Dipole estimation errors due to skull conductivity perturbations: Simulation study in spherical head models

Electroencephalogram (EEG) dipole source localization is a non-invasive technique used in the pre-surgical diagnosis of epilepsy. In the present study we investigated the dipole location and orientation errors due to skull conductivity perturbations, in seven 3-shell concentric spherical head models with brain-to-skull conductivity ratio (R-sigma) ranging from 10 to 40. Each head model was compared to the baseline head model with R-sigma = 20. It is noted that perturbations in the skull conductivity generate dipole location and orientation errors: the more R-sigma deviates from the baseline value the greater the errors and the larger the error ranges. Results show that the estimated dipole location is radially shifted away from the center of the head model if the skull conductivity is larger than that of the baseline head model (11, = 10, 15), while it is radially shifted towards the center in case the skull conductivity is less than that of the baseline head model (R, = 25,30,35,40). The dipole orientation error due to skull conductivity perturbations is not significant (maximal mean 6 mm, standard deviation = 3 mm), especially when the dipoles are near the skull the maximal mean can reach 8 mm. Therefore, accurate estimation of the skull conductivity of the head model is necessary to enhance the reliability in EEG dipole source localization.Electroencephalogram (EEG) dipole source localization is a non-invasive technique used in the pre-surgical diagnosis of epilepsy. In the present study we investigated the dipole location and orientation errors due to skull conductivity perturbations, in seven 3-shell concentric spherical head models with brain-to-skull conductivity ratio (R-sigma) ranging from 10 to 40. Each head model was compared to the baseline head model with R-sigma = 20. It is noted that perturbations in the skull conductivity generate dipole location and orientation errors: the more R-sigma deviates from the baseline value the greater the errors and the larger the error ranges. Results show that the estimated dipole location is radially shifted away from the center of the head model if the skull conductivity is larger than that of the baseline head model (11, = 10, 15), while it is radially shifted towards the center in case the skull conductivity is less than that of the baseline head model (R, = 25,30,35,40). The dipole orientation error due to skull conductivity perturbations is not significant (maximal mean 6 mm, standard deviation = 3 mm), especially when the dipoles are near the skull the maximal mean can reach 8 mm. Therefore, accurate estimation of the skull conductivity of the head model is necessary to enhance the reliability in EEG dipole source localization.P