Dynamic Cortical Gradients in Sample Entropy of Intracerebral EEG
Abstract number :
1.458
Submission category :
3. Neurophysiology / 3G. Computational Analysis & Modeling of EEG
Year :
2023
Submission ID :
1257
Source :
www.aesnet.org
Presentation date :
12/2/2023 12:00:00 AM
Published date :
Authors :
Presenting Author: Giridhar Kalamangalam, MD DPhil – University of Florida
Abbas Babajani-Feremi, PhD – University of Florida; Mircea Chelaru, PhD – University of Florida
Rationale: Local tissue-level properties of the cerebral cortex vary systematically, forming identifiable spatial gradients. An example is the whole-brain rostro-caudal gradient (RCG) in neuronal density (Calahane et al, Front Neuroanat 2012;6:28). Superimposed on RCG are other gradients radiating away from primary sensorimotor, visual and auditory regions to association areas. These include cortical thickness (Wagstyl et al, NeuroImage 2015;111:241), intracortical myelin (Huntenburg et al, TICS 2018;22:21) and intrinsic neuronal timescale gradients (INT; Mahjoory et al, eLife 2020;9:e53715). Here we identify gradients in the intracerebral EEG during different states of arousal.
Methods: Accessing the Montreal Neurological Institute Open IEEG Atlas (von Ellenrieder et al, Ann Neurol 2020;87:289), we computed sample entropy (SE) of 5s data segments as in prior work (Kalamangalam & Chelaru, Brain Connect 2021;11:850). SE is a metric characterizing the complexity of experimental time series (Richman & Moorman, Am J Physiol 2000;278:H2039). SEs for a brain region were normalized into the range [0,1], transformed to colors, and overlaid on a cortical surface model using FreeSurfer's Destrieux atlas (Destrieux et al, NeuroImage 2010;53:1)
Results: SE followed an RCG with high values anteriorly and a global minimum in posterior cortex. Primary motor cortex had the highest SE of all, declining over premotor and cingulate motor regions. SE was high over Heschl’s gyrus and lessened over posterior auditory association areas. Primary sensory cortex had higher SE than sensory association areas. SE was higher over pericalcarine cortex than over proximate visual association areas. SE declined in N2/N3 sleep in all brain regions. In REM, the prefrontal cortex exhibited SE maximum; Heschl’s gyrus retained SE value of N2 but planum temporale recovered to near-wake values; calcarine cortex retained low N2-like values, but the remaining occipital lobe recovered to wake-like values.
Conclusions: The intracerebral EEG exhibits rich spatial structure - cortical gradients - in the distribution of SE. Our results are concordant with several recent publications exploring similar themes on the same dataset (Gao et al, eLife 2020;9:e61277, Olejarczyk et al, Sc Rep 2022;12:451, Armonaite et al, Cereb Cor 2023;33:3284). SE cortical gradients are dynamic with respect to sleep-wake states. SE cortical gradients change in intriguing ways in REM that presumably reflect the neural correlates of sensory experience in the absence of the elemental sensations themselves. Our results deepen the conventional interpretation of EEG from its frequency content attributes (Berger bands, rhythmic activity, etc.) to a spatial perspective concordant with the distribution of other cortical biological variables. In particular, we show that SE is predictive of INT and thus tracks temporal processing hierarchies in cerebral cortex. The alteration of EEG gradients by pathology (e.g. epileptic foci) remains an area for further work.
Funding: Wilder family endowments to the University of Florida and NINDS R21NS128503 to GPK.
Neurophysiology