Dependency of Lower Limb Joint Reaction Forces on Femoral Anteversion

Abstract

BackgroundMusculoskeletal (MSK) models based on literature data are meant to represent a generic anatomy and are a popular tool employed by biomechanists to estimate the internal loads occurring in the lower limb joints, such as joint reaction forces (JRFs). However, since these models are normally just linearly scaled to an individuals anthropometry, it is unclear how their estimations would be affected by the personalization of key features of the MSK anatomy, one of which is the femoral anteversion angle. Research QuestionHow are the lower limb JRF magnitudes computed through a generic MSK model affected by changes in the femoral anteversion? MethodsWe developed a bone-deformation tool in MATLAB (https://simtk.org/projects/bone_deformity) and used it to create a set of seven OpenSim models spanning from 2{degrees} femoral retroversion to 40{degrees} anteversion. We used these models to simulate the gait of an elderly individual with an instrumented prosthesis implanted at their knee joint (5th Grand Challenge dataset) and quantified both the changes in JRFs magnitude due to varying the skeletal anatomy and their accuracy against the correspondent in vivo measurements at the knee joint. ResultsHip and knee JRF magnitudes were affected by the femoral anteversion with variations from the unmodified generic model up to 11.7{+/-}5.5% at the hip and 42.6{+/-}31.0% at the knee joint. The ankle joint was unaffected by the femoral geometry. The MSK models providing the most accurate knee JRFs (root mean squared error: 0.370{+/-}0.069 body weight, coefficient of determination: 0.764{+/-}0.104, largest peak error: 0.36{+/-}0.16 body weight) were those with the femoral anteversion angle closer to that measured on the segmented bone of the individual. SignificanceFemoral anteversion substantially affects hip and knee JRFs estimated with generic MSK models, suggesting that personalizing key MSK anatomical features might be necessary for accurate estimation of JRFs with these models.