Graph theoretical structural connectome analysis of the brain in patients with chronic spinal cord injury: preliminary investigation.

Abstract

Retrospective study. We aimed to characterize the convergent disruptions of the structural connectivity based on network modeling technique (i.e., graph theory) to identify significant changes in network organization/reorganization between uninjured and chronic spinal cord injury (SCI) participants. USA. Ten adult participants including 4 with chronic SCI and 6 uninjured were scanned using a multi-shell diffusion imaging on a 3.0 T MR scanner. Whole brain structural connectivity matrix was estimated by performing the quantification of the number of white matter fibers (called edges) connecting each possible pair of brain region (called nodes). Brain regions were defined according to Desikan-Killiany cortical atlas. Using connectivity matrix, connectivity strength as well as six different graph theoretical measurements were computed for each participant. They include: (1) global efficiency; (2) local efficiency; (3) degree; (4) betweenness centrality; (5) average shortest length and (6) clustering coefficient. Finally network based statistics was applied to extract nodes/connections with significant differences between groups (uninjured vs SCI). The SCI group showed significant decreases in betweenness centrality in the left precentral gyrus (T-score=2.98, p value=0.02), and the right caudal middle frontal gyrus (score = 2.35, p value=0.047). It also showed significant decrease in left transverse temporal gyrus (T-score=2.36, p value=0.046) in clustering coefficient. In addition, altered regions in the occipital and parietal lobe were also identified. These results suggest that not only local but also global alterations of the white matter occur after SCI. The proposed modeling technique has the potential to serve as a screening tool to identify any areas of the brain affected after SCI.