Camera-based instrument navigation in ENT-surgery

Algorithms for a camera-based surgical navigation system, with a localization based on the surface structure of the patients face, are developed and evaluated. The measurement technology is based on a single HD-camera that is integrated on a surgical instrument. Sophisticated and expensive state of the art systems are thus replaced by a software solution, which reduces the hardware costs and the space requirements in the operation room. A functional prototype is implemented and tested in a laboratory and an operating room. The challenge is to determine the camera's pose from a camera live stream with a feature based approach, without additional markers on the patient. The core software modules are mapping, registration and localization. The root-mean-square error of the system varies between 2.4 mm and 6.4 mm dependent on the lighting conditions. By removing a systematic error from the registration process, the accuracy is improved to 1.3 mm to 2.0 mm.