HealthCareBusiness asked me to provide an article on some of the trends that I have been seeing in Imaging IT, including RIS, Reporting, and PACS (including Enterprise Imaging, Cloud, AI, and Pricing Models), along with some other general market trends.
If you prefer to read it in the magazine’s reader app (runs in your browser), you can use this link.
Acknowledgements: I would like to sincerely thank Dr. Alex Towbin (@towbinaj), Ryan Fallon, and Jason Nagels (@jaynagels) for their time to review this information and provide valuable input.
Important Note:This article is in no way intended to dissuade organizations from sharing imaging exam data. As described within, there are many benefits to everyone, and sharing of imaging data should be consider a best practice. This article is intended to explore how vendors charge differently for storing imported imaging data in their systems.
The sharing of imaging exams across different enterprises has long been of value to patients and healthcare systems. From teleradiology, patient transfers, consults, treatment planning, to other scenarios, avoiding the repeat of an exam by sharing existing ones lowers costs, minimizes stress and frustration on the patient (and their family), and avoids any unnecessary radiation exposure.
As technology and reliable Internet has evolved, the norm of printing film, then exporting digital exams to portable media (CDs, DVDs, USB drives, etc.), has given way to cloud-based image sharing solutions. With these solutions acting as a secure broker among enterprises, imaging data can be easily moved from one organization to another. Patients can even upload their exam data using a webpage.
And while the managed movement of imaging data from one organization to another has become much easier and more common, the importation of the data to systems, like the PACS, at the destination enterprise often still requires some manual activity to get the exam data acquired at another enterprise to include all the unique data identifiers used at the destination enterprise.
The costs of the above activities are all well understood by imaging professionals, but there is another aspect that is less understood: Does the receiving imaging IT system vendor charge for imported exams?
Imaging Exam Acquisition Revenues and PACS Costs
In most jurisdictions (with private or public funded healthcare), a fee is paid by the payor for each imaging exam performed. This fee, commonly called the Technical Fee (or TechFee) in the U.S., is intended to compensate the health provider for the labor, equipment, consumables, facilities, IT systems, and other resources required to perform the exam and manage the resulting data. The TechFee amount will vary among different procedures and payors.
Note: A Professional Fee (ProFee) is also billed, but these funds may go to pay the Radiologist group, if they are not directly employed by the health system. So, even if an imported exam is read (for example, as an “over read”) and a ProFee bill generated, in some cases, the health system paying for the PACS does not receive any revenue (but do incur costs, as detailed below). Not to mention, ProFee billing for over reads can be unpopular among patients and referring physicians, as it increases overall cost of care.
It is common for PACS vendors to charge a software license for the use of the solution. The commercial model may vary between a licensed annual exam total allowed or a per use fee, but in most cases, the more exams that are acquired and stored by the organization, the more the vendor is paid. This is generally accepted as fair by both vendors and health system buyers.
Where there can be variability in vendor agreement terms is in deciding which exams are counted when determining compliance with the license terms or fees for usage. As the health system receives a TechFee for any exams they acquire, both vendor and health system provider are generally in agreement that these exams should count towards the software license.
But what about imported exams?
Some vendors treat all exams – acquired and imported – as equal and they are both included in the license compliance or usage calculations, even though the health system typically receives no compensation for imported exams. Some vendors may charge a lower fee for imported exams, while others include terms in the agreement that specify that only exams acquired by the health system, and not imported exams, are counted.
If multiple applications are involved – for example, a cloud-based image sharing solution, a local system for reconciling the imaging data with a local order, the PACS, and a Vendor Neutral Archive (VNA) – there may be the terms of solution’s software licenses to review and consider.
Do you Need to See the Images or Have the Images?
Before exploring how imaging IT vendors may count exams when determining system usage for software license calculations, let’s discuss some different ways in which imaging data are accessed and shared.
