St. Luke’s University Health Network optimizes radiology workflow with PowerScribe One and Dragon Copilot

Modernizing radiology for a growing health system

St. Luke’s University Health Network is a 16‑campus health  system serving communities across Northeast Pennsylvania and Western New Jersey. As imaging volumes increase and clinical expectations rise, leaders at St. Luke’s recognized that traditional, fully on-prem technology could no longer keep pace with the demands on radiologists and other clinicians.

The organization needed a foundation that could scale across the enterprise, support advanced AI, and integrate seamlessly into radiologists’ daily workflows.

According to Dr. Robert Fournier, Chairman of Radiology at St. Luke’s University Health Network, the strategic direction was clear. “Our mission inside of St. Luke’s was to get things to the cloud. We knew that was the direction the network was taking.” 

Moving to embrace cloud capabilities was not only about infrastructure updates. It was about enabling innovation while maintaining reliability, continuity, and trust. Radiologists needed modern solutions that could help reduce cognitive burden, improve report quality, and help address growing workforce pressures—without added disruptions or distractions.

Embracing cloud capabilities with PowerScribe One

To support its modernization goals, St. Luke’s adopted PowerScribe One, a radiology reporting solution designed with cloud capabilities. PowerScribe One uses cloud-based AI to enable accurate speech recognition without voice training, integrate FDA-cleared third-party imaging insights, and generate personalized draft impressions all directly in the report.

“So, moving to PowerScribe One for us allowed us to enable a lot downstream,” Dr. Fournier explained. “Again, if you want to introduce AI into your environment, you can actually do that through PowerScribe One.” 

By shifting to cloud-based reporting, St. Luke’s could align radiology with the broader enterprise cloud strategy while creating a standardized environment where new capabilities can be deployed easily and at scale. 

Driving adoption and efficiency with Smart Impression

One of the first AI‑powered capabilities St. Luke’s enabled within PowerScribe One was Smart Impression, which uses generative AI to draft personalized impressions and is designed to help radiologists boost efficiency and produce more complete and consistent reports.

“Smart Impression was one of the tools we turned on fairly early, and it was widely adopted by my team,” Dr. Fournier said. 

Adoption was rapid because Smart Impression integrated directly into existing workflows. Radiologists did not have to leave their reporting environment or open an additional widget. Instead, Smart Impression worked where they already spent their time—inside the report—helping to ensure that key findings were captured accurately and consistently.

The impact was both clinical and operational. “I know my team does better,” Dr. Fournier noted. “I don't worry about not reporting any findings as they are automatically picked up and put into the impression.”

By surfacing and incorporating key findings from the report, Smart Impression helped radiologists review their impressions with greater confidence, supporting more consistent and complete reporting.

From an efficiency perspective, the results were measurable. Facing radiology workforce shortages, St. Luke’s leveraged Smart Impression to help accelerate reporting and free up capacity equal to a full-time employee. “When we first turned on our impression generation, it saved us about 1.3 full time employees (FTEs),” Dr. Fournier explained. While he emphasized that each organization could calculate the cost implications differently, the productivity gain reinforced that AI in the reporting workflow must be trusted and fully integrated rather than an experimental add‑on. 

Scaling third-party imaging AI with Precision Imaging Network

With the PowerScribe One foundation in place, St. Luke’s sought to expand its integration of third-party imaging AI insights directly into the report without creating additional administrative or security burdens. The organization joined the Precision Imaging Network to streamline third-party AI model evaluation and deployment. 

“Precision Imaging Network does some sort of initial evaluation of some of these models, making sure they work in other people’s environments…The legal contracts are done with a single contract,” Dr. Fournier said. 

This approach addressed one of the biggest barriers to AI adoption in healthcare: the time and complexity required to move models through IT security, legal review, and validation. By streamlining these processes, Precision Imaging Network allowed St. Luke’s to focus on the clinical value of the third-party AI models rather than the administrative friction for implementing each one. 

