Generic scaled versus subject-specific models for the calculation of musculoskeletal loading in cerebral palsy gait: Effect of personalized musculoskeletal geometry outweighs the effect of personalized neural control

• Published in: Clinical biomechanics (Bristol) Vol. 87; p. 105402
• Main Authors: Kainz, Hans, Wesseling, Mariska, Jonkers, Ilse
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.07.2021
• Subjects: Cerebral palsy; Electromyography; Gait analysis; Magnetic resonance images; Muscuoskeletal model; OpenSim; Physical Medicine and Rehabilitation; OpenSim; Cerebral palsy; Electromyography; Muscuoskeletal model; Gait analysis; Magnetic resonance images
• ISSN: 0268-0033; 1879-1271; 1879-1271
• DOI: 10.1016/j.clinbiomech.2021.105402

Musculoskeletal modelling is used to assess musculoskeletal loading during gait. Linear scaling methods are used to personalize generic models to each participant's anthropometry. This approach introduces simplifications, especially when used in paediatric and/or pathological populations. This study aimed to compare results from musculoskeletal simulations using various models ranging from linear scaled to highly subject-specific models, i.e., including the participant's musculoskeletal geometry and electromyography data. Magnetic resonance images (MRI) and gait data of one typically developing child and three children with cerebral palsy were analysed. Musculoskeletal simulations were performed to calculate joint kinematics, joint kinetics, muscle forces and joint contact forces using four modelling frameworks: 1) Generic-scaled model with static optimization, 2) Generic-scaled model with an electromyography-informed approach, 3) MRI-based model with static optimization, and 4) MRI-based model with an electromyography-informed approach. Root-mean-square-differences in joint kinematics and kinetics between generic-scaled and MRI-based models were below 5° and 0.15 Nm/kg, respectively. Root-mean-square-differences over all muscles was below 0.2 body weight for every participant. Root-mean-square-differences in joint contact forces between the different modelling frameworks were up to 2.2 body weight. Comparing the simulation results from the typically developing child with the results from the children with cerebral palsy showed similar root-mean-square-differences for all modelling frameworks. In our participants, the impact of MRI-based models on joint contact forces was higher than the impact of including electromyography. Clinical reasoning based on overall root-mean-square-differences in musculoskeletal simulation results between healthy and pathological participants are unlikely to be affected by the modelling choice. •Modelling framework had a minor impact on joint kinematics and kinetics.•Mean differences in muscles forces between models were below 0.2 body weight.•Mean differences in joint contact forces between models were up to 2.2 body weight.•Personalized geometry, electromyography data and gait pattern affect results.•Modelling choice did not affect clinical reasoning based on overall differences.