From medical images to a personalized and predictive spatial model of the ankle joint

Personalized joint models may allow the definition of better treatments and devices and may improve gait analysis accuracy by reducing soft-tissue artefacts [1]. Spatial parallel mechanisms proved to replicate the natural tibiotalar motion with a high accuracy [2,3], featuring the main articular structures that guide the joint motion (articular contacts and isometric fibres of tibiocalcaneal (TiCaL) and calcaneofibular (CaFiL) ligaments) but they require a motion estimation to adjust the model parameters that, however, cannot be easily obtained in-vivo. A new technique was proposed that obtains the ankle natural motion by maximizing the joint congruence at all flexion angles [4]: it only requires the 3D model of articular surfaces, which can be obtained by medical images. A combination of congruence maximization and parallel mechanisms is here presented to define spatial models that include the joint constraints and predict the tibiotalar natural motion of a specific subject from standard medical images. The procedure is applied and validated in-vitro on five ankles, by comparing model predictions with experimental motion measured by bone pins.