Assessment of computational tools for MRI RF dosimetry by comparison with measurements on a laboratory phantom

This paper presents an extended comparison between numerical simulations using the different computational tools employed nowadays in electromagnetic dosimetry and measurements of radiofrequency (RF) electromagnetic field distributions in phantoms with tissue-simulating liquids at 64 MHz, 128 MHz and 300 MHz, adopting a customized experimental setup. The aim is to quantify the overall reliability and accuracy of RF dosimetry approaches at frequencies in use in magnetic resonance imaging transmit coils. Measurements are compared against four common techniques used for electromagnetic simulations, i.e. the finite difference time domain (FDTD), the finite integration technique (FIT), the boundary element method (BEM) and the hybrid finite element method-boundary element method (FEM-BEM) approaches. It is shown that FDTD and FIT produce similar results, which generally are also in good agreement with those of FEM-BEM. On the contrary, BEM seems to perform less well than the other methods and shows numerical convergence problems in presence of metallic objects. Maximum uncertainties of about 30% (coverage factor k = 2) can be attributed to measurements regarding electric and magnetic field amplitudes. Discrepancies between simulations and experiments are found to be in the range from 10% to 30%. These values confirm other previously published results of experimental validations performed on a limited set of data and define the accuracy of our measurement setup.