Salaheddine STA Thesis defense

Total Knee Arthroplasty (TKA) is a common surgical procedure used to treat knee osteoarthritis by replacing the damaged joint with a prosthesis. Despite the frequency of TKA, complications such as prosthesis loosening, wear, and infection often necessitate revision surgeries. Over the years, technological advancements, including robotic systems and patient-specific instruments (PSIs), have been introduced to improve surgical accuracy and reduce complications. However, current PSI techniques lack intraoperative tools for real-time verification of their correct positioning during surgery, which can affect outcomes. This thesis presents an innovative approach to addressing this challenge by using depth cameras and a markerless pose estimation algorithm to verify PSI placement during surgery. Unlike traditional navigation systems that rely
on markers, our method captures real-time information from the surgical site through depth cameras, providing non-intrusive feedback on PSI alignment based on preoperative planning. The core of the approach revolves around the Point Pair Features (PPF) 6D pose estimation algorithm, enhanced with novel improvements to ensure greater robustness in the surgical environment. The results demonstrate the potential of markerless depth camera-based systems for intraoperative PSI localization, though further refinements are needed to meet clinical standards. Additionally, we highlight opportunities for future improvements using advanced deep learning techniques. While initially developed for TKA, this approach could be extended to other orthopedic surgeries, such as hip and shoulder replacements.