Fig. 1. Estimation and characterization of the dynamical structure underlying transitions in intrinsic brain activity using our TDA-based Mapper approach.

Here, we present data from a representative participant (MSC-01; odd sessions). A Individualized parcellated data from the highly sampled Midnight Scan Club (MSC) individuals⁵⁷ was split into two halves: odd sessions (2.5 h) and even session (2.5 h) sets. The Mapper approach was independently run on each set to generate the underlying structure as a graph. Each graph consists of nodes and edges, where the nodes could in turn contain multiple whole-brain volumes (or TRs; the size of a node represents the number of TRs). The nodes are connected if they share TRs. B The Mapper-generated graph can be characterized in several ways. Here, we examine topological properties by annotating the graph nodes using nodal degree. C The graph can also be annotated with meta-information to characterize the mesoscale structure. Here, we show annotation using the activation of individual-specific resting-state networks (RSNs). A pie-chart-based annotation is used to reveal the proportion of time frames with each node belonging to different RSNs. D Similarly the graph can also be annotated using other available meta-information, e.g., session information.