Musculoskeletal computer models are increasingly being used in biomedical research and clinical decision making. Simulating the interaction between muscles and skeletal structures that occurs as muscles wrap around hard tissue such as bone is central to musculoskeletal models and is important for understanding the biomechanics of patient-specific models. Medical imaging can provide patient-specific bone shapes for musculoskeletal models, but traditional muscle wrapping methods use simple geometric shapes for wrapping surfaces. In this paper, we propose a new method for fast and robust simulation of muscle paths interacting directly with arbitrary mesh-based surfaces. A muscle path is modeled as a massless, frictionless elastic strand that wraps tightly around intervening obstacle surfaces. Strands are implemented as piecewise-linear curves partitioned by knot points (typically 50-100), with artificial forces applied to each knot, independently of the dynamic simulation, to pull the strand taught while avoiding obstacle penetration. Contact forces between knots and obstacles are computed quickly using a distance grid, which is interpolated quadratically to provide smoother results. The method works with multiple wrapping obstacles and non-convex surfaces, and allows paths to come into and out of obstacle contact during dynamic musculoskeletal simulations. A suite of evaluations was performed for simple cases with exact solutions as well as for more complex surfaces involving bone shaped 3D meshes. The method is also fast, with strand update times of around 0.5 msec (on an Intel Core i7 based desktop) for a variety of bone shaped obstacles. Muscle wrapping on arbitrary surfaces opens up new possibilities for patient-specific musculoskeletal models where muscle paths can directly conform to bone shapes extracted from medical image data. Our method has been implemented in the open source simulation system ArtiSynth (www.artisynth.org).
Sign-in or become an IEEE member to discover the full contents of the paper.