Demons算法在四维CT图像配准中的应用
Implementation of Demons Algorithm in 4D CT Image Registration
-
摘要: 由呼吸运动引起的器官的运动及形变是4D精确放射治疗中最主要的误差来源。利用4D CT图像和配准技术来分析不同呼吸时相中组织、器官的形变是目前研究的热点。目前常用的研究方法是基于形变模型的3D配准算法。基于光流场的Demons算法具有速度快和精度高的特点,本文将其应用到4D CT图像配准中,并在原始图像形变力的计算中引入了对称梯度。实验结果表明,结合对称梯度的Demons方法具有更快的收敛速度,而且匹配误差小。Abstract: Respiratory motion is the primary source of error in the 4D precise radiation therapy.It’s a hot topic to analyze the organ deformation between different respiratory phases by registration of 4D images.Currently,3D deformable registration algorithm is commonly used in the research.The algorithm based on optical flow has high speed and high accuracy.Demons algorithm and symmetric gradients are applied to the calculation of original deformable force of 4D CT registration.It is demonstrated that Demons algorithm using symmetric gradients converges faster to results with less error.