Abstract

Info

Title:: A fast iterative registration method based on iterative closest point algorithm with adaptive thresholds for vascular intervention

Author(s):: SUN Haoyue; HU Zhi; QI Jinlong; CUI Guohua; School of Electronic and Electrical Engineering, Shanghai University of Engineering Science, Shanghai 201620, China

Keywords:: Keywords: vascular interventional surgery iterative closest point dynamic registration adaptive threshold fast point select

Abstract:: A fast iterative registration method based on iterative closest point (ICP) algorithm with adaptive thresholds is proposed for the vascular intervention system based on augmented reality to overcome the challenge that the accuracy of master-slave registration is affected by the dynamic change of the position when the flexible guidewire moves in the vessel and to address the problem of iteration inefficiency. The physical model of blood vessel and guidewire is established to obtain the position data of the motion points. After fully considering the change of system state during the iteration, the ICP algorithm with adaptive thresholds is designed to achieve precise registration by calculating the overlapping rate of the master and slave points, thereby improving the registration accuracy. Meanwhile, the fast point selection iteration improves the registration efficiency. The simulation is carried out for performance comparison with the classical ICP algorithm, and the results show that the dynamic high-precision registration algorithm proposed in the study can improve the registration speed and accuracy by 57% and 70%. The proposed method is more suitable for augmented reality-based vascular interventions for it can enhance the dynamic characteristics of the registration method and suppress the effects of the morphological changes of the guidewire during the surgery on the registration accuracy and efficiency.