Topological Data Analysis reveals robust alterations in the whole-brain and frontal lobe functional connectomes in Attention-Deficit/Hyperactivity Disorder.

Attention Deficit Hyperactivity Disorder (ADHD) is a developmental disorder characterized by difficulty to control the own behavior. Neuroimaging studies have related ADHD with the interplay of fronto-parietal attention systems with the default mode network (DMN) (Castellanos and Aoki, 2016). However, some results have been inconsistent, potentially due to methodological differences in the analytical strategies when defining the brain functional network, i.e., the functional connectivity threshold and/or the brain parcellation scheme. Here, we make use of Topological Data Analysis to explore the brain connectome as a function of the filtration value (i.e., the connectivity threshold), instead of using a static connectivity threshold. Specifically, we characterized the transition from all nodes being isolated to being connected into a single component as a function of the filtration value. We explored the utility of such a method to identify differences between 81 children with attention-deficit/hyperactivity disorder (ADHD; 45 male, age: 7.26 - 17.61 years old) and 96 typically developing children (TDC; 59 male, age: 7.17 - 17.96 years old), using a public dataset of resting state fMRI in human subjects. Results were highly congruent when using four different brain segmentations (atlases), and exhibited significant differences for the brain topology of children with ADHD, both at the whole brain network and the functional sub-network levels, particularly involving the frontal lobe and the default mode network. Therefore, this is a solid approach that complements connectomics-related methods and may contribute to identify the neurophysio-pathology of ADHD. Significant Statement Topological Data Analysis investigates the topology of interacting nodes. It may model the connectomes as a topological process instead of a static graph, exploring the transition of all nodes being isolated to binding together, as a function of the connectivity threshold. Here, we explored three parameters to characterize the algebraic topology of individual connectomes using four different brain atlases, further exploring the subnetwork levels. Our findings showed that the area under the curve robustly differentiates children with attention-deficit/hyperactivity disorder and typically-developing children, suggesting decreased functional segregation, with the greatest effects on the frontal lobe and the default-mode network. Overall, these results support the use of the proposed methods to robustly explore topological differences in the brain connectome.