Rapid Estimation of Entire Brain Strain Using Deep Learning Models

Abstract

This study proposes a deep learning head model for rapid estimation of whole-brain strain during head impacts. A five-layer deep neural network with feature engineering was trained on 2511 head impacts obtained from finite element simulations and on-field measurements in college football and mixed martial arts. The model predicts the maximum principal strain (Green–Lagrange) for all brain elements in less than 0.001 seconds, achieving an average RMSE of 0.022 with a standard deviation of 0.001 across repeated training runs. This approach enables fast, accurate estimation of brain deformation and supports potential clinical use in real-time traumatic brain injury detection.

Yuzhe Liu(刘雨喆)

Professor of Biomechanics