By Shane Walker

This year’s RSNA conference kicked off with a plenary lecture from Dr. James Brink, chair of radiology and chief of Enterprise Radiology at Mass General Brigham. In his talk, Dr. Brink discussed the importance of radiologists engaging with and defining their role in value-based health care. If not, other specialties and organizations will take the lead and set their own initiatives. Picking up on this theme, one of our focus points was exploring the impact that striving to achieve increased value was having on hardware and software imaging procurement, including AI. The other focus point was identifying the innovation occurring in attempts to keep pace with the demands of the radiology department. As is true with most discussions regarding value-based procurement (VBP), it remains an emerging area that is still not well defined and is difficult to measure and identify. That said, there was much evidence of suppliers preparing their solutions to address the VBP concept as it continues to evolve. In this editorial they include GE Healthcare, Fujifilm, Agfa, Philips, Konica Minolta, and Shimadzu.

Increasing productivity and decreasing burnout with AI in the workflow
Speaking with Scott Miller, CMO of Global Imaging at GE Healthcare, we discussed the benefits of AI within radiology. He summarized, “our AI is intended to be an enabler to improve the patient experience, staff experience, and clinical outcomes”. For Miller, AI will allow technology to deliver a variety of outcomes and be a less expensive, faster means of upgrading technology as it advances. He discussed the need for AI, saying that today’s radiologists are facing heavy strains. As such, Miller poses that AI can assist in their workload by eliminating mundane tasks. For instance, it can prioritize findings, highlighting those that need attention quickly and conduct free reads. In doing so, it can change worklists so that critical findings are attended to first. A few related products include the SIGNA Hero MRI system with AIR Recon DL enhancements to reduce exam times; the AMX Navigate portable X-ray with ‘zero click exam’ workflow; and the Critical Care Suite 2.0 collection of AI algorithms for case prioritization, automated measurements, and quality control. As with years past, GE is still providing much of its AI embedded on device. Another product to note is the Allia Platform for interventional OR that uses GE’s Edison AI platform and Liver ASSIST Virtual Parenchyma. This platform provides 3D visualization to simulate injections dynamically. Productivity gains are also being found in cloud PACS, where the application of AI for load balancing is increasing productivity by 20%.

When asked if radiologists are now coming to GE with suggestions of what engines to include, Miller explained that the responses are mixed. Some radiologists want AI that reads and highlights the normal studies, but they still need to be involved in clinical decisions. This is because the AI currently cannot ‘decide’ what treatment is best for a patient, as it does not understand lab data, genomic data, and personalized history data. Technology is still most useful for increasing productivity while minimizing errors. Further, there is a willingness for providers to pay a premium if the AI feature is enabled within other products that are already in use. They want the technology placed into their workflow through either PACS orchestration or on device. Of this AI, some is developed in-house, and some is from a third party, incorporated into the solutions. When asked about the quantity split between the two, Miller explained that a significant share is developed in-house, but that monthly more and more third parties are being evaluated for the Edison ecosystem. As a result, customers can trial new applications without stepping outside the platform. Miller said they expect to see a significant increase in this over time. Christian O’Connor, Imaging GM EMEA at GE Healthcare, added that it is in the best interest of the industry to continue to crowdsource the development of AI. This is because it improves the speed to market and the impact on specific applications. To add to the question, he commented that there are currently 24 engines on Edison with least 50 other systems in the queue for consideration. He says that combined, this AI network is helping providers and suppliers identify the use cases where AI is actually needed.

The influence of outcome-based language in RFPs
When asked if they were seeing outcome-based language in tenders, O'Connor commented that there was an increase in non-price criteria, but this is still fairly limited in the US. He explained that the EMEA market is a 60+% tender driven market and that many companies do have some value-based intent, but these do not often manifest in the final procurement decisions. This is due, in part to the difficulty of defining value—it could be productivity, it could be financial, every institution defines it a little differently. Once the institution’s definition is clear, the next step is figuring out how to measure that value. Even though many health systems have the ambition to achieve these steps, their processes are lagging. Miller added that 80% of the US healthcare cost is labor. So, even with a significantly reduced price for equipment, health systems are missing out on a great deal of cost reduction opportunity if labor efficiency is not part of the criteria.

VBP, which is closely linked to the ‘most economically advantageous tender’, or MEAT, concept is a framework for cost-effectiveness intended to yield the ‘best’ price but not necessarily the lowest price. Utilizing a VBP framework may not cost less but will increase value to both hospitals and patients in the long run. O’Connor touched on this, saying he is concerned about the increasing prices of technologies. Those that offer value-based outcomes may be too expensive to be pursued in the inflation environment we are currently experiencing. This then slows the advancement of VBP. Today, more than 50% of focus in tenders is on price and in some instances, the product solution that would deliver the best outcome exceeds initial budget expectations. Although this may return the initial investment and more, this is a risk for some organizations.

