Patellar dislocation is a complex pathoanatomical condition that affects approximately 1:1000 children and adolescents. Up to 71% of those affected are at risk of recurrent dislocations, that can lead to long-term complications, such as persistent knee pain, decreased activity levels and impairment of knee function. The cause of patellar dislocation is believed to be multifactorial, including lower limb misalignment, abnormal patellofemoral morphology and inadequate soft tissue restraints. While anatomical risk factors and thresholds for typical measures have been well established in adults, there is a paucity of normative and pathoanatomical data that contribute to patellar dislocation in paediatric populations, where the morphology of the patellofemoral joint (PFJ), as well as lower limb alignment, change with growth. Moreover, many surgical techniques common in adults (e.g., trochleoplasty) cannot be performed in a paediatric population due to the risk of growth disruption. Given the challenges in managing recurrent patellar dislocation in paediatric populations, treatment is often unsuccessful and commonly results in poor functional outcomes for the patient while incurring substantial costs. Poor treatment outcomes may be due to limitations in current diagnosis and treatment methods. Nowadays, clinicians use static measures from medical imaging to inform diagnosis and treatment of recurrent patellar dislocation. These measures, despite providing valuable insights on the PFJ anatomy, are unable to provide understanding of the patellar dislocation mechanism during dynamic tasks. Conversely, subject-specific computational models of the musculoskeletal system, with individualised geometries and anatomical structures, have the potential to capture the complex functional relationship between multiple risk factors for patellar dislocation on an individual basis. Creating fully subject-specific models requires accurate personalisation of the joint kinematics, which will improve estimates of all the dependent quantities of interest for musculoskeletal modelling (e.g., muscle moment arms, articular contact points and ligament kinematics). Therefore, the general purpose of this thesis was to develop subject-specific rigid-body models of the tibiofemoral and patellofemoral joints in children and adolescents with recurrent patellar dislocation. The first study systematically reviewed the current literature to characterise lower limb alignment, patellofemoral morphology and soft tissue restraints of the PFJ through medical imaging measurements in paediatric recurrent patellar dislocators (RPD) and age-matched typically developing (TD) participants. Moreover, the data were synthesised to stratify the factors that influence PFJ stability and recommendations on the assessment and reporting of PFJ parameters in this patient population were provided. Results from a meta-analysis conducted on measures reported in two or more studies that included both a control and a patellar dislocator group showed that the tibial tuberosity to trochlear groove (TT-TG) distance and bony sulcus angle can be confidently used to predict the risk of recurrence in the paediatric population. These results can streamline the patient evaluation and best inform clinical decision-making. The paper describing these results was published as Barzan, M., Maine, S., Modenese, L., Lloyd, D.G., Carty, C.P., Patellofemoral joint alignment is a major risk factor for recurrent patellar dislocation in children and adolescents: a systematic review. JISAKOS 2018, 0:1-11. doi:10.1136/jisakos-2017-000189. The aim of the second study was to analyse the differences in lower limb alignment, patellofemoral alignment and trochlea dysplasia between paediatric RPD patients and TD participants using magnetic resonance imaging (MRI). This was essential as results from the previous study revealed that there is a paucity of reported radiological parameters to define normal and pathoanatomical paediatric cohorts. A prospective crosssectional study was conducted on 24 RPD children and adolescents and 25 age-matched TD participants. Significant differences between the two groups were found for acetabular inclination, tibial-femoral torsion, TT-TG distance, lateral patellar tilt, congruence angle and cartilaginous sulcus angle. TT-TG distance and cartilaginous sulcus angle were included in the final predictive model, which correctly classified 84.4% of cases of recurrent patellar dislocation. Therefore, these measures should be included in the evaluation of paediatric patients who present with recurrent patellar dislocation. The paper describing these results was submitted as Ngo-Nguyen, C., Maine, S., Barzan, M., Stockton, C.A., Modenese, L., Lloyd, D.G., Carty, C.P. Radiological predictors of paediatric patellofemoral joint dislocation from medical imaging. The Knee. The purpose of study three was to develop three subject-specific tibiofemoral (TFJ) kinematic models, with either rigid or extensible ligament constraints, and a subject-specific PFJ model for eight healthy paediatric participants. The estimated joint and ligament kinematics from the three models were also validated against in vivo kinematics measured from MRIs at four different TFJ flexion angles. The three TFJ models were created from MRIs and used to solve the TFJ kinematics: (i) 5-rigid-link parallel mechanism with rigid surface contact and isometric anterior cruciate (ACL), posterior cruciate (PCL) and medial collateral (MCL) ligaments (Δ!"), (ii) 6-link parallel mechanism with minimised ACL, PCL, MCL and lateral collateral ligament (LCL) length changes (Δ!