Remember NeuroMate? It was the first neurosurgical robot to receive CE mark and FDA approval. (Now it's part of Renishaw's product portfolio.) Here are some details about the first development of the system:
"Fig. 1 shows the design of the system tested under clinical conditions. The head of the patient is put into a stereotactic frame, within the range of a pair of Xray systems (a film holder is put in 2 positions that are nearly orthogonal).The X-ray images are processed on a digitizing board connected to an IBM PC-AT which controls the motions of a 6 joint manipulator. At the beginning of each operation, the robot carries a calibration cage made of two pairs of plexiglass planes containing 9 metallic balls, and places it around the head in order to obtain reference X-ray images (Fig. 2). Equations linking the 3D coordinates of each landmark with the corresponding image coordinates are used in a least-squares fitting method. This provides matrices which are used later on to position the robot with respect to a pair of X-ray images. Then, the surgeon selects a target and a trajectory with the help of angiograms, ventriculograms and NMR or CT sections manually matched together. To help his decision, the surgeon has the opportunity to use anatomy based routine procedures stored in the PC (functional neurosurgery). He can also ask to align the cylindrical guide held by the robot with X-rays so that the projection of the guide on an X-ray image is a small circle on which the presence of a vessel is easily checked. Finally, the system looks for a trajectory that avoids the obstacles (mainly the stereotactic For many reasons, the introduction of robots in the medical world raises some specific problems (2). In our case, the accuracy and safety problems have been studied. We use a kinematics robot modelling based on a division of the configuration space and on the assumption that the robot is "simple" in each of the 8 sub-spaces. Specific calibration procedures based on the non-linear least-squares fitting of our model with experimental 3D measurements have been performed. As we only use relative motions with respect to the plexiglass calibration cage, we have easily obtained an overall accuracy of +/- 1 mm in each direction, and we think that a more complicated model will still increase this accuracy (3). Moreover, the position of the guide is always checked on X-ray images before being accepted, and interactive small motions are easily performed when necessary. The reduction ratios of each articulation have been modified so that the motions are slow and that hardware and mechanical breakdowns become unlikely. But most of all, when the robot is in position, the motors' supply is cut off so that no incident may occur when a tool is implanted in the brain of the patient. So far, the system has been successfully used in 16 clinical cases: 3 were stimulating electrode implants (fig 3), 12 were biopsies of tumors (fig 4), and one was a radioactive iodine seed implant."
Source: IEEE A new system for computer assisted neurosurgery