模糊水平集心脏CT图像序列分割方法

Fuzzy level set segmentation method for cardiac CT image sequence

为解决心脏CT图像序列传统分割方法人工交互复杂、效率低的问题,提出一种模糊水平集分割方法。只需选定单张心脏CT图像,利用改进后的模糊聚类算法获取感兴趣区域初始轮廓,将其结果用于引导C-V模型水平集进行心脏组织精准分割。为有效减少图像序列分割时间及人工交互,由单张图像分割结果作为其空间相邻图像水平集的初始轮廓,避免重复性聚类过程,循环迭代得到每张图像最终轮廓位置。实验结果表明,该算法能准确分割出心脏各组织边缘,时间代价小、人工交互简单,分割结果能为心脏三维重建提供准确的数据集。

A fuzzy level set segmentation method was proposed to solve the complexity and low efficiency problems of the conventional cardiac CT image sequence segmentation method.This method could be realized only by selecting a single cardiac CT images,and using the improved fuzzy clustering algorithm to extract the initial outline of the region of interest（ROI）.Its results were used to guide the C-V model level set for cardiac tissues precise segmentation.To effectively reduce the time cost for image sequence segmentation and human interaction,each segmentation result was used as the initial contour of its adjacent images.This access avoided the repetitive clustering process,and the final outline of the image was obtained.Experimental results demonstrate that the algorithm can accurately segment the edge of the heart organization with less time and simple human interaction.Besides,the segmentation results can provide accurate data set for the three-dimensional heart reconstruction.