超声心动图块匹配斑点跟踪算法模板块大小的自适应估计

Adaptive Estimation of Template Block Size for Block Matching Based Echocardiography Speckle Tracking

块匹配法是利用超声心动图对心肌进行斑点跟踪定量分析最为常见、有效的方法,块匹配法中模板块大小对斑点跟踪精度有很大的影响,然而现有算法中模板块大小都是由经验来确定,不仅造成斑点跟踪精度较差而且缺乏自适应性。本文提出一种用于超声心动图块匹配斑点跟踪的模板块大小自适应估计方法。首先利用双向跟踪准确度估计跟踪误差,然后根据不同大小块的双向跟踪准确度自适应地确定块匹配中模板块的大小,解决了模板块大小自适应的问题。通过与传统固定模板块大小的方法比较,采用自适应模板块大小方法的斑点跟踪结果在精确度和稳定性上都取得了更好的结果,其中斑点跟踪平均误差由2.5±0.2下降为1.1±0.1,为心脏功能临床诊断和治疗提供了更为可靠的相关参数。

Block matching method is the most common and effective method for quantitative analysis of myocardial function based on speckle tracking echocardiography.For the block matching method,the size of template block has a great influence on the accuracy of speckle tracking.However,the size of template block in the existing algorithm is determined by experience or a large number of experiments.It not only causes poor speckle tracking accuracy,but also lacks adaptivity.In this paper, an adaptive estimation method of template block size for block matching based speckle tracking is proposed.Firstly,the tracking error is estimated by bidirectional tracking accuracy,and then the size of template block in block matching is determined adaptively according to the bidirectional tracking accuracy of different size blocks.Compared with the traditional fixed template block size method,the adaptive template block size method achieved better accuracy and stability in speckle tracking results,and the average error of speckle tracking decreased from 2.5±0.2 to 1.1±0.1,providing more reliable parameters for clinical diagnosis and treatment of myocardial function.