Spatial registration and spinal surgery navigation based on constraint of anterior-posterior and lateral views

To improve the precision, safety and usability of the robot assisted spinal operation, registration between preoperative CT images and intraoperative X-ray images is utilized. Conventional single planar registration is processed with large navigation error due to lack of depth information along the optical axis, therefore anterior-posterior (AP) and lateral views are calculated simultaneously to improve the registration and navigation accuracy. Based on the analysis on the transformation model of anterior-posterior (AP) and lateral projection spaces, initial registration parameters are automatically estimated and the mapping from the 3D surgical space to the CT space is established by using the biplanar geometric constraint. The real-time postures of surgical instrument in CT space are calculated by using stereo-vision-based positioning technique based on virtual X-ray navigation, and 3D navigation is realized. A vertebral phantom experiment validates the accuracy of the registration and navigation method, and animal lamina milling experiments prove that the operation safety could be improved and the real-time monitoring is accomplished.