Cancer detection in the cloud

27 July 2021 | Physical Sciences/Engineering

Odin Vision logo

Colorectal cancer is one of the commonest cancers in the world. When caught in the early stages, it’s straightforward to treat, with a survival rate of over 90%. But this drops to less than 10% for patients whose cancers are found at later stages. Early detection relies on endoscopy procedures, where a small camera is placed inside the body to look for abnormal growths or ‘polyps’ in the bowel.

The challenge for doctors is that images from endoscopy procedures are extremely difficult to interpret. This affects all aspects of the endoscopic procedure from identifying anatomy and navigating to detecting, diagnosing, and treating diseases. With around a quarter of early-stage cancers currently being missed there is a significant opportunity to improve patient outcomes, reduce costs and enhance the patient experience.

Now, a cloud Artificial Intelligence (AI) platform based on research from UCL is helping clinicians to detect more cancers at an earlier stage, with life-changing impacts for their patients.

Improving cancer detection with machine learning

Peter Mountney is CEO of Odin Vision, the UCLB spinout company which is commercialising this technology.

“Our system uses AI to analyse images from endoscopies during the procedure and draws boxes around areas of abnormal tissue.”

The ‘CADDIE’ (computer aided detection and diagnosis for endoscopy) system uses a database of millions of images, which have been labelled by doctors as either healthy or diseased tissue. Through machine learning, the system begins to identify the visual characteristics associated with diseased tissue. It was developed by Professor Danail Stoyanov, a leading expert in robot vision at UCL, in collaboration with Professor Laurence Lovat, a consultant gastroenterologist at UCH.

Providing instant diagnoses

Crucially, the AI doesn’t replace clinical decision-making, but allows doctors to focus on potentially diseased areas, and, with support from the technology, assess whether tissue is low or high risk of being cancerous. Because data is analysed at the time of the procedure, the doctor can give the patient an instant diagnosis that can be immediately shared with the patient, reducing the stress and anxiety associated with waiting weeks for results.

Through the NHS AI in Health and Care Award, the technology was identified as having significant potential to increase the impact of AI in the NHS. A £1m grant by the NHS AI Lab at NHSX, the Accelerated Access Collaborative (AAC), and National Institute for Health Research (NIHR) will fund Odin to generate further clinical evidence, enabling wider adoption within NHS hospitals. Early results show the potential for significant increases in detection rates of polyps when clinicians are supported by the AI system.

A vision for global growth

With over 100 million endoscopy procedures performed worldwide every year, the potential global impact of Odin Vision’s technology platform is huge.

“The challenge is to build scalable technology that can address this large market, this is why a cloud-based endoscopy platform is so important” explains Peter.

By creating a cloud-based endoscopy platform, the company has addressed three of the significant challenges in scaling healthcare solutions – hardware, speed to market and AI updates.

Using the cloud avoids having to deploy expensive, limited, and fast-ageing hardware into hospitals. It means hospitals have access to the latest supercomputing power and their adoption of technology is future-proofed. The cloud also provides the infrastructure to bring new products to hospitals at the click of a button, short-cutting the traditionally slow journey of bringing a new product to market. As AI moves fast, and algorithms are constantly improving, a final benefit of cloud technology is in enabling doctors to have the latest and best-performing systems without having to book an engineer to visit and perform a system update.

Odin Vision’s technology can be applied in diseases throughout the digestive tract and surgery, opening the prospect of multiple use-cases. A second product which can aid detection of oesophageal cancer called CADU has already been released via the platform.

Their flexible approach means they are well placed to adapt to the requirements of other healthcare systems around the world, something they are already exploring as part of the Innovate UK Global Business Innovation Programme.

 

Odin Vision

CADU. Credit: Odin Vision.

Pioneering a new spinout model

As a UCLB spinout, the company has been able to benefit from support and funding to create and grow the company.

“Odin Vision were one of the first companies to use the pioneering Portico Ventures spinout model,” explains Weng Sie Wong, Senior Business Manager at UCLB. “The model assigns UCLB a fixed equity stake in return for a licence to the spinout for UCL’s software and know-how IP assets. Other than an initial anti-dilution provision over the first £1m of external investment, the equity dilutes over time in line with other shareholders as the company raises more funds.”

The UCL Technology Fund (UCLTF) was the first investor, involved even before the company was formed. David Grimm, Investment Director at UCLTF, notes their impressive progress: “Odin have gone from a standing start to having a product on the market earning revenues in the endoscopy field in a very short space of time. And they’ve shown great capital efficiency in doing that.”

Odin Vision has received grant funding from Innovate UK, the NHS, the NIHR and the UK Space Agency and European Space Agency. These investments have allowed Odin’s team to refine its product and scale up operations.

Closing the healthcare inequality gap

From the outset, the Odin Vision team were determined to avoid the risks of their technology contributing to greater inequalities in healthcare.

“We wanted AI to be used to close the healthcare inequality gap,” says Peter. “That means it should be available to every hospital and every patient.”

Using a cloud-based endoscopy platform makes this possible, Peter explains: “We can put our systems into hospitals at a very low initial cost and they’re used as and when they’re needed. This makes it cost-effective for the hospitals and easier for them to adopt the technology.”

It adds up to an exciting future for Odin Vision, and a potential life-saving impact for thousands of patients around the world.

 

Links

https://odin-vision.com