Title: Analyse van afwijkende gewrichtsbelasting en spiercontrole bij kinderenmet botvervormingen gebruikmakend van dynamische simulaties van hetgangpatroonUnderstanding impaired joint loading and muscle control in a populationof children with bony deformities using dynamic simulations of gait
Other Titles: Analyse van afwijkende gewrichtsbelasting en spiercontrole bij kinderenmet botvervormingen gebruikmakend van dynamische simulaties van hetgangpatroon
Authors: Bosmans, Lode
Issue Date: 11-Dec-2014
Abstract: Human gait is probably one of the most frequently used motor skills during daily life. Human gait results from a complex interaction of different lower limb muscles crossing different joints, navigated by high-levelneural control mechanisms. Within a healthy population, each subject presents the same typical gait characteristics. However; differences between individual subjects are present, with each walking pattern presentingits own kinematic and kinetic characteristics. When subjects present movement disorders (e.g. subjects with Parkinson disease, stroke, a traumatic bone or muscle injury, etc), the typical gait characteristics are impaired. Therefore, in order to effectively treat movement disorders in patients, a good understanding of the individual gait impairments is fundamental. Especially in more complex disorders, such as cerebral palsy (CP), the analysis of individual gait impairments is essential to determine the most suitable treatment. Children with CP encounter several primary (muscle weakness, spasticity and impaired selective control) and secondary (muscle contractures and bony deformities) disorders that impair gait.In the past, experimental studies have contributed to insights inthe biomechanical mechanisms underlying muscle coordination patterns during normal and pathological gait. However, in order to determine causalrelationships between muscle forces underlying gait and the resulting joint loading or muscle control of joint kinematics, the use of musculoskeletal models in dynamic simulations of gait is required. Most often, generic musculoskeletal models, that describe the human musculoskeletal system of an averaged adult with normal anthropometrical properties, were used. These generic models were then scaled to closely match the anthropometry of an individual subject. For a population of children with CP presenting proximal femoral deformities, more specifically increased femoral anteversion angle and neck-shaft angle, the use of generic models is not appropriate. Therefore, musculoskeletal models that incorporate moresubject-specific detail are required. Such models can be created based on medical imaging, more specifically magnetic resonance imaging (MRI). Therefore, this thesis used subject-specific musculoskeletal models and dynamic simulations of gait to define the causal relationships between underlying muscle forces on the one hand and joint loading and muscle control of kinematics on the other hand during pathological gait in children with CP.In this thesis we addressed the following four research topics:·         Identification of the influence of three-dimensional anatomical variability of muscle point location on muscle forces during normal walking.·         Identification of the effect of including subject-specific geometry, and more specifically proximal femoral deformities, in the musculoskeletal model on magnitude and orientation of hip joint loading during gait in children with CP.·         Identification of the effect of subject-specific geometry, more specifically proximal femoral deformities, on muscle control of pelvis and hip kinematics during gait in children with CP.·         Identification of the impact of increasing levels of subject-specific detail of musculoskeletal geometry on the magnitude and orientationof hip joint loading during gait in children with CP.The first study examined inter-subject variability in musculoskeletal geometry in healthy subjects. Thereafter, the effect of this variability on calculated muscle forces was investigated using dynamic simulations of gait. We acquired medical imaging of six healthy subjects to analyze the anatomical variability of musculotendon points, which were documented in terms of the absolute (in m) and relative (percentage of bone geometry dimension) variability. Thereafter, we obtained experimental data from an instrumented gait analysis of one control subject with normal gait characteristicson a treadmill at 4 km/h. Finally, dynamic simulations of gait were generated, by varying specific musculotendon points within their documentedanatomical variability to assess the effect on calculated muscle forces. The results of the anatomical variability showed that each musculotendon point shows distinctive variability in each dimension. The largest absolute anatomical variability was noticed in the superior-inferior dimension of musculotendon points belonging to the femur segment. Relative anatomical variability was largest in the anterior-posterior dimension direction and this for each segment. Between all musculotendon points, clear differentiation of the magnitude of the anatomical variability was noticed.  Moreover, the location of several musculotendon points of the generic model showed poor correspondence with the documented anatomical variability, especially for the musculotendon points belonging tothe pelvis segment. Better correspondence with the anatomical variability was found for musculotendon points belonging to the femur segment. Best correspondence was found for the selection of musculotendon points belonging to the tibia segment analyzed in this study. Thereafter, musculotendon points of the generic model were perturbed within the documented anatomical variability to analyze the effect on the muscle forces calculated during a dynamic simulation of normal gait. The results showed thatthe anatomical variability of musculotendon point location affected thecalculated muscle forces. Muscles most sensitive to perturbations within the anatomical variability were iliacus and psoas. Perturbation of thegluteus medius anterior, iliacus and psoas induced the largest concomitant changes in muscle forces of other, unperturbed muscles. Therefore, when creating subject-specific musculoskeletal models, these attachment points should be defined precisely. However, the size of the anatomical variability