Scale-Free Bursting in Human Cortex following Hypoxia at Birth

Abstract

The human brain is fragile in the face of oxygen deprivation. Even a brief interruption of metabolic supply at birth challenges an otherwise healthy neonatal cortex, leading to a cascade of homeostatic responses. During recovery from hypoxia, cortical activity exhibits a period of highly irregular electrical fluctuations known as burst suppression. Here we show that these bursts have fractal properties, with power-law scaling of burst sizes across a remarkable 5 orders of magnitude and a scale-free relationship between burst sizes and durations. Although burst waveforms vary greatly, their average shape converges to a simple form that is asymmetric at long time scales. Using a simple computational model, we argue that this asymmetry reflects activity-dependent changes in the excitatory-inhibitory balance of cortical neurons. Bursts become more symmetric following the resumption of normal activity, with a corresponding reorganization of burst scaling relationships. These findings place burst suppression in the broad class of scale-free physical processes termed crackling noise and suggest that the resumption of healthy activity reflects a fundamental reorganization in the relationship between neuronal activity and its underlying metabolic constraints.

The heavy tail of the human brain

“…This is because even though the microscopic dynamics may be inaccessible (e.g., single neurons in humans in vivo, or magnetic domains in ferromagnetic samples), near the critical point their bulk activity yields macroscopic signals that can be detected with EEG (or in the case of Barkhausen noise, a pickup coil). Recently it was shown that human neonatal EEG following hypoxia at birth exhibits bursty activity with lengthy power laws spanning several orders of magnitude (Figure 3a) [44 ]. In this context, size is not a spatial property but rather a measure of the energy released in an event.…”

Novel methods for predicting clinical outcome in neonates from clinical electroencephalography recordings

“…Contrasts on EEG analyses were then performed, comparing: i) infants with no IVH, ii) infants with IVH grades 1-2 and iii) IVH grades 3-4. EEG data were exported to MATLAB (Mathworks, Natick, MA, USA), band-pass filtered (0.2-20 Hz), and analyzed using conventional and custom algorithms (Roberts et al, 2014).…”

“…Roberts et al 2014;Shew et al 2009). The study of criticality can be dichotomized into two classes: those systems which "self-organize" and those tuned to operate at critical points.…”

“…A. Roberts et al 2014) opens novel insights into critical states of cortical activity under the constraints of metabolic depletion. Using similar methods and more stringent statistical tests (i.e., Bayesian model selection), the present study establishes that scale invariant distributions exist in preterm EEG data -a characteristic feature recently identified as a key, but challenging objective .…”

Design and optimisation of time-frequency analysis for multichannel neonatal EEG background features in term neonates with hypoxic ischaemic encephalopathy: characterisation, classification and neurodevelopmental outcome prediction

“…Most research has focused on specific patterns: for example, seizure detection and localisation [29,[35][36][37], burst-suppression classification [24,[152][153][154] and detection of the sleep-wake cycle [38]. Classification of a broader variety of EEG background patterns will increase the system complexity.…”