TUSNet Training Simulations

We are excited to present TUSNet, the first end-to-end deep learning model capable of computing transcranial ultrasound pressure fields and phase aberration corrections with high accuracy and efficiency. Outperforming traditional methods like k-Wave by over 1200X, TUSNet achieves 98.3% accuracy in peak pressure estimation with a computation time of only 21 milliseconds, providing a fast and accurate solution for non-invasive ultrasound-based therapies. This repository provides a small sample of the k-Wave simulations that we used to train TUSNet.