of the musculotendon pointlocation was not related to the sensitivity of the calculated muscle forces: Musculotendon points with the largest anatomical variability did not necessarily show the largest sensitivity of muscle forces. In contrast, the musculotendon points which induced large concomitant changes in calculated muscles forces upon perturbation, showed rather small anatomical variability.The second study examined the effect of accounting for aberrant bone geometry in musculoskeletal models on the magnitude and orientation of the hip joint loadingduring gait in children with CP. More specifically, we discriminated the effect of bony deformities on hip joint loading from the effect of aberrant gait characteristics on hip joint loading. First, experimental gait data was collected for seven children with CP and one healthy child. Second, medical imaging data were collected, in order to create personalized musculoskeletal models for the children with CP. Subsequently, dynamic simulations of subject-specific pathological gait and imposed normal gait were generated using generic and personalized models to analyze hipjoint loading. When imposing normal gait characteristics to the musculoskeletal model, the presence of proximal femoral deformities increased the magnitude of all components of the resultant hip contact force duringthe first peak of normal gait and consequently the magnitude of the resultant hip contact force vector. This is mainly due to the increased (internal) hip extension moment at loading response, which increased muscleforces in the hip extensors. At the second peak of the resultant hip contact force, the loading was not increased. Instead, it was slightly decreased, however non-significantly. Furthermore, the resultant vector wasorientated more vertically in both the sagittal and frontal plane when accounting for the proximal femoral deformity. These results suggest that proximal femoral deformities affect both the magnitude and orientationof the resultant hip contact force during normal gait. When imposing CP-specific gait characteristics to the musculoskeletal models that include proximal femoral deformity, the combination of aberrant bone geometry and CP-specific gait characteristics reduced the hip contact force compared to those observed during normal gait. However, the orientation of the hip contact force vector changed drastically: The hip contact force was orientated even more vertically and anteriorly compared to hip contactforce orientation during normal gait. From a functional perspective, these results suggest that aberrant femoral geometry affects hip joint loading during both pathological and normal gait. Furthermore, they suggesta beneficial effect of altered gait kinematics on joint loading affecting the loading orientation. These findings are relevant to understand the development of bony deformities. Previous studies have analyzed the effect of aberrant joint loading, stating that the presented joint loadingorientation further deteriorates the degree of femoral deformity over time.The third study analyzed the effect of accounting for aberrant bone geometry in musculoskeletal models on muscle control of joint kinematics during CP gait. More specifically, individual muscle contributions to joint angular accelerations at the hip and knee joint and pelvis orientation were analyzed. For this study, a workflow similar to the second study was used. However, differences in muscle control of joint kinematics due to the presence of proximal femoral deformity and the presence ofCP gait were analyzed separately. When imposing subject-specific gait in the presence of proximal femoral deformities, both factors have a distinct effect on muscle control of the analyzed degrees of freedom. For individual muscles, negative effects (i.e. changes in muscle control that facilitate crouch gait) as well as positive effects (i.e. changes in muscles control that counteract crouch gait) are reported depending on the analyzed degree of freedom, indicating a complex, muscle-specific effectof the deformity. In general, both femoral deformity and CP gait affectmuscle control most in the sagittal plane, followed by transverse and frontal plane. Primary muscle functions are preserved, however, their relative contributions change. More specifically, the CP gait pattern countered for the loss of muscle potential induced by the proximal femoral deformity. This was found for gluteus maximus/medius in the sagittal (strengthening the potential to increase pelvis retroversion, hip and knee extension) and transversal plane (increased potential to induce external hip rotation), for the vasti in the sagittal plane (strengthening its potential to increase pelvis retroversion) and the hamstrings in the frontal plane (strengthening its potential to increase pelvis tilt and decrease hip adduction). In contrast, CP gait further compromised the muscle potential of gluteus maximus/medius transverse plane control (increased backward pelvis rotation), iliopsoas/rectus femoris frontal plane control (increased hip adduction) and the hamstrings sagittal plane (decreased pelvis retroversion, hip extension and increased knee flexion). Interestingly, a trade-off in hip extension potential between the mono-articular gluteus maximus and bi-articular hamstrings was noticed. The mono-articular gluteus maximus (and medius) potentials as hip extensor were augmented, whereas the potential of the hamstrings to extend the hip decreased and the potential to flex the knee increased. From a biomechanical pointof view, the CP gait characteristics are therefore beneficial for gluteus maximus, strengthening their role as hip extensors, but less beneficial for the hamstrings, facilitating their role in inducing crouch gait characteristics. Likewise, the hamstrings show increased potentials to generate pelvis tilt, whereas these potentials reduced for gluteus maximus. However, this interpretation always needs to be considered relative tothe actual muscle force production. Gluteus medius produces a substantial level of force during the entire single stance. However, the muscle force production of gluteus maximus reduces at the end of mid stance. Therefore, their potential effect to extend the hip at the end of mid stance is probably overestimated. For the biarticular hamstrings, muscle force production is low during single