Personalized Functional Networks in ABCD Children: Linking Topography with Socioeconomic Status
Poster No:
2205
Submission Type:
Abstract Submission
Authors:
Shaoling Zhao1, Haowen Su1, Jing Cong1, Peiyu Chen1, Guowei Wu1, Qingchen Fan1, Yiyao Ma1, Xiaoyu Xu1, Hang Yang1, Hongming Li2, Adam Pines3, Runsen Chen4, Zaixu Cui1
Institutions:
1Chinese Institute for Brain Research, Beijing, Beijing, China, 2University of Pennsylvania, Philadelphia, PA, 3Psychiatry and Behavioral Sciences, Stanford School of Medicine, Stanford University, Stanford, CA, 4Vanke School of Public Health, Tsinghua University, Beijing, China
First Author:
Co-Author(s):
Adam Pines
Psychiatry and Behavioral Sciences, Stanford School of Medicine, Stanford University
Stanford, CA
Psychiatry and Behavioral Sciences, Stanford School of Medicine, Stanford University
Stanford, CA
Introduction:
Convergent evidence has demonstrated that there is remarkable inter-individual variability in the spatial topography of functional networks, even after alignment of brain structure (Wang, Buckner et al. 2015, Gordon, Laumann et al. 2017, Kong, Li et al. 2019, Cui, Li et al. 2020). More importantly, the variability of functional topography subserves higher-order executive functions and confers to diverse psychopathologies (Sydnor, Larsen et al. 2021, Tooley, Bassett et al. 2021). If individual differences are not accounted for, the estimation of topographic variations could be aliased into between-network connectivity, potentially biasing both inference and interpretation.
Here, based on a larger-scale cohort, the adolescent brain cognitive development (ABCD) study (Volkow, Koob et al. 2018), we aim to delineated 17 personalized functional networks for ABCD children to accelerate the understanding of personalized functional topography with diverse environmental, cognitive, and psychopathological factors.
Here, based on a larger-scale cohort, the adolescent brain cognitive development (ABCD) study (Volkow, Koob et al. 2018), we aim to delineated 17 personalized functional networks for ABCD children to accelerate the understanding of personalized functional topography with diverse environmental, cognitive, and psychopathological factors.
Methods:
We delineated personalized functional topography for 3921 ABCD participants (9 and 10-year-olds) who had high-quality resting-state fMRI data with at least 20 minutes. Using an advanced machine learning technique, spatially regularized non-negative matrix factorization (NMF)(Li, Satterthwaite et al. 2017), we parcellated the cortex into 17 functional networks for each ABCD child based on their own fMRI data. As an exemplary usage, we further examining how individual variation in cortical structure of functional networks was related to childhood socioeconomic status (SES). We used partial least square regression (PLS-R) and nested two-fold cross validation (2F-CV) to evaluate whether the multivariate pattern of functional topography could be used to predict unseen individuals' SES.
Results:
We parcellated the cortex into 17 networks based on previous studies (Yeo, Krienen et al. 2011, Kong, Li et al. 2019, Cui, Li et al. 2020) for each ABCD child and publicly shared this resource (https://zenodo.org/records/10200111). By comparing the overlap with priori canonical functional atlases (Yeo, Krienen et al. 2011, Cui, Li et al. 2020), our networks were named as two visual, three somatomotor, one auditory, three dorsal attention, one ventral attention, two fronto-parietal, three default mode, one temporal-parietal, and one limbic networks (Figure 1). These networks exhibited distinct spatial topography across individuals (Figure 2A), with maximum variability in higher-order association networks and lowest in sensorimotor networks, which is consistent with prior studies in adults (Wang, Buckner et al. 2015, Gordon, Laumann et al. 2017) and youths(Cui, Li et al. 2020) (Figure 2B).
