Optimization of muscle wrapping objects using simulated annealing.

Abstract

Musculoskeletal models use wrapping objects to constrain muscle paths from passing through anatomical obstacles; however, the selection of wrapping object parameters is typically a manual, iterative, and time-consuming process. The purpose of this study was to use a data-driven optimization algorithm to determine wrapping object parameters. Wrapping parameters were determined using simulated annealing for two cases: (1) modeling the triceps at the elbow using a cylindrical wrapping object, and (2) modeling the middle deltoid using a spherical wrapping object. It was found that an optimization algorithm could be used to determine wrapping object parameters which produced moment arms that were similar to experimental data. The greatest benefit of this method is the efficiency at which model parameters were determined, thus eliminating much of the time required to manually refine the wrapping objects. Model development could be further improved by extending this method to other model parameters and combining various optimization techniques.