| 多尺度区域生长与去粘连模型的乳腺细胞分割 |
| |
| 引用本文： | 王品,胡先玲,谢文宾,李勇明,刘书君.多尺度区域生长与去粘连模型的乳腺细胞分割[J].仪器仪表学报,2015,36(7):1653-1659. |
| |
| 作者姓名： | 王品 胡先玲 谢文宾 李勇明 刘书君 |
| |
| 作者单位： | 1.重庆大学通信工程学院重庆400030；2.第三军医大学生物医学工程学院重庆400038 |
| |
| 基金项目： | 国家自然科学基金(61108086,61171089)、国家博士后基金(2013M532153)、重庆市自然科学基金(CSTC2012jjA40015)、重庆市科技攻关计划(cstc2012gg yyjs0572)、中央高校基金(CDJZR12160011,CDJZR13160008,CDJZR155507)项目资助 |
| |
| 摘 要： | 乳腺癌已经成为女性最常见的恶性肿瘤,组织切片显微图像的病理分析是诊断的主要手段,细胞的准确分割是病理分析的重要环节。该文提出了一种新的乳腺细胞显微图像的自动分割算法:首先结合小波分解和多尺度区域生长算法分离细胞和背景,实现对细胞的精确定位;然后采用改进的数学形态学对粘连细胞进行一次细分割;接着再采用基于曲率尺度空间(CSS)的角点检测分割算法对粘连细胞进行二次细分割;两次细分割方法构成了一个双策略去粘连模型,保证了去粘连的准确性和鲁棒性。将算法应用到22幅乳腺细胞显微图像上,可以对不同类型的乳腺细胞图像进行全自动分割,有较高的分割灵敏度(0.944±0.024)和特异度(0.937±0.038),且具有较好的普适性。
|
| 关 键 词： | 图像分割 乳腺细胞 小波分解 区域生长 角点检测 |
Image segmentation of breast cells based on multi scale
region growing and splitting model |
| |
| Wang Pin,Hu Xianling,Xie Wenbin,Li Yongming,Liu Shujun.Image segmentation of breast cells based on multi scale
region growing and splitting model[J].Chinese Journal of Scientific Instrument,2015,36(7):1653-1659. |
| |
| Authors: | Wang Pin Hu Xianling Xie Wenbin Li Yongming Liu Shujun |
| |
| Affiliation: | 1.College of Communication Engineering, Chongqing University, Chongqing 400030, China;
2. College of Biomedical Engineering, The Third Military Medical University, Chongqing 400038, China |
| |
| Abstract: | Breast cancer has become the most common cancer in women. An important method to make a definite diagnosis of breast cancer is biopsy pathologic examination. Accurate segmentation of the cells is the important step towards pathological analysis of microscopic images. This paper puts forward a full automatic segmentation algorithm for the breast cell microscopic images. Firstly, wavelet decomposition and multi scale region growing (WDMR) algorithms are utilized to separate the cells from background, and the precise cell localization is realized. Then, a double strategy splitting model (DSSM) is applied to split the overlapped cells for better accuracy and robustness. This model firstly uses an improved mathematical morphology algorithm to perform the first fine segmentation of the overlapped cells, and then the Curvature Scale Space (CSS) based corner detection method is adopted to perform the second fine segmentation. The algorithm was applied to 22 breast cell microscopic images, and the experiment evidence shows that the proposed method can carry out effective automatic segmentation of different kinds of breast cell images, and achieves high segmentation sensitivity of 0.944 (±0.024), specificity of 0.937 (±0.038), and has good universality. |
| |
| Keywords: | image segmentation breast cell wavelet analysis region growing corner detection |
| 本文献已被 CNKI 等数据库收录！ |
| 点击此处可从《仪器仪表学报》浏览原始摘要信息 |
|
点击此处可从《仪器仪表学报》下载全文 |
|
