The Study And Realization Of Translational Motion Artifacts Correction In MRI

Magnetic resonance imaging (MRI) technology is one of the most important tools forthe medical clinical diagnosis and research. However, due to the relatively long time inmagnetic resonance imaging process, the imaging results are vulnerable to motion effects.Artifacts would be produced due to patient movement, which degrade the quality of theimage and affect the clinical diagnosis seriously. Therefore, the study of the motion artifactsuppression method has an extremely important significance.The basic principles of MRI and the causes of motion artifacts in MRI are analyzed inthis paper. We summarize the existing motion artifact correction methods. These methods canbe broadly divided into two categories, including real-time artifact suppression methods andpost-processing artifact correction methods. We focus on the post-processing artifactcorrection methods.Firstly, we give a detailed analysis of the artifact correction method using phaserecovery algorithm with frequency domain constraints based on the extraction of region ofinterest (ROI). The image of ROI and the image with no motion have a certain degree ofsimilarity in the frequency domain. The offset of motion can be estimated by the phasedifference between the image of ROI and the image with motion effect. During the correctionprocess, the accuracy of the ROI extraction directly affects the artifact correction effect.Combing the snake algorithm with artificial extraction of ROI, we can get relatively accuratearea of the interest. Existing research are almost based on simulation.We sample the actualimaging data, and observe the effect of motion artifact correction. Experimental results showthat after correction, the image with the motion of the phase direction can be significantlyimproved, and its contour and internal details become clear. It becomes more conducive toobservation. The image with the motion of the frequency direction is also improved. Butcompared to the correction of the image with the motion of the phase direction, it is lesseffective.The other two post-processing correction methods are also analyzed in detail. Weanalyze the correction principle of each method, and the problems encountered in the test.These can provide reference and basis for the subsequent research of post-processingcorrection method.The spectrum shift method is based on that the displacement of the imaged object in thefrequency encode direction can also produce a displacement in x direction in themixed-domain signal. This method theoretically has effect in the correction for the pixel level offset. In experiments we find that it is sometimes difficult to find out the offset ofmixed-domain boundary, the subsequent correction can not be performed.In the correction method based on sub-band contour tracing, the k-space data is dividedinto several sub-bands. Through the sub-band image contour extraction, we can determine thecenter position of the sub-band image, and then we can obtain the offset of the movement. Itwas found that the sub-band low-resolution images are very vague, and can not be used totrack the movement of an object.In this paper, the MRI artifact correction algorithms are realized and a platform iscreated under the MATLAB GUI environment. By the platform, we can read the data, selecta correction method, observe the correction effect, and quantify the correction error. Itprovides a reference for the study of new algorithms, which can be added into the platform.