Active Projects:

Deep learning surrogate model for WHIM

Our lab is the first to embrace the power of Artificial Intelligence (AI), specifically, deep learning, for brain impact modeling. The latest work uses Transformer neural network (TNN), originally developed for natural language processing, for near real-time response feedback of the complete spatiotemporal dynamics of brain. To the best of our knowledge, it is the first application of TNN in the broad area of biomedical engineering and impact biomechanics.

Video below compares strains from direct model simulation (up to hours) with that from TNN (instant).


Biomechanical basis of sex differences in concussion

Male brains are larger than female brains on average. With the same head impacts, male brains will sustain larger strains, and thus, would be more vulnerable. However, clinical evidence points to the opposite. We hypothesize that it is the morphological differences of axons that lead to sex differences in injury vulnerability — female axons have fewer microtubules (like straws) and there are random gaps analogous to “material defects”. Figure below illustrates the cross-sections of a female (left) and a male (middle) axon, and a unit cell model of the axon (right).


Our models successfully reproduced microtubule undulation observed immediately after rapid stretch injury (video below), which highlights high confidence of the model dynamic behavior.


Large-scale and multiscale modeling of traumatic axonal injury (TAI)

Using a unique multimodel dataset from concussed athletes from the University of British Columbia, we are developing an efficient and scalable computational pipeline to unify the global brain injury model and axonal injury model to conduct large-scale (i.e., across the entire white matter, head impacts, and individuals) and multiscale (i.e., from organ to cell length scales) modeling of traumatic axonal injury. This would offer mechanistic insight into the causation of brain injury.

Meso-scale modeling of traumatic brain injury

State-of-the-art brain injury modeling focuses either on the global brain or a representative tiny section of a single axon. There is a substantial gap between the two, which warrants closer investigation for brain injury at the intermediate length scale.

(more to come …)


Media Report