Spatial Mapping Method of Craniosynostosis Surgical Robot Based on Point Cloud Registration

Abstract

The preoperative planned surgical path is difficult to accurately reproduce in the treatment of craniosynostosis. This paper proposes an automatic space mapping method between the robot and the preoperative Computer Tomography (CT). The method first uses the region growing algorithm to automatically segment the surgical window scanned by the structured light scanner and extract the relevant area for registration. Secondly, the coarse matching algorithm based on the Signature of Histogram of Orientation feature and the Iterative Closest Point fine registration algorithm based on the Fast Point Feature Histogram are used to complete the registration of preoperative CT and scanner. Thirdly, a registration method based on Singular Value Decomposition (SVD) is performed to match the robot and the scanner coordinate system. A surgical robot with dual arms is modeled by Denavit-Hartenberg parameters. The kinematic relationship of the dual-arm robot in spatial positioning is analyzed. Furthermore, combined with the spatial mapping method, an analytical method is used to solve the inverse kinematic of the surgical robot. To qualify the effectiveness and accuracy of the spacing mapping method, a model experiment is carried out by the surgical robot. The results show that the mean target error between the planning path on the preoperative CT and the path on the 3D printed skull model drawn by the robot is 0.2625 mm and the standard deviation is 0.2435 mm, which meets the requirement of the clinical usage.