Functional Implications and Heritability of Gyral Hubs in Gyral Morphological Networks
Poster No:
2130
Submission Type:
Abstract Submission
Authors:
Shengfeng Liu1,2, Zhang Yu2, Tianzi Jiang1,2,3
Institutions:
1Brainnetome Center, Institute of Automation, Chinese Academy of Sciences, Beijing, China, 2Research Center for Augmented Intelligence, Zhejiang Lab, Hangzhou, China, 3School of Artificial Intelligence, University of Chinese Academy of Sciences, Beijing, China
First Author:
Shengfeng Liu
Brainnetome Center, Institute of Automation, Chinese Academy of Sciences|Research Center for Augmented Intelligence, Zhejiang Lab
Beijing, China|Hangzhou, China
Brainnetome Center, Institute of Automation, Chinese Academy of Sciences|Research Center for Augmented Intelligence, Zhejiang Lab
Beijing, China|Hangzhou, China
Co-Author(s):
Tianzi Jiang
Brainnetome Center, Institute of Automation, Chinese Academy of Sciences|Research Center for Augmented Intelligence, Zhejiang Lab|School of Artificial Intelligence, University of Chinese Academy of Sciences
Beijing, China|Hangzhou, China|Beijing, China
Brainnetome Center, Institute of Automation, Chinese Academy of Sciences|Research Center for Augmented Intelligence, Zhejiang Lab|School of Artificial Intelligence, University of Chinese Academy of Sciences
Beijing, China|Hangzhou, China|Beijing, China
Introduction:
The organization of cortical folding patterns are related to brain function, cognition and behaviors [Fischl et al., 2008]. Due to the distinct complexity and high variability of cortical morphology, it has been a challenging task to quantitatively model the cortical organization patterns. To this end, we propose a graph-based representation of human cortical organization-gyral morphological network (GMN) -which is constructed by an adaptive fusion model by integrating multiple morphological features, e.g. curvature and sulcal depth. More importantly, we uncovered inheritable signatures among functional brain networks and laminar areas through analyzing the properties of the nodes of the GMN.
Methods:
Inspired by the Gyral Net (GN) recently proposed by [Chen et al., 2017], we developed a Fast and Adaptive algorithm for Constructing Gyral Morphological Networks (FAC-GMN) by integrating multiple morphological metrics such as cortical thickness, curvature and sulcal depth. The FAC-GMN model consists of five steps: (1) feature fusion, (2) gyral crest segmentation, (3) distance transform, (4) tree marching, and (5) tree connection, as illustrated in Fig. 1(B-F). In this study, we are especially concerned with nodes of the GMN, i.e., vertices where three or more edges (gyral crest lines) intersect, called gyral morphological hubs (GMHs). Given that our previous work demonstrated significant structural and functional differences between these gyral hubs and other gyral regions and sulci [Liu et al., 2022], we further analyzed the relationship between the spatial distribution of these hubs and the intrinsic functional networks and laminar areas, and their heritability (Fig.1G), thereby discovering inheritable signatures among functional brain networks and laminar areas.
·Figure 1. The whole computational analysis framework
Results:
Experiments on 1081 young adults from HCP dataset demonstrated that the proposed FAC-GMN and the GN method constructed roughly similar cortical architecture. But our method obtained better results in detailed architecture, especially in some cortical regions where the contours of gyral crests are not particularly clear, as shown in Fig. 2A. In addition, the FAC-GMN method was more flexible and efficient than the GN, because the algorithm is fully adaptive. Specifically, the FAC-GMN method revealed more gyral morphological hubs, longer connected gyral crests, better gyral network integrity, shorter execution time (i.e., faster), and higher inter-individual variability.
By analyzing the spatial distribution of gyral hubs in the functional networks [Thomas Yeo et al., 2011] and laminar areas [Mesulam, 1998], we found that both in left and right hemisphere, more gyral hubs are detected in the heteromodal areas including limbic and default mode networks (Fig 2B). Our findings support that these morphological hubs could serve as structural and functional hubs in brain networks, and are involved in more functional networks compared to other gyral areas, especially those related to high-order cognitive functions [Zhang et al., 2020].
Heritability analysis of the GMHs distributed in functional networks and laminar areas showed that, compared to the heteromodal areas, the spatial distribution of the GMHs in unimodal areas are more heritable (h2=0.26, p=0.0003 vs. h2=0.14, p=0.02).
By analyzing the spatial distribution of gyral hubs in the functional networks [Thomas Yeo et al., 2011] and laminar areas [Mesulam, 1998], we found that both in left and right hemisphere, more gyral hubs are detected in the heteromodal areas including limbic and default mode networks (Fig 2B). Our findings support that these morphological hubs could serve as structural and functional hubs in brain networks, and are involved in more functional networks compared to other gyral areas, especially those related to high-order cognitive functions [Zhang et al., 2020].
Heritability analysis of the GMHs distributed in functional networks and laminar areas showed that, compared to the heteromodal areas, the spatial distribution of the GMHs in unimodal areas are more heritable (h2=0.26, p=0.0003 vs. h2=0.14, p=0.02).
·Figure 2. Experimental Results
Conclusions:
Results suggest that the proposed FAC-GMN model outperforms the classical gyral net as well as single-feature gyral networks in terms of the length and integrity of gyral networks and capturing more gyral morphological hubs. Besides, more gyral hubs were detected in the heteromodal areas including limbic and default mode networks, which were also significantly inheritable among twins. This study provides new avenues to study the gyrification patterns of cerebral cortex in neuro-development and aging and toward better understanding the neural basis of human cognition.
Modeling and Analysis Methods:
Multivariate Approaches 2
Other Methods
Neuroanatomy, Physiology, Metabolism and Neurotransmission:
Cortical Anatomy and Brain Mapping 1
Keywords:
Cortex
Data analysis
Machine Learning
Modeling
MRI
Multivariate
STRUCTURAL MRI
Structures
1|2Indicates the priority used for review
Provide references using author date format
Fischl B, Rajendran N, Busa E, Augustinack J, Hinds O, Yeo BTT, Mohlberg H, Amunts K, Zilles K (2008): Cortical Folding Patterns and Predicting Cytoarchitecture. Cereb Cortex 18:1973–1980.
Liu S, Ge F, Zhao L, Wang T, Ni D, Liu T (2022): NAS-optimized topology-preserving transfer learning for differentiating cortical folding patterns. Medical Image Analysis 77:102316.
Mesulam MM (1998): From sensation to cognition. Brain 121:1013–1052.
Thomas Yeo BT, Krienen FM, Sepulcre J, Sabuncu MR, Lashkari D, Hollinshead M, Roffman JL, Smoller JW, Zöllei L, Polimeni JR, Fischl B, Liu H, Buckner RL (2011): The organization of the human cerebral cortex estimated by intrinsic functional connectivity. J Neurophysiol 106:1125–1165.
Zhang T, Li X, Jiang X, Ge F, Zhang S, Zhao L, Liu H, Huang Y, Wang X, Yang J, Guo L, Hu X, Liu T (2020): Cortical 3-hinges could serve as hubs in cortico-cortical connective network. Brain Imaging and Behavior 14:2512–2529.