Orthopaedic disorders involve abnormal mechanics that can sometimes be corrected by surgery. Flatfoot deformity involves collapse of the medial arch of the foot, which can lead to severe pain and difficulty in walking. Surgical corrections involve various alterations of the bones, joints, and muscle attachments with the aim of relieving pain and improving foot mechanics. Despite being a very common condition, flatfoot is one of least understood and most controversial orthopaedic problems.

A methodology was developed for creating subject-specific computer models of anatomically based foot mechanics and for utilizing them in simulations of different surgeries for flatfoot correction. Magnetic resonance imaging of an asymptomatic flatfoot subject was used to construct three dimensional geometries of individual bones for the model. Radiographs of the same subject were obtained with the foot under different loading conditions. Finite element techniques were used to model the contact and movements at the joints, reactions of ligaments, and hyperelastic stress-deformation behavior of soft tissues beneath the foot. A novel optimization technique was developed to automatically adjust certain soft tissue parameters of the finite element model such that the postures of foot bones were reproduced when standing was simulated. This tracking was accomplished by matching model bone positions to those measured in radiographs of the same subject. Custom-written code was developed to automatically perform this matching by repeatedly adjusting the finite element model, performing simulations, and interpreting results. Techniques were developed to enable the simulations to run quickly while retaining stability and accuracy. Using the resulting subject-specific model, several surgeries for flatfoot were then simulated, including medializing calcaneal osteotomy and several medial column fusions. Changes in arch height and plantar pressures were assessed.

The numerical optimization technique successfully reduced the differences between model and experimental bone positions of our subject during standing. Through automatic adjustment of unknown soft tissue properties of the model, differences between bone bounding edges were reduced to an average magnitude of 1.4 mm per edge, whereas a nominal estimate of model parameters gave average errors of 2.6 mm. Validity of the model was also supported by comparisons of plantar pressures predicted in the model to those measured experimentally from our subject. Several different types of simulated surgical arthrodesis each increased arch height by more than 3 mm and shifted plantar pressures more onto the lateral midfoot. Medializing calcaneal osteotomy also shifted plantar pressures laterally, and it slightly increased arch height.

The modeling framework developed in this research enables semi-automated creation and testing of subject-specific biomechanical foot models. A collection of such models will allow the merits of different surgeries to be systematically tested. Such testing is not possible with in vivo studies because surgical variations cannot be performed on a patient. Systematic evaluation of orthopaedic surgeries will lead to better preoperative planning that takes individual subject characteristics into account.