Influence of EEG measurement montage on source localization error bounds due to head modeling errors caused by brain lesions

EEG source reconstruction accuracy depends on numerous factors, including head modeling accuracy, the specific inverse approach and the adopted EEG measurement montage. In This work we present results of a simulation study, performed with an eccentric-spheres head model, investigating the EEC dipole source reconstruction errors bounds caused by neglecting brain lesions in the head model. To separate the effect of head modeling accuracy from errors due to specific inverse approach, we based our study on an exhaustive "goal function (GF) scan" method, in which the source parameter search space is discretized and at every scan point a GF value is computed, allowing the exhaustive determination of dipole source reconstruction error bounds and the confidence interval for inverse problem solution. Six different electrodes montages have been considered, from a minimum of 32 to a maximum of 128 electrodes, keeping spatial sampling constant; electrodes coverage increases varying minimum electrodes latitude on the scalp. Source localization and intensity error bounds obtained justify the conclusion that, in the presence of a lesion, a pathological head model must be selected to accurately reconstruct the neural source, as the systematic error due to neglecting lesion progressively increases adopting smaller EEG electrodes coverages.