基于弹性匹配活动轮廓模型心脏超声图像分割

SEGMENTING ECHOCARDIOGRAM IMAGES BY ELASTIC MATCHING-BASED ACTIVE CONTOUR MODEL

超声医学图像由于受成像机理的影响,图像对比度不高、边缘不明显。基于传统活动轮廓线模型(snake模型)的分割方法可能产生过分割或泄漏问题。由于医学图像中拓扑结构已知,因此基于先验知识的活动轮廓线分割方法是解决这个问题的一个有效途径。建立一种新的基于弹性匹配活动轮廓线模型,该方法将待分割曲线的形状与原型曲线用弹性匹配测量变形量或相似度。曲线在演化过程中,根据变形量或相似度,可以准确分割模糊的边缘,同时保持整体目标分割形状。通过对二维小儿超声心脏图像的左心房内壁进行分割,经比较,基于弹性匹配活动轮廓线分割比传统活动轮廓线分割的误差有显著减少,避免了传统活动轮廓线的过分割或泄漏问题。

Echocardiographic images often have poor image contrast due to its imaging mechanism. The segmentation algorithms based on traditional active contour model, such as the snake model, may have over segmentation or leakage problem. Because of having known the topology structure in these medical images, to incorporate the active contour segmentation method with a priori knowledge is an effective way to solve this problem. This paper introduces a new snake model, which is incorporated with elasticity matching active contour. Our method uses the criterion of elasticity matching to measure the deformation or similarity between the contour to be segmented and the prototype contour. Therefore, during the evolving process of the active contour, the diffuse boundaries can be accurately segmented based on the similarity, meanwhile the segmentation form of whole object is retained. By segmenting the inner surface of left atria in 2D echocardiographic images of a child,the comparison made indicates that the elasticity matching-based active contour model can achieve more accurate segmentation result than the traditional active contour model and avoid the over segmentation or leakage problem the traditional active contour may have.