DeepOWaves

Neural Operators for Predicting Brain Deformations in Concussions (DeepOWaves)

Description:

Concussions are considered to be mild traumatic brain injuries (mTBIs). They are a major global public health challenge resulting from head impacts due to road traffic accidents, mechanical falls, contact sports, mechanical falls, etc. They are mostly characterised using the head motion data and are further complemented with sophisticated finite element computer simulations of brain deformation caused by the corresponding head motion. However, the computational techniques can take hours to days to make any biofidelic estimate of the corresponding brain deformation and therefore cannot be deployed for real-time prediction. Furthermore, the brain injury biomechanics relationship still remains disputed, recently the formation of shear shock waves in the brain from smooth impacts was discovered which could be the primary mechanism behind the diffuse axonal injuries, the most common type of mild TBIs. However, these strong deformations cannot be studied using existing computational techniques further emphasising the need for new computational tools.

This project aims to develop neural network based operators to predict the non linear brain deformation which could be used in concussion assessment in real-time. The neural operators will learn this new physics and will be able to predict in real time. These computational tools will be thoroughly validated with real-world experiments and will have the potential for translation to industry applications like in predicting concussions in rugby and other contact sports. 

Lead Investigator: Ciaran Campbell

Funded by:

Taighde Éireann – Research Ireland under the Government of Ireland Postgraduate Scholarship Programme [grant number GOIPG/2025/7514].