Spatio-temporal evolution of interictal epileptic activity : a study with unaveraged multichannel MEG data in association with MRIs

This thesis addresses issues relating to MEG modelling, analysis and interpretation of results. A source model employing current density distributions, namely Magnetic Field Tomography (MET), is used to obtain the MEG results. The first issue of concern refers to the registration of MEG data with structural MR images in an attempt to improve the localisation capability of MEG/MET. Simulations testing some spatial and tem poral aspects of the reconstruction capability of MET are also provided. A novel way of conducting MET studies in depth is suggested and implemented: the iterative use of a source space designed to cover deep situated structures on either side of the brain. The main bulk of this thesis is concerned with research into interictal epileptic activity as recorded by means of multichannel MEG system s and analysed using MET. The major aim is to investigate whether or not MET analysis of unaveraged MEG data (single epochs) is feasible in cases of pathophysiological signals and more specifically interictal signals from patients with epilepsy of a complex partial type. The investigation is undertaken against the "traditional" view of the impropriety and absurdity of using single epoch records in the MEG analysis due to noise dominance; we provide evidence that analysis of single, unaveraged epileptic spikes is actually feasible: we demonstrate spatio-temporal coherence in the MET results of the various single interictal events and show that activity extracted from the "averaged event" is made up of activity contributions which occur intermittently and at variable latencies. Our statements are drawn from the study of both superficial and deep activity.