Some enterprises provide a web-based image viewer, often as part of a referring physician or patient portal, to allow authorized users to see and navigate images and results. Cloud-based image sharing solutions also allow images temporarily stored to the solution to be viewed by authorized users. Similarly, Health Information Exchanges (HIEs) that support image data exchange usually also provide an image viewer to see the images. Common among all of these approaches is the ability to see and interact with the images without having to move the DICOM objects to the information-consuming enterprise, or having to reconcile the data to local identifiers or terms.
This is an efficient way to “browse” imaging data managed by an outside system and organization, but is often not efficient if advanced processing is required (if the image viewer lacks the features) or if the imaging data need to be closely compared to data stored in other systems (like the local PACS). While this method may suffice for some clinicians, Radiologists likely prefer the data to be presented in their primary system (linked to the local Patient ID and all other imaging records): the PACS.
In some cases, the DICOM data is moved to the local PACS and stored temporarily. These data are available for viewing using all the PACS tools, but are not stored to the long-term archive. The system purges the data from the PACS cache based on any configured retention rules. These exams are often reconciled to local identifiers, like the local enterprise’s Patient ID, to make comparison in the system easy, but not all organizations will localize these data (see note below).
Many organizations will apply a specific labelling, such as a prefix or suffix character or string in the Study Description, to denote that the exam was acquired elsewhere. This can preempt any confusion if the acquisition protocol or image parameters are not consistent with the local policies. When used consistently, this method may also help identify the number of imported exams within an imaging IT system.
This approach is common where images are shared with an organization, through an image sharing solution or portable media, and the local PACS is the preferred viewer (due to its advanced tools and/or system familiarity). It is also used where Radiologists need to review the imaging data for scenarios like teleradiology or consults, but the patient is not being transferred or admitted to the local organization.
Like Guest Studies above, these exams are received, stored, and (typically) reconciled to include local identifiers (for example, Patient ID and Accession Number) and terms (for example, an updated Study Description). They are archived to the PACS or VNA and made part of the local enterprise’s patient medical record.
This is common where the patient is going to be transferred to the local enterprise or will be getting diagnostic or clinical services, such as a follow-up imaging exam or surgery, and the enterprise wishes to retain these records for future reference.
Documented organizational policies and procedures that define when and how outside imaging exams are accessed and imported should provide guidance to staff and affiliates.
Determining whether an outside exam is beneficial or necessary to import requires significant knowledge. Often, a Radiologist is required to make this decision, but in some cases an experienced imaging analyst with deep Radiology knowledge, like a senior Technologist, may be able to make this determination in many cases.
For Guest Studies and Archived Exams, if the data are not reconciled to local identifiers, there are some risks that the outside identifiers match local ones, which can result in a potential patient risk (exam linked to wrong patient). It can also cause issues when searching for data (some demographic values as stored in the DICOM data need to be known), resulting in some clean-up activities during a data migration.
Exam Counting Methods
Many vendors run a query against the imaging IT system to collect a count of all imaging exams stored to the system over a period (for example, a year). They count each exam that has a unique Study Instance UID value (a required DICOM attribute). This method would include all stored exams, including imported exams.
Another method is to have the health system produce a report from their RIS that provides a count of all imaging exams performed by that enterprise. This list could exclude imported exams and would generally represent exams for which the health system received revenue in the form of TechFees. The vendor would have to trust the health provider’s report, but a sanity check against the imaging IT system’s database would suggest if the information were within reason.
How Much Difference Does this Make?
So, is this aspect worth looking into? Let’s look at some numbers.
Imagine a large healthcare enterprise that stores a total of 1 million imaging exams to their PACS each year. As an academic system with lots of affiliations, this enterprise has many patients in the area routinely referred for additional imaging and treatment.
Their vendor charges $1.50 per exam stored to the system.
If we assess different percentages of the total exam volume that is imported, we see that the cost to store imported exams could vary from $150K to $375K.
Keep in mind that the health system has no TechFee revenues to offset these costs (though there may be other clinical or diagnostic services that the patient receives for which the health system can bill).
The Model Matters
In the above case, if the vendor licenses the software based on a total number of annual exams, as long as the total annual exams stored does not exceed the amount currently licensed, no additional software license needs to be purchased. If the vendor counts imported exams in the total, and image sharing increases, it may require the health system to purchase additional licenses (for example, to allow another 50K to 100K annual exams to be stored), even though they receive no revenue for the imported exams.
