On the Magnetic Control of Endoscopic Robots

Abstract

The use of magnetic fields for medical robot actuation has been demonstrated for applications in opthamology, otolaryngology, cardiology, and gastroenterology; the latter of which is the clinical focus of this dissertation. Diseases in the gastrointestinal (GI) tract account for eight million deaths worldwide annually. Cancer of the large bowel, which makes up a tenth of the world's cancers, is the second most common cancer in women and third in men. Other GI ailments may result in a significant decrease in quality of life. Flexible endoscopy is the gold-standard screening and therapeutic technique for the lower bowel. It has the drawback of inducing tissue stress that may lead to discomfort and perforation. For this reason, sedation is typically administered. Several robotic solutions have been investigated to supplant flexible endoscopy; however, none have been clinically adopted. A robotic approach to endoscopy that relies on magnetic actuation is presented in this work.

The proposed system consists of a magnetic flexible endoscope (MFE) that is actuated using a single actuating permanent magnet that is mobilized via a serial manipulator. Magnetic closed-loop control (CLC) relies on using sensory feedback from the actuated device, e.g. localization, to command motions of the endoscope. In this dissertation, an approach to magnetic CLC is described that relies on linearizing the relationship between magnetic wrench and magnet motion is described. Singularities that arise during this actuation are analyzed.

The use of task-autonomy in magnetic endoscopy is investigated. First, an algorithm for retroflexing the MFE is presented. Second, a magnetically guided microultrasound (µUS) probe is servoed using both localization and µUS image feedback; this enables robust imaging of the mucosa without the need for non-intuitive teleoperation. The efficacy of magnetic CLC is further investigated by considering the effects of localization noise on force uncertainty. A sensitivity-ellipsoid-based algorithm is proposed for minimizing this uncertainty. Finally, a sensor-free method for estimating the local shape of the MFE tip is proposed which may be used to enhance endoscope control.