Comparison of Scaled vs. Ultrasound Based Musculoskeletal Models on Knee Muscle Moments During Single-Leg Squatting

Abstract

Muscles produce force, resulting in moments about a joint, causing movement of the body. Muscle forces are estimated with a Hill-type model incorporating four parameters; optimal fiber length (OFL), tendon slack length, physiological cross sectional area (PCSA), pennation angle, and maximal isometric force (F[superscript]max) scaled to individual subjects. Purpose: The purpose of this study was to determine if subject specific musculotendon parameters estimated in vivo using ultrasound would better estimate moments produced about a joint compared to previous scaling methods. Methods: 7 recreationally active and resistance trained males and females with no history of lower extremity injury participated. Subjects performed single-leg squats while kinematic, kinetic, and muscle activation data was recorded. Two models for each subject were used in SIMM to estimate knee moments, activations, and muscle forces: scaled (SC) and ultrasound-based (US). Ultrasound imaging of the primary knee muscles were used to derive subject-specific muscle parameters. Scaled muscle parameters were scaled from the model's generic muscle parameter values. Results: The scaled model produced approximately 50% more error compared to the ultrasound model (RMSE: US= 2.71Nm vs. SC= 6.08 Nm) when comparing inverse dynamics knee moments to each model. EMG analysis showed less error in the ultrasound vs. scaled models when compared to experimental muscle activation (RMSE: US= 0.16 mV ± .07, SC= 0.23 mV ± .09) (p< .05). Hamstring activation error was not statistically different between models (RMSE: US= 0.13 mV ± .07 vs. SC= 0.11 mV ± .04) (p> .05). Correlations between model and experimental EMG were weak to modest in both models for all muscles [Quadriceps: (US r= 0.50 mV ± .45, p< 0.01, SC r= 0.55 mV ± .27, p<0.01), Hamstrings: (US r= 0.44 mV ± .25, p< 0.01; SC r= 0.23 mV ± .30, p< .01)] Conclusion: Advances in methodologies used in the field of biomechanical musculoskeletal modeling could be applied to a variety of pathological patients enabling researchers and physicians to better understand how pathology relates with muscle function. More research is warranted in the attempt of deriving a more physiological relevant muscle modeling technique.