基于图像跟踪的呼吸运动分析

Breath Motion Estimation Based on Image Tracking

呼吸运动是一种内在的周期性运动,由于人体的各异性很难获得精确的呼吸运动模型,所以目前用基于统计的概率模型来表达。传统的呼吸运动分析模型是用高阶余弦函数来描述由呼吸引起的上下运动,但并不能很好地满足实时和三维分析的临床需要。在传统模型的基础上,提出一种基于统计学和图像跟踪的呼吸运动分析方法。该方法借助于用X光机实时采集的患者胸部透视序列图像,通过基于活动轮廓的方法,跟踪图像中横膈膜的位置变化,建立呼吸运动模型。实验证明:与传统的呼吸运动模型相比,该方法更能实时地反映人体呼吸运动的规律,并可方便建立三维呼吸运动分析数值模型,具有较高的临床实用价值。

The organ motion due to respiration is an intra periodic motion. An accurate model of breath motion is difficult since breath motion varies from the different person and even different stage of the same person. So, a statistical model for the breath motion analysis has been used widely to analyze superior-inferior breath motion only, having some difficulties to meet the clinical requirements of real-time and 3D analysis. This paper presented a new technique for breath motion estimation based on the statistics and image tracking. A snake method was used first to track the position of the target, such as the diaphragm in the image set obtained from X-ray imaging machine, and therefore the position-time curve of diaphragm motion could be found. The final breath motion model was constructed as a probability distribution function through calculate the position probability in the duration of image set. The experimental results proved that the model could reflect the real-time breath motion better than the traditional method did and be used to build 3D breath motion model.