Integrating Imaging Tools Helps Radiology AI Deliver Real Value
A standards-based approach is key to successful solutions interoperability
Although often overlooked, interoperability is the unspoken hero of health care. It enables two or more different systems to exchange and use data, playing a crucial role in driving efficiency and improving patient care.
In fact, without interoperability, there would be no electronic health records (EHRs) sharing data across practices, labs sending results to hospitals, or prescriptions being electronically sent to pharmacies.
“Interoperability makes things simpler, scalable and translatable for both the vendor and the consumer,” said Ali Tejani, MD, a radiologist at UT Southwestern Medical Center in Dallas. “It’s the difference between manually reporting measurements or AI results versus auto population of information in our radiology reports.”
Making interoperability possible are standards—rules, guidelines and characterizations established and approved by consensus from a recognized body. Two standards that enable interoperability in radiology are Digital Imaging and Communications in Medicine (DICOM) and Health Level 7 (HL7).
“These standards are what makes storing, transmitting and viewing medical images and related information coming from different medical devices and systems possible,” said Kent Hutson, MD, CPE, co-chair of Integrating the Healthcare Enterprise (IHE) Radiology Planning Committee and artificial intelligence lead technical consultant at Radiology Partners.
But standards alone are not enough.
“There can be more than one way to apply the standards to a given problem, making it almost inevitable that vendor A and vendor B will have made different, incompatible choices,” Dr. Tejani explained.
Solving this incompatibility requires that these standards be integrated, undergoing the process of ensuring that products coming from different vendors can achieve a common understanding of the shared tasks and a coordinated implementation of all relevant standards.
Helping to make such integration possible is the IHE.
A Roadmap Toward Interoperable Solutions
An initiative established and supported by RSNA and the Healthcare and Information and Management Systems Society (HIMSS), IHE looks to address persistent obstacles to the effective integration of information systems.
“IHE doesn’t create standards, it tells us how to apply existing standards—DICOM, HL7 and others—to specific use cases,” Dr. Hutson explained. “In doing so, it provides a script that developers can use to design interoperable solutions.”
If a developer follows the provided script and data format, which IHE refers to as ‘profiles’, then the developed solution will be interoperable with any system (e.g., electronic medical record) that supports that same profile.
“For system vendors, an IHE profile serves as a roadmap showing them how to play their role in a multivendor solution,” Dr. Hutson added.
To make it even easier for company engineers to support the product development and testing processes, IHE regularly organizes Connectathons. These events serve as stress tests, offering a place where vendors can work to test and fine-tune their codes, with the aim of having them certified as being compliant with a particular IHE profile.
AI Poses New Challenges to the Integrated Imaging Workflow
The net result of IHE’s standards-based approach to integration has been a streamlined radiology workflow, starting from a patient’s initial imaging request to the delivery of the final report and everything in between.
The deployment of AI solutions in radiology practice is creating new challenges to this integrated imaging workflow. “Any time we add new software and systems into the workflow we create an added need for interoperability and the introduction of AI is no exception,” Dr. Tejani explained.
The good news is there’s no need to start from square one. “We may not have achieved interoperability for AI integration yet, but what we do have are DICOM, HL7 and other established standards, which are the key to integrating AI into the imaging workflow,” Dr. Tejani added.
The challenge is getting companies to adopt these standards. “There are a lot of AI companies getting into the medical arena for the first time, meaning they are either unaware of our standards or lack any meaningful experience working with them,” Dr. Hutson said.
The result is that there are many products entering the market that do not support radiology’s established standards. This makes it difficult, if not impossible, for radiologists to monitor the AI, integrate the AI results into the workflow and use the data in an interoperable way. In other words, it disrupts the workflow.
“We need AI vendors to adhere to IHE profiles so that their solutions work with ours,” Dr. Hutson added. However, for that to happen, radiologists and health care administrators—the people who work with these systems—must demand that vendors provide standards-based solutions.
A good place to start is with the request for proposal (RFP).
“If a health care system wants to be able to do all radiation monitoring in a standard way so it supports registries, etc., instead of specifying the details, they just need to require that any vendor applying to the RFP support a specified IHE profile,” Dr. Hutson said.
“Seamless integration of AI models means knowing exactly how systems interact in the imaging workflow, and the only way to know that is by getting feedback from the people who will use AI in practice.”
ALI TEJANI, MD
Radiologists also play a crucial role in the integration process, providing clinical expertise to create appropriate frameworks for AI integration and their translation into practice. “Seamless integration of AI models means knowing exactly how systems interact in the imaging workflow, and the only way to know that is by getting feedback from the people who will use AI in practice,” Dr. Tejani added.
Dr. Tejani noted there are numerous opportunities within the IHE profile lifecycle for providing feedback about one’s experience using a particular AI solution. “This information is absolutely invaluable to the development of IHE profiles,” he said.
No Longer an Issue of Why, But How
While AI promises substantial productivity gains by automating tedious and time-consuming tasks, turning this promise into practice requires that AI solutions be seamlessly integrated into the existing radiology workflow.
“We know why we need interoperability; the challenge is figuring out how to achieve interoperability while integrating AI technologies into the workflow,” Dr. Tejani concluded. “But with the right champions, organizational support and the active involvement of radiologists, clinicians and vendors, I am confident this is a challenge we will overcome.
For More Information
Access RSNA data tools and standards resources.
Learn about the Radiology Reimagined: AI, innovation and interoperability in practice demonstration happening at the RSNA annual meeting.
Read about the RadLex 20th anniversary.
Learn more about the work of the IHE.
RadLex: Two Decades of Advancing Radiology Standards
The comprehensive radiology lexicon developed by RSNA is celebrating its 20th anniversary.
RadLex has been the gold standard for radiology terminology, transforming how medical imaging is described, reported and analyzed. Today, it has evolved into a robust, structured lexicon encompassing over 75,000 terms, ensuring consistency in image annotation, structured reporting and AI applications.
As RadLex celebrates this milestone, it represents the power of standardized language in radiology—bridging past innovations with the future of precision medicine and intelligent imaging. Learn more at RSNA.org/RadLex.