The Primary Sensation Network Drives Time-resolved Functional Fluctuation Patterns during Resting
Poster No:
1517
Submission Type:
Abstract Submission
Authors:
Dezhi Jin1, Ye He1
Institutions:
1School of Artificial Intelligence, Beijing University of Posts and Telecommunications, Beijing, China
First Author:
Dezhi Jin
School of Artificial Intelligence, Beijing University of Posts and Telecommunications
Beijing, China
School of Artificial Intelligence, Beijing University of Posts and Telecommunications
Beijing, China
Co-Author:
Ye He
School of Artificial Intelligence, Beijing University of Posts and Telecommunications
Beijing, China
School of Artificial Intelligence, Beijing University of Posts and Telecommunications
Beijing, China
Introduction:
A recent study developed a temporal unwrapping method to decompose functional connectivity (FC) into frame-by-frame co-fluctuation patterns based on edge timeseries (ETS) [1,2]. The investigation revealed that moments ("events") characterized by high-amplitude co-fluctuation contained more brain network information that might drive FC than low-amplitude moments [3]. However, other studies found that a range of co-fluctuation amplitudes synchronized across subjects during movie watching [4], and could predict cognitive performance [5]. This suggests that a broad range of co-fluctuation moments may hold specific neural significance other than only the high-amplitude events are important [6]. Therefore, it is worth systematically exploring how the co-fluctuation patterns differ in encapsulating brain network information and what causes variations in whole-brain co-fluctuation across timeseries.
Methods:
We preprocessed resting-state fMRI data from 100 unrelated subjects from the Human Connectome Project. Followed the previous approach [2], we computed the element-wise product of the z-scored BOLD timeseries between pairs of brain regions, which generated a co-fluctuation matrix at each time point. Then, we calculated root sum square (RSS) of the co-fluctuation values across region pairs at each time point, quantifying the co-fluctuation amplitude（Fig. 1A）. The timepoints were ranked based on their RSS and divided into 20 bins.
FC matrices were derived by averaging co-fluctuation matrices either across the entire timeseries (static FC networks) or within each bin (bin FC networks), which recovers the Pearson correlation coefficient. Various network topological properties were calculated for each bin FC network.
To evaluate the contribution of each subnetwork to the whole-brain co-fluctuation amplitudes, we defined the ratio of subnetwork-specific RSS to the whole-brain RSS, which was normalized by the number of edges in each subnetwork. We constructed a null model by applying the cyclic shift operator to ETS to examine the significance.
FC matrices were derived by averaging co-fluctuation matrices either across the entire timeseries (static FC networks) or within each bin (bin FC networks), which recovers the Pearson correlation coefficient. Various network topological properties were calculated for each bin FC network.
To evaluate the contribution of each subnetwork to the whole-brain co-fluctuation amplitudes, we defined the ratio of subnetwork-specific RSS to the whole-brain RSS, which was normalized by the number of edges in each subnetwork. We constructed a null model by applying the cyclic shift operator to ETS to examine the significance.
Results:
Compared to static FC network, high-amplitude FC networks exhibited stronger connectivity in primary sensation networks (visual, somatomotor, dorsal attention, and ventral attention networks) while low-amplitude FC networks displayed the opposite pattern (Fig. 1D,F,G). It emphasized the role of primary sensation networks in the disparity of FC networks across different co-fluctuation amplitudes.
Examining the global topological properties, we observed that the brain network became more integrated and less segregated as the co-fluctuation amplitude decreased in Fig 2A. In local topological properties, the clustering coefficient of all regions decreased with decreasing co-fluctuation amplitude in Fig 2B. The clustering coefficient and degree findings revealed a pattern of separation between primary and high-level networks. Within the control and default mode network, regions exhibited higher degrees in low-amplitude bins compared to high-amplitude bins (Fig 2C).
Fig. 2D illustrates the contribution of subnetworks to the whole-brain co-fluctuation. The results indicated that the visual, somatomotor and dorsal attention networks drove the co-fluctuation in the high-amplitude bins, while the limbic network influenced low-amplitude bins (Fig 2D). This suggests that the events were predominantly driven by the primary sensation networks.
Examining the global topological properties, we observed that the brain network became more integrated and less segregated as the co-fluctuation amplitude decreased in Fig 2A. In local topological properties, the clustering coefficient of all regions decreased with decreasing co-fluctuation amplitude in Fig 2B. The clustering coefficient and degree findings revealed a pattern of separation between primary and high-level networks. Within the control and default mode network, regions exhibited higher degrees in low-amplitude bins compared to high-amplitude bins (Fig 2C).
Fig. 2D illustrates the contribution of subnetworks to the whole-brain co-fluctuation. The results indicated that the visual, somatomotor and dorsal attention networks drove the co-fluctuation in the high-amplitude bins, while the limbic network influenced low-amplitude bins (Fig 2D). This suggests that the events were predominantly driven by the primary sensation networks.
·Fig. 1. ETS and bin FC Networks
·Fig. 2. Network Topological Properties and Contribution of Subnetworks to Whole-Brain Co-Fluctuation
Conclusions:
Our study systematically examined the functional organization patterns of the brain across varying states of co-fluctuation amplitudes and revealed that the primary sensation network drove the high amplitude of whole-brain co-fluctuation. These findings provided new insights into understanding the mechanism underlying functional connectivity and establishing a solid foundation for further exploring time-resolved functional fluctuations.
Modeling and Analysis Methods:
Activation (eg. BOLD task-fMRI) 2
Connectivity (eg. functional, effective, structural) 1
fMRI Connectivity and Network Modeling
Methods Development
Physiology, Metabolism and Neurotransmission :
Neurophysiology of Imaging Signals
Keywords:
FUNCTIONAL MRI
Somatosensory
Source Localization
Systems
1|2Indicates the priority used for review
Provide references using author date format
[2] Betzel, R.F. (2022), 'Individualized event structure drives individual differences in whole-brain functional connectivity, NeuroImage, vol. 252, p. 118993
[3] Zamani Esfahlani, F. (2020), 'High-amplitude cofluctuations in cortical activity drive functional connectivity', Proceedings of the National Academy of Sciences, vol. 117, no. 45, pp. 28393-28401
[4] Tanner, J.C. (2023), 'Synchronous high-amplitude co-fluctuations of functional brain networks during movie-watching', Imaging Neuroscience, vol. 1
[5] Wehrheim, M.H. (2022), 'Few temporally distributed brain connectivity states predict human cognitive abilities', NeuroImage, vol. 277, p. 120246
[6] Ladwig, Z. (2022), 'BOLD cofluctuation ‘events’ are predicted from static functional connectivity', NeuroImage, vol. 260, p. 119476