Age-related differences in control energy of brain state transitions
Poster No:
1132
Submission Type:
Abstract Submission
Authors:
Vibin Parakkattu1, Lorenzo Lorenzo-Luaces1, Youngheun Jo1, Joshua Faskowitz1, Alina Podschun2, Sebastian Markett2, Richard Betzel1
Institutions:
1Indiana University, Bloomington, IN, 2Humboldt-Universität zu Berlin, Berlin, Berlin, Germany
First Author:
Co-Author(s):
Introduction:
The brain's physical wiring – the connectome – shapes ongoing patterns of activity, tracing out a trajectory over time through a high-dimensional space (Parkes et al., 2023). Recent work has leveraged network control theory to understand how control inputs can be delivered to the brain to alter its dynamic trajectory to traverse a desired set of brain states (Cornblath et al., 2020). Control theory provides a useful measure for quantifying the effort needed to move from one brain state to another – the so-called "control energy" (Pasqualetti et al., 2014). Here, we study a lifespan dataset to understand how control energy varies across the human lifespan.
Methods:
We studied the Nathan Kline Institute, Rockland Sample dataset (https://fcon_1000.projects.nitrc.org/indi/enhanced/; Nooner et al., 2012). We focused on N = 458 participants (ages 7-85 years) with high-quality and low-motion functional and diffusion MRI data. For each participant, we reconstructed a subject-specific connectome (Tournier et al., 2019) and processed the resting-state fMRI using fMRIprep (Esteban et al., 2019).
To estimate brain states, we identified local peaks in the amplitude of brain-wide activity (root mean square) and extracted whole-brain activation patterns during those instants. We repeated this procedure for all participants. We then aggregated across participants and clustered them using the k-means algorithm (correlation distance and number of clusters, k = 6). For each subject, each frame (from trough to trough on either side of a given peak) was then labelled as belonging to 1 of 6 brain states.
Network control theory enables us to estimate the "effort" associated with state transitions. With this framework, we then calculated, using subjects' own connectomes, the energies necessary to transition between every pair of states. We calculated the correlation of these energies with age after regressing out biological sex, intracranial volume, number of usable (low-motion) frames, as well as residual motion.
To estimate brain states, we identified local peaks in the amplitude of brain-wide activity (root mean square) and extracted whole-brain activation patterns during those instants. We repeated this procedure for all participants. We then aggregated across participants and clustered them using the k-means algorithm (correlation distance and number of clusters, k = 6). For each subject, each frame (from trough to trough on either side of a given peak) was then labelled as belonging to 1 of 6 brain states.
Network control theory enables us to estimate the "effort" associated with state transitions. With this framework, we then calculated, using subjects' own connectomes, the energies necessary to transition between every pair of states. We calculated the correlation of these energies with age after regressing out biological sex, intracranial volume, number of usable (low-motion) frames, as well as residual motion.
Results:
We found that the k=6 clusters corresponded to three pairs; each pair consisted of an activation pattern and its near-perfect anti-correlate (Fig. 1a & b). Clusters 1 and 2 corresponded to activation/deactivation of the visual and dorsal attention networks; clusters 3 and 4 correspond to activation/deactivation of default mode with salience/ventral attention networks; clusters 5 and 6 correspond to activation/deactivation of somatomotor and visual networks with the control network (Fig. 1c).
We found that, on average, control energy was greatest when transitioning between paired patterns (Fig. 1d) and, more generally, was anticorrelated with the spatial similarity of activation patterns (r = -0.93; Fig. 1e).
Further, we found heterogeneity across control sites – regions/nodes where control signal was "injected" – in terms of their regional control energies. When considering all possible transitions, we found evidence of three broad clusters, each corresponding to a different brain-wide pattern of regional control energy. These patterns were closely aligned with target states and, again, were grouped by pairs.
Finally, we showed that both regional and whole-brain control energies were correlated with age (Fig 2a). In particular, control energy associated with transitioning from states 3 and 4 into 2 and 1, respectively, decreased significantly with age after regressing out covariates (r = -0.24, p < 10^-8; Fig. 2b & c). Relatedly, we found evidence of distinct regional differences in control energy with age. Broadly, these changes implicated sensorimotor systems-e.g. visual, somatomotor, and dorsal attention networks (Fig 2d & e).
