Systematic Artifacts in Current-Induced Magnetic Field Measurements by MRI

Computational volume conductor models of the human head are increasingly used in neuroscientific research to estimate induced electric fields in non-invasive brain stimulation methods or for source localization in electro- and magnetoencephalography. However, the anatomical complexity of the human head makes accurate head modeling challenging. To create reliable head models it is important to validate their accuracy.
A good candidate for non-invasive validation is magnetic resonance current density imaging (MRCDI). MRCDI measures small perturbations of the phase caused by the magnetic field from injected currents. MRCDI methods that are sensitive enough to detect injected current as low a 1 mA in the human brain have recently been demonstrated.
The work presented in this thesis aimed at reducing systematic artifacts in MRCDI. Magnetic stray fields from the currents in the lead wires are detrimental in MRCDI. Although lead stray fields can be corrected, residual errors can still influence the data if the leads are too close to the brain. Due to safety risks when highly conductive materials are used in the MR scanner, strict guidelines for lead wire positioning have to be followed, making commercial leads impractical for MRCDI. To circumvent the impractical safety guidelines, new low-conductivity lead wires optimal for MRCDI were constructed and a safety study was performed using RF simulations and temperate measurements.
The second aim was to reduce physiological noise in MRCDI. Although the MR sequences used in human in-vivo MRCDI up to now have demonstrated good sensitivity to current-induced magnetic fields, low-frequency spatial noise patterns arising from physiological noise in the measurements have been prevalent in the measured magnetic fields.
Echo planar imaging (EPI), the fastest MR imaging method, was used to increase robustness to physiological noise. Physiological noise in MRCDI measurements acquired with EPI was analyzed and further physiological noise reduction was attempted with physiological noise correction techniques. However, due to the robustness of EPI-based MRCDI, physiological noise correction had a negligible impact.