The problem to solve is to register the surgical instrumentation coordinate system with the image coordinate system. At the upper-left image the surgeon uses an electromagnetic sensor to determine the coordinates of the patient markers. These markers are shown in red in the upper-right image. A similar process is performed on the three-dimensional model of the patient to get the blue markers at the upper-right image. Then we perform two steps to minimize the difference between the two clouds of points (already done at the right). In the first step the VNL library is used to adjust the two sets of points by matching their mass centers and eigenvectors. This phase also serves to put in correspondence the two sets of points. The second step is a fine tuning, which is performed using an Iterative Closest Point algorithm (ICP), which has been widely used in Computer Graphics and other applications. Registration error was 0.95 mm rms. With a minimum of 0.44 mm and a maximum of 1.3 mm. An axial plane is added in the figure below.