Correcting corrupted-phases by blood perfusion, diffusion and body motion in MRI imaging.

在摄取小流量的灌注或弥散图像时，由于身体运动，容易产生伪影。在这项研究中，我们开发了伪影校正算法的平移，旋转和膨胀收缩的身体运动。(1)平移刚体运动：利用第n个视图（相位编码）中MRI信号f（x，n）的傅立叶变换F（x，n）显示对象沿频率轴（x）沿着的位置这一事实，我们通过身体运动校正了损坏的相位。我们还开发了一种方法，它可以独立于x轴校正身体运动alosng y轴。(2)旋转刚体运动：对于旋转中心和旋转角度未知的步进旋转，首先估计旋转角度，使得重建图像外部区域的能量最小。然后采用相位恢复算法来消除旋转中心偏离原点所带来的相位干扰。(3)膨胀-收缩可变形体运动：在例如腹部在x轴和y轴上成比例地膨胀和收缩的情况下，我们已经表明，我们可以通过以下方式减少伪影：首先从频率轴上的投影获得与（1）相同的x方向运动，然后校正对应于y的相位。这些算法也适用于儿童和老人因身体运动而不能安静地躺下拍摄几到几十张图像时所引起的伪影校正为了使算法更接近实际应用，对更多自由度、更灵活的运动进行建模是未来的任务。

When taking perfusion or diffusion images of small flow, artifact is easily caused by body motion. In this research, we developed artifact correction algorithms for translational, rotational, and expanding-shrinking body motions.(1) Translational rigid body motion : Using the fact that Fourier transform F (x, n) of MRI signal f (x, n) in the n-th view (phase encode) shows the object position along the frequency axis (x), we corrected the corrupted phases by a body motion. We have also developed a method which can correct independently of x-axis the body motion alosng y-axis.(2) Rotational rigid body motion : For a step rotation with unkown rot ation center and rotation angle, first we estimated the rotaion angle so that energy in outside region of the reconstructed image becomes minimum. Then we applied the phase retreaval algorithm to reduce the phase disturbance caused by the off-origin of the rotation center.(3) Expanding-shrinking deformable body motion : In case of such as abdomen expanding and shrinking proportionally in x-and y-axs, we have shown that we can reduce the artifact by ; first obtain x-directional motion from the projection onto the frequency axis as the same as (1), and then correct phases corresponding to y-axis by the phase retreaval algorithm.These algorithms can also be applied to the artifact correction caused by body motion of child and old person who cannot lay still for the image taking time of several to several ten minutes with conventional MRI.In order to make the algorithm more close to the practical use, it's a future task to model the more flexible motion with more number of freedom.