Towards markerless computer assisted surgery: Application to total knee arthroplasty

Area of Science:

• Orthopaedic Surgery
• Computer Assisted Surgery
• Surgical Instrument Localization

Background:

• Conventional Computer Assisted Orthopaedic Surgery (CAOS) solutions face limitations in surgical instrument localization.
• Accurate real-time tracking of surgical instruments is crucial for enhancing precision in orthopaedic procedures.

Purpose of the Study:

• To propose and evaluate a novel depth-sensor based approach for markerless localization of surgical instruments in orthopaedics.
• To overcome the limitations of existing CAOS localization techniques.
• To compare the performance of different depth sensors for this application.

Main Methods:

• The Point-Pair Features (PPF) algorithm was employed for 6D pose estimation of Patient-Specific Instruments (PSIs) during Total Knee Arthroplasty (TKA).
• Four different depth sensors were evaluated.
• Performance was assessed using Depth Fitting Error (DFE), Pose Errors, and Success Rate metrics.

Main Results:

• The Microsoft Kinect Azure demonstrated superior performance regarding Depth Fitting Error (DFE).
• The Occipital Structure core exhibited better results in terms of Pose Errors and Success Rate.
• Comparative analysis revealed sensor-specific strengths for instrument localization.

Conclusions:

• This research presents the first depth-sensor based solution for intraoperative markerless localization of surgical instruments in orthopaedic surgery.
• The findings highlight the potential of depth sensing technology to advance CAOS.
• The study provides valuable insights for selecting appropriate depth sensors for surgical applications.