Graph-based algorithms for the analysis of functional connectivity in zebrafish habenula and telencephalon

The human brain is a huge and complex network in which smaller structures communicate and process information at different scales of time and space. In recent years brain connectivity has become one of the main interests of many leading neuroscience centres all over the world, which investigate different aspects of brain connectivity, in particular the branch called functional connectivity, for many different purposes: from clinic applications in neurological diseases, to basic research both on humans and on experimental models. As a matter of fact, animal models allow the study of much simpler brain networks and structures compared to humans, and they open the possibility to exploit imaging or genetic tools which can’t be performed on humans for technical or ethical reasons. Yaksi lab, the laboratory within which this study has been carried out, is a research centre in the field of systems neuroscience, in which neuroscientists, physicists and engineers are interested in how behavioural tasks and sensory computation affect the neural circuits and pathways within the brain of zebrafish. These animal models allow two-photon calcium imaging, an imaging tool which records fluorescence signals related to the activation of single neurons. Two of the main brain areas the lab has been interested in are telencephalon, the homolog of mammalian amygdala and hippocampus, and habenula, which can be found also in mammals. The aim of this thesis is to investigate two aspects, related to spatial and temporal organization respectively, of functional connectivity in these two zebrafish brain regions, exploiting some graph-based algorithms that, as confirmed in literature, are particularly suitable for representing and visualizing brain networks. On one hand spectral clustering algorithm was used in order to cluster synchronous neurons during ongoing activity. A heuristic method was used to estimate the optimal number of clusters, but it resulted in a poor ability to recognize small clusters. On the other hand temporal development of neural activation and its changing between ongoing activity and odour stimulation were analysed, leading to clues on how different the neuronal activity patterns are between the two behavioural states.