Scaling properties and heterogeneous dynamics of epileptic activity
Lucas G. S. França · UCL Discovery (University College London) · 2020
Epilepsy is a brain disorder characterised by recurrent seizures with a wide range of cognitive, psychological, and social consequences. Seizures are the transient occurrence of signs and symptoms due to abnormal excessive synchronous neural activity. Epilepsy is currently treated with medication or by a resective surgical procedure; however, about 30% of patients still do not achieve seizure control. The unpredictability of when seizures occur means that even infrequent seizures can have a devastating effect on someone’s life and has hampered progress in response mode treatments. There has, therefore, been a major research drive for the development of seizure prediction/detection techniques. These have suffered from poor sensitiv- ity and/or specificity. Moreover, many of the advanced mathematical methods for seizure detection/prediction do no more than reproduce the results of linear and simpler approaches. This thesis focuses on the scaling properties of intracranial electrophysiologi- cal measures from both humans and rats. Firstly, a novel approach to evaluate mul- tifractal properties of brain signals is presented and its properties are demonstrated. Secondly, the developed approach is applied in a series of clinical challenges: distinguishing sleep phases; a study on characteristic time for seizure detection; a pu- tative classification scheme for focal epileptic seizures; and an evaluation of data from an animal model of epileptogenesis. The proposed technique was capable of extracting distinct information from the EEG signals that has clinical utility. Moreover, the multifractal nature of the brain signals hints at how neural structures work and lends support to results from previous research.