Automated Analysis of Microscopic Images and Time-Lapse Microscopy Videos

Automated Analysis of Microscopic Images and Time-Lapse Microscopy Videos

Spatial and temporal information on infection processes can only be obtained from image and video data of microscopy experiments. In the context of infection biology research, we are mainly interested in the analysis of confrontation assays between immune cells and human-pathogenic fungi. To characterize infection processes in a high-throughput manner, automated analysis of the image and video data is essential and comprises the steps (i) preprocessing, (ii) segmentation of regions of interest (ROIs), (iii) ROI classification, and (iv) ROI tracking in the case of time-lapse microscopy. These steps require the development of algorithms tailored to the experimental setup.

In collaboration with our experimental colleagues, we develop algorithms for the automated image analysis of phagocytosis assays, e.g. for alveolar macrophages and Aspergillus fumigatus conidia, as well as of invasion assays, e.g. for Candida albicans and oral epithelial cells. This framework can be used for systematic large-scale studies such as comparative mutant-screenings. Time-lapse microscopy is an important technique to study the dynamics of biological processes. Methods for automated segmentation and tracking are often limited to certain cell morphologies, cell staining, and/or motion-models. We develop an automated segmentation and tracking framework that does not have these restrictions and meets the requirement that highly variable cell shapes can be handled without relying on any cell staining. This allows tracking immune cells, such as neutrophils, and investigating time-dependent changes in their migration behavior due to the presence of fungal pathogens.

In the context of natural product research, droplet-based microfluidics replaces traditional culture flasks. After having introduced triggered imaging with real-time image-based droplet classification as a universal method for label-free and growth-dependent droplet sorting, we are now focusing on individually barcoding droplets for unique identification.