26 April 2016 Network inference from functional experimental data (Conference Presentation)
Author Affiliations +
Functional connectivity maps of neuronal networks are critical tools to understand how neurons form circuits, how information is encoded and processed by neurons, how memory is shaped, and how these basic processes are altered under pathological conditions.  Current light microscopy allows to observe calcium or electrical activity of thousands of neurons simultaneously, yet assessing comprehensive connectivity maps directly from such data remains a non-trivial analytical task. There exist simple statistical methods, such as cross-correlation and Granger causality, but they only detect linear interactions between neurons.  Other more involved inference methods inspired by information theory, such as mutual information and transfer entropy, identify more accurately connections between neurons but also require more computational resources.   We carried out a comparative study of common connectivity inference methods.  The relative accuracy and computational cost of each method was determined via simulated fluorescence traces generated with realistic computational models of interacting neurons in networks of different topologies (clustered or non-clustered) and sizes (10-1000 neurons). To bridge the computational and experimental works, we observed the intracellular calcium activity of live hippocampal neuronal cultures infected with the fluorescent calcium marker GCaMP6f. The spontaneous activity of the networks, consisting of 50-100 neurons per field of view, was recorded from 20 to 50 Hz on a microscope controlled by a homemade software. We implemented all connectivity inference methods in the software, which rapidly loads calcium fluorescence movies, segments the images, extracts the fluorescence traces, and assesses the functional connections (with strengths and directions) between each pair of neurons.  We used this software to assess, in real time, the functional connectivity from real calcium imaging data in basal conditions, under plasticity protocols, and epileptic conditions.
Conference Presentation
© (2016) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Patrick Desrosiers, Patrick Desrosiers, Simon Labrecque, Simon Labrecque, Maxime Tremblay, Maxime Tremblay, Mathieu Bélanger, Mathieu Bélanger, Bertrand De Dorlodot, Bertrand De Dorlodot, Daniel C. Côté, Daniel C. Côté, "Network inference from functional experimental data (Conference Presentation)", Proc. SPIE 9690, Clinical and Translational Neurophotonics; Neural Imaging and Sensing; and Optogenetics and Optical Manipulation, 969019 (26 April 2016); doi: 10.1117/12.2212726; https://doi.org/10.1117/12.2212726

Back to Top