改进的自适应呼吸信号均值漂移跟踪方法研究

Improved Adaptive Mean Tracking Method for Respiratory Drifting

利用呼吸门控技术进行放射治疗已成为一个热门研究问题,它能确保对健康组织细胞造成最小程度的伤害,而其中的关键问题之一是研究呼吸信号均值漂移跟踪方法,以保证门控技术在开关放射线方面的准确性。本文基于Kanatani高精度椭圆拟合分析方法,并在广义特征分解问题基础上,提出了一种改进的自适应呼吸信号均值漂移跟踪方法,利用我们提出的改进算法可以在状态空间中用椭圆对呼吸信号进行估计。在本文中我们可以得到椭圆的中心值,它对应于我们所要估计的呼吸信号均值,其对有信号损失的数据以及伪数据都具有鲁棒性,在实验中我们验证了由于利用了Kanatani提出的方法,对小数据量信号进行估计时,我们的算法同样具备良好的抗噪性,这修正了仅利用广义特征分解问题中寻找对应特征向量的最大特征值方法求解椭圆估计参数方面的不足。

Respiratory gating technology for radiation therapy has become a hot research problem, which can ensure the healthy tissue involving minimal damage. One of the key problems of the respiratory gating technology is mean tracking method for respiratory signals drifting, to ensure the accuracy of gated area on radiation switch. Based on Kanatani hyperaccuracy ellipse fitting method, and the generalized eigenvalue decomposition problem we proposed an improved adaptive mean tracking method for respiratory signals drif- ting. Using our proposed algorithm, the respiratory signal can be estimated by ellipse model in state-space. In this paper, we can a- chieve the coordinates of the ellipse center, which corresponds to the mean of the respiratory signal which we need to estimate. It is ro- bust to the lost data or pseudo data. In the experiment, we demonstrated that anti-noise performance of our algorithm is brilliant in the estimation of the small amount data, due to using the Kanatani＇ s proposed method. It has also fixed the problem of ellipse fitting meth- od in using only the generalized eigenvalue decomposition to find the corresponding eigenvectors of the largest eigenvalue.