Butterfly Network Raises $100 Million to Bring Deep Learning to Medical Imaging

Butterfly Network launches today with $100 million in funding to make a small handheld medical-imaging device that aims to make ultrasounds and MRIs faster, cheaper and smarter.

Investors include Aeris Capital and Jonathan Rothberg, who previously sold his gene-sequencing startup Ion Torrent for more than $375 million. Butterfly is the first company to debut from Rothberg’s 4Combinator startup medical technology incubator.

Butterfly’s device aims to “replace a room of imaging equipment that costs $2 million to $6 million dollars, and takes eight seconds to acquire an image,” Rothberg said. “We can do 30 to 60 times more per second, and be 10 times more precise in how we direct the energy.”

It may sound a bit pie-in-the-sky for now, but this concept is parallel to what Rothberg has already done to make DNA sequencing cheap and accessible.

Butterfly’s goal is to launch the device in as soon as 18 months — pending development and regulatory approval — with a price tag of hundreds of dollars, Rothberg said.

How? By moving these machines onto a dedicated integrated circuit, and by hiring minds from different fields (for instance, an astronomer working on creating high-resolution images of the universe) to look at the problems with new eyes and lots of resources.

Then there’s the next step: Analyzing lots and lots of images. Rothberg’s hope is that Butterfly can apply deep learning — the artificial intelligence technique used by companies like Google to hone features like speech recognition by modeling neural networks to handle massive amounts of data — to medicine.

If all that goes well, it could mean smarter research and diagnoses without ever having to put a knife to a patient.

And that is why Rothberg is approaching the broader tech community with his plans, he told Re/code. “Goldman Sachs and Google get all the resumes,” he said. “I want people to know they can come and build devices that could affect the life of someone you love.”

This article originally appeared on Recode.net.