Supported by the National Natural Science Foundation of China under Grant Nos.60602050, 60805004, 60675021 (国家自然科学基金); the National High-Tech Research and Development Plan of China under Grant No.2006AA01Z120 (国家高技术研究发展计划(863))
为了充分利用心脏核磁共振图像(magnetic resonance image,简称MRI)中关于左心室的解剖和功能信息,必须先分割左室壁内、外膜.提出一种基于Snake模型的左室壁内、外膜分割方法.首先提出了Snake模型的卷积虚拟静电场外力模型CONVEF(convolutional virtual electric field),该外力场捕捉范围大、抗噪能力强、在C形凹陷区域等问题上性能突出,而且基于卷积运算,采用快速Fourier变换可以实时计算.就左室壁内膜的分割而言,考虑到左室壁的形状近似为圆形,引入基于圆形约束的能量项.对于左室壁外膜的分割,充分挖掘了左室壁内、外膜形状上的相似性和位置上的相关性,构造了形状相似性内能和一个新的边缘图,该边缘图用来计算新的外力场.基于所有这些策略并采用内膜的分割结果初始化,可以自动、准确地分割外膜.通过对一套活体心脏MR(magnetic resonance)图像进行分割并和手工分割结果和GGVF(generalized gradient vector flow) Snake模型的分割结果进行比较,结果表明该方法是有效的.
In order to make a thorough use of the anatomical and functional information derived from cardiac magnetic resonance images, the epicardium and endocardium of the left ventricle should be extracted in advance. This paper presents a method for segmentation of the endocardium and epicardium of the left ventricle in cardiac magnetic resonance images using Snake models. It first proposes an external force for active contours, which is called convolutional virtual electric field (CONVEF). This CONVEF external force possesses the advantages of enlarged capture range, noise resistance and C-shape concavity convergence and can be implemented in real time by using fast Fourier transform since it is based on convolution. Considering that the left ventricle is roughly a circle, a shape constraint based on circle is adopted for segmentation of the endocardium. As to locating the epicardium, an internal energy based on shape similarity is proposed, and an edge map is coined to calculate the new external force by exploiting the resemblance between the endocardium and epicardium in shape and position. With these strategies, taking the final contour for endocardium as initialization, the Snake contour is reactivated to locate the epicardium automatically and accurately. This paper demonstrates the proposed approach on an in vivo dataset and compare the segmented contours with that of the GGVF (generalized gradient vector flow) Snake and manual collections. The results show its effectiveness.