Comparison of shape representation methods for dynamic cell analysis  

作　　者：李衡 刘志文 安兴 时永刚 

机构地区：[1]School of Information and Electronics,Beijing Institute of Technology

出　　处：《Journal of Beijing Institute of Technology》2014年第4期541-548,共8页北京理工大学学报（英文版）

基　　金：Supported by the National Natural Science Foundation of China(61271112)

摘　　要：To evaluate the performance of basic shape representation methods for the description of dynamic cellular morphology, several frequently-used shape descriptors are compared. The methods are examined by using 50 lymphocyte video clips including two kinds of lymphocyte cells. Our goal is to represent cell shape in each frame accurately, meanwhile precisely classify the two groups of cells based on the cellular morphological variations in the video clips. Experimental results illustrate that in general the region-based shape descriptors outperform the contour-based ones, since the contourbased methods are excessively sensitive and ignorant to cellular internal information. Due to their robustness to noise, the region-based shape descriptors are suitable for dynamic cell representation. Although region-based methods are more time-consuming, they analyze the entire cell area.To evaluate the performance of basic shape representation methods for the description of dynamic cellular morphology, several frequently-used shape descriptors are compared. The methods are examined by using 50 lymphocyte video clips including two kinds of lymphocyte cells. Our goal is to represent cell shape in each frame accurately, meanwhile precisely classify the two groups of cells based on the cellular morphological variations in the video clips. Experimental results illustrate that in general the region-based shape descriptors outperform the contour-based ones, since the contourbased methods are excessively sensitive and ignorant to cellular internal information. Due to their robustness to noise, the region-based shape descriptors are suitable for dynamic cell representation. Although region-based methods are more time-consuming, they analyze the entire cell area.

关 键 词：cellular morphology dynamic cell images shape descriptors 

分 类 号：Q24[生物学—细胞生物学]