Given the pace of technology advancement, artificial intelligence (AI) in radiology has moved past the nascent stage toward maturity. Commercial adoption is on the rise, while the industry continues to see an increasing number of startups emerging to serve radiology AI needs. However, the market is fundamentally different today than before the pandemic. Here are the top trends in radiology AI that all market stakeholders should consider.

  1. Solution Development

Comprehensive End-to-end, Value-adding Solutions

Gone are the days of developing a simple reading room solution that assists radiologists in identifying areas of interest. Solutions that support the full care continuum for current disease conditions are more likely to gain a foothold. These solutions will suggest the best imaging tests to be performed based on symptoms and the optimal scan settings to obtain the necessary images. They will guide clinicians in making the appropriate diagnosis and treatment decisions.

  • RapidAI – Beyond stroke detection, this workflow solution helps in-app communication with stroke team members to assemble an interventional team at a moment’s notice, reducing the time from detection to intervention.
  • VIDA – In pulmonology, VIDA Insights (erstwhile LungPrint) helps with early disease detection, optimizes the interpretation time of complex conditions like COPD and ILD, and helps make the right treatment decisions.

Multiple Anomaly Detection Solutions

Care providers increasingly prefer solutions that can detect multiple anomalies in a single scan, which can reduce image reading time and minimize human error. These solutions could be valuable in emergency trauma cases, where an incidental finding could help save a life. Although this is notably a smaller trend, such solutions are perceived to provide a higher return on investment. One example is Annalise.AI, which can detect over 120 abnormalities in a chest X-ray.

Risk-based Screening Stratification Solutions

Artificial intelligence solutions designed to help reduce the volume of scans by “weeding out” normal patients and flagging abnormal ones for radiologist review are becoming increasingly popular. Recent evidence presented at radiology conferences points to how these solutions can significantly reduce the radiologist’s workload, especially with mammography screening (in developed countries) and tuberculosis screening (in developing countries), which typically see large volumes of scans. These solutions also help prevent, or at least reduce, unnecessary diagnostic testing such as biopsies and associated costs.

  1. Demand for Radiology AI Solutions

APAC Warming Up to Radiology AI

There is a marked increase in radiology AI adoption among emerging markets, especially in the APAC region. Care providers are procuring solutions from local vendors, like Synapsica.AI and Rises.AI in India and Advanced Abilities Solutions in the Philippines. Solutions providers from the region are also foraying into other emerging markets by partnering with local vendors. For example, Lunit (South Korea) entered Indonesia in partnership with INFINITT, while Vuno (South Korea) entered Latin American markets in partnership with Visual Medica (PACS).

Self-developed AI Solutions Remain in Demand

Despite many artificial intelligence solution vendors out there, the demand from hospitals to develop their own AI solutions persists. Some vendors also help in this regard; Gradient Health offers annotated images for training AI solutions, Encord offers DICOM annotation software, and Paxera Health has partnered with Penn Medicine to help develop their own AI solutions.

Platforms and Marketplaces Grow but Face Lukewarm Response

Since 2017, several companies have established an AI marketplace offering a gamut of AI services from a variety of vendors. Blackford Analysis, Nuance Health (now with Microsoft), and Envoy AI (now under Symphony AI group, rebranded as Eureka) were among the first to adopt this approach. Over the years, several others emerged, such as Arterys, IBM, Fovia.AI, Incepto (Europe), CARPL (Mahajan Imaging, India), Doctor Net (Japan), Vizyon (blockchain and teleradiology), and Wingspan (China). More recently, we have Enlitic (Curie platform) and QMenta, which focuses on neurology.

However, this approach still faces adoption barriers, leading key players to shift to a platform model, which offers end-users a host of value-add solutions to help manage administrative and operational processes along with their imaging workflows. These are better positioned to demonstrate higher value and ROI, but only time will tell if this approach sways radiology departments and hospital CFOs in its favor. Philips (HealthSuite) and GE (Edison) are leading this transition, although their strategies extend beyond radiology.

  1. Interesting Vendor Shifts and Dynamics

Novel Partnerships

Over the past five years, the evolving complexity of radiology AI and its commercial ecosystem has shown that no solution provider can succeed alone. Emerging use cases and intense competition will force unprecedented partnerships among AI solutions providers in their pursuit to capture market share. Partnerships can range from basic integrations (ScImage and DiA Imaging Analysis for viewer integration), distribution (Fujifilm X-ray with Annalise.AI), and R&D (Mayo Clinic with Vuno for precision oncology R&D), to providing end-to-end solutions (AstraZeneca’s partnership with Oxipit.AI).

New Players’ Entry Somewhat Balanced by Consolidation

The competitive landscape is currently witnessing interesting developments. On the one hand, a host of new vendors like Artyra, Mireye, and Vinbrain have entered the fray, especially from emerging markets. On the other, a few key players like Radnet and Nanox have consolidated the market to a certain extent through a spate of acquisitions.

Interestingly, a few companies have exited the space altogether. MaxQ.AI pivoted into other businesses, whereas IBM sold Watson Health to a private equity firm.

  1. Growth Barriers: Reimbursements, Funding & IPOs

The broader barrier to adopting the reimbursement model has not been addressed, even in developed markets. The market saw no major developments beyond the “NTAP” payments for Stroke AI, which caused quite a flutter.

The sector is also witnessing an uneven funding pattern and a significant drop in the number of deals being made, potentially fueled by the global economic slowdown. Pre-pandemic times saw an equitable distribution of funding across all kinds and stages of AI startups. Now, only a few big players like Viz.AI and Aidoc are snagging lucrative deals, while smaller vendors are receiving much lower amounts. However, Lunit, which performed quite well over the past two years, has filed for an initial public offering (IPO).

Conclusion

Massive strides in computer technology have helped artificial intelligence make serious inroads in the field of radiology. These developments are worth noting for radiology stakeholders and the broader healthcare space in general as AI continues to redefine care delivery with more companies constantly pushing the boundaries of what’s possible. We believe the next three years will see radical changes to this space, and we’ll closely track.

To discover more about the latest trends for this market and learn how Frost & Sullivan can help you promote your Radiology AI solutions, contact us.

About Frost & Sullivan

For six decades, Frost & Sullivan has been world-renowned for its role in helping investors, corporate leaders, and governments navigate economic changes and identify disruptive technologies, Mega Trends, new business models, and companies to action, resulting in a continuous flow of growth opportunities to drive future success.

Your Transformational Growth Journey Starts Here