By Aaron Krol

December 22, 2015 | In a development lab at the University of Pittsburgh Medical Center (UPMC), radiologists are learning to work with a new, cloud-based system to manage the image files produced by CT scanners, MRIs, ultrasound machines, and other medical imaging devices. By moving storage and processing capabilities to the cloud, UPMC hopes to make these images more accessible to clinicians at all stages of care. A radiologist working in this lab might take a CT scan of a suspected tumor, pull up the scan on a tablet while consulting with an oncologist, and run a 3D visualization program on the spot to better interpret the image.

“We’ve had upwards of a hundred radiologists rotate through that lab,” says Dr. Rasu Shrestha, a radiologist and Chief Innovation Officer at UPMC. “For these difficult cases that we often deal with in a large academic medical center, we’re able to have radiologists collaborate with pathologists and surgeons and oncologists. That’s powerful. That has specific implications for patient outcomes.”

UPMC is the development partner for the GE Health Cloud, a medical imaging platform that GE Healthcare plans to roll out commercially in the spring of 2016. GE’s ambition is for the Health Cloud to become the standard way for hospitals to store, share, and analyze their imaging data―not just from GE devices, but from their competitors’ machines too.

“It would free us from the confines of the client workstation,” says Shrestha, “these three- or four-monitor workstations that radiologists like myself have in our reading rooms, where we’re locked down for the duration of the day.”

Data mobility is one of the biggest challenges in 21^st century healthcare. The technology now exists to quickly send patient information between doctors and hospitals, but our fragmented care system makes it hard to set up structures to handle this data sharing. To usefully get information across hospital walls, there need to be standard data formats, and common frameworks for access and analysis.

This has been a huge headache for electronic health records (EHRs), a relatively new industry in which dozens of providers have built systems, often mutually incomprehensible, to digitize patient histories. But imaging may be less of a steep climb. Since the 1990s, almost all medical imaging devices have followed the DICOM file type standard, which embeds patient data along with each image. That means that even proprietary platforms, built by a host of different vendors, are already speaking a common language.

“Today we have all these powerful MRI, CT, X-ray imaging machines, but we’ve just scratched the surface in terms of the insight we can generate from these images,” says Justin Steinman, Chief Marketing Officer for GE Healthcare IT. “These images and these algorithms require massive computing power to process.”

Cloud Applications

GE Healthcare is one of the country’s largest manufacturers of imaging machines, counting 500,000 of its instruments installed in American hospitals. But the Health Cloud is meant less as an extension of these devices, and more as an expansion of a radiology department’s compute capabilities. Accepting DICOM files from any instrument, and featuring an open software development kit for third-party programs, the Health Cloud will somewhat resemble a large, agnostic cloud computing platform like Amazon Web Services or the Google Cloud Platform, but with a menu of built-in apps for medical imaging.

This model will displace some of GE Healthcare’s current business. Like its competitors, GE sells enterprise software for radiologists, most notably its Advantage Workstation suite of visualization tools. Now, the company is porting algorithms from that software into a Health Cloud app, Centricity Cloud Advanced Visualization, which might sit in the GE Health Cloud’s “app store” alongside competing programs with some of the same functions.

“We’ve shifted to a cloud-first mentality,” says Steinman, “because that’s where we think our customers are going over the long haul.”

When the Health Cloud launches next year, it will feature four Centricity apps built by GE, and probably a small number of third-party programs. Steinman hopes that outside groups, including GE’s competitors, will eventually provide the large majority of apps. GE will take a cut of any revenue these apps bring in, but their developers will get access to all of GE’s customers working in the cloud. “We see a world where other imaging analytics providers, whether they’re big hardware providers or two guys in a garage with a great idea, choose to run their applications on the GE Health Cloud,” Steinman says.

Not all of the Health Cloud’s early apps are adaptations of current software. At the UPMC development lab, Shrestha is most excited about brand-new tools that take advantage of the cloud’s mobility and ease of access. “In the past, if we needed to share an imaging study with a colleague who might be across town, we would have had to burn a CD or DVD,” he says. “That’s such an antiquated way of doing things. Today, you’re able to click a button and essentially have studies be in possession of the right end users, all securely, all seamlessly through the cloud.”

One of the apps available on launch, Centricity MDT Virtual Meetings, will allow screen sharing and live analysis with an entire care team, all signed into the app from their own devices. Another, Centricity Case Exchange, is built to share images with physicians outside one’s own hospital, including those who aren’t using the GE Health Cloud themselves.

“These are new apps we’re building to leverage the capabilities of the cloud,” says Steinman. “With Centricity Case Exchange, I can publish an image in the cloud, send an email to a clinician in New York City, and ask him to do a remote consult. We’re obviously big believers in connectivity.”

A Holistic System

While GE wants to build the first complete home for medical imaging in the cloud―and, with its underlying Predix cloud platform, do the same for other industries as well―healthcare is already trending this way on its own momentum. “The economics of cloud make a lot of sense to not just us, but any organization,” says Shrestha. “With the cloud, with the touch of a button you’re able to deploy solutions across large areas in your healthcare institution.”

“One thing we’re learning is that the power of cloud opens up all kinds of new scenarios that we haven’t thought through fully yet,” says Steinman. Sharing images through the cloud makes it easy to consult with doctors in distant institutions, for example, but it’s not clear how insurance companies will reimburse for their time. And while granting more members of a care team access to medical images is a net win for patients, it introduces privacy concerns about which hospital staff are opening patient files, and for what purposes.

In the long term, the compute power of the cloud may also do much more than speed up existing algorithms. One of GE’s early partners on app development is NeuroQuant, whose software examines brain MRIs to monitor the course of Alzheimer’s disease. With the large datasets on tap in the Health Cloud, that software might gradually become more predictive, spotting signs of disease earlier.

GE, too, is building an app to work with longitudinal records of patient data, called Centricity Image Access Portal. The DICOM standard has already laid the groundwork for tying together a single patient’s imaging history, but in the high-performance computing environment of the cloud, that record can more easily be tapped for unexpected insights. The data-rich field of medical imaging is especially fertile ground for machine learning, finding patterns impenetrable to radiologists looking at a handful of scans at a time.

Not all GE’s ambitions for the Health Cloud are so futuristic; as an industrial giant, the company is also mindful of more pedestrian goals. “Our vision is that this is going to have clinical applications of all sorts, but also financial applications, and operational applications,” says Steinman. “Ultimately, you’ll be able to manage the lifecycle of your imaging devices in the GE Health Cloud, and do cost of care analysis or revenue cycle management.”

That’s not as exciting as creating predictive diagnostics that scan through thousands of medical images, but it can still make a difference in quality of care. As hospitals adapt to more value-based models of care, as opposed to charging for each procedure or MRI image, they’ll need to do a better job tracking how their costs relate to patient outcomes. That can only be done effectively through systems that span whole hospital networks and every stage of care―just the kind of task the cloud was made for.