In the case of total knee arthroplasties (TKA) a better outcome may be achieved by using patient-customized implants. The implant design is based on the actual bone surface obtained from medical imaging. Due to the degeneration of bone and cartilage, osteophytes may develop which show up in the acquired images and thus in the segmented surface data. The removal of these osteophytes (to be removed by the surgeon during operation) is necessary for implant design and is cumbersome and time consuming.
Therefore, we developed a method for automatic osteophyte removal. From a set of manually segmented bone surfaces as well as their original surfaces we annotated a mean shape with average sites of osteophytes. Osteophytes can then be automatically removed from a new unknown shape by morphing the annotated mean shape into the target shape while omitting sites of osteophytes. Due to the underlying regularization in the morphing procedure, these areas are virtually interpolated and thus the osteophytes are removed.