An Effective “Traffic Map” of the Human Brain via Unified Structural and Functional Connectivity

1516 

Abstract Submission 

Arzu C Has Silemek^1,2, Haitao Chen², Pascal Sati^1,2, Wei Gao²

¹Department of Neurology, Cedars-Sinai Medical Center, Los Angeles, CA, ²Biomedical Imaging Research Institute (BIRI), Cedars-Sinai Medical Center, Los Angeles, CA

Arzu C Has Silemek  
Department of Neurology, Cedars-Sinai Medical Center|Biomedical Imaging Research Institute (BIRI), Cedars-Sinai Medical Center
Los Angeles, CA|Los Angeles, CA

Haitao Chen  
Biomedical Imaging Research Institute (BIRI), Cedars-Sinai Medical Center
Los Angeles, CA

Pascal Sati  
Department of Neurology, Cedars-Sinai Medical Center|Biomedical Imaging Research Institute (BIRI), Cedars-Sinai Medical Center
Los Angeles, CA|Los Angeles, CA

Wei Gao  
Biomedical Imaging Research Institute (BIRI), Cedars-Sinai Medical Center
Los Angeles, CA

Conventional studies of the human brain connectome utilize either structural or functional connectivity measure, but few studies have combined them for a better understanding of how the brain is effectively wired (Zamani Esfahlani, Faskowitz et al. 2022). In this study, we viewed structural connections (SC) as the brain's "road" system, with strong SC representing efficient "road segments" over short distances and weaker SCs indicating greater "cost" over longer distances. Functional connections (FC) were considered as "effective traffic" between different pairs of regions. To bridge the two, each "traffic"/FC would choose to use the most efficient "route" defined by one of more steps of direct SCs that collectively have the least "cost" (i.e., anatomical distance/SC). By unifying structural and functional connectivity (USFC) and defining all structural "routes" that support most efficient functional communication, we created an effective "traffic map" of the human brain. The mostly used routes/road segments of this traffic system were highlighted and we further characterized the relationship between SC and FC in these routes as well as their global efficiency properties.

Data from a cohort of 394 subjects [59% female, age range: 22 – 36] from the Human Connectome Project were used. After constructing SC and FC measures based on the AAL template(Jeurissen, Tournier et al. 2014, Fenske, Liu et al. 2023), the most "efficient" pathway for each FC in each subject was identified by summing the cost of each "step" (i.e., the Euclidean distance divided by the strength of direct SC between a pair of regions) along all potential routes (up to 4 steps were searched) and choosing the one with the least "cost". A USFC value for each "road segment"/direct FC was then calculated as the sum of all FC values that use this segment in their respective routes, essentially quantifying the amount of "traffic" on this "road segment". Next, we examined nodal-level functional-structural coupling at the group level across all routes, focusing on those that are consistent in over 50% of the subjects. Finally, global efficiency of the whole brain system was calculated and compared across each metric (i.e., FC, SC and USFC) via Dijkstra algorithm.

The effective "traffic map"/ "route system" of the human brain was shown in Fig. 1A while the top 10 "most heavily used" segments were shown in Fig. 1b. These segments concentrate on the default mode network, the salience network, along with subcortical and visual regions (Fig. 1B). For SC-FC relationships across all routes, slightly positive (for 1-step routes) (Fig. 1D) or non-significant correlations (for 2 and 3-step routes) (Fig. 1F & Fig. 1H) were observed for routes supporting positive FCs, which is consistent with previous findings. Intriguingly, more significant, and stronger negative associations were identified for routes underlying negative FCs for all routes raging from 1 to 3 steps (Fig. 1C, Fig. 1E & Fig. 1G), which has never been reported. Note no common patterns (i.e., shared by >50% of subjects) emerged for 4-step connections so they were not evaluated. When the global efficiencies were quantified for the whole brain connectome, those supported by USFC were significantly higher than those based on either FC or SC alone (p < 0.001) (Fig. 1I).

Overall, we introduced USFC as a novel approach and created the brain's first effective "traffic map" highlighting most heavily used "road segments" from the brain's default-mode network, salience network, and visual/subcortical regions. Our analysis revealed a notable negative correlation between negative FC and underlying SC for the first time, suggesting that stronger negative FC needs support from stronger structural connectivity. Finally, the USFC system showed much elevated global efficiency of the whole brain connectome compared to FC or SC alone, underscoring the brain's better efficiency when characterized by this novel effective "traffic map".

Connectivity (eg. functional, effective, structural) ¹

Other Methods ²

ADULTS

FUNCTIONAL MRI

MRI

STRUCTURAL MRI

Other - Effective traffic map