Directed Network Mapping (DNM) models the electrical propagation within the atria using a directed graph. With DNM, reentries of different atrial flutter (AFL) mechanisms are associated with closed path in the network. Since the noise in the endocavitary recordings may hamper the identification of cycles, this study aimed to quantify whether graph-based centrality measures could identify nodes associate to reentries without the need of actual cycle identification. Endocavitary recordings from 10 patients with complex AFL of five distinct mechanisms were considered. Centrality measures included betweenness (B), harmonic centrality, and two derived from PageRank and Hyperlink-Induced Topic Search. A specific visualization called centrality map was defined by depicting the measures at all nodes. Correlation coefficients were computed between the centrality map and the counting of the cycles passing through each node. Also, centrality maps were evaluated for detecting reentries using ROC analysis. Moderate to strong correlations (>0.5) were found between centrality and cycle count maps. Here, B resulted the most correlated and accurate measure in detecting reentries across different patients (AUC: 0.84 ± 0.05) and mechanisms (AUC: 0.81 ± 0.06). Integrating centrality measures in DNM may hold potential to characterize AFL mechanisms.