TeraRecon Acquires Machine Learning Platform Developer McCoy Medical

TeraRecon Acquires Machine Learning Platform Developer McCoy Medical – Our Take

Written by Simon Harris

  • On 1st June, TeraRecon, a US developer of advanced visualisation and enterprise medical image viewing solutions, announced the acquisition of a majority share in McCoy Medical Technologies, which provides a developer platform and a vendor-neutral application program interface (API) for machine learning algorithms used in medical imaging.
  • The two companies have formed a new entity, initially called WIA Corporation, that will operate independently from TeraRecon. The new company retains McCoy Medical’s advisory board, which includes Dr. Eliot Siegel (University of Maryland), Dr. Paul Chang (University of Chicago) and Dr. Khan Siddiqui (American College of Radiology’s IT and Informatics Committee).
  • WIA’s strategy is to provide an open platform that enables algorithm developers to commercialise their products, and that makes it easier for end-users to find and purchase algorithms from multiple developers in one location.

Here’s our analysis of the deal, likely outcomes and what this will mean for the imaging IT market.

An App Store for Radiology

WIA Corporation is addressing two of the main challenges for specialist developers of image analysis solutions – the need for an efficient distribution channel and integration with clinical workflows. Along with the challenges of obtaining regulatory approval, these factors have restricted the number of commercially available products and hindered the uptake of machine learning in radiology. As Dr. Keith Dreyer, Vice Chairman of Radiology at Massachusetts General Hospital, pointed out in the opening session at SIIM 2017, there are 2,613 radiology findings and 23,373 conditions. However, the commercialised machine learning algorithms on the market today address only a fraction of these. This is partly because of the time and expense associated with integrating algorithms, typically via PACS or advanced visualisation workstations, into the clinical workflow. Moreover, algorithm developers need an effective route to market. For most health care providers, the time and administrative burden of purchasing algorithms piecemeal from multiple vendors is not a viable option.

With its open platform and vendor neutral API, WIA is creating an online marketplace for image analysis products. The company is in discussion with more than 20 algorithm developers to distribute their products and plans to announce the first ones in the coming weeks. These will likely include commercial image analysis companies and academic researchers. TeraRecon also plans to host its own algorithms on the WIA platform.

For algorithm developers, there is no upfront fee to distribute their products on the WIA platform and, according to WIA, the initial integration typically takes less than a day, depending on the complexity and market readiness of the algorithm. WIA plans to generate revenues from two streams – sales of algorithms through its platform and subscription payments from health providers to gain access to the platform. Providers will likely also pay an upfront fee to cover initial set-up and integration of WIA’s platform with their existing imaging IT systems. The level of commission paid to developers will depend on a variety of factors, including the popularity of the algorithm. A higher rate of commission will be paid to the most popular algorithms. The commission levels are structured to ensure that the algorithm developers receive the largest share of any sales.

Not the First to Market

As was highlighted in our last article on machine learning in medical imaging (Partnerships are King for Machine Learning in Radiology), a number of the leading medical imaging companies have launched open platforms and are establishing ecosystems of companies that provide specialist healthcare applications. Examples include GE Health Cloud, Siemens Healthineers Digital Ecosystem, IBM Watson Health Core and NTT DATA Unified Clinical Archive. Each of these features a mix of products developed by the platform owner and from specialist third party developers. For example, GE Health Cloud features image analysis products from Arterys, Imbio and Pie Medical Imaging, alongside products developed in-house. So, is there room for WIA?

The success of these open platform ecosystems will depend on a variety of factors, including the availability of a wide range of applications and algorithms. For developers, these open platforms are an effective route to market and give access to the platform owner’s installed base of customers. Once the WIA platform has been fully integrated, algorithm developers will have access to TeraRecon’s global customer base of more than 2,000 sites, which should help WIA to attract partners for its platform. As the deal with TeraRecon is not exclusive, WIA is also courting partnerships with other imaging IT vendors to further expand its reach.

The major medical imaging companies that have launched their own open platforms, as mentioned above, could also integrate with the WIA platform to gain access to its partners’ products. A broader range of products will increase the attractiveness of their online marketplaces to health care providers. For the platform owners, a partnership with WIA negates the need to have separate commercial agreements and integrations with each of the developers on WIA’s platform.

WIA is placing a strong focus on academic researchers and universities that have developed image analysis tools but do not have the knowhow or resources to commercialise them. This could lead to the WIA platform having content that is not available from other online marketplaces. As Jeffrey Sorenson, president and CEO of TeraRecon, pointed out at the time of the McCoy acquisition, “The algorithms work, but the business model of a single algorithm doesn’t work. The overhead to commercialise a single algorithm exceeds the value of a single algorithm.”

Beyond Medical Imaging

Some health care providers will see WIA’s exclusive focus on medical image analysis as a positive, but others may prefer the broader scope of the open platform ecosystems from the major health technology vendors. These feature applications for a wider range of clinical needs, including image analysis, analytics, dashboarding, integrated workflow, and in some cases, population health and telehealth. The overarching industry trend towards more integrated care will likely drive demand for enterprise platforms that form a base for all diagnostic and acute clinical care management – a clinical IT equivalent to an EMR.

Over the longer term, these cross-discipline open platforms may limit the market potential for dedicated image analysis platforms. However, at this early stage, there is certainly room for both – any efforts that encourage vendor partnerships and to accelerate the commercialisation of machine learning-based image analysis tools can only be a good thing.