基于瓶颈检测和分水岭算法的重叠宫颈细胞图像分割方法  

Overlapping Cervical Cell Image Segmentation Based on Bottleneck Detection and Watershed Algorithm

作　　者：段鹏 程文播[2] 钱庆[2] 章强[2] 杨任兵 潘宇骏 DUAN Peng;CHENG Wenbo;QIAN Qing;ZHANG Qiang;YANG Renbing;PAN Yujun(University of Science and Technology of China,Hefei,230026;Suzhou Institute of Biomedical Engineering and Technology,Chinese Academy of Science,Suzhou,215163)

机构地区：[1]中国科学技术大学,合肥市230026 [2]中国科学院苏州生物医学工程技术研究所,苏州市215163

出　　处：《中国医疗器械杂志》2020年第1期7-12,共6页Chinese Journal of Medical Instrumentation

基　　金：中国科学院科研仪器设备研制项目（YJKYYQ20170067,YJKYYQ20170068）;吉林省与中国科学院科技合作产业化专项资金（2018SYHZ0007）;中国科学院科技服务网络计划（KFJ-STS-SCYD-007）

摘　　要：针对宫颈细胞图像存在大面积重叠的问题,该研究提出一种基于瓶颈检测和分水岭算法的图像分割方法。首先使用多边形近似得到细胞轮廓上的所有特征点,再通过瓶颈检测和椭圆拟合得到正确的分裂点对,从而确定重合区域的大致范围,对该区域的梯度图像使用分水岭算法得到内部的边界信息,最后与外轮廓进行叠加,得到重叠细胞的分割结果。实验结果表明:该算法能够从重叠的宫颈细胞图像中分割出单个细胞轮廓,具有很好的准确度和完整度。与医生手工标记的分割结果接近,与现有其它算法相比,分割效果更好。This study proposes an image segmentation method based on bottleneck detection and watershed algorithm to solve the problem of overlapping cervical cell image.First,we use polygon approximation to get all feature points on the cell contour and then use bottleneck detection and ellipse fitting to obtain the correct split point pairs.Therefore,the approximate range of the overlapping region was determined.The watershed algorithm was used to obtain the internal boundary information for the gradient image of the region.Finally,the segmentation results of the overlapped cells were obtained by superimposing with the outer contour.The experimental results show that this algorithm can segment the contour of a single cell from the overlapping cervical cell images with good accuracy and integrity.The segmentation result is close to that of doctors'manual marking,and the segmentation result is better than other existing algorithms.

关 键 词：宫颈细胞分割 瓶颈检测 重叠细胞 分水岭算法 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]