Abstract
In this work with the assistance CST finite difference time domain solver and Hugo human body/SAM phantom models we investigate SAR and B1 profile of multichannel array at frequencies of 100, 200 and 300 MHz. The goal of the investigation is to optimize for the given transmit array B1 homogeneity with simultaneous SAR reduction. Standard optimization approach include two steps – simulation of B1/SAR profiles followed by tunning amplitude and phase of each source for meeting the goal. CST Microwave studio does good the first step job but CST does not offer built-in field field optimization by variation of port source power and phase. It would be no problem because such optimization could be done by third party software for example Matlab but lack of fast data export to a binary file or directly to another application like Matlab is a huge problem on the way to use third party software in connection to CST.
Another open question is tissue properties frequency dependence definition for Hugo model. In case of using time domain solver present default constant definition is not realistic if simulation is performed in rather wide frequency band that is necessary for multi nuclear magnetic resonance imaging experiments.
Another our research activity is to study influence of tissues anisotropy on SAR and B1 value. In the beginning we use a simple model - an anisotropic tissue insert, which dimension was varied, that is placed inside a single tissue isotropic cubic shape or SAM phantoms. Different types of tissue anisotropy are investigated – only electrical conductivity, only relative permittivity, both electrical conductivity and relative permittivity. As a reference simulations of isotropic tissue insert, which electrical conductivity and relative permittivity were set to maximum or minimum value of corresponded anisotropic insert, are performed. The results show significant influence of tissues anisotropy on SAR but rather weak dependence of B1 map in case of small insert dimension. As expected the effect is stronger if tissue anisotropy axis is parallel to electrical field. Unfortunately it is impossible to simulate a rather complex anisotropic object like human brain because CST limit anisotropy definition to only global coordinate system.
Our experience shows that it is possible to decrease RAM memory requirement significant if power loss density calculation will be performed as postprocessing step. Also it should be noted that stability and performance of both 64 and 32 bits Linux distributed solvers are much worse than Windows solver.
