Dissertation
Dissertation > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Information processing (information processing) > Pattern Recognition and devices > Image recognition device

PROPELLER MRI -based sampling affine motion artifact reduction method

Author LiuXiaoWu
Tutor ChenWuFan;FengYanQiu
School Southern Medical University,
Course Biomedical Engineering
Keywords Magnetic Resonance Imaging PROPELLER Motion artifact correction Affine transformation Affine motion estimation
CLC TP391.41
Type Master's thesis
Year 2010
Downloads 37
Quotes 2
Download Dissertation

MRI (Magnetic Resonance Imaging, MRI) may noninvasive conditions in the human body or other biological structure and function in vivo imaging, it has a high image resolution, image many parameters, faults can be in any direction, the human body without ionization radiation damage and so on. Therefore, magnetic resonance imaging can be widely used in clinical practice and clinical and scientific research become increasingly important imaging method. However, magnetic resonance imaging data acquisition process due to relatively long time, the patient's autonomy or independence movement is often difficult to avoid, which resulted in the formation of the final reconstructed image artifacts, thus affecting the doctors to make a correct diagnosis . Uncertainty due to movement and movement is difficult to obtain a priori knowledge of motion artifact correction will be very difficult, seriously affect and hinder the development of magnetic resonance imaging technology and applications. Therefore, how to effectively eliminate motion artifacts on the impact of magnetic resonance imaging, has attracted wide attention from scholars at home and abroad, is currently in the field of medical imaging research focus. In this paper, familiar with and master the principles and techniques of magnetic resonance imaging, based on sampling methods and reconstruction algorithms PROPELLER conducted in-depth research. PROPELLER method for rigid motion artifact elimination with good results, and magnetic resonance imaging in the head to get a successful application. But generally only in the head rigid motion imaging, imaging other parts of the body is often accompanied by varying degrees of soft tissue stretching deformation. For this soft tissue deformation must be based on non-rigid motion model to accurately describe and make corrections. PROPELLER Methods Although to some extent related weighted eliminated some rigid motion artifacts outside, it does not describe the non-rigid motion effectively, and thus of very limited. Therefore, this article rigid motion artifact reduction algorithm based on PROPELLER method is proposed based on the affine motion artifact reduction algorithms. The algorithm PROPELLER sampling for each K-space through the inverse Fourier transform of the temporary image reconstructed by the affine motion model based image registration algorithm to obtain the non-rigid motion information, and then the affine transformation frequency domain properties, the use of The resulting estimate affine motion parameters on PROPELLER sampling space of each K-correction, and finally put all the corrected K-space grid reconstruction of combined final image can be obtained. Simulation data and real data experimental results show that, compared to the existing PROPELLER reconstruction algorithm for rigid motion and affine motion artifacts caused all have a good correction. To improve the affine motion parameter estimation accuracy and stability, this paper presents a frequency domain phase correlation algorithm based on affine parameter estimation method. Original method for solving affine motion parameters directly to a third-order unit matrix as the initial value substituted into the nonlinear least squares, because the affine transformation parameters are more likely to correspond to the same kind of deformation of a variety of parameters combinations, So bad initial selection will directly affect the convergence result of uncertainty, while the new method first proposed the use of frequency-domain phase correlation algorithm derived for each K-space of the rigid motion parameters, and then the rigid motion parameters as the initial value is substituted into the affine motion parameter estimation which, after motion compensation performed by the grid reconstruction of the final result. Experimental results show that the proposed new method for the affine motion parameters estimation accuracy, better stability for affine motion artifact elimination method is better than the original.

Related Dissertations
More Dissertations