Empirical limitations of current low-intensity focused ultrasound simulation platforms

Transcranial low-intensity focused ultrasound (LIFU) is an emerging tool for noninvasive neuromodulation, with potential therapeutic applications across a range of neuropsychiatric disorders. As the field advances toward clinical translation, individualized simulations of sonication parameters are critical for estimating dosing to ensure safety, efficacy, and reproducibility. Currently, k-Plan (Brainbox Ltd.) and BabelBrain (open-source) are two widely used software packages for sonication simulation, yet their outputs have not been empirically compared. While CT scans remain the gold standard for capturing fine anatomical details of the skull, their use carries radiation risk. As an alternative, algorithms generating MRI-based pseudo-CTs have been developed to approximate skull bone properties.

In this study, we directly evaluated the consistency of simulation results from the two platforms, k-Plan and BabelBrain. We also examined the impact of skull modeling using CT vs. pseudo-CT on LIFU simulation results.

We compared LIFU simulation results between the two platforms when applied to the same individuals, trajectories, and using the same sonication protocol parameters. To assess the validity of using pseudo-CTs for simulations, we also compared simulation outputs generated using CT versus pseudo-CT inputs.

Our results reveal substantial discrepancies between software platforms (k-Plan vs. BabelBrain), underscoring the need for additional ground-truth validation of current sonication simulation software, particularly regarding how simulations estimate the skull's acoustic properties, a main limitation of this study. Additional discrepancies between CT and pseudo-CT skull modeling approaches highlight the need to further optimize pseudo-CT algorithms before widespread clinical adoption.