If the vendor licenses the imaging IT solution on a per-exam usage fee, and they count imported exams, the health system would have to pay the vendor for each exam imported (at whatever rate they specify in the agreement, which may be the same as an acquired exam, or at a lower rate).
Other Costs and the Cloud
Today, most healthcare providers procure and manage their own hardware (for example, servers, storage, and network equipment) and the solution vendor supplies the software, technical and professional services, along with ongoing support. So, in addition to any software license costs paid to the imaging IT vendor, the health system also must supply the hardware to store and manage the imported exams, and the labor to perform the importation. If a data migration is performed sometime in the future, the health system will have to pay for any imported exams that are included in the migration.
This is all in addition to any fees paid for an image-sharing solution.
The above details are important to consider if a cloud-based PACS solution is considered.
When a system is on-premises (for example, hosted in the health system’s data center), the vendor may be willing to acknowledge that the imported exams do not count towards the total exam volume when calculating the software license, but when they also provide all the hardware infrastructure, they will incur costs for all data stored. This may result in some fee being charged to the health system.
Sidenote: Enterprise Imaging
While beyond the focus of this article, many of the same questions should be posed when storing clinical images, also called Enterprise Imaging (EI) content, to the imaging IT system. Most of these data sets are not captured in response to an ordered service and have no billing associated with their capture or storage. If the imaging IT solution vendor counts these data sets (which may be one or more image, video, audio file, document, etc.) as an exam when determining software license compliance or system usage, costs (with no associated revenues) could be higher than expected.
Sidenote: Labor for Importing and Reconciling Exam Data
Another significant cost of image-sharing is the labor involved in reviewing, importing, and reconciling the imaging data to use local identifiers. This cost is not hidden, but is not measured by most enterprises, and these costs can be larger than many people understand. Industry vendors have worked on solutions to auto-generate the necessary order information to perform reconciliation automatically. New data standards, like HL7 FHIR® (https://www.hl7.org/fhir/), and APIs that support them developed and deployed in EHRs make this approach more feasible. Careful attention needs to be paid to how data elements like Study Description are applied in the reconciled data, as how procedures are labelled vary among enterprises. Inconsistent Study Description values can affect relevancy rules for pre-fetching and routing and may necessitate maintaining complex terminology mapping tables (if supported by the imaging IT solution). Series descriptions, which often affect features like display protocols, also vary among modalities and enterprises. Terminology normalization is a complex challenge involving many unmanaged data variables that some believe may be solved more effectively through the application of Artificial Intelligence (AI).
Tips for Minimizing Costs
The first step is understanding your current situation with the following factors to consider:
Do an inventory of acquired (check the RIS) and imported exams for each year for the past three years. Understand how many exams you are importing (with no offsetting revenue) and look for any trends. Are exam imports increasing year-over-year?
Read the agreement with your current vendor to understand the terms around how exams totals are counted. Understand whether they differentiate between acquired and imported exams. Understand whether Guest Studies (as described in this article) are counted, even though they are not archived.
If your imaging IT vendor does count imported exams when calculating the software license or usage, make sure you account for these costs when budgeting capital and operating budgets.
Understand the fees for any image-sharing solution you use. Does the vendor charge a different amount for exams that are viewed from the cloud, but not imported to the local PACS? Do they charge both the sender of the data and the receiver, or just one of the parties?
Review your policies and procedures for exam importation. Are they clear so that the necessary exams get imported, but ones that are not relevant are not?
If your organization is looking to replace an imaging IT system, like a PACS or Vendor Neutral Archive (VNA):
Ask the bidding vendors how they determine which exams are included in the count for software license calculations. If some do count imported exams and some do not, include an estimate of the additional costs for those that do in your Total Cost of Ownership (TCO) analysis.
The purpose of this article is not to dissuade enterprises from sharing imaging exam data. As stated, there are many benefits of this activity to all stakeholders, most notably the patient. The purpose here is to explore variability in how imaging IT solution vendors count exams for software licensing, and how this can add up to represent significant costs, over time.
