Medical imaging is a broad term that includes several different techniques. After working on AI-driven tools to enhance X-ray and mammography, French startup greamers are now aiming to work on magnetic resonance imaging (MRI).
Instead of starting from scratch, Gleamer has already worked on Caerus Medical, an AI-powered MRI analytics, and has acquired a startup that is merging with Pixil.
Greamers are part of a second wave of startups seeking to improve medical imaging using artificial intelligence. A few high-tech founders who created a startup on this topic in 2014 or 2015. Most of it went anywhere, but this space had some consolidation. For example, the medical vision and arteries of Zebra were acquired by Nanox and Tempus, respectively.
Founded in 2017, Gleamer builds a kind of co-pilot AI assistant for medical imaging. Greamers allow radiologists to theoretically improve the accuracy of their diagnosis when interpreting medical images.
Startups are persuading 2,000 institutions from 45 countries to use software solutions. Overall, Gleamer has processed 35 million exams. The company is CE and FDA certified for bone trauma interpretation products. Europe also offers products with a special focus on chest x-rays, orthopedics and bone age measurements with CE certification.
“Unfortunately, a one-size-fits-all approach to radiology doesn't work,” Greamer co-founder and CEO Christian Aulusch told TechCrunch. “It's extremely complicated to have a big model that covers all medical imaging and provides the level of performance that doctors expect.”
So the company created a small internal team focused on mammography and CT scans. “Three weeks ago we released a mammography product that we've been working on for 18 months,” says Allouche. It is based on a unique AI model trained with 1.5 million mammography.
“We have a partnership with Jean Zay, the French government's GPU cluster,” says Allouche. The company is also working on CT scans for cancer.
But what about the MRI? “MRI is a different technical space,” says Allouche. “MRI has many tasks. It's not just detection, it's segmentation, detection, characterization, classification, and multisequence imaging.”
So Gleamer has acquired a small startup (Caerus Medical) and is moving faster by merging with bigger moves (Pixil). These two companies have been working in this field for several years. Greamer does not disclose terms of the transaction.
“These two companies will be two MRI platforms with clear ambitions to cover all use cases over the next two to three years,” says Allouche.
Preventive medical imaging
The Greamer model shows promising results, but it is not perfect yet. For example, the company's new mammography model claims that startups can detect four of five cancers. In comparison, human radiologists without AI support usually identify cancer in three out of five cases.
However, increasing productivity from tools like Greamer can fundamentally change medical imaging. Missing the tumor may appear in follow-up trials in a few months.
“I think in the not-so-distant future, we are all getting the everyday full-body MRI that is paid by our insurance companies.
However, in some cities, there are already too few radiologists to meet the demand for reactive imaging. As the industry shifts to preventive imaging, AI tools become essential.
Greamer CEO believes AI could become a “orchestration and triaging” tool. Most medical imaging tests are performed as a way to rule out some diagnoses. “So there's a need to automate all of this with a very robust AI model with a much higher level of sensitivity than humans,” says Allouche.