EEG source estimation accuracy in presence of simulated cortical lesions

Methods to reconstruct the neuroelectrical activity in the brain source space can be used to improve the spatial resolution of scalp-recorded EEG and to estimate the locations of electrical sources in the brain. This procedure can improve the investigation of the functional organization of the human brain, exploiting the high temporal resolution of EEG to follow the temporal dynamics of information processing. As for today, the uncertainties about the effects of in-homogeneities due to brain lesions preclude the adoption of EEG functional mapping on patients with lesioned brain. The aim of this work is to quantify the accuracy of a distributed source localization method in recovering extended sources of activated cortex when cortical lesions of different dimensions are introduced in simulated data. To this purpose, EEG source-distributed activity estimated from real data was modified including silent lesion areas. Then, for each simulated lesion, forward and inverse calculations were carried out to localize the produced scalp activity and the reconstructed cortical activity. Finally, the error induced in the reconstruction by the presence of the lesion was computed and analyzed in relation to the electrode sampling and to the size of the simulated lesion. Results returned values of global error in the whole cortex and of error in the non-lesioned area which are strongly dependent from the number of recorded scalp sensors, as they increase when a lower spatial sampling is performed on the scalp (64 versus 32 EEG channels). For increasing sizes of the lesion, statistical analysis showed that only a lesion involving 1000 dipoles induces significantly higher errors level with respect to the lesion-free condition.