Non-invasive patient registration based on 3D feature points of vein

Computer assisted surgical system (CASS) development is one of the most vigorously pursued research fields in biomedical engineering because of its numerous advantages. Nevertheless, patient registration is still a heavy burden for both patients and surgeons. Meanwhile, it is well known that veins can be seen under the skin in near infrared images because of the properties of hemoglobin. In this paper, we propose a novel registration method based on this principle. In our registration method, we extract feature points from veins using an infrared stereo vision system, and then match them with the feature points extracted from preoperative medical images. Experiments were performed to verify the feasibility of this vein-based registration. As a result, we confirmed the availability of our proposed method, and showed that it can be utilized for numerous applications.