Augmented Reality Calibration with Stereo Image Registration for Surgical Navigation

Augmented reality (AR), an important approach for humans to access more information, has been widely used in various fields, such as medical, military, entertainment, and educational fields. In the medical field, AR can superimpose virtual preoperative patient information on the real intraoperative scene to guide the surgery. Currently, many AR methods are used in the medical field, but most of them require complicated tracking markers or demanding apparatus. This work proposes an AR calibration method with a binocular stereo camera. The accurate feature matching pairs are used to reconstruct the point cloud. Thereafter, the 3D virtual mesh model is superimposed on the real scene by image-patient registration. A head phantom, a liver phantom, and three volunteers participated in our experiments. The accuracy of our method is significantly higher than the other two AR methods.