Research On Three-dimensional Registration Techniques Of Surgical Navigation

Surgical navigation system is a new research field generated by a combination ofdisciplines, such as image processing, three-dimensional positioning, precisionmachinery, computer science and surgery. One of key technology in the surgicalnavigation is the spatial registration technology, which seeks to a kind of transform inthe space coordinates between patient space and three-dimensional image space, so thatthey could unified to one coordinate system.Firstly, the surgical navigation system was briefly introduced. Through theintroduction and analysis of the principles on surgical navigation systems, spatialregistration process, and study on surgical navigation system,summarized the basicrequirementsofthesurgical navigation systemregistrationof spatialtechnology, and somedefects of spatial registration technology at this stage, which led to the needs and goalsof this study: to find a suitable spatial registration algorithm for non-matching set ofpoints.Subsequently, this paper separately described the spatial registration methods basedon the paired base point and unpaired set of points. The registration based on the pairedbase point is according to much corresponding point coordinates obtained by thethree-point method, the quaternion, and SVD matrix decomposition methods tocompute the best transform between them. For the spatial registration method based onthe non-paired set of points, the PCA-based main direction method and the iterativeclosest point (ICP) algorithm were introduced. After analysis of the effectiveness,stability and computational complexity in a variety of different algorithms on the pointset space registration, the paper proposed the improved ICP registration algorithm forthe unpaired point set. At first, by the improved fit alignment registration method of thePCA-based main direction, to achieve a pre-registration; and then, by the ICPregistration algorithm based on the Delaunay triangulation, to achieve the secondaccurate registration. The algorithm achieved the spatial registration in two space basedon unpaired set of points.Finally, for the verification of adaptation, timeliness, and accuracy of matchingalgorithm based on improved ICP, this paper designed a series of simulation experiments for analysis. A series of experimental results showed that the proposedregistration algorithm can achieve higher accuracy at the case of errors interference. Thepaper also designed a series of experiments to compare the efficiency on the traditionalICP registration algorithm and the ICP registration algorithm based on Delaunaytriangulation. The results showed that the time efficiency of the ICP algorithm based onDelaunay triangulation is better. At last, the use of an actual scan data of the algorithmfor the validity had achieved good registration results. Though the registration of theabove simulation and real data, the study proved that the improved ICP registrationalgorithm could accurately achieved the spatial registration in two space based onunpaired set of points...