Autonomous Retroflexion of a Magnetic Flexible Endoscope

Abstract

Retroflexion during colonoscopy is typically only practiced in the wider proximal and distal ends of the large intestine owing to the stiff nature of the colonoscope. This inability to examine the proximal side of the majority of colon folds contributes to today's suboptimal colorectal cancer detection rates. We have developed an algorithm for autonomous retroflexion of a flexible endoscope that is actuated magnetically from the tip. The magnetic wrench applied on the tip of the endoscope is optimized in real-time with data from pose detection to compute motions of the actuating magnet. This is the first example of a completely autonomous maneuver by a magnetic endoscope for exploration of the gastrointestinal tract. The proposed approach was validated in plastic tubes of various diameters with a success rate of 98.8% for separation distances up to 50 mm. Additionally, a set of trials was conducted in an excised porcine colon observing a success rate of 100% with a mean time of 19.7 s. In terms of clinical safety, the maximum stress that is applied on the colon wall with our methodology is an order of magnitude below what would damage tissue.

Development of a therapeutic capsule endoscope for treatment in the gastrointestinal Tract: Bench testing to translational trial

“…In addition, future capsules will combine positioning and localization functionality to allow precise spatial targeting of reagents. This could be achieved by controlling the capsule with magnets, similar to Slawinski et al [15] who use a magnetic robotic arm to propel a capsule endoscope. Future work will aim to combine the UmTDD system described in this paper with fluorescence imaging, wireless communications and positioning/localization facilities to allow fully autonomous therapy in the GI tract.…”

Sensitivity Ellipsoids for Force Control of Magnetic Robots With Localization Uncertainty

“…Second, we assume that slow motions of the IM which simplifies the mathematical formulations. This assumption has been used in our previous works with success [26]. Third, we assume ideal knowledge, i.e.…”

“…To visualize the relationship between these values in the presence of localization noise, we conducted a set of 5000 numerical simulations where a uniform localization noise of 15 mm and 15 • was applied with a single stationary EM and a stationary IM. This linear noise translates to approximately 135% of the size (11.11 mm diameter and length) of the IM used in our previous experimental work [9], [26]. In each of the 5000 simulations, the EM's pose was randomly chosen within a distance between magnets of 15 cm with a STD of 1 cm, or in a bound of approximately 7% of the separating distance.…”

“…The motion of an IM is dictated by device mechanics and dynamics, as well as environmental interactions. Rather than precisely modeling environmental effects which may be cumbersome, prior works have successfully implemented closed-loop control without environmental modelling [7], [9], [26] which we have done in this work as well. To choose a desired EM motion, a desired infinitesimal change in force must be determined, which we refer to as δ f des .…”

New Robotic Technologies in Cancer Colon Screening

“…This is related to the orientation of the magnetic field that controls the applied force. Most of the solutions with external magnets produce a force with a vector component that attracts the device to the external magnet [43], thus against the colonic wall. This force needs to be estimated because if too high, it can stretch the colonic wall and cause pain.…”