摘要
人工检测和计算有丝分裂细胞的过程非常冗长,而且不同病理医生之间的诊断结果有较大差异性,因此,在临床中迫切希望有定量的计算机辅助自动检测方法。提出了一种基于低秩表示的计算机辅助自动检测有丝分裂方法,将有丝分裂细胞看做低秩表示中的稀疏部分,非有丝分裂部分看做低秩部分。在ICPR 2012有丝分裂竞赛提供的有丝分裂图片数据库上的实验结果表明,和已有的基于模式识别的检测方法相比,该方法能够获得较高的F-measure和recall值,分别为0.59和0.56。
It was usually time consuming and tedious to manually detect and count the number of mitosis and the results might vary largely among pathologists. This paper presented a low rank representation based computer aided method for mitosis detec- tion. Mitosis cells could be considered as the sparse section while the non-mitosis could be seen as the low rank section. Ex- perimental results on ICPR 2012 contest mitosis image set show that this method can achieve F-measure 0.59 and recall 0.56, respectively. The evaluation results are better than the traditional pattern recognition based methods.
出处
《计算机应用研究》
CSCD
北大核心
2015年第1期280-283,311,共5页
Application Research of Computers
基金
国家自然科学基金资助项目(61273259)
江苏省"六大人才高峰"高层次人才资助计划项目(2013-XXRJ-019)
关键词
低秩表示
病理图像
细胞有丝分裂检测
low rank representation
histopathological image
mitosis detection
