School of Computer Science and Engineering
The Hebrew University of Jerusalem, Israel
Parallel robot (Technion) |
Fluoroscopic image-based registration |
We have started developing new devices and procedures for precise minimally-invasive image-guided orthopaedic surgery. Our technical focus is in two areas: (1) the development of novel miniature robotic devices, and (2) the development of new fluoroscopy-based modeling and registration techniques. The key concept behind our technical focus is to use corrected and calibrated fluoroscopic images containing specially shaped robot tip and the anatomy of interest to perform accurate intraoperative modeling and registration without the need of implanted fiducials or direct contact.
This image-based technique is essential for percutaneous procedures and has the potential to reduce the morbidity of fiducial-based and open procedures, and improve the outcomes of the traditional procedure. Our clinical focus will first be on spine procedures, including percutaneous and transpedicular screw placement and percutaneous discectomy. Later, will also explore improvements to illiosacral screw placement, dynamic hip screw placement, knee arthroscopy, total knee replacement, and elbow and shoulder procedures.
The novel medical robots are based on two ideas. One is the development of new parallel robot structures, which by their very nature will better fit the medical requirements of surgery. Parallel structures are much more compact than the commonly used serial ones. They are more accurate and their reduced work volume is a big safety advantage in medical applications where the required motions are small. The second idea is the development of a miniature robot that attaches directly to the bone. The miniature robot, which is more than one order of magnitude smaller than a conventional PUMA, will simplify and even eliminate the need for a separate registration procedure and for tracking bone motion.
The motivation to use C-arm X-ray fluoroscopy for intraoperative bone modeling, spatial location, and registration is that it is immediate, non-invasive, and ubiquitously available. Although it produces radiation, controlled use of several dozen static shots per case (as opposed to continuous fluoroscopy or several hundred static shots as it is common now) will greatly reduce the surgeon's cumulative exposure to radiation while retaining the benefits of this imaging modality. To be useful for precise procedures, the camera must be calibrated and the images must be corrected from geometric distortion. Then, the images must be correlated and registered to each other. Preliminary results of our on-going work in this area demonstrate that sub-millimetric accuracy is obtainable in fluoroscopic image processing. We are developing develop robust and practical algorithms for fluoroscopy image processing, modeling and anatomy-based registration with CT surface models. We will emphasize integration, simplicity, full automation, and practicality.
![[Back]](arrow2.gif)
Back to the CAS Lab Page
Last modified: January 31st, 2002.