Compelling evidence suggest that lower SES in childhood impacts brain development and potentialy increases risks for mental disorders (Luciana, Barch et al. 2023, Sydnor, Larsen et al. 2023). Therefore, we also evaluated brain-wide association between the multivariate pattern of functional topography with individuals' SES. For 3,198 participants who have complete SES measures, we employed PLS-R and 2F-CV to predict individual's SES and found that the personalized brain networks could significantly predict unseen individuals' SES (median Pearson's r=0.26, permutation p<0.001, Figure 2C). By examining the total contribution weights of each cortical vertex, we observed that the occipital-temporal, parietal, and prefrontal areas contributed most to the prediction (Figure 2E), and these multivariate patterns of feature weights were constrained by individual topographic variability (Spearman correlation, r=0.45, spin test p<0.001)(Figure 2F).
Compelling evidence suggest that lower SES in childhood impacts brain development and potentialy increases risks for mental disorders (Luciana, Barch et al. 2023, Sydnor, Larsen et al. 2023). Therefore, we also evaluated brain-wide association between the multivariate pattern of functional topography with individuals' SES. For 3,198 participants who have complete SES measures, we employed PLS-R and 2F-CV to predict individual's SES and found that the personalized brain networks could significantly predict unseen individuals' SES (median Pearson's r=0.26, permutation p<0.001, Figure 2C). By examining the total contribution weights of each cortical vertex, we observed that the occipital-temporal, parietal, and prefrontal areas contributed most to the prediction (Figure 2E), and these multivariate patterns of feature weights were constrained by individual topographic variability (Spearman correlation, r=0.45, spin test p<0.001)(Figure 2F).
·Figure 1. Group atlas of 17 functional networks in ABCD children
·Figure 2. Personalized functional topography is associated with individual differences in SES
Conclusions:
Overall, our results provide an open resource to improve the exploration of brain function for basic and clinical research that accounts for individual difference in functional network topography, with SES as an example to offer a potential explanation for how environmental factors impacts on later-life outcomes.
Lifespan Development:
Early life, Adolescence, Aging 2
Modeling and Analysis Methods:
fMRI Connectivity and Network Modeling
Segmentation and Parcellation
Neuroinformatics and Data Sharing:
Brain Atlases 1
Databasing and Data Sharing
Keywords:
Cortex
Machine Learning
Open Data
Other - Adolescence; Personalized functional network topography; Adolescent Brain Cognitive Development; Socioeconomic status
1|2Indicates the priority used for review
Provide references using author date format
Gordon, E. M., et al. (2017). "Precision Functional Mapping of Individual Human Brains." Neuron 95(4): 791-807 e797.
Kong, R., et al. (2019). "Spatial Topography of Individual-Specific Cortical Networks Predicts Human Cognition, Personality, and Emotion." Cerebral Cortex 29(6): 2533-2551.
Li, H., et al. (2017). "Large-scale sparse functional networks from resting state fMRI." Neuroimage 156: 1-13.
Luciana, M., et al. (2023). "Adolescent brain cognitive development study: Longitudinal methods, developmental findings, and associations with environmental risk factors." Dev Cogn Neurosci: 101311.
Sydnor, V. J., et al. (2021). "Neurodevelopment of the association cortices: Patterns, mechanisms, and implications for psychopathology." Neuron.
Sydnor, V. J., et al. (2023). "Intrinsic activity development unfolds along a sensorimotor-association cortical axis in youth." Nat Neurosci.
Tooley, U. A., et al. (2021). "Environmental influences on the pace of brain development." Nat Rev Neurosci 22(6): 372-384.
Volkow, N. D., et al. (2018). "The conception of the ABCD study: From substance use to a broad NIH collaboration." Developmental Cognitive Neuroscience 32: 4-7.
Wang, D., et al. (2015). "Parcellating cortical functional networks in individuals." Nat Neurosci 18(12): 1853-1860.
Yeo, B. T., et al. (2011). "The organization of the human cerebral cortex estimated by intrinsic functional connectivity." J Neurophysiol 106(3): 1125-1165.