Towards surgical subtask automation — Blunt dissection

Abstract

Robot-assisted Minimally Invasive Surgical techniques are becoming standard-of-care in the surgical practice. With the rapid advancement of surgical robotics, many believe that the next step is subtask automation, since it has the potential to safely improve one element of a surgery, which might be most suitable for design and execution supported by artificial intelligence. The outcome of surgery may significantly improve from partial automation already, patient given better accuracy and targeting. This technology can also help surgeons efficiently with time-consuming operations. In this paper, we demonstrate the automation of blunt dissection, which is a common element in almost all abdominal procedures. Blunt dissection is regularly performed during Laparoscopic Cholecystectomy, e.g., when the surgeon exposes the Calot triangle. Automating this episode would streamline the procedure, and multiple similar automated sub tasks can be integrated into the surgical workflow. The presented method only relies on the images of a stereo camera system, and therefore does not put any additional overhead on the Operating Theater. The method was successfully tested in vitro and ex vivo in simplified simulated environment.