What is the best hip center location method to compute HKA angle in computer-assisted orthopedic surgery? In silico and in vitro comparison of four methods.

ABSTRACT

BACKGROUND: In computer-assisted orthopedic surgery, the hip center (HC) can be determined by calculating the center of rotation of the femur in relation to the pelvis. Several methods are available Gamage, Halvorsen, Pivot or Least-Moving Point (LMP). To our knowledge, no studies have compared these four methods. We therefore conducted in silico and in vitro experiments to assess whether their accuracy and precision in locating the HC and calculating the hip-knee-ankle (HKA) angle were equivalent. HYPOTHESIS: The four methods show similar accuracy and precision. PATIENTS AND METHODS: The in silico experiment assessed the independent influence of four parameters (camera noise, acetabular noise, movement amplitude, and number of circumductions) on accuracy. The accuracy and precision of the four methods and the impact on HKA ankle calculation were assessed in an in vitro study on six cadaver limbs. RESULTS: In the in silico experiment, all differences according to method were significant (p<0.0002). The Pivot method was the most accurate for acetabular and camera noise, number of circumductions, and movement amplitude. With the LMP, Pivot, Gamage and Halvorsen methods, error was respectively 23.07±8.40 (range 2.10-54.67) mm, 1.98±081 (0.15-4.89) mm, 28.18±3.42 (18.57-37.60) mm and 2.84±1.46 (0.11-9.44) mm depending on camera noise, 1.65±0.72 (0.13-4.80) mm, 0.52±0.22 (0.05-1.23) mm, 3.02±0.57 (0.60-4.78) mm and 0.61±0.27 (0.04-1.82) mm depending on movement amplitude, 0.50±0.20 (0.05-1.34) mm, 0.18±0.08 (0.01-0.44) mm, 0.36±0.14 (0.03-0.80) mm and 0.21±0.09 (0.01-0.55) mm depending on number of circumductions, and 11.30±5.77 (0.56-37.87) mm, 2.78±1.47 (0.10-8.77) mm, 88.08±8.85 (60.59-117.79) mm and 24.33±9.82 (1.40-66.17) mm depending on acetabular noise. In the in vitro experiment, differences were non-significant between the Pivot and LMP methods (p>0.98) and between the Gamage and Halvorsen methods (p>0.65). With the LMP, Pivot, Gamage and Halvorsen methods, precision was respectively 8.2±4.6 (3.3-23.6) mm, 7.3±3.6 (3.4-14.1) mm, 33.6±19.1 (4.7-111.4) mm and 35.0±25.0 (4.7-132.5) mm. Accuracy was 13.5±8.2 (3.2-40.7) mm. 12.3±6.4 (3.2-23.6) mm, 47.0±33.3 (6.2-176.6) mm and 40.3±27.8 (6.1-130.3) mm. The LMP and Pivot methods were thus more accurate and more precise than the Gamage and Halvorsen methods. HKA angle error was 1.1±0.9° (0.1-3.7) and 0.9±0.8° (0.0-2.5) with the LMP and Pivot methods, and 3.2±2.7° (0.0-12.7) and 3.8±3.5° (0.0-13.3) with the Gamage and Halvorsen methods. DISCUSSION: The study highlighted differences between the four methods of HC location in computer-assisted surgery; the Pivot method was the most accurate and precise, thus falsifying the study hypothesis. LEVEL OF EVIDENCE III, prospective comparative in silico and in vitro study.