Backbone Extraction Reveals Structural Organization in Multicellular Functional Networks

Functional connectivity and network analysis provide a widely used framework for quantifying interactions in complex systems, including multicellular tissues whose coordinated activity supports normal physiological function. In this study, we construct functional networks of pancreatic β-cells within islets of Langerhans using multicellular calcium imaging data. A correlation matrix derived from experimentally recorded signals captures pairwise interactions between cells. Our primary objective is to extract functional connectivity patterns using a metric backbone approach and compare it with conventional threshold-based network construction. The backbone identifies structurally relevant connections by removing redundant links while preserving shortest-path relationships. Results obtained across six independent pancreatic islets show that both approaches yield comparable average degree and clustering, whereas backbone networks show higher efficiency, a larger connected component, and more homogeneous degree distributions, reflected by lower degree Gini coefficients and hubness values. Modular structure is nevertheless largely preserved, indicating that backbone extraction reduces redundancy without strongly distorting the underlying functional architecture. Overall, our findings support backbone extraction as a data-driven framework for constructing functional cellular networks while avoiding limitations associated with arbitrary threshold selection. This approach may improve the quantitative analysis of multicellular dynamics and be broadly applicable to other biological systems exhibiting complex interaction patterns.