Automated image-based screening of cell cultures for cell therapy

The goal of this project is to develop an automated computer vision system to acquire and analyze images of adherent cell cultures. These cultures are involved in the establishment of a cell therapy protocol using primary cells. Cultures are stained to facilitate the image segmentation based on color information. Nevertheless, the high density and the uneven dispersion of cells favor nuclei clustering which are identified by a nonparametric classification. This approach automatically fits the individual nuclei features which vary according to cultures. A split and merge post-processing is used to divide aggregates into individual nuclei. This automatic system is applied to the improvements of cell therapy protocol and to control the characteristics of cells during the clinical trial