Research On Dynamic Magnetic Resonance Imaging Based On Compressed Sensing

The dynamic MR imaging of time-varying objects is a series of continuous imaging operations.The imaging object is time-varying,so more rapid MR imaging technique is critical for dynamic MR imaging.In order to achieve the critical goal,there are two methods.The first is accelerating speed of scanning and shortening the time of data collection through improving hardware,which enables fast dynamic MRI.However,due to hardware constraints,the development of this method has entered a bottleneck period.The second is to reconstruct the image from under-sampled data,which reduces data acquisition time.The first method is limited by the hardware or other factors,so technique based on the second method of MRI has become the focus of recent research.If the signal has a sparse representation in a transform domain,using compressed sensing technology can make it recovery when the sampling rate is far below the Nyquist sampling theorem.MR cine has good sparse characteristics in both time and space direction.Therefore,the compressed sensing can be applied to dynamic MRI properly,which greatly reduces the amount of data collected and shortens the scanning time of MRI.Therefore,this paper studies dynamic MRI methods based on compressed sensing.The main research contents include the following aspects:Study on the basic principles of MRI,cognitive magnetic resonance imaging radically.Delving into compressed sensing theory,and familiar with the compressed sensing method in the use of magnetic resonance imaging.Delving into the traditional methods of off-line dynamic MRI based on compressed sensing theory,k-t SPARSE and k-t FOCUSS.And using multi-feature constrained compressed sensing theory to improve traditional methods.Compared with the traditional compressed sensing method,the multi-feature constrained compressed sensing can better describe the edge features of the image.And the fusion process of multiple feature sets is equivalent to smooth noise reduction,which make the method have good robustness and anti-noise ability.The experimental results show that the multi-feature constrained compressed sensing method can be applied to traditional dynamic MRI method,which can effectively improve the quality of reconstructed image.Research on real-time dynamic magnetic resonance imaging.According to thecharacteristics of cine frame difference image,we proposed a real-time dynamic magnetic resonance imaging method using elastic network model.Our method is based on the compressed sensing theory,adds adaptive weights constraint on the basis ofsparse constraint,and uses the elastic network model to solve quickly.Compared with the Kalman filter method,our method does not need to assume that the residual image is Gaussian white noise,and the adaptive weight can better describe the residual image structure.In this paper,simulated heart cine and real heart cine were used for the experiment.And the results show that our method has better reconstruction quality than traditional Kalman filter method in both cines.