REPORTING FROM THE 2019 AGA TECH SUMMIT
SAN FRANCISCO – Artificial intelligence (AI) can improve video recording and physician support during colonoscopy procedures, and the data being collected could eventually pave the way to AI systems powerful enough to detect polyps on their own.
AI has been touted as a means to reduce costs and improve patient outcomes, but there’s another benefit that is sometimes overlooked – physician satisfaction. Examining colonoscopy after colonoscopy can get a little overwhelming. “There’s only so much the human eye can see, and when you’re doing multiple colonoscopies a day, there’s a possibility that you’re just seeing too much. [AI] could help lessen the burden in making those diagnoses,” said Kurt Heine in an interview.
Mr. Heine is vice president of the endoscopy division at Olympus America. He joined other speakers at a panel on artificial intelligence at the 2019 AGA Tech Summit sponsored by the AGA Center for GI Innovation and Technology.
Physicians are under an increasing burden with the aging population and recommendations from the American Cancer Society that colonoscopies should begin at age 45. “We need a more efficient way to perform colonoscopies. Ideally AI would aid in the adenoma detection rate and perhaps someday in diagnosis. We’re trying to build a support tool to assist in that procedure,” said Heine.
That point was echoed by Matt Schwartz, CEO of Virgo, which specializes in video recording of colonoscopies. In an effort to ease physician burden, it is automating some simple tasks, like starting and stopping video recording during a procedure. That’s a time-consuming process, and “it’s the sort of task that AI is really good at, so it’s a natural fit to get our first foray of AI on the market,” said Mr. Schwartz in an interview. Virgo currently provides a small, Apple TV–sized box that connects directly to an existing endoscopy system that independently records procedures.
The system uploads captured video into the cloud, and automatically creates highlights for easy viewing. That feature is available now. In the future, Virgo hopes to offer advanced analysis of withdrawal time, cecal intubation rate, and other features. “We want to provide video analytics that can help doctors and physician groups make informed decisions on how to improve quality,” said Mr. Schwartz.
Developing AI poses distinct challenges, and can be costly and time-consuming, said Jason Tucker-Schwartz, PhD, director of marketing for NinePoint Medical. During the session, he outlined the development of the AI component of the company’s NvisionVLE Imaging System, which is the only Food and Drug Administration–approved AI product for upper GI imaging. It reveals layers inside the esophageal tissue wall with high resolution, where 90% of cancers begin.
“It’s a new type of data that gastroenterologists are not used to seeing,” said Dr. Tucker-Schwartz. But that has presented a challenge to physicians attempting to identify and process all that novel information.
That called for AI, but because NinePoint is a small company with limited resources, it needed to control development costs. To streamline matters, they developed a system that focuses on three features that are most useful in making esophageal diagnoses. “The AI algorithms find those features as a function of depth so we can flatten them and use them to create a roadmap that physicians can use to guide their [interpretation],” said Dr. Tucker-Schwartz.
The resulting system has garnered lots of positive feedback, according to Dr. Tucker-Schwartz. The experience highlights the need to incorporate physician input into product development. “You need to involve them in all the steps along the path to end with a product that meets not only their goals but your business goals as well,” he said.
The long-term goal for AI in colonoscopy is automated polyp detection, a so-called optical biopsy, but that vision lies well in the future, said Mr. Schwartz. The primary issue is that only still images are available as training sets, and these don’t capture the diversity of patients, endoscopy systems, and operators that will be required to create a robust, generalizable polyp detection system. Existing efforts have shown promise on training sets, but struggle in real-world tests. “AI is good at tricking you into thinking it’s a working system when it’s only looking at retrospective data,” said Mr. Schwartz.
Olympus signed an agreement last year with ai4gi, a commercial initiative applying deep learning to gastrointestinal cancer, to combine its AI systems with Olympus’ colonoscopy line, but Mr. Heine agreed that optical biopsies won’t appear any time soon: “We’re not ready right now to launch anything that’s making a diagnosis claim. It’s not about optical biopsies at this point. It’s about supporting the physician,” he said.
Along with improving video capture and quality-control efforts, Mr. Schwartz believes that Virgo’s systems can help solve the problem of limited training data. By capturing and storing video data from a wide range of procedures, it is generating a resource that could boost the field and may one day make optical biopsies a reality. “It becomes the training set to build the AI video systems of the future,” he said.
Mr. Heine is an employee of Olympus. Dr. Tucker-Schwartz is an employee of NinePoint Medical. Mr. Schwartz is an employee of Virgo.
Virgo was recently selected to join the Summer ’17 NYC Techstars accelerator program, and we are launching our first product later this month. With this in mind, I thought it would be beneficial to explain a bit about what we are working on and why we are so excited to be working on it...