利用纹理信息的带标记线心脏核磁共振图像分割

针对带标记线的心脏核磁共振图像分割受到标记线强梯度影响的现象,提出了一种基于图像细节信号能量的纹理分析方法,该方法合理地利用了标记线本身的信号特征,进而达到了去除标记线影响的效果。由于心脏核磁共振图像边缘较弱,噪声相对较强,本文应用测地线活动轮廓模型定义演化曲线,并用水平集方法求解,从而很好地提取了左心室的心内膜。在提取左心室心外膜时,为克服边界断裂、缺省的情形,添加了距离约束能量项,解决了依靠梯度和纹理信息无法准确分割的问题。实验与误差分析结果表明,本文方法能较好地提取带标记线的左心室内外心膜,且具有较高的准确率。

Tagged Cardiac Magnetic Resonance Images Segmentation Using Texture Information

Aiming to solve the problem of the strong gradient coming from the tagged lines in the tagged cardiac magnetic resonance images,a texture analysis method is proposed based on a detailed signal energy which uses the signal feature provided by the tagged line itself.It can effectively remove the influence of the tagged lines.The cardiac magnetic resonance images are characterized by the weak edges and strong noises.Geodesic active contour model to define the evolving curve,which is solved by the level set method.To extract the outer edges of the left ventricle,a distance constraint energy term is added to solve the problem of the edges' reak or default.Depending on this energy term,the feature edges which cannot be directly extracted by gradient or texture information can be well extracted.Experimental results and error analysis show that the proposed method in this paper can effectively extract the inner and outer boundaries of the left ventricle and also has high accuracy.