I am modeling an MRI transmit coil with a gel phantom. The resonance frequency observed in a broadband simulation changes when I change the type of voxeler used: With the topological voxeler, resonance frequency is at about 127.6 MHz, whereas with the intersection voxeler on the same model, same grid and same simulation settings, resonance frequency is at about 126.5 MHz. Given the difference in B1 field distribution observed in any of these simulations at frequencies 1 MHz apart, this difference is not negligible.
Strangely, increasing the resolution of the grid does not seem to reduce the difference in resonance frequency between the two voxelers.
Which of the two voxelers is more likely to produce the correct result?
Is there any other setting I should check to improve results?