作　　者： ; ; ; ;

机构地区： 浙江大学生物医学工程与仪器科学学院

出　　处： 《浙江大学学报（工学版）》 2006年第2期262-266,共5页

摘　　要： 为了获取独立的三维耳蜗模型,提出了一种交互式、半自动、由粗到细、结合三维显示反馈的耳蜗分割方法.浏览二维切片,调节图像强度区间并选取感兴趣区域,从颞骨螺旋CT(computed tomography)图像中对耳蜗进行粗分割;采用基于阈值区间的三维水平集(level set)分割算法对耳蜗进行细分割.应用可视化技术,对分割结果进行实时显示,显示结果作为用户进行参数调节的参考依据,经过多次人机交互,直至获得满意的分割结果.利用临床颞骨螺旋CT图像进行耳蜗分割实验,结果表明该方法适合于分割结构复杂且表面光滑的目标. To obtain the 3D separate model of the cochlea, an interactive, semiautomatic, coarse-to-fine image segmentation approach with 3D visual feedback was developed for the cochlea segmentation. The intensity range and the volume of interest were defined to roughly segment the cochlea by viewing the spiral computed tomography （CT） image volume slice by slice. Then the threshold-based 3D level set algorithm was used to separate the cochlea from other anatomic structures. Visualization technology was used to render the segmented results. The user can modify the intermediate parameters based on the real-time 3D visual feedback until getting Satisfying result. The test results using spiral CT images of the temporal bone prove that the method is applicable for segmenting complicated structures with smoothing surface.

关 键 词： 耳蜗 图像分割 水平集算法 可视化

领　　域： [电子电信] [电子电信]