Fast processing of label-free video microscopy movies of human and bacterial cell populations growing in vitro during chemical exposure
Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
A fast computational framework for large-scale parallel processing of label-free video microscopy movies of human and bacterial cell populations growing in vitro during chemical exposure was developed in MATLAB®. The overarching aim was to quantify and study time evolving morphological effects due to chemical perturbations caused by single drugs and combinations. Using this framework, a previously reported method for characterization of differences in time evolving morphologies of human cell populations, based on pixel histogram hierarchies of phase-contrast microscopy images, was re-implemented, refined and subsequently optimized with respect to method-specific tuning parameters. This implementation was also generalized for time-lapse microscopy movies of bacterial cell cultures, generated by the oCelloScope™ system, which acquires multiple series of images of non-adherent cell populations in the cell culture medium. In addition, a separate computational framework for large-scale parallel quantification of the bacterial growth was deployed as an alternative to the growth kinetics analysis provided by the integrated commercial software of the oCelloScope™ system. The potential of the implemented frameworks was demonstrated on experimental data by processing time-lapse movies from different human and bacterial cell populations, while being exposed to different single chemical compounds and combinations. These novel computational tools are compatible with either single high-end multi-core computers or cloud-based distributed computing infrastructures offered via MapReduce, and Hadoop® MapReduce, respectively. This enables fast and fault-tolerant processing of huge video microscopy datasets and opens for optimization of user-defined tuning parameters.
Place, publisher, year, edition, pages
2016. , 118 p.
Medical Image Processing
IdentifiersURN: urn:nbn:se:uu:diva-303946OAI: oai:DiVA.org:uu-303946DiVA: diva2:974708
Master Programme in Bioinformatics
2016-09-20, C4:301, BMC, 13:00 (English)
Gustafsson, Mats, Professor in Medical Bioinformatics