Now showing 1 - 1 of 1
ItemVirtual intracranial electroencephalography for epilepsy surgery: an ictal magnetoencephalographic studyCao, Miao ( 2020)Pharmaco-refractory focal epilepsy is a serious clinical problem. Epilepsy surgery is an effective approach to treat pharmaco-refractory focal epilepsy, particularly for complex cases with no clear lesion or an extensive lesion. However, surgical treatment is currently under-used and does not always render favourable outcomes. Invasive intracranial electroencephalography (iEEG) is the pre-surgical gold-standard to localise and circumscribe the epileptogenic zone (EZ). However, iEEG has several important limitations, such as constrained spatial sampling and invasiveness. More importantly, it is not always guaranteed that iEEG electrodes cover the entire EZ, which is believed to be one of the main reasons for unsuccessful surgeries. Non-invasive neuroimaging techniques mitigate the risks and limitations of iEEG by imaging brain structure and neural activity in a whole-brain fashion. Recent advances in electroencephalography (EEG), magnetoencephalography (MEG) and magnetic resonance imaging (MRI) combined with source imaging techniques allow to investigate neural dynamics at comparable temporal and spatial resolutions to iEEG but non-invasively. Solving forward and inverse problems are the two major missions of EEG and MEG source imaging. In this thesis, a study using realistic head model derived from individual MRI of a healthy subject and an epilepsy patient is conducted to understand the operating regime and limitations of constructing EEG and MEG forward models with compromise from brain lesion. Simulations using forward and inverse modelled ictal iEEG time-series and ictal MEG signals also offer crucial insights into reliably reconstructing ictal source signals that preserve important clinical characteristics, such as morphology and spatial patterns. Attempts have also been made to construct functional networks using ictal source signals reconstructed from MEG. There is a pressing need for non-invasive approaches that objectively characterise the EZ in presurgical evaluation. Dynamical network models using iEEG have demonstrated multiple successes in predicting the EZ and surgical outcomes. Combining non-invasive neuroimaging techniques with sophisticated dynamical network modelling approaches may offer valuable information to the current clinical localisation of the EZ such as iEEG. A novel approach, virtual iEEG (ViEEG), is proposed to non-invasively investigate ictal dynamics like iEEG without its limitations. The proof-of-concept study using 36 seizures captured MEG from 12 patients suggest dynamical network models applied to ictal ViEEG provide the valid characterisation of the EZ and non-ictogenic brain areas that are less likely to overlap the EZ. Moreover, solutions from ViEEG and dynamical network models using MEG alone predicts the iEEG seizure onset zone and the optimal source localisation solution that can only be offered using simultaneous EEG and MEG. The proposed approach and its findings demonstrate the feasibility of non-invasively and objectively characterising the EZ and motivate future work to optimise the current methods. The successful implementation of the proposed approach in the clinical setting would lead to significant benefit to people with refractory focal seizures: making surgery more available, minimising invasive recordings and therefore mitigation of risks, as well as improved surgical outcomes.