Analysis Featured

Seeing Blood Flows in the Human Body in 4D With Computer Vision

Five years ago, a man entered a small hospital emergency room with puzzling symptoms that seemed heart-related. Tests showed there had not been a heart attack. A sonogram revealed odd, broccoli-like growth in one part of the heart and a fringed flap waving when the heart beat. Just in case the growth was fungal, professionals administered high-level antibiotics during an hour-long ambulance ride to a city hospital, and many tests later, the diagnosis was made: An unsuspected genetic heart defect had caused severe blockage in one valve and excess pressure had blown out another.

This medical emergency took hours to diagnose. With Arterys, that hospital could see 4D imagery of the heart and blood flowing through it in minutes just by interfacing their own equipment with a cloud-based app.

VisionAR caught up with one of Arterys’ founders, John Axerio-Cilies, to find out how this unique clinical care game-changer came into existence. “We all met at Stanford,” Axerio-Cilies said. “I was working on my PhD in Computational Fluid Dynamics (CFD).”

The Genesis of a Revolutionary Medical Imaging Service for Hospitals

Before starting Arterys, Axerio-Cilies used his knowledge to work on aerodynamic flows for Formula 1 racing cars. CFD is a branch of knowledge that be used to simulate flow of any kind, including airflow around a car or fluid through a human body.

Fabian Beckers was obtaining his business degree from Stanford when he met Axerio-Cilies in 2010 and experiencing health issues he thought CFD could help with. At that time, cloud and machine learning was very new. Deep learning hadn’t been invented yet.

Around the same time, they met Albert Hsiao and Shreyas Vasanawala, both Stanford students, who were working on how to visualize and quantify blood flow identification in the heart using MRI. The four decided to start Arterys, a private company, incorporating in 2011 and, in 2013, raising their $2.5 million in their first funding round followed by a $12 million series A round in 2015 and 2016.

“It takes a lot of computation to produce blood flow in 4D in the heart,” Axerio-Cilies said. “The only way to service the product was to use the cloud. We had to build out specific software requirements. We basically built out a platform with GE Healthcare, our supporter and partner, to interpret images anywhere in the world.”

Cardiac MRI is their first app, but the company is well on the way to employing many apps as healthcare professionals see the benefit of 4D flow and analytics to provide value-based care. The ability to use AI, machine learning, and cloud computation to augment diagnostic services frees the physician in everyday life by providing comprehensive, accurate information quickly so the patient can be diagnosed correctly and treated immediately.

How Arterys Helps Patients, and Hospitals

When asked how the business model works, Axerio-Cilies said, “Think of it as an app store with Software-as-a-Service (SaaS). Hospitals are the customer. They buy licenses for a particular application and are charged on a price-per-user basis. It could be $10 to $60,000 depending on the features, access, data per user, etc.”

The advantage of the cloud is that you don’t need physical hardware to process, to render, quantify, or segment any data. Gigabytes of data can’t be rendered on a laptop. Access can be through a web browser, so laptops, workstations, smartphones, or any device at home or work can be used. From a tech perspective, anything with a web browser can access Arterys, but the FDA-cleared product has to be on a high resolution monitor and regulation equipment if performing diagnostic care.

The backbone of the system is micro-service architecture built on GPUs. It leverages off Amazon. Arterys is transitioning to include Google and Azure. Clusters of data are sent from scanners in the hospital to the cloud, processing images while the physician logs in to view data in real time. Tools are provided to render and quantify for clinical workflow. The system can image the heart, measure linear or area distances, and perform advanced quantification for blood flow anywhere in the body.

Arterys’ core building blocks are GPUs for machine learning. One GPU takes an hour to process what more than 20 GPUs can do seamlessly in real time, therefore, GPU clusters distribute the work faster and more efficiently.

“All the big players are trying to build out medical imaging platforms,” Axerio-Cilies said. “IBM Watson has a big presence. GE, Phillips, and Siemens are all trying to build out analytics platforms while smaller companies are trying to do similar things with different applications.”

Three out of four of the big names deploy local server-based apps. Arterys uses its infrastructure to do complex rendering that no one else can do with 2D images.

“We have a unique app called 4D Flow in collaboration with GE Healthcare,” Axerio-Cilies said. “The technology makes it all clear. It’s data-driven medicine so the diagnostic technicians can do their jobs.”

What is the Future of Arterys?

Arterys is working hard to expand their portfolio of apps. They just got a second FDA clearance on a deep learning product and hope to have many more applications coming soon. Their typical customer is any medical professional who owns a medical imaging device. This includes outpatient facilities, radiology, medical centers, clinics, and more.

Deep learning is quickly changing medical imaging and medical practice. Radiologists may be leery of the way AI automation will change their day-to-day life since in the last five years there’s been a dramatic shift in technology. Workstations are moving from local-based software to cloud-based intelligence software. That’s a learning curve the healthcare professional has to deal with. Arterys is demonstrating the superiority of their cloud-based product and is primed to take advantage of that.

“As a cloud operation, we’re collecting data on a day-to-day basis,” Axerio-Cilies said. “It’s similar to Tesla and Google. Tesla has driven 200 million miles, Google has only driven 2 million miles. Tesla has much more information to train their machine learning model. They are so far ahead that Google will never be able to surpass them. It’s the same thing here: Our competition will not be able to catch us because as we get more customers we also get smarter.”

Asked about future plans, Axerio-Cilies said Arterys is going beyond cardiovascular and looking at oncology.

“Oncology has very tedious and time-consuming workflows,” he said. “We can use machine learning to identify, segment, and track. It’s a big opportunity.”

Currently, Arterys needs capital. Axerio-Cilies said they have the infrastructure in place along with the machine learning tools, code base, and necessary tools. But additional capital will help them hire teams to create more apps in parallel.

“We are the first company to have an FDA-cleared cloud-based deep learning application,” he said. “That’s very unique—a game changer in the world of medicine. We can change clinical care from a regulatory and technological perspective.”


Written with Nicki Jacoby.

Allen Taylor

About the author

Allen Taylor

An award-winning journalist and former newspaper editor, I currently work as a freelance writer/editor through Taylored Content. In addition to editing VisionAR, I edit the daily news digest at Lending-Times. I also serve the FinTech and AR/AI industries with authoritative content in the form of white papers, case studies, blog posts, and other content designed to position innovators as experts in their niches. Also, a published poet and fiction writer.

Add Comment

Click here to post a comment

Your email address will not be published. Required fields are marked *