A computer vision framework for quantifying the social behaviour of Drosophila Melanogaster
3 February 2017
Campus Drie Eiken, Promotiezaal Q0.02 - Universiteitsplein 1 - 2610 Antwerpen-Wilrijk (route: UAntwerpen, Campus Drie Eiken
PhD defence Tanmay Nath - Faculty of Science
Social isolation profoundly influences social behavior. Drosophila melanogaster is a powerful model to study small-scale, temporally-transient social behavior. However, longer-term analysis of large groups of flies is hampered by the lack of effective and low-priced tools. Under the aegis of FLiACT, a European training network, we built a low-cost computer vision framework for quantifying the social behavior of flies.
We constructed the imaging platform and a highly robust tracking pipeline to track up to fifty flies simultaneously. Thereafter, we developed an automatic classification system and constructed social interaction networks to quantify the social behaviour of flies. Based on a graph matching technique, we proposed a novel methodology to compute an average social network obtained from different experiments.
We quantified average social networks obtained from social isolation experiments. We observed that isolation significantly and swiftly altered individual and local parameters depicting near-neighbour relationships. In summary, we describe a powerful computer vision approach to unravel the behavioral impact of isolation on the social network dynamics in Drosophila melanogaster.