Revealing structure-function relationships with computational models of the human body

点击翻译按钮获取中文摘要

In this research program, I will dramatically improve the speed, fidelity and predictive power of computer models of the body that involve an interaction of hard and soft tissue structures, including the spine, shoulder, and face. My breakthrough advances in modeling will create unprecedented insights into how these complex biomechanical systems function, and help to guide rehabilitation for Canadians suffering from painful musculoskeletal injuries. Healthy back function requires the correct interaction between hard and soft tissue structures of the spine: the vertebrae interacting with the soft discs that separate them and the muscles and ligaments that move and stiffen them. The same interplay between soft and hard tissues occurs throughout the body. One notable example is the face, where our most exaggerated expressions, and our most subtle, are accomplished by soft facial tissues supported underneath by the skull and jaw bones. Another example is the shoulder, where a very unstable "ball-and-socket" is supported by a concert of muscles and ligaments. The spine, shoulder, and face are each incredibly intricate and complex biomechanical systems, which are difficult to understand by measurements alone. Computer models can help our understanding by allowing us to visualize movements and forces in these systems and revealing the underlying relationships between structure and function. However, current methods for biomechanical modeling fall short for coupled soft-hard tissue systems, like the spine, face, and shoulder. I will focus on advancing three modeling techniques. I will investigate fast finite-element methods, which are the current bottleneck in soft-tissue simulation. I will develop new ways to handle dynamic constraints, which are needed to capture contact between joint surfaces and the mechanical coupling of soft and hard tissues. Finally, I will develop novel learned models of face, shoulder and spine movements, trained with millions of computer simulations, to generate reliable predictions in real-time. These techniques will be used to create the world's most expressive virtual face animations, a shoulder model with contact simulation in the joint where injuries are most frequent, and a spine model that simulates the interaction between vertebral bones, soft-tissue discs, and hundreds of muscles in real-time. I will train 16 HQP in 9 projects focusing on advanced simulation and machine learning techniques for human modeling. Their research will lead to breakthrough discoveries of the mechanisms of the body and invention of new technologies for simulating those mechanisms. This program will benefit Canada by advancing our high-tech sectors in machine learning, automation, robotics, medical devices, computer animation, and video games. It will also positively affect the lives of Canadians by inventing new computer tools that can guide personalized rehabilitation for individuals suffering from back and shoulder problems.