Effect of robotic manipulator calibration quality on instrument positioning accuracy in a surgical navigation system: an experimental phantom study

Authors:
  • МГМагомедов Ильяс Гайдарович
    Магомедов Ильяс Гайдарович
  • ТСТрапезников Денис Сергеевич
    Трапезников Денис Сергеевич
  • ЧСЧаплыгин Сергей Сергеевич
    Чаплыгин Сергей Сергеевич
  • ЗВЗотов Денис Витальевич
    Зотов Денис Витальевич
Open Access
Publisher website
Google Scholar
Abstract
Aim: to evaluate influence of robotic manipulator calibration accuracy and the spatial configuration of calibration points on the resulting target-point localization accuracy in a robot-assisted surgical navigation system. Material and methods. We conducted an experimental study with the use of an anthropomorphic head phantom made of polyethylene terephthalate glycol (PETG) using FDM. The phantom contained nine fiducial markers and an integrated measurement unit equipped with a monocular optical system. For the study we used integration of a collaborative robotic manipulator (REDS-C10) with a surgical navigation system (Autoplan). The research conducted a complete factor experiment investigating the relative accuracy impact of spatial configuration points (7 calibration configurations A–G of differing number of points and calibrator plans) rated for three calibration options (standard calibration, calibration with a prescribed calibration needle length error of +0.5 mm and calibration with incorrectly determined calibration point positions) and two surgical approaches (90° and 45° relative to the horizontal plane trajectories). A total of 5 repetitions were performed for each case, amounting to 210 measurements in all. The biopsy needle (1.9-mm-diameter) deviation from the aiming point was determined by pixel-based image analysis (pixel size: 0.029 mm). The vertical and horizontal components were measured separately. We used the Kruskal–Wallis and Mann–Whitney U tests with Holm correction to compare the results. Paired comparisons of the components were performed using the Wilcoxon signed-rank test. Results. Under standard calibration, the deviation was 0.81 (0.40) mm (95% CI, 0.71–0.90), compared with 1.21 (0.44) mm for a calibration needle length error and 1.72 (0.54) mm for incorrectly determined calibration point positions (H = 84.7; p 0.001; ε² = 0.40; all pairwise comparisons, p 0.001). The proportions of measurements exceeding the clinically acceptable threshold of 2 mm were 0%, 1.4%, and 30.0%, respectively. The two types of calibration error differed in the direction of the resulting deviation. A needle length error produced predominantly horizontal deviation (1.12 [0.43] vs. 0.35 [0.31] mm; p 0.001), whereas errors in calibration point determination resulted in an approximately isotropic deviation (1.09 [0.53] and 1.19 [0.60] mm; p = 0.25). The calibration point configurations also differed significantly under standard calibration (p 0.001). Configuration B yielded the smallest relative increases in deviation (+42.4% and +56.3%) while maintaining no measurements exceeding 2 mm. Conclusions. Calibration quality is an independent determinant of accuracy in robot-assisted surgical navigation. Incorrect determination of calibration point positions has a greater effect on accuracy than a calibration needle length error. The direction of the bias is specific to the type of calibration error and may be used to detect errors during calibration quality control. A uniform distribution of 12 calibration points across the four calibrator planes provides an optimal compromise between baseline accuracy and robustness to calibration errors.

This publication has 15 references indexed in Scilit: