Estimation-Based Control of a Magnetic Endoscope without Device Localization

Magnetically controlled catheters and endoscopes can improve minimally invasive procedures as a result of their increased maneuverability when combined with modern magnetic steering systems. However, such systems have two distinct shortcomings: they require continuous information about the location of the instrument inside the human body and they rely on models that accurately capture the device behavior, which are difficult to obtain in realistic settings. To address both of these issues, we propose a control algorithm that continuously estimates a magnetic endoscope’s response to changes in the actuating magnetic field. Experiments in a structured visual environment show that the control method is able to follow image-based trajectories under different initial conditions with an average control error that measures 1.8 % of the trajectory length. The usefulness for medical procedures is demonstrated with a bronchoscopic inspection task. In a proof-of-concept study, a custom 2[Formula: see text]mm diameter miniature camera endoscope is navigated through an anatomically correct lung phantom in a clinician-controlled manner. This represents the first demonstration of the controlled manipulation of a magnetic device without localization, which is critical for a wide range of medical procedures.