Differences Between MEG and High-Density EEG Source Localizations Using a Distributed Source Model in Comparison to fMRI

Abstract

Electroencephalography (EEG) and magnetoencephalography (MEG) are widely used to localize brain activity and their spatial resolutions have been compared in several publications. While most clinical studies demonstrated higher accuracy of MEG source localization, simulation studies suggested a more accurate EEG than MEG localization for the same number of channels. However, studies comparing real MEG and EEG data with equivalent number of channels are scarce. We investigated 14 right-handed healthy subjects performing a motor task in MEG, high-density-(hd-) EEG and fMRI as well as a somatosensory task in MEG and hd-EEG and compared source analysis results of the evoked brain activity between modalities with different head models. Using individual head models, hd-EEG localized significantly closer to the anatomical reference point obtained by fMRI than MEG. Source analysis results were least accurate for hd-EEG based on a standard head model. Further, hd-EEG and MEG localized more medially than fMRI. Localization accuracy of electric source imaging is dependent on the head model used with more accurate results obtained with individual head models. If this is taken into account, EEG localization can be more accurate than MEG localization for the same number of channels.

Multimodal effective connectivity analysis reveals seizure focus and propagation in musicogenic epilepsy

“…Euclidean distances between modalities indicated good concordance between modalities and were within the methodological limits described in our previous study (Klamer et al, 2014). This is reassuring and highlights the ability of our chosen multimodal technique to correctly localize the epileptogenic focus, especially in combination with the findings of video-EEG, neuropsychological testing and especially the results of the invasive EEG recordings, that all indicated a right mesial temporal focus.…”

“…Source analysis of the onset of each seizure as well as the average of all seizure initiating spikes in MEG was performed using BESA Research 5.3.7 (Brain Electrical Source Analysis, BESA GmbH, Gräfelfing, Germany) and a spherical head model. Hd-EEG data was analyzed using the Cartool software (brainmapping.unige.ch/ cartool) and a simplified realistic head model called LSMAC (Locally Spherical Model with Anatomical Constraints) derived from the individual MRI, as we were able to show in a previous study that the use of individual realistic head models improves EEG source analysis results (Klamer et al, 2014). For both data sets a distributed source model (LORETA, low resolution brain electromagnetic tomography) was applied.…”

From connectome to cognition: The search for mechanism in human functional brain networks

“…Note that similar results might be expected with EEG, given recent use of anatomical MRI images and sophisticated models of skull tissue conductivity to create “individualized” head models. Such head models have been observed to yield equivalent EEG source modeling accuracy as with MEG (Klamer et al, 2015). …”

From bird to sparrow: Learning-induced modulations in fine-grained semantic discrimination

“…The present analyses were based on distributed source estimations; a similar approach is routinely used in magnetoencephalographic studies (Ahlfors et al, 2010;Michel and Murray, 2012;Dannhauer et al, 2013;Egorova et al, 2014;Klamer et al, 2015). Here, we used the "LAURA" inverse solution, which is the combination of a minimum norm inverse solution and a local auto-regressive average regularization approach (Grave de Peralta Menendez et al, 2001;Grave-de Peralta et al, 2004;Michel et al, 2004).…”