Increased Safety in Robotic Skull Base Surgery with Redundant Navigation and Automated Registration

Introduction: Safety mechanisms have become a major issue in robotic surgery today. We present an advanced version of our robotic setup interconnected with a redundant navigation system for increasing intraoperative safety while performing telemanipulatory as well as fully automated maneuvers.

Methods: In contrast to the previous “all-in-one” version, we built a modular three-component setup. The robot is placed on a pedestal and can be immobilized with the operating table by 2 mechanical arms. An optical camera for the computer navigational control is placed on a mobile tripod.

The basic feature of the computer navigation system is the CAPPA station for ENT surgery. The system references by automatically detecting a referencing frame (DRF) mounted on a noninvasive upper jaw mouthpiece. Software components of both systems—navigation and robotics—were combined into one user interface. Accuracy and clinical applicability studies were carried out. The system's operational gross accuracy was tested by maneuvering the robot's tool tip, which operates as a stereotactic pointer, onto CT markers (1 mm in diameter) placed on the phantom head's outer skull and bony paranasal sinus walls. Resulting positions were categorized as absolute accuracies and graded as either adequate or inadequate. Obtained absolute accuracies were assessed by measuring distances from the tool tip to the pinpointed marker, both on the screen and on the head. Absolute accuracies were deemed adequate when both measurements were equivalent or both measurements were inadequate, with either distance showing an offset of larger than 1 mm. Maneuvers were performed either fully automated or telemanipulatory on cadaveric heads.

Results: A modular setup was designed and was deemed feasible in its size and weight dimensions as well as its maneuverability for application in a routine operating room environment. The camera in combination with the registration frame allows an automated patient referencing. The navigational feedback is integrated in real-time in the robot's user interface. In case of blocked visibility to the DRF, the robot powers down and activates the force torque sensor, thus softening all articulating joints.

In the telemanipulation mode, using only navigational feedback without visible control both on the head's surface and in the paranasal sinuses, we found solely inadequate accuracies in pinpointing a specific marker. Accuracies with both navigational control and endoscopic view were found adequate both on the skull's surface and within the paranasal sinuses. Fully automated procedures were all performed with adequate accuracy on the head's surface and paranasal sinuses.

Discussion: The challenge was to integrate a redundant computer navigational control system with the robot's operating system under 1 control module. This enables the navigational system to actively interfere with the robotic execution if certain safety standards are not met during surgery. By using redundant navigation feedback, we were able to add another safety feature, the “loss-of-control” function, which shuts down any robotic action. However, no increase of the absolute accuracy was observed by adding this feature. We conclude that redundant navigational control does not make the robot more accurate, but it adds a potent safety feature to the system.