Probing the Neural Circuits Underlying the Social Brain Using Diffusion Tractography and Graph Theory

Background: Recent functional imaging, lesion and neurophysiological studies have identified a group of functionally associated brain areas as the neural basis for social cognition, which are commonly referred to as the social brain. While social deficits are the hallmark of autism spectral disorders (ASD)(Ditcher, 2012), work on systematically mapping the underlying networks connecting these brain regions and studying the topological characteristics of the networks is lacking.  

Objectives: The objectives of the current study are to dissect the aforementioned anatomical network using diffusion tractography and to probe its structural architecture using graph theory methods.  

Methods:   Subjects: 20 healthy right-handed human subjects (age: 39.3±11.9 yrs, 10 females) were included in the analyses. MRI data acquisition:  MRI data were obtained using a 3T Siemens Trio Tim scanner. The imaging parameters for T1w and diffusion MRI can be found in our previous work(Li, 2013). Social brain network reconstruction: As no single task can concurrently activate all brain regions underlying social processing, we chose our nodes based on the collated coordinates of activated brain areas in previous task fMRI studies(Overwalle, 2009). The fusiform face area (FFA), the superior temporal sulcus (STS), the inferior parietal lobe (IPL) and premotor cortex (PMC), the temporoparietal junction (TPJ), precuneus (PC), ventromedial prefrontal cortex (vmPFC) and amygdala (AMG) were included in the analyses. Spherical regions-of-interest (ROIs) centered at each coordinate were generated,  projected onto a surface-based template in FreeSurfer, and then transformed to each individual’s space. We used probabilistic diffusion tractography, implemented in FSL, to track the anatomical connections among the sixteen social brain ROIs. After the connections between each ROI pair were derived, the thresholded tract volumes were employed as an index of connectivity strength between seed ROI pairs(Li, 2012). In graph theoretic analyses, we used four complementary measures to indicate whether a node plays a central role in a network (also called hub)(Li, 2013). Nodes that were ranked in the top percentile in the majority of the four centrality measures were identified as putative hubs in the network. Results:  

Results:  The spatial trajectories of the pathways mapped using diffusion tractography were consistent with those identified using invasive tracer methods in monkeys. Based on the four centrality measures, the most frequently occurring putative hub under varying thresholds was the left AMG, followed by the right PC and right AMG.

Conclusions:   We mapped the anatomical connections among sixteen social brain areas using diffusion probabilistic tractography and then analyzed the structural architecture of the network using graph theory methods. We identified bilateral AMG and the right PC as the most frequently identified hubs. Our AMG findings are consistent with electrophysiological, lesion, and functional imaging studies demonstrating the critical role of the amygdala in social perception and cognition. Our network approach may provide a new avenue for examining function and dysfunction among the social brain network in ASD.