As we could glean from HIMSS17, seven major imaging and IT vendors are setting out to be the first to build and operationalize the next-generation cloud-based platform for advanced imaging analytics. The opportunity to become the ‘Android’ or the ‘Apple Store’ of imaging informatics – where providers can access and use a host of applications all from a single unified user interface, through non-capital intensive pricing models and independently of the hardware they use – is immense, indeed.
These large public or hybrid cloud platforms must compete against one another to attract value-adding content from the best-and-brightest independent software vendors (ISV) out there, to, in turn, deliver them as software-as-a-service (SaaS), likely through various flavors of subscription-based and pay-per-use models. These will serve to complement the vendors’ native applications to form a comprehensive and diversified portfolio.
Three Vendors Leading the Charge
- Siemens Healthineers Digital Ecosystem: Arguably, the most significant industry development for imaging informatics at HIMSS17 was the debut of the “Siemens Healthineers Digital Ecosystem.” The cloud-based platform
expands on the native teamplay portfolio which Siemens reported at RSNA 2016 was already operational at more than 160 sites. Now evolved into an ecosystem, it incorporates a variety of third-party apps for providers to choose from, ranging from advanced MRI (Arterys, SyntheticMR) to teleradiology (Second Opinions, USARad, TMC (Europe)), business analytics (Viewics, Dell-EMC) – among others and with more to come.
The ecosystem appears to be quite mature and is expanding at a rapid pace globally, benefiting from the productive partnership of Siemens Healthineers with Microsoft Azure. As a case in point, the pair has even been able to ‘crack open’ the German market, notorious for its stringent data privacy laws and, arguably, the most difficult to penetrate with cloud solutions in healthcare.
Siemens Healthineers may have made this introduction more than a year after its rival GE Healthcare soft-launched the GE Health Cloud, but the level of advancement and integration of the solution makes it anything but a “me-too” play for Siemens Healthineers, who seem to have operationalized the Digital Ecosystem quickly and robustly. This exciting, new development could not be timelier for Siemens Healthineers, as the company readies its spin-off later this year.
- IBM Watson Imaging: HIMSS17 was quite significant on the imaging front for IBM Watson, as the company showcased its first market-ready native application: the Clinical Image Review application is the brainchild of the partnership IBM Watson entered with Memorial Sloan Kettering in 2012. Limited in its current use case to arterial stenosis, it is an un-regulated application that helps flag up discrepancies between medical imaging reports, in this case from echocardiography, and the related clinical notes in the medical records. This allows providers to recall at-risk patients who may have been discharged without the proper follow-up. As such, in its current marketable form, the application works with imaging reports and not with the images themselves at the pixel level. The latter use case would more likely fall under IBM Watson’s regulated product line of applications, where project codename Avicenna belongs, and which is still demo’ed as work-in-progress.
Unfortunately from IBM Watson, this successful productization off of the Sloan Kettering partnership was somewhat over-shadowed during HIMSS17 by bad press
about the much less successful outcomes, at least to-date, of the ongoing partnership with MD Anderson, Sloan Kettering’s oncology luminary rival.
As far as building an open imaging analytics ecosystem, the company seems still somewhat shy of fully opening-up IBM Watson Imaging to third-party ISV’s applications. However, the integration of Medymatch’s deep learning application announced at RSNA 2016 is proof that IBM perceives the future of Watson Imaging as a platform – not just a home to its native applications.
- GE Health Cloud: Now well into its second year since it was announced at the RSNA 2015, the operationalization of the GE Health Cloud is still advancing, with good progress so far in the U.S. and the UK markets. However, it is fair to note the GE Health Cloud momentum, as it transpired from RSNA 2016 and HIMSS17, is nowhere near that of its debut a year when the ecosystem seemed to be building up
at record speed. Indeed, although it is part of the corporate-wide ‘Industrial Internet’ initiative of GE, the Health Cloud is evolving at its own pace due to healthcare-specific regulations. Working with cloud partners Microsoft Azure and Amazon Web services (AWS), GE Healthcare is phasing out the Health Cloud incrementally, country by country.
Two Contending (and Complementary?) Runner-Ups
- NTT Data: The new owner of Dell Services, which include the legacy of InSiteOne, the industry pioneer for cloud-based medical image archiving, is in a somewhat good position to become one of the go-forward medical imaging analytics ecosystems. Already, at the time of the NTT Data acquisition, the Unified Clinical Archive (UCA) had started to grow into a managed cloud-based platform featuring several value-adding analytics applications, notably from medical imaging deep learning experts Zebra Medical Vision and Imbio. Now, NTT Data wants to leverage the UCA to grow the company’s presence in the North American marketplace. But, for this to happen – and despite being very well-known across Asia – it will be crucial for NTT Data to build its footprint and brand name stateside.
- Philips Healthcare: The enterprise imaging developments of Philips Healthcare are quite impressive, notably how the company is able to bring together the full imaging IT solution suite into a unified platform under a unique, managed services model. On the other side of the house, the Philips HealthSuite cloud solutions are advancing steadily, as well, yet it is still focused mainly outside of the acute care setting and in support of Philips’s care continuum strategy. These two pillars of healthcare informatics at Philips are yet to converge, to create what can be deemed a cloud-based medical ecosystem from Philips. Perhaps, it will be along these lines that Philips Healthcare will make its next major partnership or acquisition.
Two Additional Elephants in the Room
- Google: On the busy HIMSS17 booth of Google, the cognitive giant exhibited imaging-related works-in-progress from two distinct sides of the house. Google Brain on one hand, and Google DeepMind on the other, each showed their latest advances in computer vision applied to medical images, notably in the field of ophthalmology. The algorithm of the former
can automatically detect diabetic retinopathy in retinal fundus images, while the latter
can provide early detection of age-related macular degeneration from optical coherence tomography (OCT) images.
Eye imaging may well be the first of many more clinical applications to come from Google, and perhaps an ideal area pilot its cutting-edge computer vision methods given the wide global availability of eye fundus and OCT imaging. Indeed, Google Brain, for example, is already active in other areas, such as radiotherapy planning. Most certainly, these recent developments at Google mark a decisive re-entry of Google into the health arena, after it chose to discontinue its Google Health offering years ago.
- Amazon: Coincidentally – or maybe not –, eye disease is also the area where Amazon chose to jumpstart its own machine learning medical imaging applications. Similar to Google’s, Amazon’s GPU-based algorithm
running in the Amazon EC2 cloud can automatically detect diabetic retinopathy from fundus images. This project may still be investigational and niche, but the fact is, Amazon is already extensively called upon to help cloud-enable medical imaging applications from its various partners through Amazon Web Services. Therefore, it is only a matter of timing before Amazon decides to promote its own cloud platform hosting a larger suite of native and partner applications.