Vamstar has seen an increase of VBP language in imaging tenders over the last ten years, albeit still a relatively small share of tenders in total (see figure). Vamstar’s NLP can identify this language embedded in the documentation, despite the fact that these tenders are not labeled as VBP. The language that GE has been seeing includes: an end customer with a focus on value, a transformed and developed procurement practice completely aligned to the customer and measurement systems (discussed below). O’Connor commented that despite this increase, the quantity is not yet enough to overcome the significance of pricing. Some tenders have 5-10% of criteria centered on outcomes but 50-70% is on price. VBP remains a rarity, with only some exceptions. Countries in Europe, including Wales, Spain and Scotland, are launching “Innovation Projects” as a part of the European Union’s Recovery and Resilience funding initiative. The focus of these discussions is care pathway reform, for cardiology and oncology specifically. Although currently difficult to achieve, O’Connor expects that VBP will increase in use over time. For example, when discussing the minor dent in the growth trajectory of non-price criteria that Vamstar noted, O’Connor was certain that the dip was a result of the pandemic and would be temporary. He emphasized that, to create sustainable health systems, we must move past price focus and into outcome-based focus, both of which VBP can achieve.

Expanding on his previous point, O’Connor spoke about the resultant impact VBP demand has had on innovation in the imaging product line. Currently, there is increased clinical accuracy, efficacy or patient flow in all of GE’s business units. Recent market surveys have shown GE a significant increase in the needs for operational efficiency and assistance of health networks in their strategic growth ambitions. The company has developed image-based and operational-based innovations. Operational improvements include: ensuring products are upgradeable – increasing the speed of implementation, reducing scan times with AI reconstruction (Air Recon DL, True Fidelity), reducing read times, errors and missed diagnoses with the Edison AI Orchestrator, reducing post processing times with improved advanced visualization and providing fast image acquisition to ensure productivity gains are achieved. Patient-centered improvements focus on: reducing patient absences and opening additional exam slots with Smart Scheduling – allowing more appointments and greater access to care, using encounter-based procedures to capture missed reimbursements and using unique product solutions like MR Air coils improve patient experience and comfort.

O’Connor further commented that, to deliver VBP contracts, there must be VBP digital capture of outcomes. Doing so would provide correct quantifications of achievable outcomes, thus ensuring that the value delivered meets expectations. GE has developed the tool, Imaging Insights, for this purpose. It gathers exam data from imaging machines and combines it with RIS data to deliver actionable intelligence; resulting in improved equipment utilization, patient dose exposure, schedule availability, referral patterns, and staff training.

Pivoting AI from trial to monetization, improving the patient experience
We also spoke with William Lacy, SVP Medical Informatics for Fujifilm. He told us that while Fujifilm is developing its own engines for analytics, most of the third party AI is still in trials. They have also been working to build bridges between third party vendors and buyers, integrating AI directly into PACS, organizing third party vendors, and presenting results to physicians within their workflow. The applications radiologists are requesting for AI are those that enable case prioritization. Many of these engines, according to Lacy, are in trial phases and not yet monetized. Vendors want radiologists to test them to identify their value and then gauge their willingness to pay. That activity is starting to ramp up, with engines getting into community hospitals. Here, many radiologists want to put screening programs together; Lacy says this is clearly an opportunity to deliver big value. Customers want to be able to, for instance, conduct a chest screening, a lung nodule detection; they want multiple applications of AI not discrete functionality.

Eventual monetization is obviously important for AI vendors. Working with Fujifilm PACS provides third party vendors with validation, which improves their likelihood of landing contracts. There is fear of redundancy however, and many AI engine vendors are struggling with this. Issues for them include small vendors focusing on the same application area and creating more competition. In addition, FDA approval is expensive and larger vendors are far more capable of funding the process. Despite monetization issues, Lacy says, “I believe the most important part is the actual value for the radiologist, the physician and the health system, and that's what I think is being explored right now”.

Fujifilm is looking at applications related to patient engagement and revenue cycle management; tying the entire continuum from registration all the way through to the report delivery in billing and covering all the points in between that can affect ROI. Lacy emphasized the importance of the patient experience, saying, “scheduling, getting to the office, registration, wait times, how much were they able to do in advance, do they like using the systems, did it make the imaging center experience easier?”. Because the US is a high deductible insurance environment, it would also be useful for the patient to have exam estimates upfront. This would allow them to make provider decisions and then to pay at a kiosk at the time of the study. Fujifilm is involved in technologies that improve the overall patient experience, particularly for outpatient environments, but in also identifying how much the entire continuum of care costs for the provider. This includes identifying opportunities to save money such as, better engagement with patients, bringing in money faster, using staff more efficiently and understanding the overall ROI impact. A product note for Fujifilm at RSNA is the Velocity MRI System. It is an anatomy-conformable, high-field open MRI system designed to streamline workflow and improve the patient experience using its integrated radiofrequency coils and reconstruction technologies. According to Fujifilm, the open gantry offers a more a spacious feeling for claustrophobic, bariatric, geriatric and pediatric patients.

