Patient-Specific Prediction of Total Knee Arthroplasty Surgical Exposure Using a Statistical Shape Model Augmented...

Knee osteoarthritis is a leading cause of joint degeneration, often treated with Total Knee Arthroplasty (TKA). Surgical exposure, essential for implant positioning, varies depending on the chosen approach (e.g., medial parapatellar, subvastus, midvastus), impacting soft tissue preservation and recovery. Optimizing exposure is also crucial for developing personalized solutions like robotic systems or patient-specific instrumentation (PSI). We present a Statistical Shape Model (SSM)-based approach to predict the portion of the knee joint surface exposed during TKA. The method leverages a new semi-automatic annotation technique of preoperative models paired with intraoperative RGB-Depth images captured during TKA. Augmented SSMs of the femur and tibia are constructed from bone meshes and annotated exposed areas. The accuracy of the predictions is assessed on 10 patients by comparing predicted and manually annotated exposure regions. Good similarity was observed, with dice scores and average symmetric surface distance values of 0.87 and 0.74 mm respectively for femur, and 0.90 and 0.21 mm for tibia. Inter-observer variability between two experts was used to assess the impact of manual bone annotation on RGB images, with high similarity - dice scores of 0.98 for femur and tibia - indicating minimal impact. These promising results illustrate the possibility of patient-specific prediction of surgical exposure.Clinical relevance- This approach has the potential to support a wide range of orthopedic applications. It can enhance understanding of TKA surgical exposure and facilitate comparisons between different surgical approaches. Preoperatively, the augmented SSM can refine bone segmentation, improve surgical planning for implant sizing and positioning, and help in the design of PSIs. It could also help improve the design of navigated or robotic solutions. Additionally, predicted surgical exposure could be visualized in virtual reality or on phantoms to help in training young surgeons.