The Information Flow of TMS-evoked Brain Network Dynamics
Poster No:
64
Submission Type:
Abstract Submission
Authors:
Xiao jinming1, Li Lei2, Yating Ming1, Xujun Duan3
Institutions:
1University of Electronic Science and Technology of China, Chengdu, Sichuan, 2University of Electronic Science and Technology of China, Chnegdu, China, 3UESTC, Chengdu, Sichuan
First Author:
Co-Author(s):
Introduction:
Understanding how local perturbation in neural activity influence brain dynamics is a compelling way to inference the information flow of large-scale brain network. The concurrent transcranial magnetic stimulation (TMS) and electroencephalography (EEG) is the best technology to support this non-invasive perturbation-based analysis for inferencing the cortico-cortical directed connectivity in human brain (Biabani, Fornito, Mutanen, Morrow, & Rogasch, 2019; Gollo, Roberts, & Cocchi, 2017; Rogasch & Fitzgerald, 2013). By conducting TMS-EEG, we can track the TMS-evoked activity originating from a target brain region to propagate throughout the whole brain(Momi et al., 2021; Thut & Miniussi, 2009). By using source-localized TMS-EEG analysis and whole-brain connectome-based computational modelling, Moni et al indicated that the initial EEG signal changes was caused by local dynamics in stimulation regions, while later EEG signal changes were influenced by activity within a wider connected network(Momi, Wang, & Griffiths, 2023). However, the information flow of TMS-evoked activity was unclear.
Methods:
In this study, we used concurrent TMS-EEG dataset which was collected and provided to community by the Rogasch group (https://figshare.com/articles/dataset/TEPs-_SEPs/7440713). The dataset consisted of a total 20 healthy individuals (24.50±4.86 years; 14 females), all of whom received single-pulse TMS stimulation on primary motor cortex (M1) while brain activity was recorded by density EEG. The detailed description of the dataset and steps for preprocessing can be found at(Biabani et al., 2019). We performed source reconstruction by using MNE software library. Finally, the brain activity was extracted through Schaefer 200 parcellations atlas(Schaefer et al., 2018).
The analysis pipeline was as follow: (1) By conducting sparse non-negative matrix factorization (sNMF) model, the TMS-EEG activity was decomposed into co-activation modules and time-varying weights. (2) By calculating phase slope index (PSI) for the time-varying weights, we can inference the directional information flow among co- activation modules. The PSI measures the asymmetry in phase differences between signals, providing insights into the directed interactions among brain regions(Basti et al., 2018).
The analysis pipeline was as follow: (1) By conducting sparse non-negative matrix factorization (sNMF) model, the TMS-EEG activity was decomposed into co-activation modules and time-varying weights. (2) By calculating phase slope index (PSI) for the time-varying weights, we can inference the directional information flow among co- activation modules. The PSI measures the asymmetry in phase differences between signals, providing insights into the directed interactions among brain regions(Basti et al., 2018).
Results:
Our results indicated that TMS-evoked brain activity can be decomposed into 10 co-activation modules (Fig 1). We summarize these 10 modules into 3 modes – 'Left hemisphere dominant module', 'Right hemisphere dominant module', and 'Bilateral modules'. (1) Left hemisphere dominant module include module 1, 2, 3, 4. Module 1 and 2 mainly involves left Somatomotor network (SMN) which can be regarded as stimulation regions. Module 3 mainly involves left Default network (DMN) and Visual network (VN). Module 4 mainly involves left DMN. (2) Right hemisphere dominant module include module 5, 6, 7, 8. (3) Bilateral modules include module 9,10.
By conducting PSI analysis (Fig 2), our results indicated that (1) TMS-evoked brain activity propagates from the left hemisphere dominant module to the right hemisphere dominant module. This outcome suggests that TMS stimulation extends beyond the stimulated region, transmitting across the network to the contralateral brain areas. (2) Although Module 2 serves as the stimulated region, it concurrently acts as a recipient in the information flow (high in-degree), which validate previews research demonstrating the recurrent, re-entrant activity of stimulation region. Additionally, we observed the pathway from Modules 3 and 4 on the ipsilateral hemisphere transmitting to Module 2, instead of the transmission originating from modules on the contralateral hemisphere to Module 2.
By conducting PSI analysis (Fig 2), our results indicated that (1) TMS-evoked brain activity propagates from the left hemisphere dominant module to the right hemisphere dominant module. This outcome suggests that TMS stimulation extends beyond the stimulated region, transmitting across the network to the contralateral brain areas. (2) Although Module 2 serves as the stimulated region, it concurrently acts as a recipient in the information flow (high in-degree), which validate previews research demonstrating the recurrent, re-entrant activity of stimulation region. Additionally, we observed the pathway from Modules 3 and 4 on the ipsilateral hemisphere transmitting to Module 2, instead of the transmission originating from modules on the contralateral hemisphere to Module 2.
Conclusions:
This finding is instrumental in enhancing our understanding of how signals propagate in the brain, providing a novel connectome perspective to the clinical application of TMS.
Brain Stimulation:
Non-invasive Magnetic/TMS 1
Modeling and Analysis Methods:
Connectivity (eg. functional, effective, structural)
EEG/MEG Modeling and Analysis 2
Keywords:
Machine Learning
Modeling
1|2Indicates the priority used for review
·Fig 1. TMS-evoked brain activity can be decomposed into 10 co-activation modules. These modules into 3 modes – ‘Left hemisphere dominant module’, ‘Right hemisphere dominant module’, and ‘Bilateral mod
·Fig 2. The Phase slope index analysis elucidated the TMS-induced transmission pathways.
Provide references using author date format
Biabani, M., Fornito, A., Mutanen, T. P., Morrow, J., & Rogasch, N. C. (2019). Characterizing and minimizing the contribution of sensory inputs to TMS-evoked potentials. Brain Stimulation, 12(6), 1537-1552. doi:10.1016/j.brs.2019.07.009
Gollo, L. L., Roberts, J. A., & Cocchi, L. (2017). Mapping how local perturbations influence systems-level brain dynamics. Neuroimage, 160, 97-112. doi:10.1016/j.neuroimage.2017.01.057
Momi, D., Ozdemir, R. A., Tadayon, E., Boucher, P., Di Domenico, A., Fasolo, M., . . . Santarnecchi, E. (2021). Perturbation of resting-state network nodes preferentially propagates to structurally rather than functionally connected regions. Scientific Reports, 11(1). doi:10.1038/s41598-021-90663-z
Momi, D., Wang, Z., & Griffiths, J. D. (2023). TMS-evoked responses are driven by recurrent large-scale network dynamics. Elife, 12. doi:10.7554/eLife.83232
Rogasch, N. C., & Fitzgerald, P. B. (2013). Assessing cortical network properties using TMS-EEG. Human Brain Mapping, 34(7), 1652-1669. doi:10.1002/hbm.22016
Schaefer, A., Kong, R., Gordon, E. M., Laumann, T. O., Zuo, X. N., Holmes, A. J., . . . Yeo, B. T. T. (2018). Local-Global Parcellation of the Human Cerebral Cortex from Intrinsic Functional Connectivity MRI. Cerebral Cortex, 28(9), 3095-3114. doi:10.1093/cercor/bhx179
Thut, G., & Miniussi, C. (2009). New insights into rhythmic brain activity from TMS-EEG studies. Trends in Cognitive Sciences, 13(4), 182-189. doi:10.1016/j.tics.2009.01.004