| Because of its non-ionizing radiation and multi-parameter imaging and other advantages, Magnetic Resonance Imaging has been widely used in clinic. But there are still shortcomings of the long scanning time. In recent years, there have been many approaches to improve the imaging speed, including rapid pulse sequences, half Fourier techniques, parallel imaging and so on. Although these methods largely accelerated the speed of Magnetic Resonance Imaging, but it still can not meet the clinical requirements. The new Compressive Sensing theory, breaks through the limitations of traditional sampling theorem, only needs to collect a small amount of data, and then uses the nonlinear reconstruction algorithm to obtain the original signal. At present, the field of Magnetic Resonance Imaging in Compressed Sensing is mainly based on simulation of image reconstruction algorithm. In order to be applied in engineering, two problems must be solved: the pulse sequence design and the online reconstruction. The main work of this paper is to solve the problem of the sequence design of Compressed Sensing Magnetic Resonance Imaging technology, so as to realize the imaging mode in the MR scanner.Firstly, we study the principle of Magnetic Resonance Imaging and the Compressed Sensing, explore the conditions that the Compressed Sensing theory applied in Magnetic Resonance must be met. The key of applying this theory to pulse sequences is how to realize the under-sampling of the K space data. There are many ways to sample and under-sample. Taking into account the actual instrument hardware conditions, sample trajectories, and under-sampling in frequency encoding direction cannot save the scanning time; we select the random sampling mode in phase encoding direction of the Cartesian trajectory.Secondly, we study the design of the principle of imaging sequence from the RF pulse, level selection, phase encoding and frequency encoding in detail. And then analysis the design of SE sequence based on Compressed Sensing, and indicate a detailed research of the phase encoding of variable density sampling design. Also, according to the clinical application, we introduce of the principles and methods of the water fat separation, and design the water fat separation sequence, then make the application of compressed sensing in the pulse sequence.And then, we use the PPL language to program the Compressed Sensing Spin Echo pulse sequence and Compressed Sensing Water-Fat separation sequence in the 0.35 T permanent magnet scanner of Shenzhen BASDA Company. And we debug and test the sequences. We select imaging parameters, set several different sampling factor and different sampling module, acquire the water phantom, human head transverse, and knee sagittal full sampling and under-sampling data. We use Fourier reconstruction to the full sampling data and as a control image. We use the Compressed Sensing reconstruction algorithms of our laboratory written in the VC6.0 environment based on the nonlinear conjugate gradient descent method, to offline reconstruct the under-sampled data. The results of experimental data analysis show that the sequence has reached the requirements of Compressed Sensing, the best under-sampled acceleration factor of head and knee is 2.5. At the same time, the image quality can be accepted, and is close to the full sampling image, but the acquire time is shorten to 1/2.5. It greatly saves the data acquisition time.In this paper, the MRI pulse sequences based on Compressed Sensing is realized; which can improve our level of Research and Development in the MR scanner. |
PDF Full Text Request