated data and the in vivo human data to verify the correctness of the theoretical method.We also use the mean square error,signal-to-noise ratio and signal intensity time curves individually to describe the performance of these reconstruction methods.With different under-sampled data,we compare the advantages and disadvantages of these methods and explore the more excellent algorithm for perfusion imaging and other dynamic imaging.