Prediction of the Shape of Severely Fractured Distal Tibia by Using Statistical Shape Modelling

The design of tibial components of ankle implants is critical for proper functioning. The tibial component design uses the shape of the distal tibial bone as reference. However, when the distal tibia is severely damaged or fractured, the design of tibial components becomes very difficult. In this paper, we aim to study prediction of the distal tibia shape based on the remainder of tibial bone. We use statistical shape modelling technique to create a model of the tibial bone, and then we assess its shape variability. We extract the relationships between the shape variations to produce the predictions. A dataset of 22 female bone samples and 30 male bone samples were acquired. A statistical shape model per gender was produced by using a part of the dataset population. The first set of principal component analysis modes accounted for at least 95% of the shape variations were adopted. For the rest of the bone samples, we attempted to predict their distal tibia shapes by feeding the shapes of their proximal tibia and tibial shafts into the models. The prediction was done for roughly up to 20 mm above the distal tibial articular surface. The study was done in a 5-fold cross validation setting. Root-mean-square errors of reconstruction of the samples excluded in the model development were 1.91 ± 0.61 mm and 2.01 ± 0.39 mm for females and males, respectively. The prediction errors of the distal tibia, when only the shape of the proximal tibia and tibial shafts were known, were in average 1.96 mm for female’s bones and 2.46 mm for male’s bones. These small errors can show that the distal tibia shape can be reconstructed based on the proximal and shaft shapes. This is a preliminary result bringing new insights into treatment of ankle orthopaedic diseases. It can pave the way for reconstruction of lost distal tibia.