Abstracts

Rationale: Assessment on MTLE lesion epileptogenicity is crucial in early diagnostic stages of surgery candidates. EEG and source analysis (SA) techniques provide a good non-invasive option. Their drawback is that SA methods are demanding from the user as it implies (a) spikes selection, (b) spike type separation (clustering) and (c) careful judgment in SA results interpretation to exclude clinically unreasonable solutions. We evaluated EPIGAUSS – a new epileptologist-friendly software for spikes SA – in the assessment of the EEG focus location over a population of 7 patients with MTLE. Methods: For each patient the user just has to provide a spikes selection group (from visual or automated detectors) that fits the clinical conventions. All spikes (average 203 detections per patient) are then processed automatically by EPIGAUSS. The system starts by fitting spikes with a single moving dipole algorithm using a standard headmodel in a 1 seconds interval centred in the spike selection. For each patient, dipole solutions are time aligned using two criteria based on dipole solution features: (a) best goodness of fit (GOF) and (b) maximum magnitude (MAG) latencies. For both latencies, dipoles are spatially clustered using an automatic cluster analysis algorithm. The dipoles and clusters are then co-registered onto the standard model as dipole densities and both the dipole and clusters maximum density were compared with both EEG focus information and the MTLE information extracted from anatomical MRI analysis to assess their clinical lateralization value. A 75% of overall cluster density threshold was used.Results: The dipole density at GOF provided a correct identification of the side of the lesion in all but one case where no laterality could be established. The identified dipole clusters were also in the side of the MTLE lesions (5 in the depth). Both the dipole density maximum and the dipole clusters associated at MAG latency were at the surface in temporal and frontal areas in the side of EEG focus in all patients. Clusters involving non adjacent brain areas were discarded from the analysis.Conclusions: By combining the dipole density and clusters information on individual spikes at both GOF and MAG latencies correct information on the laterality of the lesion and EEG focus was obtained. The method produces results in an automated way, with no user involvement in the tedious spike type separation task of other methods. This protocol is a pragmatic non-invasive alternative to more complex approaches based on SA. Furthermore, by keeping the spike time relation, it may provide further information on the source stability and spread of the epileptogenic activity. Finally, given the visualization scheme used (colour coding over MRI), result interpretation is very straightforward and similar to other brain imaging techniques (fMRI, etc.).