It would be unwise to select a new imaging IT solution simply based on whether they count imported exams toward the software license or not, but it should be considered in your cost projection calculations.
And it may be worthwhile to educate your imaging IT vendor partner on the economics of providing diagnostic imaging services and to align their compensation to reflect the scale of the enterprise’s revenues.
When IT professionals from Radiology and other image-generating departments merge with the enterprise IT team that provides IT infrastructure and support for enterprise-wide solutions, employee angst and culture friction can occur.
The same thing can happen when imaging IT teams historically working at different sites are merged to a single group, or when imaging informatics professionals (IIPs) from different healthcare service lines, like Radiology and Cardiology, are put into the same team.
Over the years, I have found that many working in imaging informatics understand the basics of how patient identity is managed in healthcare information systems. In today’s Consolidated Enterprises, where the historic imaging exam data for a given person may be indexed by many different patient identities, and perhaps cross-referenced by an Enterprise Master Patient Index (EMPI), ensuring that all a patient’s records are accurately associated with the correct Patient Identity (ID) requires understanding of the information models and values that may be in use at the enterprise. Along with the strategy and vision for how Patient IDs will be managed in future.
New data acquired in other enterprises continues to come into an enterprise’s imaging IT systems. From data migration associated with system consolidation (for example, as part of organizational merger or acquisition activities) to programs that involve data exchange—such as importing outside exams prior to appointments, teleradiology, or the sharing of systems with affiliates—ensuring that your imaging IT systems can safely and reliably handle all the potential complexities is important.
The intent is to establish a simple, hierarchical model by which both healthcare system operators and vendors can easily match needs with capabilities.
In addition to the proposed model, the paper provides an overview of many patient identity management concepts, data standards, along with challenges and potential methods of addressing them. Those interested in simply learning more about patient ID management in imaging informatics and IT systems may find the paper informative.
Medical Imaging Informatics and Teleradiology (MIIT) is once again virtual, but the format of this year’s meeting is new. Instead of the traditional full-day of content, MIIT21 will be organized into half a day’s content available in spring and again in the fall. Sessions will be pre-recorded, with a live faulty roundtable for each session.
There will also be a live Keynote session on Thu 25-Mar-2021 at 4 pm ET. Dr. Rasu Shrestha, Chief Strategy and Transformation Officer at Atrium Health, will give a talk titled “Confronting Reality – Reimagining the Next Normal for Healthcare”.
The live keynote and two faculty roundtables will be recorded and available to registered attendees after they are complete.
Our speakers will cover topics such as AI and Fairness, updates in data standards, updates on gender in HL7, AI deployment, research programs in data science, global health outreach, structured reporting in Enterprise Imaging, and an expert panel on the progress and vision of Diagnostic Imaging Repositories (DIR) in Canada.
As someone that has mostly worked from home for over seven years, I understand the value of digital communication and collaboration. In healthcare, we have had many digital tools for more virtual experiences, but have been slow to adopt them. The COVID-19 pandemic has caught many organizations with paper-based and analog processes flat footed.
Education is no different than healthcare and other organizations that have heavily depended on in-person settings. While there is no doubt an in-person conference is more personal and provides social networking opportunities, content can easily be delivered online (the slides and audio are all digital, right?).
This year’s Medical Imaging Informatics and Teleradiology (MIIT) meeting was scheduled to take place in-person in Hamilton, Ontario, just like it has for well over a dozen years. Given the current situation we are all facing, my Co-Chair Dr. David Koff (@koff14) and I were faced with a choice: cancel or try to continue with an online version. As we believe strongly in the importance of imaging informatics education, and the value in this year’s excellent program, we chose the latter.
The first virtual MIIT will take place on Friday May 15, 2020. We have lowered attendee and sponsor rates in consideration of the lower costs of operating a virtual meeting.
