Objective: Assessing the localization accuracy of electric and magnetic source imaging (ESI/MSI) for deep brain sources using a 3D-printed head phantom.
Methods: We developed a realistic pediatric head phantom preserving brain, skull, and scalp properties with implanted sources in clinically relevant deep brain locations. Localization accuracy of ESI/MSI was assessed across varying noise levels using dipole fitting and dynamic statistical parametric mapping (dSPM).
Results: The phantom generated realistic MEG and EEG data resembling actual epilepsy patient recordings. MSI showed superior accuracy to ESI for the deep tangential insular source (dipole: ∼17 vs. ∼33 mm; dSPM: ∼24 vs. ∼32 mm). While ESI-ECD localized some radial sources well (e.g. ∼9 mm for brainstem), its dSPM struggled to localize deep sources (e.g. insula and amygdala). Both modalities found the radial thalamus source most challenging to localize.
Conclusions: MSI outperformed ESI for localizing deep tangential sources; yet, both techniques struggled to localize deep radial sources. For point-like sources, dipole fitting delivered the highest accuracy (∼9 mm, ESI for brainstem), whereas averaged dSPM was superior for sources with distributed-source behavior (∼13 mm, MSI for orbital gyrus).
Significance: 3D Printed realistic head phantoms can aid assessing the accuracy of ESI/MSI and selecting appropriate methods for different clinical scenarios.
Keywords: 3D printing; Electric source imaging; Magnetic source imaging; Phantom; Source localization.
Copyright © 2026 The Authors. Published by Elsevier Ltd.. All rights reserved.