Hierarchy of Neural Organization in the Embryonic Spinal Cord: Granger-Causality Graph Analysis of Calcium Imaging Data

International audienceThe recent development of genetically encodedcalcium indicators enables monitoring in vivo theactivity of neuronal populations. Most analysis ofthese calcium transients relies on linear regressionanalysis based on the sensory stimulus applied or thebehavior observed. To estimate the basic properties ofthe functional neural circuitry, we propose a networkbasedapproach based on calcium imaging recorded atsingle cell resolution. Differently from previousanalysis based on cross-correlation, we used Grangercausalityestimates to infer activity propagationbetween the activities of different neurons. Theresulting functional networks were then modeled asdirected graphs and characterized in terms ofconnectivity and node centralities. We applied ourapproach to calcium transients recorded at lowfrequency (4 Hz) in ventral neurons of the zebrafishspinal cord at the embryonic stage when spontaneouscoiling of the tail occurs. Our analysis on populationcalcium imaging data revealed a strong ipsilateralconnectivity and a characteristic hierarchicalorganization of the network hubs that supportedestablished propagation of activity from rostral tocaudal spinal cord. Our method could be used fordetecting functional defects in neuronal circuitryduring development and pathological conditions