Patient Registration of Intelligent Surgical Navigation System--Based on Improved ICP Algorithm

Abstract

Aiming at the problem of inaccuracy of traditional iterative closet point (ICP) algorithm, an accurate registration strategy base on multi-feature point registration and point-to-surface ICP algorithm is proposed. Firstly, a simple region growth segmentation algorithm was used to preprocess CT images, then 3D virtual brain model was established by 3D reconstruction technology. The patient space and image space were roughly registered by multi-feature point method to obtain good initial values. Finally, several points are established in the image control to form a plane, and the improved ICP algorithm is used to accurately register the feature points obtained from rough registration with the plane. The experiment result shows that this method combined with separation theory, matrix theory, matrix theory, multi-feature point method and point-to-surface ICP algorithm can make the registration accuracy meet the precision requirements of surgical navigation system.