Skip to content

Coupling and Stochasticity in Mesoscopic Brain Dynamics

You are here: Home / PhD thesis / Coupling and Stochasticity in Mesoscopic Brain Dynamics

Coupling and Stochasticity in Mesoscopic Brain Dynamics

Maciej Jedinak, Presentation date: March 27, 2017

Author: Maciej Jedinak
Title: Coupling and Stochasticity in Mesoscopic Brain Dynamics
Director: J. García-Ojalvo and A. J. Pons
Presentation date: March 27, 2017
Link to text:

Abstract: The brain is known to operate under the constant influence of noise arising from a variety of sources. It also organises its activity into rhythms spanning multiple frequency bands. These rhythms originate from neuronal oscillations which can be detected via measurements such as electroen-cephalography (EEG) and functional magnetic resonance (fMRI). Experimental evidence suggests that interactions between rhythms from distinct frequency bands play a key role in brain processing, but the dynamical mechanisms underlying this cross-frequency interactions are still under investigation. Some rhythms are pathological and harmful to brain function. Such is the case of epileptiform rhythms characterising epileptic seizures. Much has been learnt about the dynamics of the brain from computational modelling. Particularly relevant is mesoscopic scale modelling, which is concerned with spatial scales exceeding those of individual neurons and corresponding to processes and structures underlying the generation of signals registered in the EEG and fMRI recordings. Such modelling usually involves assumptions regarding the characteristics of the background noise, which represents afferents from remote, non-modelled brain areas. To this end, Gaussian white noise, characterised by a flat power spectrum, is often used. In contrast, macroscopic fluctuations in the brain typically follow a `1/f b ¿ spectrum, which is a characteristic feature of temporally correlated, coloured noise. In Chapters 3-5 of this Thesis we address by means of a stochastically driven mesoscopic neuronal model, the three following questions. First, in Chapter 3 we ask about the significance of deviations from the assumption of white noise in the context of brain dynamics, and in particular we study the role that temporally correlated noise plays in eliciting aberrant rhythms in the model of an epileptic brain. We find that the generation of epileptiform dynamics in the model depends non-monotonically on the noise correlation time. We show that this is due to the maximisation of the spectral content of epileptogenic rhythms in the noise. These rhythms fall into frequency bands that indeed were experimentally shown to increase in power prior to epileptic seizures. We explain these effects in terms of the interplay between specific driving frequencies and bifurcation structure of the model. Second, in Chapter 4 we show how coupling between cortical modules leads to complex activity patterns and to the emergence of a phenomenon that we term collective excitability. Temporal patterns generated by this model bear resemblance to clinically observed characteristics of epileptic seizures. In that chapter we also introduce a fast method of tracking a loss of stability caused by excessive inter-modular coupling in a neuronal network. Third, in Chapter 5 we focus on cross-frequency interactions occurring in a network of cortical modules, in the presence of coloured noise. We suggest a mechanism that underlies the increase of power in a fast rhythm due to driving with a slow rhythm, and we find the noise parameters that best recapitulate experimental power spectra. Finally, in Chapter 6, we examine models of haemodynamic and metabolic brain processes, we test them on experimental data, and we consider the consequences of coupling them with mesoscopic neuronal models. Taken together, our results show the combined influence of noise and coupling in computational models of neuronal activity. Moreover, they demonstrate the relevance of dynamical properties of neuronal systems to specific physiological phenomena, in particular related to cross-frequency interactions and epilepsy. Insights from this Thesis could in the future empower studies of epilepsy as a dynamic disease, and could contribute to the development of treatment methods applicable to drug-resistant epileptic patients.