Redmond-based EchoNous threads AI technology into ultrasound medical devices
Seventy-five years ago, an Austrian neurologist named Karl Dussik embarked on an experiment that used ultrasound — high-frequency and noninvasive sound waves — to identify brain tumors in humans. He sent an ultrasound beam through a patient’s skull, sparking an interest in early medical sonography. Today, ultrasound is a valuable tool for hospitals, and the technology is used to monitor fetal health during pregnancies, or to check vital internal organs for diseases and other maladies.
Medical sonography has seen many technological improvements over the years, but Kevin Goodwin, CEO of EchoNous, hopes to further advance the field by incorporating Artificial Intelligence (AI) into Uscan, the ultrasound machines the Redmond company develops.
“A segment of the healthcare industry that holds immense opportunity to apply AI and machine learning is ultrasound,” said Goodwin. “It’s ripe for innovation.”
Goodwin, 60, grew up in Chicago. In 1986, he was hired by Bothell-based ATL Ultrasound, where he worked his way up from a sales representative to running the company’s Asia Pacific and Latin America lines of business. The CEO approached him in 1996 with a business problem. ATL Ultrasound was in the early-stage development of a portable, hand-held ultrasound device, and asked Goodwin to commercialize it. Goodwin spent the following year asking doctors what they would like to see in such a device. By 1998, ATL Ultrasound spun off a new company, SonoSite, and selected Goodwin as the company’s CEO.
“We had no revenues, no product, and it was right in the middle of the dot-com boom,” said Goodwin. “But we grew that company from a standing start to $300 million in revenue over several years, and transformed medicine by creating what is now called point-of-care ultrasound.”
Looking back, it’s easy to presuppose point-of-care ultrasound has always existed in hospitals. But there was a time when bulky ultrasound machines were anchored permanently inside hospitals, and operated by healthcare professionals where they worked — not where the patient might show up, such as emergency rooms.
“SonoSite’s products were shrunk down to the size of a laptop and were moved into emergency, anesthesiology, and critical-care departments,” Goodwin said. “We created a brand that was easy to use, and it did very well.”
Fujilfilm purchased SonoSite for $995 million in 2012, and Goodwin worked for the company’s new owner for the next two years.
When he left, Goodwin had very little interest in leading another startup medical-device company. That changed, however, as he learned more about deep learning and AI. He started to consider ways to thread that technology into highly miniaturized ultrasound-based devices.
One year after leaving SonoSite, Goodwin was hired by Signostics, an Adelaide, Australia-based company that spent nearly a decade producing a small ultrasound scanner device. Goodwin wanted to improve the company’s existing Uscan device and create an intelligent, extremely miniaturized ultrasound device that used computer algorithms and AI to more accurately scan a bladder and measure its volume, helping doctors, nurses, and patients.
Eventually, Goodwin and a former SonoSite colleague created a parent company, EchoNous (a Greek term for “intelligent ultrasound”), and KKR, a New York-based investment company, invested $70 million ($35 million in 2015, and $35 million last year) in Goodwin’s idea.
Goodwin conceded that, on their surface, AI infused bladder-scanning devices aren’t the most enticing products to incorporate machine-learning. But for doctors, nurses, and patients alike, Uscan is an improvement over conventional bladder-scanning tools, which Goodwin described as “older fish-finder-like tools and technologies” that guesstimate the size of the bladder, and produce less robust data.
“It’s an amazing tool. I’ve been a biomedical engineer for 40-plus years, servicing hospitals all over the West. When I saw this, I thought, ‘Whoa!’”
The probe on EchoNous’ hand-held Uscan device is about the size of an electric razor. A healthcare provider applies gel to a patient’s abdomen, places the probe over the bladder, and the device is initiated. “The device is then slowly pivoted from hip to hip in a fanning motion, capturing 32 times more imaging data than conventional bladder scanners, recording 256 bladder data slices in each fan, then running this rich data set through a computer vision-based algorithm to produce a high-resolution 3D model of the bladder,” Goodwin explained. “This gives healthcare providers more accurate volume measurements than the industry standard, as well as assesses excess fluid or other issues of concern.”
Goodwin recalled a recent situation where a patient’s urine contained blood, a potentially serious situation that could signal cancer.
“The doctor used the Uscan device to image the bladder and found there was no problem with the bladder structure or presentation,” said Goodwin. “The doctor put the patient on medication and sent him home. Now, they would have otherwise run an MRI, done a bunch of expensive stuff, and had the patient in the hospital for a couple more days — probably worried the patient, too, only to have the patient find out two days later they are OK. That all takes time and money, and it can be overkill.”
EchoNous released Uscan in late-2016 and, according to Goodwin, more than 20 percent of healthcare systems in the United States have incorporated the device into their hospitals, such as Cleveland Clinic and Johns Hopkins Medicine. Goodwin said the device is generating revenue that, in addition to KKR’s $70 million investment, is spurring other product development.
EchoNous is developing a new vascular access probe that is designed to help nurses more easily locate veins for peripheral IV placement, relaying real-time images of veins and arteries on the screen of a tablet. Those days of healthcare providers tying off one of your biceps and blindly feeling around for a vein from which to draw blood could soon be limited.
“I’m really stoked about this,” added EchoNous service manager Kile Jackson, who demonstrated the device. “For me to see this, I’m a Stage IV throat cancer survivor. I’ve been stuck a lot over the years. Working in hospitals all my life around MR and CT scanners and bladder scanners, this is cool stuff. It’s hugely valuable. It’s an amazing tool. I’ve been a biomedical engineer for 40-plus years, servicing hospitals all over the West. When I saw this, I thought, ‘Whoa!’”
The peripheral IV device could be released as early as this spring, pending FDA approval.
And a slim, portable AI workstation with an “AI hub” to house these tools is in development.
All of this work is being done at EchoNous’ headquarters, a two-story, 27,000-square-foot Redmond building that houses approximately 80 software, electrical, and manufacturing engineers; product designers; project managers; and administrative staff.
“It is simply a better mousetrap in what is otherwise an unsexy area of medicine,” Goodwin added. “The idea was to prove out the concept of using AI techniques and to get that process started in medicine. I think we have tremendous potential to make technology more valuable without making it more expensive.”