Digital models of human anatomy, already common elements of motion pictures and computer games, are now increasingly being incorporated into diverse "serious" applications such as medical diagnostics, surgical planning and vehicle design. Both current and emerging applications demand improved photorealism, enhanced biomechanical accuracy, better subject specificity and faster simulation algorithms. As these demands often outpace the evolution of computer hardware, new algorithms for biomechanical modeling and simulation are necessary to ensure that upcoming computational platforms are utilized to the best of their capacity.
With respect to computational performance, physics-based simulations of virtual materials in interactive applications have demonstrated inferior performance in terms of cost per degree of freedom when compared to large-scale simulations of similar phenomena in HPC settings. This is partly attributable to the regularity and economy of scale associated with large models, compared to the pronounced irregularity and heterogeneity at lower resolutions. But it is also due to the fact that the commonly-used algorithms and data structures for interactive virtual materials reflect design compromises with respect to features, accuracy and parallelization potential. In this project the PI will endeavor to show that it is possible to reinvent both algorithms and data structures so as to achieve optimal computational efficiency for nonlinear virtual materials even under the constraints of interactive simulation, and also to achieve improvements of orders of magnitude in runtime, resolution and accuracy. The work will leverage expertise from computer systems, scientific computing, continuum mechanics and numerical analysis.
Biomechanical simulation has provided a great opportunity for transformative advances in medical practice using virtual models of the human body for disease prevention and treatment. These emerging applications mandate an increased level of attention to the unique demands of interactivity, resolution and anatomical accuracy for clinical uses of biomechanical modeling and simulation. Algorithmic improvements that yield two or three orders of magnitude have the potential to transform clinical training or operation planning tasks from off-line processes to practical interactive experiences. In addition to being a valuable opportunity to improve patient care, the use of anatomical modeling as a testing ground for effective and scalable simulation algorithms will have a lasting legacy that extends well beyond the clinical field. The high degree of irregularity in shape and function that is inherent in human tissues yields an excellent and challenging benchmark for the validity of material models, the accuracy of discretization techniques and the efficiency of numerical solvers. Considerations such as the accommodation of topology change and facilitation of parallel processing make simulating virtual nonlinear tissue models an excellent opportunity to refine core computational physics and numerical analysis techniques.
Broader Impacts: Computer simulations have been established as an educational tool in many fields (e.g., driving and flight training), helping to improve the safety record of operators without risk of physical harm. They present a great opportunity to affect the quality of patient care, considering the fact that most surgical residents currently sharpen their skills while operating on actual patients. Anatomical simulators could also reduce the need for animal testing, and help with knowledge transfer across the international clinical community, especially in developing countries. As part of this project, the PI will develop new academic course offerings that will serve to connect computer engineers, applied mathematicians, and clinical practitioners.
