AbstractIndividual cell recognition is a relevant task to be accomplished when single-ion microbeam irradiations are performed. At INFN-LNL facility cell visualization system is based on a phase-contrast optical microscope, without the use of any cell dye. Unstained cells are seeded in the special designed Petri dish, between two mylar foils, and at present the cell recognition is achieved manually by an expert operator. Nevertheless, this procedure is time consuming and sometimes it could be not practical if the amount of living cells to be irradiated is large. To reduce the time needed to recognize unstained cells on the Petri dish, it has been designed and implemented an automated, parallel algorithm. Overlapping ROIs sliding in steps over the captured grayscale image are firstly pre-classified and potential cell markers for the segmentation are obtained. Segmented objects are additionally classified to categorize cell bodies from other structures considered as sample dirt or background. As a result, cell coordinates are passed to the dedicated CELLView program that controls all the LNL single-ion microbeam irradiation protocol, including the positioning of individual cells in front of the ion beam. Unstained cell recognition system was successfully tested in experimental conditions with two different mylar surfaces. The recognition time and accuracy was acceptable, however, improvements in speed would be useful.
Glycosphingolipid-Glycosphingolipid Interaction: A Model for a New Type of Cell Recognition System.