Digital image analysis for morphological state characterization of a culture of filamentous microorganisms in production of antibiotics

Abstract

The paper deals with the application of specific methods of digital image analysis for the monitoring of morphological changes in cultures of filamentous microorganisms. First, a sequence of digital image preprocessing and processing steps is proposed for the treatment of microscopic images of a filamentous culture. The preprocessing step include band pass filtering by the Difference of Gaussians filter featuring a novel approach to the task of parameters tuning, as well as the optimization of image porosity and image objects separation quality. In the processing step, the resulting enhanced images are subject to morphological state characterization using a set of several standard and modified morphological parameters. Descriptions of morphological states of different complexity are then discussed varying from the standard set of mean values of parameters to the set of parameters in their full histogram of value frequencies (distribution) form. For such complex descriptions also new fashions of graphical representation of results without loss of information are compared. The potential of the proposed full description of morphological behavior of the culture is demonstrated using a set of microscopic images taken during an industrial antibiotics production cultivation using a microorganism belonging into the Streptomyces genus. Finally, the cluster analysis is proposed for further automatic quantitative classification and interpretation of complex description of metabolic states.