ABSTRACT
Historically, stereotactic neurosurgery has been probably the first surgical discipline to benefit from computer-integrated surgery and therapy (CIST) technology. The fact that some accurate mechanical systems had been used for a long time (stereotactic frames) and the conjunction of the advent of CT imaging and sufficiently powerful low-cost computers may explain that phenomenon. The criteria corresponding to a given CIST system are many and dependent on the type of surgery. However, it is possible to define a list of global objectives at which any CIST system designer should aim. All the multimodality data that have been acquired must be modeled to provide appropriate representations for the subsequent steps of the perception/decision/action loop. There are two main technical issues at this level: calibration and segmentation. In many applications, using raw sensor data is usually insufficient for subsequent processing. Extraction of high-level representations such as the surface of a particular anatomical structure is very useful for registration and surgical planning steps.
