Numerical Validation of mSOUND Using Fully Heterogeneous Skull Models

In this work, we present a wave solver incorporated into mSOUND, an open-source MATLAB toolbox for acoustic simulations. The solver achieves an accuracy-computational efficiency balance for acoustic simulations through strongly heterogeneous media. We demonstrate its usefulness in simulating transcranial focused ultrasound through fully heterogeneous skull models, in which each voxel is assigned a unique speed of sound, density, and absorption coefficient. We compare the solver's computation times and accuracy against the widely-used full-wave solver k-Wave, achieving computational savings by two orders of magnitude while maintaining comparable accuracy, particularly in the focal region. To obtain a fully heterogeneous skull model, conversions from Hounsfield unit (HU) maps obtained via computed tomography (CT) scans to acoustic variables are needed. These conversions require a maximum speed of sound parameter to be assigned as a known input. In this work, we also investigate the effect changing this parameter has on the acoustic field as well as mSOUND's accuracy relative to k-Wave. Over four skull samples and three maximum speeds of sound, our results indicate a decrease in focal pressure and increase in error for mSOUND in the focal region as the maximum speed of sound increases.