2.6.2. LMN integration

In order to analyze the integration in the observed LMN community structures, we calculated the roles of the regions within each LMN configuration in healthy participants. Topological roles were assigned to each node based on its intra- and inter- modular connections. To that end we calculated normalized intra-modular degree (z, Meunier et al., 2009) whose value is higher if a node has a large number of intra-modular connections in comparison to other nodes in the same module. We measured inter-modular connectivity with the participation coefficient (Pc, Rubinov & Sporns, 2010). Due to the narrow distribution (Schedlbauer & Ekstrom, 2019) and dependency on the number of modules (Fornito et al., 2016), the Pc value was standardized within given community partition (Pcs). As in previous studies, the intra- and inter-modular plane was divided into four domains due to the smaller number of nodes within LMN (Meunier et al., 2009, Schedlbauer and Ekstrom, 2019). Nodes were considered as connector hubs if they had both high z (≥0) and Pcs (≥0) and as provincial hubs if they had high intra-modular connectivity (z ≥ 0), but low Pcs (<0). Nodes that had low z (<0) were considered as satellite nodes if they had high Pcs (≥0), or as peripheral nodes if their Pcs was low (<0) (Guimerà and Amaral, 2005, Meunier et al., 2009, Schedlbauer and Ekstrom, 2019). The roles were determined based on final group modular partition for each state and corresponding across-subjects mean FC matrices for group-level roles and corresponding individual FC matrices for individual-level roles. Nodes and their respective roles were grouped based on the RSN network nodes belonged to. State reconfiguration in terms of integration was tested by analyzing the number of connector hubs between two states in healthy participants using the Mann-Whitney for each network using the individual level data (i.e. number of connector hubs in each participant). We also analyzed the distribution of roles in the “mover” regions in each group using Chi-square Goodness of Fit test.

The same analysis was performed in TLE patients to test the reorganization of LMN configurations and state reconfiguration. Moreover, to test the difference between LMN state reconfiguration and its reorganization in terms of integration, we compared the change in the number of connector hubs within each RSN network between the groups. To this end, the change in the number of connector hubs was calculated as the difference between number of connector hubs during the rest and during the recall task for each network and each participant. Finally, the Mann–Whitney U test was used to test the differences in the connector hub change between the healthy participants and TLE patients for each network separately. The results were FDR corrected for multiple comparisons.

For the purpose of analyzing in more details a possible disorganization of LMN integration in patients, we calculated a specific graph theory parameter, the hub disruption index (HDI, for details on calculation of this index see Achard et al., 2012, Roger et al., 2020a) with Pcs values for LMN configurations. The HDI can indicate whether the integration property of a specific region or node is increased or decreased. The HDI was first calculated on a group level using in order to compare groups and check if there is a general reorganization or disruption of inter-modular integration in TLE patients. The groups were compared with a two-sample t-test. Additionally, we calculated the HDI on regional level in order to identify regions that show the highest increase/decrease of inter-modular connectivity between the groups.