Accuracy of EEG dipole source localisation in presence of brain lesions

Inverse solution techniques based on electroencephalographic (EEG) measurements are a powerful mean of gaining knowledge about brain functioning, being used to estimate location, orientation and strength of neural electrical sources of brain activity. A model of the head, a model of the source and an electric-field computational method are necessary to describe the EEG problem mathematically. Volume conductor models commonly used to describe the EEG neglect the presence of brain lesions. We evaluated the need of considering brain lesions in head models for precise mapping of neural activity nearby the lesion, as it is requested for neurosurgical preoperative planning. A systematic evaluation of the effects of neglecting brain lesions in EEG dipole source localisation accuracy has been performed by computer simulations for different pathologic conditions, source types and source positions. Simulations of EEG measurements were carried out using a modified eccentric-spheres model of the head in which an eccentric bubble approximates effects of actual brain lesions. A three concentric spheres model has been used in the inverse dipole fitting procedure to quantify source localisation errors caused by ignoring the presence of lesions in the head model. 64 different situations have been analysed. Source reconstruction errors resulted negligible only for some relative positions of source, lesion and electrodes. The largest errors, up to 2.5 cm, have been found for a lesion placed between source and electrodes and for a source internal to the lesion (not circumscribed tumours). We conclude that source localisation process is largely affected by a nearby lesion and thus electrical mapping of brain activity must be performed embedding brain lesions in the head model.