Transcranial ultrasound simulation with uncertainty estimation

TL;DR

This paper presents a computationally efficient method to estimate uncertainty in transcranial ultrasound simulations caused by variability in skull and brain tissue properties, validated against Monte Carlo simulations.

Contribution

It introduces a linear uncertainty propagation approach for predicting acoustic field uncertainty in transcranial ultrasound simulations, improving efficiency over traditional methods.

Findings

Uncertainty estimates closely match Monte Carlo results.

Method reduces computational cost significantly.

Applicable to focused bowl transducers at 500 kHz.

Abstract

Transcranial ultrasound simulations are increasingly used to predict in situ exposure parameters for ultrasound therapies in the brain. However, there can be considerable uncertainty in estimating the acoustic medium properties of the skull and brain from computed tomography (CT) images. Here, we show how the resulting uncertainty in the simulated acoustic field can be predicted in a computationally efficient way using linear uncertainty propagation. Results for a representative transcranial simulation using a focused bowl transducer at 500 kHz show good agreement with unbiased uncertainty estimates obtained using Monte Carlo.