Quantification of the interaction dynamics between brain regions through graph neural network method
Poster No:
1405
Submission Type:
Abstract Submission
Authors:
Shaoxian Li1, Debin Zeng2, Xiaoxi Dong1, Yirong He1, Shuyu Li1
Institutions:
1State Key Laboratory of Cognitive Neuroscience and Learning, Beijing Normal University, Beijing, China, 2Beihang University, Beijing, China
First Author:
Shaoxian Li
State Key Laboratory of Cognitive Neuroscience and Learning, Beijing Normal University
Beijing, China
State Key Laboratory of Cognitive Neuroscience and Learning, Beijing Normal University
Beijing, China
Co-Author(s):
Xiaoxi Dong
State Key Laboratory of Cognitive Neuroscience and Learning, Beijing Normal University
Beijing, China
State Key Laboratory of Cognitive Neuroscience and Learning, Beijing Normal University
Beijing, China
Yirong He
State Key Laboratory of Cognitive Neuroscience and Learning, Beijing Normal University
Beijing, China
State Key Laboratory of Cognitive Neuroscience and Learning, Beijing Normal University
Beijing, China
Shuyu Li
State Key Laboratory of Cognitive Neuroscience and Learning, Beijing Normal University
Beijing, China
State Key Laboratory of Cognitive Neuroscience and Learning, Beijing Normal University
Beijing, China
Introduction:
In computational neuroscience, one can characterize the brain's dynamic activities using potential causal mechanisms and complex dynamics analysis(Corinto & Torcini, 2019; Xu et al., 2023). One approach to analyzing brain dynamics involves the computational models of large-scale brain network dynamics(Panahi et al., 2021). Network-based approaches(Byrne et al., 2020; Deco et al., 2008; Demirtaş et al., 2019; Dunstan et al., 2023) posit that the states of brain region nodes are dynamic over time, while connections are comparatively static. It may introduce oversimplification issues. Some researchers attempt to express connective edges as dynamically evolving time series(Betzel et al., 2023). Information transmission between brain regions should occur through structural connections, so we must combine them. Interaction networks (INs) (Cranmer et al., 2020)based on message passing methods enable the dynamic inference of interaction information between nodes within the network. In this study, we introduced a model based on INs to quantify inter-regional interaction (IRI) and we found it can distinguish AD patients from healthy controls.
Methods:
Dataset. This study utilizes two batches of data, obtained from the Human Connectome Project (HCP) by the National Institutes of Health (NIH)(Van Essen et al., 2012) and from the Neurology Department at Xuanwu Hospital, Capital Medical University in China. We enrolled 41 patients diagnosed with AD and 42 healthy individuals as normal controls. We acquired resting-state fMRI and diffusion magnetic resonance imaging (dMRI) data.
INs Models. The model comprises two equations: an interaction equation and a node state equation. Using the current state as input, the interaction equation calculates the interaction information that each brain region exerts on other regions. The interaction information received by each brain region, combined with the brain region's own state, serves as the input to the node state equation, which outputs the subsequent state of each brain region. Both equations are implemented using multilayer perceptrons, and the overall model is constructed based on a graph neural network (GNN) framework utilizing a message passing architecture.
Inter-regional interaction (IRI). The dynamic outputs of the interaction equation are extracted as the interaction information between each pair of brain regions.
INs Models. The model comprises two equations: an interaction equation and a node state equation. Using the current state as input, the interaction equation calculates the interaction information that each brain region exerts on other regions. The interaction information received by each brain region, combined with the brain region's own state, serves as the input to the node state equation, which outputs the subsequent state of each brain region. Both equations are implemented using multilayer perceptrons, and the overall model is constructed based on a graph neural network (GNN) framework utilizing a message passing architecture.
Inter-regional interaction (IRI). The dynamic outputs of the interaction equation are extracted as the interaction information between each pair of brain regions.
Results:
This study constructed individual models for each participant, calculating the Mean Absolute Error (MAE) between the predicted and the actual fMRI signals. The whole-brain average MAE is 0.04 and our model can recreate the functional relationships between different brain regions across the whole brain.
As for the IRI, we found that longer fiber bundle lengths corresponded to larger absolute values of information transmission between brain regions, potentially indicating the importance of long-range connections. The proportion of negative values in the IRI within different functional sub-networks decreases with hierarchical descent. This may suggest that the positive and negative values of the IRI can reflect the excitatory/inhibitory to some extent.
The prediction error of the model was larger in the AD group compared to the NC group, which may be due to the abnormal brain function caused by AD. The high-frequency information transmission between regions was notably elevated in patients compared to the control group, especially in DMN and FN.
As for the IRI, we found that longer fiber bundle lengths corresponded to larger absolute values of information transmission between brain regions, potentially indicating the importance of long-range connections. The proportion of negative values in the IRI within different functional sub-networks decreases with hierarchical descent. This may suggest that the positive and negative values of the IRI can reflect the excitatory/inhibitory to some extent.
The prediction error of the model was larger in the AD group compared to the NC group, which may be due to the abnormal brain function caused by AD. The high-frequency information transmission between regions was notably elevated in patients compared to the control group, especially in DMN and FN.
