AI Medical Service, a Tokyo-based startup announced Monday that the Food and Drug Administration has given its machine learning algorithm breakthrough device designation for its ability to analyze endoscopy images for potential diagnosis of gastric cancer.
Since the start of the program in April 2015 through till the end of 2018, FDA has granted 112 Breakthrough Device Designations to novel products that receive expedited and priority review for speeding up their approvals process. Compare that with the same designation for novel drugs that numbered 130 as of December 2018. [The Breakthrough Therapy designation for drugs formally began in 2013 with none being offered in 2012 though the program launched that year.]
So how does the AI Medical Service’s machine learning algorithm work? The company applies neural network algorithms that are trained on large real-world datasets of images of biopsy-proven cancer lesions, benign lesions, and normal tissue that were captured using standard endoscopes. The goal is to help doctors performing endoscopies easily detect lesions suspicious of cancer. In a news release, AI Medical claimed that its “technology is further differentiated as it is able to detect gastric, colorectal, and esophageal cancer, especially for concurrent usage whereas other entities’ technology tends to focus only on colorectal cancer.”
The startup, backed by Sony announced in October that it has raised $42.9 million in a Series B funding round. The money comes from Globis Capital Partners, World Innovation Lab (WiL) and Sony Innovation Fund by IGV (Innovation Growth Ventures), as well as others who were not identified by the company.
In response to questions in an email, Dr. Tomohiro Tada, CEO of AI Medical Services, explained that most companies training AI algorithms focus on colorectal cancer, whereas AI Medical has focused on gastric cancer.
“Most of our competitors developing endoscopy AI have started with colorectal cancer (probably because it is easier than gastric cancer to detect.) On the other hand, we have started with gastric cancer, which is said to be more difficult compared to colorectal cancer, and are developing an AI that can cover stomach, colon, esophagus and so on — ultimately the whole digestive tract,” Tada wrote in an email forwarded by a company representative.
AI Medical also seems to believe that winning this Breakthrough Device Designation is especially key for a Japanese company that doesn’t enjoy the benefits of the U.S. entrepreneurial ecosystem.
“To invent a revolutionary medical device/drug has become too expensive for a small company and too high-risk for an established company. That is why in the US, you have [an] ecosystem [where] VCs provide [the] financial support for many seed-[stage companies and] eventually a few of [them] would be sold to large medical companies,” Tada said. “But we do not have such a system in Japan.”
AI Medical has 60 employees in Japan. Its competitors are ai4gi, a comercial joint venture between Satis Operations and Imagia Wision AI, both Canadian companies; Docbot, LPixel, based in Tokyo, and Cybernet, also based in Tokyo.