IMT-Atlantique Bretagne- Pays de la Loire, Brest, France.
LaTIM, INSERM, SFR IBSAM, Brest, France.
Int J Med Robot. 2021 Oct;17(5):e2296. doi: 10.1002/rcs.2296. Epub 2021 Jun 11.
A new approach is proposed to localise surgical instruments for Computer Assisted Orthopaedic Surgery (CAOS) that aims at overpassing the limitations of conventional CAOS solutions. This approach relies on both a depth sensor and a 6D pose estimation algorithm.
The Point-Pair Features (PPF) algorithm was used to estimate the pose of a Patient-Specific Instrument (PSI) for Total Knee Arthroplasty (TKA). Four depth sensors have been compared. Three scores have been computed to assess the performances: The Depth Fitting Error (DFE), the Pose Errors, and the Success Rate.
The obtained results demonstrate higher performances for the Microsoft Kinect Azure in terms of DFE. The Occipital Structure core shows better behavior in terms of Pose Errors and Success Rate.
This comparative study presents the first depth-sensor based solution allowing the intraoperative markerless localization of surgical instruments in orthopedics.
提出一种新的方法来定位手术器械，用于计算机辅助骨科手术（CAOS），旨在克服传统 CAOS 解决方案的局限性。该方法依赖于深度传感器和 6D 位姿估计算法。
使用点对特征（PPF）算法估计全膝关节置换术（TKA）中患者特定器械（PSI）的位姿。比较了四个深度传感器。计算了三个分数来评估性能：深度拟合误差（DFE）、位姿误差和成功率。
在 DFE 方面，获得的结果表明 Microsoft Kinect Azure 的性能更高。Occipital Structure core 在位姿误差和成功率方面表现更好。
这项比较研究提出了第一个基于深度传感器的解决方案，允许在骨科手术中进行无标记的手术器械术中定位。
