Glasgow researchers harness AI to combat colon cancer

Published: 20 June 2019

Engineers have shown that it is technically possible to use an AI system to guide a tiny robotic capsule inside the colon to take microultrasound images.

Engineers have shown that it is technically possible to use an AI system to guide a tiny robotic capsule inside the colon to take microultrasound images. 

Known as a ‘Sonopill’, the device could spell the end for painful endoscopic examinations for patients, where a semi-rigid scope is passed into the bowel. 

The micro ultrasound images captured by the Sonopill are also better able to identify some types of cell change associated with cancer, enabling patients to promptly receive life-saving treatment.

Bowel cancer is the UK's second biggest cancer killer and is on the rise in among those aged under 50. Diseases of the gastrointestinal tract account for approximately 8 million deaths a year across the world.

The robotic arm uses magnetic forces to guide the Sonopill through colon. Picture credit: University of Leeds

The Sonopill is the culmination of a decade of research by an international consortium of engineers and scientists, and the results of their feasibility study are published in the journal Science Robotics.

The consortium has developed a technique called intelligent magnetic manipulation. Based on the principle that magnets can attract and repel one another, a series of magnets on a robotic arm that passes over the patient interacts with a magnet inside the capsule, gently manoeuvring it through the colon.

The magnetic forces used are harmless and can pass through human tissue, doing away with the need for a physical connection between the robotic arm and the capsule.

The AI system ensures the capsule can position itself correctly against the gut wall to get the best quality microultrasound images. The feasibility study also showed that if the capsule gets dislodged, the AI system can navigate it back to the required location.

Sandy Cochran, Professor of Ultrasound Materials and Systems at the University of Glasgow and lead researcher, said: “We’re really excited by the results of this feasibility study. With an increasing demand for endoscopies, it is more important than ever to be able to deliver a precise, targeted, and cost-effective treatment that is comfortable for patients. Today, we are one step closer to delivering that through the unique combination of sensing light and sound.
We hope that the Sonopill will be available to all patients in the near future as part of regular medical check-ups, effectively catching serious diseases at an early stage and monitoring the health of everyone’s digestive system.”

Professor Pietro Valdastri, who holds the Chair in Robotics and Autonomous Systems at the University of Leeds and was senior author on the paper, said: “The technology has the potential to change the way doctors conduct examinations of the gastrointestinal tract.

“Previous studies showed that micro ultrasound was able to capture high-resolution images and visualise small lesions in the superficial layers of the gut, valuable information about the early signs of disease.

“With this study, we show that intelligent magnetic manipulation is an effective technique to guide a micro ultrasound capsule to perform targeted imaging deep inside the human body in a non-invasive way.

“The platform is able to localise the position of the Sonopill at any time and adjust the external driving magnet to perform a diagnostic scan while maintaining a high quality ultrasound signal. This discovery has the potential to enable painless diagnosis via a micro ultrasound pill in the entire gastrointestinal tract.”

The Sonopill is a small capsule – with a diameter of 21mm and length of 39mm which the engineers say can be scaled down. The capsule houses the microultrasound transducer, an LED light, camera and a magnet.

A very small flexible cable is tethered to the capsule and that also passes into the body and sends ultrasound images back to a computer in the examination room.


First published: 20 June 2019

<< June