High Content Screening (HCS) platforms allow screening living cells under a wide range of experimental conditions and give access to a whole panel of cellular responses to a specific treatment. The outcome is a series of cell population images. Within these images, the heterogeneity of cellular response to the same treatment leads to a whole range of observed values for the recorded cellular features. Consequently, it is difficult to compare and interpret experiments. Moreover, the definition of phenotypic classes at a cell population level remains an open question, although this would ease experiments analyses. In the present work, we tackle these two questions. The input of the method is a series of cell population images for which segmentation and cellular phenotype classification has already been performed. We propose a probabilistic model to represent and later compare cell populations. The model is able to fully exploit the HCS-specific information: "dependence structure of population descriptors" and "within-population variability". The experiments we carried out illustrate how our model accounts for this specific information, as well as the fact that the model benefits from considering them. We underline that these features allow richer HCS data analysis than simpler methods based on single cellular feature values averaged over each well. We validate an HCS data analysis method based on control experiments. It accounts for HCS specificities that were not taken into account by previous methods but have a sound biological meaning. Biological validation of previously unknown outputs of the method constitutes a future line of work.