By Don Dennison

Over the past couple of years, IT groups have been challenged to provide new value for organizations forced to adapt their enterprise to operate with staff working remotely and with sudden and wide variations in business workload. Imaging IT is no different. While imaging modalities have remained on-premises at healthcare provider facilities to acquire imaging exam data, many of the functions, such as diagnostic reading, have moved to work remotely.

Working as an imaging IT consultant, and volunteer with many imaging informatics organizations, I am often exposed to the evolving needs of healthcare provider organizations, as well as how industry is adapting their solutions to meet these needs.

I attended the RSNA in Chicago in person in 2021, meeting with several vendors, along with healthcare providers seeking new imaging IT solutions, like a PACS, VNA, or other.

This article provides a summary of trends I have observed not only at the RSNA, but from discussions throughout the past year or so.

State of RIS
In early 2020, I wrote an article on the two prevailing types of RIS: stand-alone and as part of the EMR. A year later, and the observations still ring true. Healthcare systems are overwhelmingly opting for a RIS provided as part of their EMR (for which the healthcare system already has a contract in place), sharing many of the same database tables (with fewer interfaces to build) and application servers. They use the embedded RIS to place imaging exam orders, schedule appointments, and support the department’s operations. Depending on the solution, it may also provide the DICOM Modality Worklist and the Radiologist’s reading worklist. While standalone RIS may provide more functions, the appeal of maintaining a single system often wins out.

Standalone RIS continue to evolve and provide many modern features to operate in today’s environment, such as automated appointment reminders and exam prep instructions by text message, or apps (or text message-based options) to allow patients to wait in their car until it is time to come in for their appointment. Analytics in these solutions are often robust, as the buyer — such as a chain of imaging centers — is often counting on this data to optimize their productivity and revenue stream.

State of reporting
For years, I have heard thought leaders extol the benefits of structured, multimedia reporting for radiology. While more common in cardiology and the breast imaging subspecialty, this has not gained wide adoption in the broad Radiology community. Some of the challenge is radiologist habits (many were trained to produce narrative reports), but there have been real and significant technical challenges, especially in a mixed-vendor solution environment. Developing highly structured reports has been possible for some time (although getting Rads to all agree on and adopt this structure is not always easy), but getting the structured values from the modality and PACS (created during the diagnostic review) into the right report fields was limited, due to a lack of APIs to pass this data in real time. The same can be said of embedding images, tables of values, or other multimedia content into the report. Single-vendor solutions that provide methods to receive and process structured data from the modality, and provide structured reporting and image display are available on the market, but most organizations have adopted separate applications for image display (PACS) and reporting.

As new web-based APIs, such as HL7 FHIR and FHIRcast, become mature and adopted by industry, the creation of the new form of multimedia report becomes more feasible.

Challenges remain, however.

Persistence and distribution of Radiology reports in a form other than plain-text HL7 ORU messages is technically possible. But interface, EMR, and other application teams are not always ready to adopt them. The tried-and-true (yet limited) method of a v2.x HL7 interface for results distribution remains common.

Another challenge is standardizing the values within the structured fields. Structure without standardized terminology or coded values prevents data analysis at scale, such as what is required for disease-specific surveillance and public health.

State of PACS
This is a big topic, so I will break it down into several areas.

Vendor strategies for solution scope - Industry builds solutions based on their prediction (or observation, if they are reactive to market trends) of where customer needs will develop or appear to be headed.

Vendors will invest R&D developing different modules of a solution — such as a reading worklist, reporting module, or image sharing solution — if they believe customers will want to source those functions from them.

One area that has received lots of attention over the past few years is enterprise imaging, including imaging data from phones, scopes, and dedicated ancillary department equipment in service lines such as wound care, dermatology, and pathology. Some vendors have made significant investments in functions to capture and manage non-diagnostic, non-DICOM data within the PACS. Other vendors have invested less, resulting in a wide variation of capability across solutions in this area.

When health systems make buying choices, they are often confronted with a choice between the preferred solution for Radiology and the preferred solution of those looking to use the new system to manage all the enterprise’s medical imaging data.

Each enterprise will make their own choices based on their strategy and the decision-making roles involved, but choosing a solution that is not preferred (or even accepted) by Radiology is a risky option.

Cloud - Vendors also vary widely on their adoption of cloud-based hosting of the PACS. Popular solutions range from on-premises, an option for either on-premises or cloud-based deployments, a hybrid with some parts being installed on-premises and some in the cloud, and some that are only available in the cloud.

