Construction and assessment of a computer graphics-based model for wheelchair propulsion

Upper limb overuse injuries are common in manual wheelchair using persons with spinal cord injury (SCI), especially those with tetraplegia. Biomechanical analyses involving kinetics, kinematics, and muscle mechanics provide an opportunity to identify modifiable risk factors associated with wheelchair propulsion and upper limb overuse injuries that may be used toward developing prevention and treatment interventions. However, these analyses are limited because they cannot estimate muscle forces in vivo. Patient-specific computer graphics-based models have enhanced biomechanical analyses by determining in vivo estimates of shoulder muscle and joint contact forces. Current models do not include deep shoulder muscles. Also, patient-specific models have not been generated for persons with tetraplegia, so the shoulder muscle contribution to propulsion in this population remains unknown. The goals of this project were to: (i) construct a dynamic, patient-specific model of the upper limb and trunk and (ii) use the model to determine the individual contributions of the shoulder complex muscles to wheelchair propulsion. OpenSim software was used to construct the model. The model has deep shoulder muscles not included in previous models: upper and middle trapezius, rhomboids major and serratus anterior. As a proof of concept, kinematic and kinetic data collected from a study participant with tetraplegia were incorporated with the model to generate dynamic simulations of wheelchair propulsion. These simulations included: inverse kinematics, inverse dynamics, and static optimization. Muscle contribution to propulsion was achieved by static optimization simulations. Muscles were further distinguished by their contribution to both the push and recovery phases of wheelchair propulsion. Results of the static optimization simulations determined that the serratus anterior was the greatest contributor to the push phase and the middle deltoid was the greatest contributor to the recovery phase. Cross correlation analyses revealed that 80% of the investigated muscles had moderate to strong relationships with the experimental electromyogram (EMG). Results from mean absolute error calculations revealed that, overall, the muscle activations determined by the model were within reasonable ranges of the experimental EMG. This was the first wheelchair propulsion study to compare estimated muscle forces with experimental fine-wire EMG collected from the participant investigated