Unveiling functional heterogeneity in anatomical functional areas: a framework for fine-grained functional...

Objective. Conventional functional connectivity (FC) analysis of wide-field calcium imaging (WFCI) data relies on the assumption of homogeneity within predefined anatomical functional areas (FAs), where the signal averaged within each FA serves as the foundation for inter-FA connectivity modeling. However, accumulating evidence suggested significant intra-FA functional heterogeneity with functionally distinct subregions. This study aims to systematically examine the existence of the heterogeneity and the consequences of violating the homogeneity assumption on the FC analysis.Approach. We propose a three-step approach: (1) The spatiotemporal clustering of pixels with similar activity dynamics into functional parcels (FPs). (2) Classifying FPs according to their cortex-wide connectivity profiles into four categories: cross-hemispheric (CH), unihemispheric (UH), cross-modal (CM) and unimodal (UM). (3) Gauging the consequences and utility of above analysis.Main results. Analysis of adult mice WFCI data (n= 6) shows that pixels can be reliably clustered into FPs, and that FPs fall into different categories and form distinct subregions, unveiling the functional heterogeneity within each FA. Crucially, fine-grained FC analysis for different categories of subregions uncovered significant differences compared to the results from the conventional method. Application of the analysis to longitudinal WFCI data on mouse brain development (n= 17) demonstrates increases in CM and CH subregions and decreases in UM and UH subregions over time, in line with expectations based on prior research into neurodevelopment and network reorganization.Significance. The present study develops a fine-grained FC analysis framework, leveraging the high spatiotemporal resolution of WFCI to more precisely characterize large-scale cortical network dynamics. The observed differences between FCs derived using the proposed framework and those from conventional methods highlight the need for caution regarding the functional homogeneity assumption in conventional FC analysis. Furthermore, the developmental regularities revealed in longitudinal mouse brain data demonstrate the utility of fine-grained FC analysis.