Abstract:

This paper deals with the registration between preoperative CT images and intraoperative physical objects in the three-dimension computer-assisted spinal surgical navigation technology,and proposes a two-step registration strategy,which realizes the registration by an electromagnetic positioning system on the three-dimension visualization platform based on Visualization Toolkit（VTK）.In this strategy,several feature points are extracted on a spine bone surface respectively in the virtual space and in the physical space for coarse registration,an electromagnetic positioning probe is used to extract points on the spine surface in physical space,and the corresponding points are extracted in the virtual space.Moreover,the ICP（Iterative Closest Point） algorithm is used to complete a fine registration.The ICP algorithm is then validated by simulation and the accuracy of it is also analyzed.Finally,a registration experiment is performed with a spine model to evaluate the accuracy of the proposed method,followed with an error analysis.The results show that the proposed registration method is simple and reliable,and that the final registration accuracy for spine model meets the finical surgery requirements well.

Key words: spine, surgical navigation, image registration, Visualization Toolkit, iterative closest point algorithm