Exploring brain network features in borderline personality disorder: a graph-based analysis of MR images

A network approach to the modelling and the analysis of functional and structural Magnetic Resonance (MR) images is an increasingly popular technique, because of the solidity of the mathematical theory at its basis, the graph theory, which allows to explore a wide range of network properties. For this work, both Functional Connectivity (FC) and Structural Covariance (SC) networks were constructed. The cohort of participants to this study was composed of 27 subjects affected by the Borderline Personality Disorder (BPD), a mental disorder causing behavioral and emotional dysregulation, and a matching group of 28 healthy controls. Brain networks were analyzed using the methods provided by graph theory, both at global and nodal level, by exploring their topological and organizational properties mainly in terms of centrality, efficiency in information transfer and modularity. The outcomes obtained from such measures, in patients and controls separately, were compared in order to find statistically significant differences between the two groups, that may be characteristic of the disease. Additionally, the outcomes of the topological quantities were correlated with a series of clinical scores, evaluating the neuro-psychological condition of the subjects. The results show significant differences between patients and controls mostly in the FC networks and especially located in the limbic system of the brain, which indeed has a fundamental role in emotion regulation. Node-specific variations tend to involve the amygdala, the caudal anterior cingulate cortex, the entorhinal cortex and the temporal pole. Such evident results were not retrieved from the SC networks, though they still supported a greater variability within the limbic system. Therefore, the analysis of brain graphs allowed to achieve the detection of topological alterations in a young psychiatric population, which would be interesting to monitor in time.