$%&) and (iii) 6-link parallel mechanism with prescribed ACL, PCL, MCL and LCL length variations (Δ!$'()*). The Δ!" and Δ!$'()* models compared best against MRI-measured data, with errors below 7° and 7 mm for joint angles and displacements, respectively, and below 2 mm for ligament lengths. Therefore, these models can be used to estimate passive three-dimensional paediatric TFJ, PFJ and ligament kinematics and can be incorporated into lower-limb models to estimate joint kinematics and kinetics during dynamic tasks. The paper describing these results was submitted as Barzan, M., Modenese, L., Carty, C.P., Maine, S., Stockton, C.A., Sancisi, N., Lewis, A., Grant, J., Lloyd, D.G., Brito da Luz, S. Development and validation of subject-specific paediatric multibody knee kinematic models with ligamentous constraints. Journal of Biomechanics. The fourth study assessed the TFJ and ligament kinematics during gait using a rigid-body lower limb model incorporating a fully subject-specific paediatric kinematic TFJ model with articular contacts and minimally deformable ligaments. This was essential to extend the findings of the previous study to dynamic tasks. To address this aim, eight healthy participants underwent MRI and three-dimensional gait analysis. For these participants, the TFJ was implemented in OpenSim, based on optimised MRImeasured geometrical parameters, as a 5-rigid-link parallel mechanism with spherical articular contacts and three knee ligaments (ACL, PCL and MCL). For each participant, TFJ angles and ligament lengths were calculated by tracking experimental markers while minimising ligament elongation using the least squares multibody optimisation (MBO) tool available in OpenSim. The kinematic results from MBO were compared against those obtained using the implicit 5-rigid-link mechanism Δ!", with significant differences found for TFJ ab/adduction, ACL and PCL strains. The developed subject-specific TFJ kinematic models are promising tools to investigate the gait biomechanics of healthy children and future studies will extend their use to the analyses of pathological gait. The manuscript describing these results will be submitted as Barzan, M., Carty, C.P., Maine, S., Sancisi, N., Stockton, C.A., Edwards, J., Brito da Luz, S., Lloyd, D.G., Modenese, L. Implementation of a subject-specific paediatric kinematic model of the knee with minimally deformable ligaments in OpenSim. Journal of Biomechanics. The purpose of study five was to develop subject-specific PFJ kinematic models to evaluate passive patellar tracking in paediatric RPD patients and TD controls. The resulting PFJ kinematics for RPD patients were also validated against in vivo kinematics measured from MRIs at four different TFJ flexion angles. Finally, the estimated PFJ kinematics between RPD patients and TD participants were compared. For RPD patients, we modelled the PFJ from MRIs using two different hinge mechanisms, which described (i) the lateral to medial translation of the patella into the trochlear groove from approximately 0° to 30° of TFJ flexion, and (ii) the motion of the patella after it reached a more congruent position in the trochlear groove. When compared to MRI data, the proposed PFJ models were able to characterise different patterns of patellar maltracking in RPD patients. Moreover, RPD patients exhibited a more externally rotated and lateralised patella than TD participants between 0° and 30° of TFJ flexion. These models provided accurate estimations of pathological PFJ kinematics and might be used to inform surgery planning and evaluation. The manuscript describing these results will be submitted as Barzan, M., Maine, S., Brito da Luz, S., Modenese, L., Stockton, C.A., Conconi, M., Sancisi, N., Lloyd, D.G., Carty, C.P. Development and validation of subject-specific patellofemoral joint kinematic models for children and adolescents with recurrent patellar dislocation. Journal of Orthopaedic Research. In conclusion, this thesis examined the anatomical and functional differences at the knee between paediatric patients with recurrent patellar dislocation and age-matched controls. The unique TFJ and PFJ kinematic patterns observed in our paediatric cohort suggest that personalised musculoskeletal models are necessary to understand potential patellar instability mechanisms. Moreover, the developed and validated knee kinematic models provide a platform for muscle force integration, thereby enabling the calculation of PFJ contact forces. Importantly, informing surgical planning with personalised musculoskeletal models that can assist in elucidating the mechanisms of PFJ instability should result in better long-term surgical outcomes, such as reduced costs from surgical revision and improved patient quality of life.