Objective: To improve electrical source imaging (ESI) dipole localization precision using electrical stimulation potentials (ESP) as ground truth. Methods: Via implanted depth electrodes, we applied serial electrical single pulse stimulation to 72 stimulation sites in four temporal lobe regions (hippocampus, amygdala, anterior and posterior temporo-lateral cortices) of seven patients and computed ESI of scalp EEG correlates of intracranial ESP. Based on the spatial offsets between the calculated ESP-dipoles and the known stimulation coordinates, we determined region-specific error vectors (rEV). Their inverse (riEV) was then applied to ESI dipoles of ESP and interictal epileptic discharges (IED). Results: riEV application reduced the mislocalizations between calculated ESP-dipoles and stimulation coordinates. At the optimal skull scalp conductivity ratio (SSCR) of 1:8, localization errors reduced on average from 15.2 (SD 4.5) to 8.2 (SD 1.8) mm. Error volumes, defined by a hull over stimulation coordinates to all error vector tips, reduced from 507 (IQR 306) to 373 (IQR 296) mm³ . In five anterior temporo-lateral IED clusters of four patients, riEV application reduced the mislocalizations between calculated IED-dipoles and IED-onset contacts from 20.6 (IQR 5.7) to 10.2 (IQR 5.7) mm. Conclusions: riEV application is inter-individually applicable which potentially improves a precision of ESI dipole localization. If further studies with more patients, other brain regions, different EEG recording and stimulating parameters as well as ESI analysis methods validate clinical benefits, riEV application could be included into clinical ESI protocols, independently of intracranial electrode implantation. Significance: Our data suggests a new, clinically applicable method to improve ESI localization.
Unnwongse et al. (Thu,) studied this question.