A Quantification Tool to Analyse Stained Cell Cultures

Abstract

In order to assess the efficiency of culture media to grow cells or the capacity of drugs to be toxic, we elaborated a method of cell quantification based on image processing. A validated approach makes segment on stained nuclei by thresholding the histogram of the best adapted color component. Next, we focus our attention on the classification methods able to distinguish isolated and aggregated nuclei because the aggregation of nuclei reveals a particuliar cell function. Two decision trees have been designed to consider the different shape features of two types of nuclei : coming a) from bone marrow and b) from immature muscular cell cultures. The most relevant characteristics are the concavity, the circularity and the area of binary objects. © Springer-Verlag 2004.