Robust and Automatic Correspondence Matching for Application in Minimally Invasive Surgery

Minimally invasive surgery offers major benefits for the patient compared to open surgery, like less pain, less trauma and shorter hospitalization durations. However, laparoscopic surgeries are a difficult procedure even for experienced surgeons and can additionally be life-threatening for the patient, especially in case of organs with a high degree of vascularization such as the liver or the kidney. Thus, preoperative 3D CT or MRI models are used nowadays as a planning support of the surgery aiming at decreasing the risk for the patient.

The overall goal of this project is to enable a navigation tool for the surgeons which shows the preoperative 3D model including important sub-surface structures like vessels and tumors. For this purpose, a registration of the preoperative 3D model and the intraoperative laparoscopic video has to be performed. During this project, the different single components of the registration pipeline will be implemented and specialized for the laparoscopic setting. Moreover, an evaluation of the registration pipeline considering typical disturbances will be performed.