Computational protocol: Automatic detection and visualisation of MEG ripple oscillations in epilepsy

Similar protocols

Protocol publication

[…] Each MEG recording was co-registered with a T1-weighted structural magnetic resonance image (MRI) of the patient with surface matching software developed by one of the authors (AH). This resulted in a co-registration error of approximately 4 mm (). A single sphere, which fitted best to the outline of the scalp, was used as volume conductor model. This model was used for the beamformer analysis described below.We used the same T1 MRI to reconstruct virtual sensors in the grey matter. This was done by segmenting the grey matter in SPM12 in Matlab (version 8.5.0; Mathworks Inc., Natick, MA, USA), down sampling the grey matter voxels to get a minimum inter-sensor distance of 5 mm, and excluding all voxels below the nose. Cerebellar grey matter voxels were excluded, but deep structures like the hippocampus and interhemispheric grey matter were maintained. The remaining voxels were used as virtual sensor locations. The coverage of virtual sensors was visually checked for each patient. Each patient had between 2060 and 2788 virtual sensor locations (average 2421, ).Fig. 1Fig. 1 [...] The results of the ripples after automatic detection and review were visualized on axial slices of the patient's MRI, and in a 3D figure. The concordance between the area(s) with ripples and the area(s) with spikes in the MEG was assessed visually and was classified as good (+) if all ripples were located in the same lobe as the spike dipoles, moderate (=) if any ripple was located in the same lobe as the spike dipoles, and bad (−) for discordance. A similar classification strategy was used to assess the concordance between the area with ripples and the resected brain area for those patients who had undergone surgery. Concordance was good (+) if > 50% of the ripple locations were included in the resection, at lobar level, moderate (=) if < 50% ripple locations were included in the resection, and bad (−) for discordance. We classified the concordance between the MEG spike dipole locations and the resected brain area by using the same criteria as for ripples.Twelve patients (14–25) had already been included in a previous study in which we visually marked ripples in a predefined area of interest using the same MEG recordings (). Here, we were therefore able to compare the number of automatically identified ripple-times to the number of visually marked ripple-times in these patients.Statistical analyses were performed using IBM SPSS Statistics 23 (IBM Corp., Armonk, NY, USA); a p-value < 0.05 was considered significant. […]

Pipeline specifications

Software tools SPM, SPSS
Applications Miscellaneous, Magnetic resonance imaging
Organisms Homo sapiens
Diseases Epilepsy