stance. Therefore, the reported potentials of the biarticular hamstrings are probably overestimated. In contrast, for biceps femoris short head a substantial amount of muscle force is present. On the other hand, presence of proximal femoral deformities during CP gait does not always negatively influence muscle function therefore predisposing the muscle control towards a crouch gait pattern. Often, a positive effect on the muscle function in one plane causes concomitant effects in other planes. For example, the positive effect on gluteusmaximus to extend the hip and knee is accompanied with a reduced potential to induce pelvis tilt and an increased potential to induce backward rotation of the contralateral pelvis. Therefore, CP gait in the presenceof proximal deformity is a well-balanced situation optimizing muscular control in the different planes.The fourth and last study examined the representativeness of musculoskeletal models that incorporate different levels of subject-specific detail to estimate hip joint loading during gait in children with CP. All models were compared against hip joint loading calculated using MRI-based musculoskeletal models. Similar to thesecond study, experimental gait data was collected in seven children with CP. In addition, MRI data was collected. In total, five musculoskeletal model types were created, with each model type accounting for a different level of subject-specific detail: First, an MRI-based model was created which included image-based definition of muscle paths, joint centers and local reference frames. Second, two model types were created usingmore automated methods, i.e. a morphed generic and a deformed generic model. Both model types matched the generic femoral bone geometry with the subject-specific femoral bone geometry using a dedicated transformation. This transformation was then applied to the musculotendon points associated with the femur segment. For both of these two model types, an additional type was created, including subject-specific hip joint centers. This resulted in five model types. Subsequently, dynamic simulations of pathological gait and imposed normal gait were generated using the five model types to analyze the representativeness of the hip joint loading estimation. Results were compared to the estimations of the MRI-based model, as this model type incorporates the highest level of subject-specific detail. When the models did not account for the subject-specific hip joint center, hip joint loading is overestimated at the first peak by morphed and deformed generic models. As the (internal) hip extension momentis overestimated at loading response, some hip extensors (gluteus maximus anterior/middle and gluteus medius posterior) show increased muscle forces. At the second peak, hip joint loading is underestimated by morphed and deformed generic models. As the (internal) hip flexion moment is underestimated at toe off, some hip flexors (rectus femoris and iliacus) show decreased muscle forces. All together, these factors contribute to the fact that the morphed and deformed generic musculoskeletal models fail to reproduce comparable orientation and magnitude of the hip joint loading. However, when these model types did account for the subject-specific hip joint center, a decreased (internal) hip extension moment at loading response and increased (internal) hip flexion moment at toe off arenoticed. At the first peak, decreased muscle forces of gluteus maximus anterior/middle and gluteus medius posterior were noticed. Despite the fact that the orientation of the resultant hip contact force in the transversal plane better approximated the MRI-based model, the resultant hip contact force was overestimated by both model types. However, at the second peak, increased muscle forces of iliacus, psoas and rectus femoris were noticed. This increase in muscle force of the hip flexors generated comparable joint loading, mainly due to a better estimation of the superior-inferior component. The orientation in the transversal plane also improved towards MRI-based values. In conclusion, the inclusion of the subject-specific hip joint centers in the morphed and deformed generic models clearly improved hip joint loading estimates, but these estimates remained significantly different compared to the MRI-based model. These remaining differences therefore can be attributed to the difference in representativeness compared to the MRI-based musculoskeletal models, which included additional information about musculotendon locations, muscle lines of action as well as local axis frames. This study therefore suggested that is it necessary to include image-based muscle paths, local reference frames and joint centers in order to obtain representative estimatesof hip joint loading. It is clear that the automated definition of muscle paths based on mathematical algorithms to match bone meshes were not suitable to sufficiently personalize musculoskeletal models to allow a representative estimate of joint loading. Therefore, these model types - although interesting given the reduced time required for model definition - should be used with caution.Overall, this thesis contributes to a better understanding of the influence of the presence of proximal femoral deformities during gait in children with CP. More specifically, we acquired a better insight in the causal relationships between individual muscle forces during gait and their effect on the magnitude and orientation of joint loading and muscle control of joint kinematics due to the inclusion of subject-specific information in musculoskeletal models. The findings of this doctoral thesis are therefore relevant for the better understanding of the development of bony deformities and gait deviations in children with CP as well as for further customization of the musculoskeletal models underlying dynamic simulations of gait in these patients.
Publication status: published
KU Leuven publication type: TH
Appears in Collections:Human Movement Biomechanics Research Group
Production Engineering, Machine Design and Automation (PMA) Section
Organ Systems (+)

Files in This Item:
File Status SizeFormat
20141116_phd_v39_Procopia_v2.pdf Published 8126KbAdobe PDFView/Open Request a copy

These files are only available to some KU Leuven Association staff members


All items in Lirias are protected by copyright, with all rights reserved.