FV 4 Mouse resting-state functional neuroimaging displays signatures of criticality

Background: Brain dynamics shows a remarkable diversity of activity
patterns on a wide range of spatiotemporal scales. Unraveling the
mechanisms that underlie such a vast array of activities is a major
challenge in neuroscience. Criticality, a key concept from statistical
physics, has been proposed as a unifying mechanism to explain the
multi-scale complexity of brain dynamics (Chialvo Nat. Phys.,
2010). Previous studies on humans have demonstrated the occurrence
of criticality in healthy brain dynamics and proposed that,
while functioning at criticality, the brain exhibits optimal information
encoding, dynamic range, and storage capacity. However, it is
still unknown whether the mouse whole-brain fMRI dynamics exhibits
signatures of criticality. Furthermore, the relationship between
the critical brain dynamics and the underlying structural brain features
is poorly understood.
Objective: Here, we initially investigated whether the mouse fMRI
statistics derived from the whole-brain activity exhibits signatures
of criticality. Next, we sought to elucidate the impact of two crucial
topological features of the brain’s structural architecture namely
fiber directionality and network sparsity, on mouse fMRI statistics
and the functional organization of resting-state activity.
Methods: To study criticality in the empirical data, we analyzed the
cluster statistics of fMRI dynamics. Next, we used a stochastic cellular
automaton model (Rocha et al., Nat. Comms., 2022) to reproduce
the empirical cluster statistics and simulate the large-scale mouse
brain dynamics based on the empirical mouse structural connectome
(Oh et al., Nature, 2014). Then, we evaluated the model’s ability
to reproduce the empirical functional connectivity and resting-state
networks obtained from resting-state fMRI recordings in 38 anaesthetized
adult male C57Bl6/J mice (Gutierrez et al., Curr. Biol.,
2019). Finally, we systematically studied the direct impact of the
brain’s structural features of fiber directionality and network sparsity
on the simulation of brain dynamics.
Results: We first found that the empirical cluster-size distribution is
scale invariant, which is a hallmark of criticality. Next, we found that
the optimal working point of the model where to replicate the
empirical cluster statistics is at the critical dynamical regime.
Moreover, at criticality, the model best fitted the empirical functional
connectome and best reproduced the resting-state networks,
which is consistent with the idea that mouse dynamics at rest may
be critical. Then, using the model, we evaluated the functional relevance
of network sparsity and fiber directionality on the emergence
of critical dynamics and the generation of spatiotemporal patterns of
spontaneous brain dynamics, such as resting-state networks. We
found that fiber directionality and intermediate levels of sparsity
provide a better fit to empirical cluster statistics, functional connectivity
and resting-state networks.
Conclusion: Our findings indicate that the spontaneous brain
dynamics of mice is critical, and that network directionality and density
are essential to achieving critical dynamics and reproducing
empirical functional data.