A Combined Multibody and Plantar Pressure Approach to Estimate and Predict Foot Kinematics Applied to 3D-printed Insoles

In Europe, over 200 million people suffer from disabling foot and ankle pain caused by a variety of pathologies as overloading injuries, degenerative joint disorders and systemic diseases (e.g. diabetes). Foot orthoses (FOs) or insoles are a popular conservative treatment to alleviate pain and prevent further deterioration of the pathological condition; the goal is to optimize a patient’s foot function by relieving symptoms and slow down or even arrest the progression of the pathology (e.g., rearranging pressure distribution in diabetic patients to prevent foot ulcers). Currently, the process of patient assessment for insole design and manufacturing relies on subjective decision making and time-consuming handicraft work by the clinician. The design of these orthoses is primarily based on the capture of the foot shape using traditional techniques such as plaster casting.

3D-printed insoles reduce the manufacturing time and allow efficient production of subject-specific insoles. It allows for the local adjustment of mechanical properties so that the insole can optimally guide the movement of the foot. Therefore, it is crucial to understand the biomechanical effect of the insole on the foot. The overall aim of this thesis was to develop the different steps necessary for the creation of a fully objective and automated digital workflow that goes from foot biomechanics assessment to the production of subject-specific insoles.

Computer multibody models can describe the causal relations between input parameters (i.e., kinetics) and model response (i.e., kinematics) given a known geometrical structure. Only a limited number of detailed musculoskeletal foot models for use in dynamic simulations were described in literature. However, the number of accessible detailed musculoskeletal foot models is limited, this is a major limitation to reproduce results or to use the reported models in different studies. In the present work, two detailed OpenSim 3D multibody foot-ankle models generated based on CT scans using a semi-automatic tool, are described. The proposed models consist of five rigid segments (talus, calcaneus, midfoot, forefoot, and toes), connected by five joints (ankle, subtalar, midtarsal, tarsometatarsal and metatarsophalangeal), one with 15 DOF and the other with 8 DOF. The calculated kinematics of both models were evaluated using motion capture and compared against literature, both presenting realistic results. An inverse dynamic analysis was performed for the 8 DOF model, again presenting dynamic results similar to literature.

The 8 DOF foot model was then used with in-vivo gait analysis measurements of flat feet and control subjects using different footwear while walking. The differences in kinematic and kinetic parameters between control and flat feet subjects were estimated in barefoot and shoe walking. In parallel, the influence of subject-specific 3D-printed insoles on kinematics and kinetics of flat feet subjects was compared to the influence of a subject-specific Ethylene-Vinyl Acetate (EVA) molded insole. The flat feet subjects presented an increase in forefoot dorsiflexion and abduction while walking barefoot. The use of EVA and 3D-printed insoles corrected the aberrant flat feet-related kinematics and kinetics presented during the minimalistic shoe condition: arch height increased, as reflected by the decreased forefoot dorsiflexion and abduction. Both insoles provided similar correction, thereby confirming the suitability of 3D-printed insoles in the correction of flat feet kinematics and kinetics.

The use of plantar pressure measurement systems to evaluate foot and ankle pathologies is well established in clinical practice. Although relevant to detect local tissue (over-)loading, to date, plantar pressure data cannot be used to evaluate ankle-foot kinematics. In this thesis, we present a least squares optimization algorithm that minimizes the weighted difference between simulated and measured plantar pressure using different marker-sets data. Both, marker positions and plantar pressures are simulated in OpenSim using the previously described 8 DOF foot model coupled with an ellipsoid based elastic foundation contact model. It was concluded that a minimum of four markers combined with pressure data was needed to estimate the kinematics with accuracy comparable to the full marker approach.  The ability of estimating full foot kinematics using a simplified set-up that relies on a limited number of reflective markers combined with plantar pressure measurements has significant time and economic implications for both research and clinical applications. It will allow a more accessible, objective clinical evaluation of foot pathologies.

With the introduction of 3D-printed insoles came the ability to manufacture insoles with mechanical properties that are tailored to specific subjects in a time efficient manner. However, our understanding of how mechanical insole properties influence the dynamic behaviour of the foot during movement is limited. So, a torque driven forward simulation framework was developed to evaluate the effect of different insoles’ properties on the kinematics of the foot.  In this thesis the effect of different insole stiffnesses on foot kinematics during walking was evaluated. The torque driven forward simulations were computed in OpenSim again using the 8 DOF foot model. The model was coupled with an elastic foundation foot-ground contact model. The insoles were modeled using bushing forces connecting the calcaneus to the forefoot. Increased bending stiffness mostly independent from torsional stiffness led to kinematic adaptations that have the potential to partially correct low foot longitudinal by increasing forefoot adduction.

This thesis demonstrates the potential of detailed foot musculoskeletal models to be used for kinematic acquisition in a clinical environment and for the improvement of insoles design. These advances can lead to improved orthopaedic patient care, improving both the evaluation as well as the treatment prescription process.