Just as importantly, AI insights surfaced exactly where radiologists needed them. “It also ends up in our workflow. It’s there in my report, and that’s where we want it all to live,” Dr. Fournier explained. “That saves us time.” 

Supporting better care with AI‑powered detection

One of the earliest third-party AI solutions deployed through Precision Imaging Network was a fracture‑detection model from Gleamer. The impact was immediate and tangible, especially across St. Luke’s geographically distributed network.

“Our first big win was with a fracture detection system,” Dr. Fournier said. He described how a senior leader received care at a remote urgent care location and experienced the same diagnostic quality as if they were treated at the main campus. “He got the same level of care there as he would in the main campus because of that fracture detection system.” 

Gleamer’s AI flagged the fracture as positive for review, triggering rapid follow‑up. “They called us immediately, got it done, and it was a big win for him,” Dr. Fournier recalled. 

For leadership, consistency and accountability were essential. “Everything we’ve deployed to date has some ROI (return on investment),” Dr. Fournier said.

Accelerating validation and preparing for the future with Dragon Copilot

“When you talk about challenges in our field, I think the biggest one you’re going to see is workforce,” Dr. Fournier said. “As medical knowledge doubles beyond what we can comprehend, we need something to help support us.” 

As a companion to PowerScribe One, Microsoft Dragon Copilot is an AI clinical assistant designed to optimize workflows with intelligent automation and real-time summarization. Within the Dragon Copilot radiology experience, St. Luke’s is using capabilities such as prior report summarization, which distills relevant prior reports and associated metadata into concise statements to provide essential context for radiologists as they interpret current studies. 

St. Luke’s began adopting Dragon Copilot, initially deploying it to a small group of radiologists and refining it in real clinical settings. “We first deployed it to about 10 radiologists,” Dr. Fournier explained, describing how those early users became “super users” who advocated expanding access to the rest of the team. 

Over time, performance improved significantly. “It got significantly better in our environment,” he said.

As workforce challenges intensify, St. Luke’s is looking at the broader clinician experience. Increasing imaging volumes, expanding medical knowledge, and ongoing staffing shortages made it clear that clinicians need intelligent support to help them remain focused on patient care.

Beyond supporting radiologists, Dragon Copilot is built to simplify and transform the way care teams work by streamlining physician and nursing charting, notes, and flowsheets—helping teams work more efficiently and confidently across the day.

Designed with role-based experiences at its core, Dragon Copilot delivers patient data, trusted clinical content, and partner-powered AI insights into a single, contextual experience within the clinical workflow.

Broadly, St. Luke’s plans to expand its use of Dragon Copilot to other members of the care team.

The organization now views Dragon Copilot as the path forward to restoring the joy in practicing medicine. “You make a clinician, a clinician again… They’re not a data entry specialist,” says Dr. Fournier.  

A strategic partnership built on trust and scale

Underlying St. Luke’s technology decisions is a broader enterprise partnership with Microsoft. “We have very key strategic partners we are with. Microsoft is clearly one of them,” Dr. Fournier said. 

The move to Microsoft Azure addressed early concerns about cloud security and performance. “You’re moving all your patient data to the cloud. That scared us a little bit at first,” he acknowledged. “But you realize having your patient data out in a safe, secure environment in multiple level data centers should make you feel better, not worse.” St. Luke’s EHR (electronic health records), Epic, and their PACS (picture archiving and communication system), Sectra, are both hosted in Azure. 

For St. Luke’s, the partnership with Microsoft extends beyond radiology. It supports an enterprise vision where AI, cloud computing, and scalable infrastructure work together across the continuum of care. 

Together with Microsoft, St. Luke’s University Health Network is addressing today’s workflow challenges while laying the groundwork for what comes next. By combining cloud-based clinical applications like PowerScribe One and Dragon Copilot with the security, performance, and scalability of Azure, St. Luke’s has built a foundation that supports teams now and adapts as care delivery evolves.