| Summary: | This thesis considers the temporal structure of burst dynamics in early brain activity through the analysis of human very preterm electroencephalograph (EEG) recordings and computational neural network models. A novel algorithm for the detection of the discontinuous bursts of activity in the preterm EEG is developed and the temporal structure of burst occurrence, size and duration are assessed. The dynamics are shown to exhibit long-range temporal correlations (LRTCs) indicating a temporal complexity within early brain activity not previously appreciated. This result is replicated in a larger population of preterm children and the effect of gestational age and postnatal age on the degree of LRTCs is examined. A possible mechanism underlying the generation of burst activity that exhibits LRTCs is investigated in a stochastic excitatory neural network model. It is shown that burst dynamics occur in the model when there is a balance between the activity of an individual neuron and the number of neurons it in turn activates. Furthermore, it is shown that correlations in the temporal statistics of these bursts exist over a wide range and extend across an infinite range in the limit of system size. The behaviour of the model with respect to different network topologies is also investigated. In summary, it is shown that complex temporal dynamics exist even in early brain activity and such dynamics can be observed in a simple model. In light of this, the evidence that the brain exhibits self-organised criticality - a theoretical framework suggested by previous authors as an explanation for LRTCs in a systems dynamics - is discussed. Overall, the observation of complex temporal structure of activity in the early developing brain suggests that the temporal organisation of this activity may play an important developmental role. This thesis therefore provides strong motivation for future work in this area.
|
|---|
