高级检索

    基于BPNN局部位移场拟合的心脏形变计算模型

    A Model Based on Local Displacement Fitting with BPNN for Calculating Myocardial Deformation

    • 摘要: 加标记的心脏核磁共振图像(tagged MRI)提供了一种非介入性的方法来跟踪心肌组织的运动,通过对MRI中tag线的分割可以获得在心动周期不同时刻的多帧稀疏的tag线位移信息.如何借助这些信息近似恢复整个心脏的运动是一个具有挑战性的课题.提出了一种新颖的心脏形变计算模型:由tag线交点信息通过BP神经网建立描述连续位移场的局部拟合函数;再通过对位移场函数的迭代求解心肌质点的运动,实现心脏形变计算.实验表明模型具有物理意义明确、算法简单和计算精度高的特点.

       

      Abstract: Tagged MRI affords a non-invasive method for tracing the motion in myocardium. Some displacement information from the spare distorted tagging curves sampled in certain given times of the heartbeat period, can be obtained by the segmented result in tagged MRI. It is a challenge to recover nearly all the LV motions with the information. In this paper a model for calculating cardiac deformation is innovated. At first, the intersection points of tagging lines, which are obtained from the segmented result in tagged MRI, are used for fitting the local myocardial displacement with BPNN. Then the material points motion in heart are calculated by iterative computation of the displacement functions and the myocardial deformation is calculated. Experiments show that this model has advantages of explicit physical meaning, simple algorithm and high precision in calculation in the model.

       

    /

    返回文章
    返回