Towards new standards for interictal MEG source imaging: comparison between dipole fitting (ECD) and distributed source imaging using coherent Maximum Entropy on the Mean (cMEM)
Abstract number :
3.119
Submission category :
3. Neurophysiology / 3D. MEG
Year :
2016
Submission ID :
195651
Source :
www.aesnet.org
Presentation date :
12/5/2016 12:00:00 AM
Published date :
Nov 21, 2016, 18:00 PM
Authors :
Giovanni Pellegrino, McGill University - Montreal Neurological Institute, Montreal, Canada; Tanguy Hedrich, McGill University - Montreal Neurological Institute, Montreal, Canada; Jeffery A. Hall, Montreal Neurological Institute and Hospital, McGill Univer
Rationale: MEG source imaging of interictal spikes is clinically useful in the presurgical workup of epilepsy patients, as it can improve the positioning of invasive EEG electrodes or even guide resection. Current clinical guidelines recommend the technique of the Equivalent Current Dipole (ECD) as inverse solution and highlight that other methods are not widely accepted for clinical purposes (Bagic, 2011). Nonetheless, distributed source imaging has gained popularity in the last twenty years, because it overcomes some of the limitations of ECD and provides maps of activations, while some of them also offer the property to account for the spatial extent of the focus. In this study we sought to qualitatively and quantitatively compare ECD vs cMEM (coherent Maximum Entropy on the Mean), a distributed source imaging approach specifically developed to recover the spatial extent of the generators of interictal discharges along the cortical surface (Grova, 2016; Pellegrino, 2016). Methods: 340 MEG source localizations from 49 refractory focal patients with a well-defined epileptic focus (invasive EEG/MRI lesion/ Surgery) were included. For temporal lobe epilepsy, the focus was usually involving both mesial and neocortical temporal regions. A qualitative comparison based on the evaluation of sublobar concordance with the epileptic focus was complemented by a quantitative assessment of: a) actual distance in millimetres between the source and the epileptic focus; b) intra-subject reproducibility. Multiple head models, all built from individual MRI, were evaluated. Results: cMEM outperformed or showed a similar performance to ECD for all the investigated measures. cMEM sublobar concordance was higher than ECD (277/340 vs 235/340, Chi-Square =13.9453, p < 0.001), especially in neocortical epilepsy (neocortical: cMEM 174/206 (84%); ECD 137/206 (67%), Chi-Square 17.96, p < 0.001). The actual distance from the focus was very good for both methods (median < 5mm). However it was significantly lower for cMEM than ECD (median of the difference: 1.31mm, z=4.827, p < 0.001), especially for neocortical generators (Neocortical: z=-4.250, p < 0.001; Temporal: z=-2.332, p=0.020). The generators localized by ECD were significantly deeper than cMEM z=2.850, p=0.004). The intrasubject reproducibility did not differ between the two methods (p>
Neurophysiology