School of Computer Science and Engineering
The Hebrew University of Jerusalem, Israel

Master of Science Thesis, 2001.

Multi-modal registration is an essential step in computer-based medical applications. It consists of finding a transformation that co-aligns common features in two modalities, so that their spatial locations coincide. Registration combines data from different sources and enables to find relative positions of surgical tools and anatomical structures.

This thesis studies methods for rigid registration, and describes phantoms and in-vitro experiments testing the accuracy and reliability of algorithms and protocols. The work was conducted in the framework of the FRACAS project, whose aim is to develop systems and algorithms for computer-aided orthopaedic surgery.

We investigate two types of registration: contact-based and image-based. Our research comprises algorithms, phantoms, and protocols for tool tip calibration, camera calibration, contact-based registration, and fluoroscopic X-ray image-based registration. In our experiments, we used a commercial optical tracking equipment.

In our in-vitro accuracy and reliability experiments, we found sub-millimetric to millimetric accuracy for tool tip calibration and contact-based methods (landmarks and cloud of points). For image-based registration, the average error is 2.75mm. Its main cause is most likely the sensitivity of the camera calibration process. We conclude the thesis with suggestions for improvements to our methods and materials to reach millimetric accuracy.