Skip to Main Content
During a minimal invasive neurosurgery it is necessary to know precisely the 3D position of the surgical instruments and their relationship and interaction with the anatomy of the human brain. Today, on developing countries many surgeons still operate without 3D visual aids, managing instruments in the blind. With this idea we developed a new technique of Computer Assisted Surgery (CAS). We developed a new neuronavigator called Neuropanacea. It consists of our image processing software, the electromagnetic tracker for the localization of each one of the surgical instruments in the surgical room, and a personal computer (PC). The procedure for registration of these points captured using an electromagnetic tracker with the 3D space coordinates of the volume might be too complex. The main problem consists on finding the geometrical homogeneous transform to apply it to the acquired points with the electromagnetic tracker, and the registration of the fiducial markers visualized on the 3D volume. If we have the sorted correspondence of each one of the points in both systems, it is always possible to find a homogeneous transform into another system. However, in most of these cases we do not have this orderly set. One method to solve this problem is the iterative closest point (ICP). This work refers to the original algorithm of ICP. From this implementation we proposed two improvements for its correct convergence. The mean square error using the original algorithm was reduced from 2.1 ± 0.5 mm, to less than 0.5 mm.