In the past decade, the face of cardiac surgery has been changed by a number of technologic advances. One such advance is the development of minimally invasive cardiac surgery where the surgical intervention is done through small incisions ( key hole surgery ) rather than through conventional open chest surgery. With this technique, small endoscopic cameras and specialized instruments are inserted through small artificial incisions. For the patient this reduces morbidity and leads to faster, less painful recovery from surgery. However, for the surgeon this poses a number of problems such as the limited tactile feedback and restricted visual feedback.The purpose of this project is to develop a image-guided minimally invasive cardiac surgery system using a surgical robot. The key research issue which we will pursue in this project is the development of novel model-based 2D-3D registration techniques for matching 3D deformable objects such as the heart as well as vessels to 2D endoscopic video images. This registration will enable us to establish the relationship between the 2D endoscopic video images seen by the surgeon and the patient's anatomy. We propose to superimpose models derived from preoperative image data onto the live 2D endoscopic video images. This will allow the surgeon to utilise preoperative tomographic data to accurate localise and track targets during the operation.