Higher-order interaction of brain microstructural and functional connectome

1584 

Abstract Submission 

Hao Wang¹, Hui-jun Wu², Yang-Yu Liu³, Linyuan Lü⁴

¹School of Physics and Optoelectronic Engineering, Hainan University, Haikou, Hainan, ²School of Media & Communication, Shanghai Jiao Tong University, Shanghai 200240, P. R. China., Shanghai, Shanghai, ³Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, ⁴University of Science and Technology of China, Hefei, Anhui

Hao Wang  
School of Physics and Optoelectronic Engineering, Hainan University
Haikou, Hainan

Hui-jun Wu  
School of Media & Communication, Shanghai Jiao Tong University, Shanghai 200240, P. R. China.
Shanghai, Shanghai

Yang-Yu Liu  
Brigham and Women’s Hospital and Harvard Medical School
Boston, MA

Linyuan Lü  
University of Science and Technology of China
Hefei, Anhui

Despite a relatively fixed anatomical structure, the human brain can support rich cognitive functions, triggering particular interest in investigating structure-function relationships (Honey, Sporns et al. 2009). Myelin is a vital brain microstructure marker; however, most myelin studies have constructed a structural covariance network at the population level, making individual cognitive or behavioral predictions impossible. Therefore, examining the myelin microstructural and functional relationship at the individual level is urgently needed but is still elusive. Recently, higher-order representations (beyond the node or edge level) emerged, including simplicial complexes (Giusti, Pastalkova et al. 2015), persistent homology (Liang and Wang 2017), neural network (Suárez, Richards et al. 2021), subgraphs (Przulj 2007), and motifs (Benson, Gleich et al. 2016), which have proven to be extremely useful in understanding and comparing complex networks. Nevertheless, few studies have examined individual myelin microstructure-function relationships using higher-order representations. Here, we quantify the individual-level microstructure-function relationship using a higher-order framework and explore the microstructure-function higher-order relationship across individual cognitive scores, development and network scale.

MRI data. Dataset: We downloaded unprocessed MR data of 213 participants from the Human Connectome Project (HCP) "S1200" new subjects release (Van Essen, Smith et al. 2013). We excluded eight participants who met any of the following criteria: (a) mean of framewise displacement (mFD) > 0.25 mm; (b) more than 20% of the FDs were above 0.2 mm; and (c) if any FDs were greater than 5 mm (Parkes, Fulcher et al. 2018). Seven participants aged > 36 years were excluded from the study. Finally, we obtained quality-controlled resting-state fMRI (rfMRI), T1-weighted, and T2-weighted images of 198 participants (108 males and 90 females). Here, we constructed the individual level myelin (T1w/T2w) structural network using probability distribution function (PDF)-based morphological methods (Wang, Jin et al. 2016) with Gordon parcellation of 333 brain regions (Gordon, Laumann et al. 2016), and constructed functional networks with the same parcellation as the myelin-structural network. Exploring the brain microstructure-function relationships using a higher-order framework, derived from 2- to 4-node subgraphs (Figure. 1)

Global (network-level) higher-order microstructure-function relationships were negatively correlated with male participants' personality scores and declined with age (P = 0.005). Nodal (node-level) higher-order microstructure-function relationships are not aligned uniformly throughout the brain, being stronger in association cortices and lower in sensory cortices (Figure 2a-b), showing gender differences (P < 0.001, FDR correction), see Figure 2c. Notably, higher-order microstructure-function relationships are maintained from the whole-brain to local circuits, which uncovers a compelling and straightforward principle of brain structure-function complex interactions (Figure. 2d-f). Additionally, targeted artificial attacks can disrupt these higher-order relationships. The main results are robust against several factors, including (I) effect of sparsity thresholding; (II) effect of sample size; (III) split-half reliability for nodal higher-order interactions; (IV) higher-order relationships on other types of networks.

Our results advance our understanding of higher-order structural-function similarity that underlies cognition, individual differences, and aging, as well as provide a framework for measuring the similarity between complex systems.

Connectivity (eg. functional, effective, structural) ¹

fMRI Connectivity and Network Modeling ²

Cortical Anatomy and Brain Mapping

Anatomical MRI

BOLD fMRI

Cognition

Computational Neuroscience

Computing

Data analysis

Modeling

MRI

Myelin

Statistical Methods