基于运动跟踪获取人体呼吸曲线的初步研究

摘要

4D-CT的实现需要在采集CT图像的同时,同步获取患者的呼吸曲线,再据此对CT图像进行排序和三维重建.本研究提出一种基于运动跟踪获取人体呼吸曲线的方法.即在人体腹部贴上标记物,用两台摄像机在CT扫描过程中成正交位同步跟踪呼吸运动,将得到的视频提取为单帧图像,利用区域生长算法提取标记物的质心,并跟踪质心的运动,描绘坐标的任意维度与时间的关系曲线,即得到的呼吸曲线.通过此方法获得了正常人在深呼吸时的呼吸曲线,并在该曲线上确定了屏气模式下7个时相点的位置,腹部表面在z轴方向上的运动幅度最大,约为3.5 cm.实验结果表明该方法切实可行,并且相比于其他的呼吸信号提取方法,具有简单易行的优点.%The realization of 4D-CT typically requires the synchronous recording of the respiratory curve of the subject while CT scanning, and then sort and three-dimensional reconstruction of CT images accordingly. This paper presents a method based on motion tracking for human respiratory curve. Initially, a few markers were attached to the abdomen of the subject. Two cameras were placed orthogonally and used to monitor the respiratory motion synchronously in the CT scanning process. Then the single frames were extracted from the videos. The centroid of each marker in these images was determined using region growing algorithm and its movement was tracked. Finally, the arbitrary coordinate dimension of the centroid vs time curve, I. E. The respiration curve, was depicted. Through this method, the respiration curve of a normal person in deep breathing was acquired, and the locations of seven discrete breathing phases in breath - holding mode were determined. The motion of abdominal surface in the z axis direction has the maximum amplitude of 3. 5 cm. The experimental results indicated that this method was feasible and easy to realize, compared with some other respiratory signal extraction methods.