- A robotic magnetic flexible endoscope system can make colonoscopies more efficient and cost-effective
- Using the semi-autonomous system doesn’t require specialised training
- This is another step towards more machine reliance in the medical field
With any cancer, early detection is vital for survival. This is especially true for colon cancer, which came under the spotlight last month when filmstar Chadwick Boseman succumbed to the illness.
The best way to detect colon cancer and other diseases in this area is through a colonoscopy, but the procedure can be costly, requires special expertise and can be a painful experience. Demand for colonoscopies also exceeds the supply, and the technology used in most procedures is quite outdated.
That might soon all change. Scientists from the University of Leeds have created a semi-autonomous robotic system that can overcome many of these obstacles.
This system is set to replace the almost-obsolete flexible endoscope which requires continual maintenance and sterilisation, and can be painful as it pushes through a patient’s colon. Doctors also have to be specially trained to use it, thus limiting availability and access.
Better technology has been developed, like magnetically actuated endoscopes, that addresses some of these issues, but certain barriers remain when it comes to speed, clinical analysis and special training.
“Developing advanced control strategies capable of assisting and offering an intuitive user experience with reduced procedure times would serve to enable the clinical translation of magnetic colonoscopy, with the overarching goal of widening and improving patient care,” explain the scientists.
Robotic MFE system
This is where the robotic magnetic flexible endoscope (MFE) comes in. The magnetic endoscope is equipped with a camera and LED light, manipulated by a KUKA LBR Med robotic arm with its own magnet. The video feed is projected onto a monitor that also shows the speed of the robot and how far the two magnets are from each other.
However, it’s not as easy as just giving the robot arm a pre-defined trajectory to move the endoscope. The body’s internal landscape and colon differ from person to person, with various obstacles and the soft tissue of the colon easily changing shape.
With this in mind, the robotic MFE system does require some form of intelligent control and autonomy, similar to self-driving cars.
“Our contribution to the field of machine intelligence is the ability to explore how different levels of computer assistance may improve the procedure and reduce user workload in robotic colonoscopy,” say the scientists, who have been working on the system for the past 12 years.
“With robotic assistance in navigation, training resources can thus be directed towards the cognitive aspects of endoscopy such as recognition of pathology, differential diagnosis and creation of treatment plans."
In the case of this system, the video feed is continuously translated by the robot to make real-time calculations on where to move the endoscope. The operator can then solely focus on the clinical side of the procedure, approve manoeuvres before they are made and may override the system if needed.
Testing it out
To test it, the scientists recruited 10 users untrained in colonoscopies to perform the procedure on a latex dummy, using various autonomous levels – ranging from full control by the user to greater control by the robot.
The users were asked to indicate how difficult it was to use the system.
The novice users fared far better with the semi-autonomous system than the direct-control system, with a success rate of 100% compared to 58%. They were also more frustrated when they were in control, and found the task far less demanding when they had more of a monitoring role.
Only 8% of the 50 semi-autonomous procedures required some form of override.
The scientists then went on to see how quickly and how far a novice user could navigate a live pig’s colon, which is more difficult to navigate than a human’s.
“During the semi-autonomous repetitions, the MFE was navigated in autonomous mode for (on average) 87% of the time required to reach the marker for user 1 (total distance of 45 cm) and 78% for user 2 (total distance of 85 cm).”
Together, the results of the study proved that endoscope teleoperation and semi-autonomous navigation were far quicker than conventional colonoscopies when done by novices and newly-trained operators.
This system could be even faster as the technology improves, and could signal the use of semi-autonomous robotics in other medical endoscopies.