Automatic detection of yeast and pseudohyphal form cells in the human pathogen candida glabrata

Abstract

The morphological switching shown by Candida glabrata between oval-shaped budding yeast cells and elongated pseudohyphal growth structures has been found to be related to its ability to undergo invasive growth, and thus, predicted to affect virulence. Therefore, the morphological analysis of C. glabrata cell cultures is a procedure of clinical relevance. In the present study, a pioneering algorithm was developed to automatically detect both yeast and pseudohyphal form structures from a database of 82 phase contrast microscopy images of C. glabrata cell cultures. The algorithm produced robust results, despite some limitations in the quality of the images, having detected correctly 84.56 % and 51.94 % of the yeast form cells and pseudohyphal structures, respectively. Future work should focus on improving these results by incorporating the analysis of cell nuclei or septa positions, extracted from fluorescence images.