Towards visual matching as a way of transferring pre-operative surgery planning

Abstract

This paper presents a case study towards intra-operative visualization of pre-operative medical images and surgery planning information. We describe a two stage procedure to provide a surgeon with valuable information during surgery, by augmenting live video with a 3D visualization of surgery planning. A first stage consists of automatic determination of the intrinsic parameters of the video camera by means of one image of a calibration object. It comprises an automatic ellipse extraction and a solution to the 2D-3D correspondence problem. With the intrinsic camera parameters known, we perform a second stage to compute the extrinsic camera parameters with respect to the patient under surgery, allowing us to position and visualize the medical images in the right place. Techniques are investigated to visualize and manipulate the pre-operative medical images in order to cope with changes in the surgery scene.