Analysis of Structural Network Topology in Depression using Graph Theory
Neuroimaging studies have suggested a difference in structural brain connectivity in depression. Recently, structural brain connectivity and psychopathology have been studied using graph theory analysis, which provides metrics on properties of brain organization. While there have been some studies applying this analytic technique to study depression, these have largely been done using categorical rather than dimensional approaches to psychopathology. This study applied both a traditional categorical approach and a dimensional approach to examine the relation between commonly used graph theory measures and depression. The dimensional analysis included 439 subjects and the categorical approach included 357 subjects with depressive symptoms and 82 subjects without any diagnoses. Anatomical co-variance matrices were constructed using 9 morphometric features and matrices were analyzed to produce the following metrics: normalized clustering coefficient, normalized path length, small-world parameter, normalized global efficiency, and normalized local efficiency. The categorical approach utilized an ANCOVA and the dimensional approach utilized multiple regressions. The categorical analysis did not suggest a significant difference between the “depressed” and “healthy” group with regards to any of the graph theory metrics. In the dimensional analysis a significant positive relation was identified between depressive symptom counts and both normalized local efficiency and normalized clustering coefficient. This shows some concordance with previous studies, and suggests that global features of white matter microstructure may be relevant for depression when examined dimensionally. Future studies using other types of neuroimaging data and applying graph theory techniques may yield additional insight into which graph theory metrics are most relevant for depression.