Modelling Shear Shock Waves in the Brain with Machine Learning (DiscoSoft)

Description:

Traumatic brain injury due to vehicular accidents is one of the most common causes of disability and death worldwide. The violent head impact acts like a sudden inertial loading on the brain and can lead to diffuse axonal injuries (DAIs), the most common type of direct impact brain injury, which occurs deep inside the brain. Still, the underlying biomechanics causing DAIs are poorly understood. Recently observed shear shock waves in the brain could be the primary cause behind DAIs. The soft tissue of the brain exhibits nonlinear and viscous behaviour, i.e., a smooth excitation at the brain surface quickly transforms into shear shock waves (which have very high local acceleration) deep inside the brain. Therefore, it is crucial to understand the material response of these tissues to evaluate the risk of damage that could happen during the impact. With this project, we propose a data-driven discovery of the wave physics of brain matter to determine a novel theoretical and computational model. We have two specific aims: 1) The development of a sparse-regression-based machine learning model to discover the underlying governing equations using custom numerical simulations for shear shock waves in the brain and the experimental ultrasound data, and 2) The development of an artificial neural network-based surrogate model. These will be developed with the help of finite element computer simulations and experiments to be performed at the host institution. This project will develop novel paradigms for learning and modelling the mechanics of the brain matter, along with novel
simulation techniques which will offer computationally inexpensive solutions for real-world simulations in contrast to the current simulation tools.  

Funded by:

Government of Ireland Postdoctoral Fellowship 2023 Grant Number GOIPD/2023/1552

#LoveIrishResearch 

 

irc-logo