A structured, member-driven approach helps health systems assess value, protect PHI & scale AI responsibly
Radiology departments were among the earliest adopters of artificial intelligence (AI), using the new capability to automate image segmentation and measurements, detect critical findings and triage cases. That doesn’t mean, however, that adoption has been quick or without scrutiny. The rapid growth of AI has increased pressure on health systems to ensure new technology delivers measurable value, protects patient health information (PHI) and includes appropriate physician oversight.
St. Luke’s Health System is in the early stages of adopting AI. “We are approaching it in a thoughtful and intentional way,” says Kandis Pedersen-Romero, MSRS, RT(R)(CT), RDMS, Area. Director of Medical Imaging for the Idaho-based integrated delivery network (IDN). A few months ago, St. Luke’s stood up an AI advisory group to focus on enterprise governance and prioritizing targeted and high-value use cases for scalability of AI-enabled devices.
Pedersen-Romero, a member of the HealthTrust Radiology Advisory Board, joined the HealthTrust delegation at the 2025 Radiological Society of North America (RSNA) annual meeting to evaluate suppliers’ offerings and assess emerging solutions.
The mission for HealthTrust delegates
The RSNA conference in Chicago includes upwards of 40,000 attendees each year. HealthTrust brings internal leaders and radiology subject matter experts from its Radiology Advisory Board to meet with suppliers and evaluate technology. In 2025, the delegation selected and visited 18 suppliers to review current AI offerings and learn what is in development. The goal: Equip members with deeper insight before they make purchasing decisions.
“RSNA is always overwhelming, but doing our rounds with the HealthTrust Radiology Advisory Board was focused because of the agenda,” says Pedersen-Romero. She has attended RSNA previously and finds the structured discussions valuable. “The explosion of AI suppliers this year was very evident,” she explains, noting that she did not see any new radiology machines on the market that were not using AI.
George Puente, RT(R)(CT), MBA, Director, Medical Equipment, Design and Construction at HCA Healthcare, also says AI dominated the conversation at RSNA. He points to photon-counting CT as “the new chapter in computed tomography,” along with advances in cardiovascular imaging, voice-enabled workflow tools and reduced helium use in MRI systems. Those developments, he says, have implications not only for image quality and workflow, but also for facility planning, downtime and cost.
Radiology remains one of healthcare’s most active areas, with the FDA approving more than 1,000 AI-enabled products. “When planning a technology purchase, it is vital for health systems to understand the role of AI, where new equipment might overlap with existing products, and identify any enhanced benefits, so they do not pay for applications they don’t need,” says Catheryn Pasco, MBA, BS, RT(R)(M) RDMS, HealthTrust’s Imaging/Radiology Portfolio Director.
Addressing AI concerns
“Healthcare providers need to ensure that suppliers of AI-enabled devices are able to keep related data private and secure,” explains Beth Loughrige, MBA, BS, Senior Director of Strategic Sourcing at HealthTrust. In addition to data privacy concerns, governance, security and compliance should also be part of a thorough supplier evaluation. Contracts often require expanded terms and conditions to support the responsible use of AI and to safeguard PHI, Loughrige adds.
For St. Luke’s, access to PHI is a key concern. “Many AI platforms want access to PHI, so we have to approach that carefully,” Pedersen-Romero says. HealthTrust supports members throughout the supplier evaluation process. “We use HealthTrust for supplier vetting, which includes checking financial stability and evidence-based performance reviews.” These resources assess whether the supplier actually provides what it says it will provide.
Puente echoes that need for balance. At HCA Healthcare, he says, AI is being used in imaging for patient positioning, Y90 planning, lesion volumes, semi-automatic LV measurements and dose reduction, while multiple departments evaluate how tools can improve processes and streamline data collection. But he cautions against treating AI as a replacement for expertise. “Radiology is a science and an art,” he says. “AI can be a powerful addition, but it should be used in conjunction with current processes and ongoing education.”
He also notes that a structured evaluation process is essential. With so many products now marketed as AI-enabled, organizations need a methodical way to determine whether a tool truly adds value.
HealthTrust also shares benchmarking data and education on best practices, helping members evaluate supplier claims and providing summaries to explain operational requirements—facilitating faster adoption of AI-enabled devices. AI can improve workflow and radiologists’ reading time, but requires a lot of physician oversight. “We’re focused on being responsible and making sure the AI is sustainable. We want it to improve patient care without introducing unmanaged risk,” says Pedersen-Romero.
The power of aligned decision-making
“HealthTrust sourcing and contracting decisions are memberdriven, utilizing an aligned decision-making process,” shares Jennel Lengle, MS, BS, RN, Vice President, Supply Chain Board & Clinical Operations at HealthTrust. Members of our clinical advisory boards, in collaboration with HealthTrust Strategic Sourcing and Clinical Operations subject matter experts, recommend all products and services. Advisory Board recommendations are then presented to the organization’s Supply Chain Board for approval, Lengle says.
About 18 months prior to a contract expiring, the HealthTrust team begins supplier negotiations, shares Kathy McCardell, BSN, RN, Director of Radiology & Cardiovascular Services, Clinical Operations at HealthTrust. “Face-to-face discussions with suppliers at conferences allow advisory board members to evaluate emerging technologies firsthand.”
“The general membership would be surprised at the depth of the work we do around choosing which suppliers to participate in an advisory board review process,” says Lengle. “I don’t think all GPOs are able to send 20 people to hear from suppliers about current and future technology and their approach to providing enhanced patient care through diagnostic imaging.”
Puente says that from a member perspective, HealthTrust’s role in AI evaluation should center on image quality, throughput and diagnostic value. For modalities that use radiation, he says, members need to understand which vendors improve image quality while reducing dose, and which tools meaningfully optimize workflow at a time when many organizations continue to face staffing challenges.
Looking ahead
St. Luke’s is currently using AI for stroke detection and workflows, automated breast ultrasound support and heart flow analysis for cardiovascular disease. It is considering AI technology to help identify Alzheimer’s disease in patients, and to help draft radiologist impressions and recommendations to reduce workload burdens.
Puente sees similar promise in AI’s ability to improve throughput and the patient experience. In radiology, he says, better patient positioning, scan planning and camera-based recommendations in CT can reduce scan time, lower dose and support image quality. Those gains can mean less time waiting, less time on the table and faster reports back to physicians.
Pedersen-Romero sees AI shifting from innovation to operational integration. “AI is moving from novelty to operationalization in every way,” she says. “Now the question is how to responsibly scale it, keep PHI safe and improve the patient care experience.”
Questions IDNs should ask when evaluating AI products
- What AI technology do we currently have?
- Where will we use it, & what operational benefits will it deliver?
- How will the new product address gaps, such as image quality or workflow?
- Who benefits most—radiologists, technologists or administration?
- Will it replace existing technology?
- Will it standardize with our current applications & hardware?
- How will we define & communicate its value?
YOUR TURN
Tell us how you are using AI to enhance your radiology service line by emailing thesource@healthtrustpg.com or posting to the HealthTrust Huddle
