Abstract

Microswimmers usually move in three dimensions, but camera sensors are two-dimensional. This can restrict the scope of experiments using standard techniques. In the past, video microscopy data has been analysed by the most intuitive approach, simply viewing video images as we do everyday objects. Teaching a computer to track objects of a certain shape or size then yields information about their dynamics. This approach is the same used in the macroscopic domain, where ‘machine vision’ approaches to object recognition and tracking have been very successful. The work in our lab takes a different approach by using aspects of classical optics and signal processing to design new image processing algorithms, to ‘mine’ more information out of digital images. This has two advantages: (i) We can extract three dimensional imaging data from two-dimensional images; (ii) by moving away from traditional ‘machine-vision’ ideas, we can redesign imaging systems that are cheaper and more lightweight. I will present several examples from recent work in holographic imaging of microswimmers that use image-processing algorithms to obtain three-dimensional data on microorganism swimming trajectory and shape. We have adapted our technology to investigate the physics of swimming in a diverse range of single-celled microorganisms including eukaryotes (Chlamydomonas, Plasmodium), archaea (Haloarcula, Halorubrum), and bacteria (Pseudomonas, Shewanella). In particular, the ability to follow hundreds or thousands of individual swimming bacteria in volumes of up to a cubic millimetre allows us to address questions on the statistics and variability of cell swimming trajectories. The figures below show examples of the swimming trajectories of bacterial cells – E. coli in this case – rendered to scale.

References: 1. Jikeli et al., Nat. Commun. 6 p.7985 (2015). 2. Wilson et al., Proc,. Natl. Acad. Sci. USA 110 (47) p.18769 (2013). 3. Wilson & Zhang, Opt. Express 20 p.16735 (2012).