Conclusions:
Our study developed a whole-brain activity model based on INs and proposed IRI to evaluate information exchange between different brain regions. Using this metric, we found a potential correlation between the output information of distinct brain regions and the E-I ratio. Furthermore, aberrant IRIs in AD patients may underlie their cognitive impairments. This study provides novel approaches and perspectives for investigating the dynamic changes in brain region connectivity.
Disorders of the Nervous System:
Neurodegenerative/ Late Life (eg. Parkinson’s, Alzheimer’s)
Modeling and Analysis Methods:
Classification and Predictive Modeling 2
Connectivity (eg. functional, effective, structural)
Diffusion MRI Modeling and Analysis
fMRI Connectivity and Network Modeling 1
Keywords:
Computational Neuroscience
Degenerative Disease
Immitation
Machine Learning
Modeling
1|2Indicates the priority used for review
Abstract Information
By submitting your proposal, you grant permission for the Organization for Human Brain Mapping (OHBM) to distribute your work in any format, including video, audio print and electronic text through OHBM OnDemand, social media channels, the OHBM website, or other electronic publications and media.
The Open Science Special Interest Group (OSSIG) is introducing a reproducibility challenge for OHBM 2025. This new initiative aims to enhance the reproducibility of scientific results and foster collaborations between labs. Teams will consist of a “source” party and a “reproducing” party, and will be evaluated on the success of their replication, the openness of the source work, and additional deliverables. Click here for more information. Propose your OHBM abstract(s) as source work for future OHBM meetings by selecting one of the following options:
Please indicate below if your study was a "resting state" or "task-activation” study.
Healthy subjects only or patients (note that patient studies may also involve healthy subjects):
Was this research conducted in the United States?
Were any human subjects research approved by the relevant Institutional Review Board or ethics panel? NOTE: Any human subjects studies without IRB approval will be automatically rejected.
Were any animal research approved by the relevant IACUC or other animal research panel? NOTE: Any animal studies without IACUC approval will be automatically rejected.
Please indicate which methods were used in your research:
For human MRI, what field strength scanner do you use?
Which processing packages did you use for your study?
Provide references using APA citation style.
Byrne, Á., O’Dea, R. D., Forrester, M., Ross, J., & Coombes, S. (2020). Next-generation neural mass and field modeling. Journal of Neurophysiology, 123(2), 726–742. https://doi.org/10.1152/jn.00406.2019
Corinto, F., & Torcini, A. (Eds.). (2019). Nonlinear Dynamics in Computational Neuroscience. Springer International Publishing. https://doi.org/10.1007/978-3-319-71048-8
Cranmer, M., Sanchez-Gonzalez, A., Battaglia, P., Xu, R., Cranmer, K., Spergel, D., & Ho, S. (2020). Discovering Symbolic Models from Deep Learning with Inductive Biases (No. arXiv:2006.11287). arXiv. https://doi.org/10.48550/arXiv.2006.11287
Deco, G., Jirsa, V. K., Robinson, P. A., Breakspear, M., & Friston, K. (2008). The Dynamic Brain: From Spiking Neurons to Neural Masses and Cortical Fields. PLoS Computational Biology, 4(8), e1000092. https://doi.org/10.1371/journal.pcbi.1000092
Demirtaş, M., Burt, J. B., Helmer, M., Ji, J. L., Adkinson, B. D., Glasser, M. F., Van Essen, D. C., Sotiropoulos, S. N., Anticevic, A., & Murray, J. D. (2019). Hierarchical Heterogeneity across Human Cortex Shapes Large-Scale Neural Dynamics. Neuron, 101(6), 1181-1194.e13. https://doi.org/10.1016/j.neuron.2019.01.017
Dunstan, D. M., Richardson, M. P., Abela, E., Akman, O. E., & Goodfellow, M. (2023). Global nonlinear approach for mapping parameters of neural mass models. PLOS Computational Biology, 19(3), e1010985. https://doi.org/10.1371/journal.pcbi.1010985
Panahi, M. R., Abrevaya, G., Gagnon-Audet, J.-C., Voleti, V., Rish, I., & Dumas, G. (2021). Generative Models of Brain Dynamics—A review (No. arXiv:2112.12147). arXiv. http://arxiv.org/abs/2112.12147
Van Essen, D. C., Ugurbil, K., Auerbach, E., Barch, D., Behrens, T. E. J., Bucholz, R., Chang, A., Chen, L., Corbetta, M., Curtiss, S. W., Della Penna, S., Feinberg, D., Glasser, M. F., Harel, N., Heath, A. C., Larson-Prior, L., Marcus, D., Michalareas, G., Moeller, S., … WU-Minn HCP Consortium. (2012). The Human Connectome Project: A data acquisition perspective. NeuroImage, 62(4), 2222–2231. https://doi.org/10.1016/j.neuroimage.2012.02.018
Xu, Y., Long, X., Feng, J., & Gong, P. (2023). Interacting spiral wave patterns underlie complex brain dynamics and are related to cognitive processing. Nature Human Behaviour. https://doi.org/10.1038/s41562-023-01626-5