We found that, on average, control energy was greatest when transitioning between paired patterns (Fig. 1d) and, more generally, was anticorrelated with the spatial similarity of activation patterns (r = -0.93; Fig. 1e).
Further, we found heterogeneity across control sites – regions/nodes where control signal was "injected" – in terms of their regional control energies. When considering all possible transitions, we found evidence of three broad clusters, each corresponding to a different brain-wide pattern of regional control energy. These patterns were closely aligned with target states and, again, were grouped by pairs.
Finally, we showed that both regional and whole-brain control energies were correlated with age (Fig 2a). In particular, control energy associated with transitioning from states 3 and 4 into 2 and 1, respectively, decreased significantly with age after regressing out covariates (r = -0.24, p < 10^-8; Fig. 2b & c). Relatedly, we found evidence of distinct regional differences in control energy with age. Broadly, these changes implicated sensorimotor systems-e.g. visual, somatomotor, and dorsal attention networks (Fig 2d & e).
·Fig. 1
·Fig. 2
Conclusions:
In conclusion, our work shows that transitions between empirically derived brain states vary with age. These findings open the possibility that observed age-related differences in cognition/behavior may be owed to differences in successfully navigating brain state transitions.
Lifespan Development:
Aging 1
Modeling and Analysis Methods:
Connectivity (eg. functional, effective, structural)
fMRI Connectivity and Network Modeling 2
Multivariate Approaches
Task-Independent and Resting-State Analysis
Keywords:
ADULTS
Aging
Computational Neuroscience
Development
FUNCTIONAL MRI
Modeling
Multivariate
NORMAL HUMAN
Systems
Tractography
1|2Indicates the priority used for review
Provide references using author date format
Esteban, O., Markiewicz, C. J., Blair, R. W., Moodie, C. A., Isik, A. I., Erramuzpe, A., Kent, J. D., Goncalves, M., DuPre, E., Snyder, M., Oya, H., Ghosh, S. S., Wright, J., Durnez, J., Poldrack, R. A., & Gorgolewski, K. J. (2019). fMRIPrep: a robust preprocessing pipeline for functional MRI. Nature Methods, 16(1), 111–116. https://doi.org/10.1038/s41592-018-0235-4
Nooner, K. B., Colcombe, S. J., Tobe, R. H., Mennes, M., Benedict, M. M., Moreno, A. L., Panek, L. J., Brown, S., Zavitz, S. T., Li, Q., Sikka, S., Gutman, D., Bangaru, S., Schlachter, R. T., Kamiel, S. M., Anwar, A. R., Hinz, C. M., Kaplan, M. S., Rachlin, A. B., … Milham, M. P. (2012). The NKI-Rockland Sample: A Model for Accelerating the Pace of Discovery Science in Psychiatry. Frontiers in Neuroscience, 6. https://doi.org/10.3389/fnins.2012.00152
Parkes, L., Kim, J. Z., Stiso, J., Brynildsen, J. K., Cieslak, M., Covitz, S., Gur, R. E., Gur, R. C., Pasqualetti, F., Shinohara, R. T., Zhou, D., Satterthwaite, T. D., & Bassett, D. S. (2023). Using network control theory to study the dynamics of the structural connectome. BioRxiv, 2023.08.23.554519. https://doi.org/10.1101/2023.08.23.554519
Pasqualetti, F., Zampieri, S., & Bullo, F. (2014). Controllability metrics, limitations and algorithms for complex networks. IEEE Transactions on Control of Network Systems, 1(1), 40–52.
Tournier, J.-D., Smith, R., Raffelt, D., Tabbara, R., Dhollander, T., Pietsch, M., Christiaens, D., Jeurissen, B., Yeh, C.-H., & Connelly, A. (2019). MRtrix3: A fast, flexible and open software framework for medical image processing and visualisation. NeuroImage, 202, 116137. https://doi.org/10.1016/j.neuroimage.2019.116137