We have a great program this year with Dr. Tessa Cook from UPenn, Dr. Vamsi Narra from BJC, Dr. Cree Gaskin from UVa, Les Folio from the NIH, Dr. Adam Prater from Emory, Michael Toland from UMMS, Ted Scott from HHSC, along with Kevin O’Donnell providing an update on the DICOM standard at MIIT 2020 and MIIT’s own Britt Tomlin providing an update on Alberta’s province-wide Connect Care CIS program.
At only CAD$80 per person (which, at the current exchange rate, is only US$57 for our American friends), and with accreditation for CE credits, it is very high value. There are no travel costs and imaging informatics professionals can join from anywhere in the world.
Register using the “Register Now” button at miit.ca.
Sponsorship of MIIT is still available, starting as CAD$1,000 (~US$715 with today’s exchange rate). Higher tier sponsors are given high visibility with attendees and the opportunity to give a brief talk during the lunch hour.
Be sure to stay top-of-mind among imaging IT decision makers and influencers (from anywhere now) to make up for lost contact at cancelled meetings, conferences, and trade shows!
Medical Imaging Informatics and Teleradiology (MIIT), the one-day educational meeting on imaging informatics, returns to beautiful Liuna Station in Hamilton, ON on Friday May 15, 2020. MIIT has been going for 15 years and this year has a great program with speakers from leading Canadian and U.S. institutions speaking on a range of timely topics, including hot topics like AI, multimedia reporting, change management, and building effective teams, as well as a panel discussion.
We are excited to have Drs. Tessa Cook from UPenn, Vamsi Narra from BJC, Cree Gaskin from UVa, Les Folio from the NIH, Adam Prater from Emory, as well as Michael Toland from UMMS and Dr. Christian van der Pol and Ted Scott from HHSC, providing talks in person or remotely, along with Kevin O’Donnell providing an update on the DICOM standard and MIIT’s own Britt Tomlin providing a talk on the Alberta province-wide Connect Care EMR project.
How diagnostic images get from server hard drives to the screen is a topic of great interest to both industry and buyers of imaging IT solutions. Speed of image access is critical for quality of service and care, along with productivity and user satisfaction.
Different solutions take different approaches to optimize image delivery over varying networks. Many solutions combine more than one software design method to achieve the best possible performance. In many cases, industry or buyers will use jargon, like “streaming”, to apply a simple term to these sometimes complex technical methods.
In a recent article by AXIS Imaging, I describe four common techniques that are used in (and occasionally between) different imaging IT systems to maximize image display speed. The article length limit prevented coverage of additional methods and intentionally excluded IT infrastructure optimizations (for example, faster networks, CPUs, and drives) and the use of irreversible lossy image compression of the images on disk.
Important Article Corrections
Although I followed up with the author about some transcription errors they made in preparing the article language, they were not corrected (at least at the time of posting this), so I am going to note some corrections here.
Where the article states “…radiology practices and departments have options when designing high-speed image display…”, it should state “…radiology practices and departments have options when choosing the solution for high-speed image display…”. Radiology practices choose a solution and that solution will use one or more of the design methods (or additional ones not listed), but the Radiologists don’t choose the methods within the solution.
In subsection #3, the second bullet refers to a condition where pre-caching is typically not possible (the article states the opposite). If the worklist is a separate application from the image display application, the image display application often has no method of knowing which exams listed are in the worklist, so is unable to pre-cache the images to the workstation. This is not always true, as some worklist applications can expose this information to the image display application through an API, but this is not a universally available capability and does require that a specific integration be developed to support this across the applications.
In subsection #4, where it states “visual design infrastructure”, it should instead state “virtual desktop infrastructure”. People commonly refer to it as VDI.
A lot of attention is paid to imaging IT systems, like PACS and VNAs, and EMRs these days, but Radiology Information Systems (RIS) play a very important role in the success of the Radiology service line within an enterprise.
The industry and market for RIS has changed a lot since their introduction, with two core markets (with different needs) evolving.
Remote reading is a hot topic for many working in diagnostic imaging. It is highly desirable for many Radiologists who want to telecommute like the rest of us. And, as cloud-based imaging IT solutions become more popular, the ability to read remotely (Rads won’t be going to the data center to read) will become the standard.