摘要
目的从成像空间出发,提出了一种新的纠正MRI刚性平移运动伪影的方法。方法利用梯度读出方向和相位编码方向的运动特性不同,分别采用不同的方法来消除刚性运动伪影。首先读出方向的运动,通过追踪频谱边缘非零区域和零区域的偏差进行估计,然后在频谱反方向移动相同的量来消除;利用改进的Snake算法,即梯度向量场方法提取目标区域的边界,然后利用相位迭代恢复算法消除残留的读出方向亚像素级伪影和相位编码方向的伪影。结果按国际通用方法生成SL头颅模板,通过对模板的仿真试验,证明修正后图像的信噪比大大提高,验证了方法的有效性和可靠性。结论本文提出的方法能够有效消除MR图像平移运动伪影,与传统的相位迭代恢复算法相比,对于较大运动伪影修正效果更好。
Objective In this paper, a new algorithm for suppressing MRI artifacts due to the translational motion in image plane is described. Methods By analyzing the different features of read-out and phase-encoding axis motions, the different axis motion artifacts can be canceled by different algorithms in two steps. The motion is estimated by tracing the edges between the non-zero area and zero area of spectrum, and the read-out directional motion is suppressed by shifting the spectrum in an reverse direction. The Snake method is used to extract the object boundaries from artifacts images. Then, the phase retrieval method is employed to eliminate the remaining sub-pixel motion of the read-out axis and the entire motion of the phase-encoding axis. Results The experiments results showed the obvious efficiencies and effectiveness of this method with the improved SNR of the corrected image. Conclusion Using this algorithm, the quality of the image is improved greatly and the corrected image is satisfied even when motion is bad comparing with using classical iterative algorithms.
出处
《北京生物医学工程》
2008年第1期15-19,共5页
Beijing Biomedical Engineering
基金
国家973计划(2003CB716102)资助