Buyer strategies are equally diverse, with differing attitudes and priorities when it comes to adopting cloud-hosting for their next PACS. Even within a healthcare organization, attitudes can vary widely, from skeptical among IT infrastructure management staff to curious (or even eager) among IT executives.

In enterprises where clinical stakeholders perceive the value of their organization’s IT staff to be low or unreliable, there is often a desire to “outsource” the IT infrastructure to the PACS vendor (who may be brokering the hosting to a third party or providing it themselves). Internal-facing IT organizations will need to be very customer-service focused and responsive to compete with these new options.

Assessing cost over time, such as calculating a Total Cost of Ownership (TCO), is important in determining both projected capital and operating expenses. Comparing these to existing, internal costs — including IT staff labor — is important to determine the benefit of cloud-hosting to an enterprise.

State of AI in imaging IT - This topic is too big to cover in depth here (and would be outdated by the time the article is published, with the speed of change in this area), but here are a couple of observations.

If you ask PACS (or other imaging IT solution) vendors about artificial intelligence (AI), be prepared to hear about their “AI Marketplace”. This is a term used to describe a strategy of using a variety of methods to incorporate the vendor’s own-developed AI-based apps along with some from third parties. Some third-party, AI-based apps may be resold by the PACS vendor, and for others only an integration may be offered (the buyer needs to source the third-party app for themselves). Almost every vendor has a roadmap with “several announcements pending”.

With investment in early stage, AI-focused imaging IT solution startups at a high level, we should anticipate some consolidation (and outright failures) in the marketplace over the next couple of years.

When I ask buyers what specific problem they want an AI-based app to solve, they often lack a well-defined response. In general, they want whatever value proven AI-based apps can provide. They often lack a defined process to validate and assess the value of adopting an AI-based solution.

Pricing Models - Another rapid change in PACS is the model by which the solution is priced. Historically, most PACS were purchased as a capital expense — including the software license and system implementation services — with recurring annual Support and Maintenance Agreement (SMA) fees as an operating expense. Some vendors offered solutions in more of a “subscription” model with the most common approach being a per-exam fee.

Many vendors are shifting to offering a subscription model. This reduces the upfront capital expense for buyers, but increases their operating expenses. The finance process of some enterprises are not prepared for this shift, while others (those with limited capital now) welcome it.

One obvious advantage of subscription models is that the cost of the solution scales with the business; costs are correlated to the number of exams acquired (and, thus, revenues) each year.

Some aspects for buyers to consider are whether:
• Imported exams (no revenue) count as part of the per exam subscription fees; this is an important and often under-discussed topic.
• Enterprise imaging content (often with no revenue) stored also counts as an exam
• There are terms defining limits on per-exam cost increases over time

Other observations and trends
Here are some other thoughts on imaging IT:
• Academic radiology is increasingly investigating the option of providing subspecialty reading to other enterprises than their own. Affiliated hospitals that are dissatisfied with their private reading group often express interest in contracting with an academic radiology department to provide these services. They are interested in imaging IT solutions that support this type of cross-enterprise reading without having to install PACS (and reporting and other solution) clients for each hospital.
• Organizations are looking into new options for automating imaging exam data importation using an HL7 FHIR API provided by their EMR to auto-generate an appropriate order to link to the imported images. While not as captivating a solution as one based on AI, solutions like this will save millions of hours of manual effort across health systems.
• At RSNA 2021, I found startup companies build new PACS solutions. I have often felt that there were already plenty of options available in a mature marketplace (like the U.S.), where replacement and consolidation are the remaining growth opportunities. At first, I questioned the wisdom of this investment, but many large imaging modality and IT companies have a habit of failing to foster and deliver innovation, opting to acquire companies every so often to provide something appealing to customers. With disruptive technologies like web, cloud, and AI, a startup needs only to reach a reasonable level of maturity and adoption and they can expect a larger company caught flat-footed to get out their check book. Of course, the pace of innovation will die shortly after acquisition (life in a big company is not what attracted the talent to the company in the first place, and leaders have cashed out) and the cycle will repeat.

About the author: Don Dennison, CIIP, FSIIM, is a consultant that has worked in the medical imaging informatics industry for over 15 years. A Fellow of the College of the Society of Imaging Informatics in Medicine (SIIM), he is a frequent speaker and panelist at SIIM, RSNA, and other conferences on topics ranging from medical imaging record interoperability, integration of imaging data within the EMR, multifacility integration, and others.