Bringing advanced image processing into price sensitive markets
Vamstar spoke with George Curley, Director of Medical Imaging Marketing North America for Agfa, about the company’s new Valory digital radiography (DR) system. “The idea behind Valory is for Agfa to address the needs of productivity in every market sector…we created Valory to be an affordable solution that still has all the productivity gains of Musica’s image processing”, said Curley. Agfa customers say their exams are completed up to four minutes faster with Musica supported casework; “we want to bring that experience into the more price sensitive marketplaces”. Valory has a single interface and features a 15-hour battery, ceiling-suspended X-ray tube, fixed or elevating table, and cesium iodide detectors—combined with Musica image processing the system enables dose reduction of 30% to 50% according to the company.

When asked if the company had seen increased price sensitivity since the start of the pandemic, Curley noted that there was a rush in 2020 for very specific technology, such as mobiles, with a more recent shift to the value-based care areas. But the value emphasis has actually been in the works for about three years, prior to the pandemic, but is aligning well with the Valory launch. While the big centers mostly have the equipment they need, the smaller facilities are struggling to keep up, so in that regard “the answer is yes, but a qualified yes”. Regarding VBP and outcome-based criteria, Curley said that the total cost of ownership has been a key consideration with Valory, accounting for the hardware, service and remote support. It also spans patient satisfaction, with decreasing exam times meaning less time on the table and a quicker diagnosis.

First time right, more clinical value from existing technology
Annemarie Blotwijk, MRI Global Marketing Director for Philips, said that as imaging volumes continue to increase, there is limited time to provide feedback to new techs. So, Philips has developed the Radiology Smart Assistant, an AI-based solution to improve accuracy of X-ray image acquisition. This solution helps improve acquisition accuracy for upright chest X-ray exams through continual quality analysis and feedback about position accuracy, given at the point of image acquisition. The continuous feedback provides X-ray technologists with consistent, immediate visual feedback for every chest exam. It can help improving performance of well-educated X-ray technologists through positive reinforcement. Blotwijk summarized, “what we see is that there is a big need in hospitals for coaching guidance, better results, first time right. And based on that we have developed the Radiology Smart Assistant”.

In the same vein, the message from Konica Minolta at RSNA was about getting more information, more clinical value, from existing technology. Guillermo Sander, Director of Digital Imaging Innovation at Konica Minolta said that one of their innovations is dynamic imaging. This is a series of standard 2D X-ray images acquired at high speed and very low dose which can be played back as a loop, to provide body motion. In one example, this allows for an orthopedic doctor to analyze joints in movement in a weight bearing position. Sander noted that modalities like MRI have explicit spatial resolution, but they do not provide temporal resolution. What happens as you move your shoulder, your elbow, or when you breathe? In standard imaging the radiologist may see that the patient's lungs are at 70% typical capacity, but not if the right lung is working near full capacity and the other one is working on 60% capacity. Sander explained, “when you see the movement of both lungs working you can see that one is obstructed, so using the same X-ray machine we are getting much more clinical information that allows for better clinical decision-making before moving to another modality”.

At the Shimadzu booth, Wayne Stecker, Applications Manager for Shimadzu, provided a demonstration of the company’s SLOT Advance technology. It delivers images with long fields of view – such as for full spine or full leg images – by utilizing a simple workflow with high accuracy and minimal X-ray dose. SLOT Advance collimates the X-ray beam, so that it only exposes a narrow central slit field without distortion caused by oblique rays. These central slit images are captured using the company’s Sonialvision image processing and hi-res FPD by moving it in parallel with the X-ray tube. These successive images are then automatically reconstructed to create one long image in the Sonialvision G4 digital imaging unit (DR-300). This enables accurate measurements without distortion, while utilizing a simple workflow.

2021’s RSNA conference highlighted the potential for leading value-based care initiatives from the radiology department. Across multiple companies, AI continues to be a tool to move forward, enabling faster, more accurate imaging, better use of staff, and improving the patient experience. And while there were certainly many examples of innovation in high-end imaging systems, there was also much interest in using AI, analytics, and advanced visualization solutions to help deliver more information and clinical value to radiologists using existing or affordable equipment.

About the author: Shane Walker is a senior director at Vamstar Ltd., an AI-based B2B healthcare and life sciences marketplace. As part of his role at Vamstar, Mr. Walker serves as a healthcare industry analyst and subject matter expert with a wide range of system level and component level technology coverage including investigative, diagnostic, and therapeutic equipment and software for clinical, ambulatory, and laboratory environments. Prior to joining Vamstar (www.vamstar.io), Mr. Walker served as a lead consultant for healthcare projects at IHS Markit. He can be reached at shane.walker@vamstar.io.