This summer marks the entrance of new reporting software programs into radiology departments, mainly because Microsoft ended the renewal and maintenance of PowerScribe 360 (PS360) software on August 31, 2026. For close to a decade, PS360 had been the dominant program for radiology reporting.
Microsoft will allow PS360 to reach its end-of-life on August 31, 2027. The announcement has given hospitals, clinics, and doctors across the United States an opportunity to explore and define new preferences about communicating imaging results. The end of PS360’s reign could mean new programs will be more efficient and responsive to radiologists’ needs.
“Microsoft’s announcement in March prompted many organizations to have conversations about what we’d like,” says Roni Talukdar, MD, president of Radiologic Associates of Fredericksburg (RAF) in Virginia. “Due to PS360’s dominance, reporting has been very similar for at least a decade. That’s an eternity in technology. The new options may allow us to do more, with different partners.”
Tessa Cook, MD, PhD, an associate professor and vice chair of informatics in the department of radiology at the Perelman School of Medicine at the University of Pennsylvania, says a reporting system is a “mission-critical” solution for a radiology practice. Cook is also a member of Radiology Today’s editorial advisory board.
“The newer systems are more expensive, although the actual increase in cost depends on the size of the organization,” she says.
Cook says there are more options in the market than there were 15 years ago, especially since the advent of generative AI. “Most modern systems include features like impression generation (a summarization of the radiologist’s key findings) and ‘ambient mode.’ [The latter] allows radiologists to dictate in whatever order they prefer and inserts the findings into the right places in a template,” she says. “There are also better integrations between reporting systems and PACS for exchange of measurement data, and also between reporting systems and the RIS.”
Radiologist Concerns
Microsoft introduced a new program in place of PS360, called PowerScribe One. Some health systems will adopt this software because it offers continuity within an existing ecosystem.
Yale New Haven Health System (YNHHS) moved away from Microsoft’s radiology reporting software because the system had already been seeing friction in its reporting workflow. The concerns included system slowness, template challenges, and other issues that affected the day-to-day work of its radiologists.
Melissa Davis, MD, MBA, is an associate professor of radiology and biomedical imaging, as well as the vice chair of imaging informatics for radiology and biomedical imaging at Yale University School of Medicine. She says YNHHS evaluated solutions based on usability, integration across its radiology environment, scalability, and a vendor’s ability to evolve with the health system over time.
“We also listened closely to radiologists to understand what was slowing them down and what would make the biggest difference in their daily work,” Davis says. Ultimately, YNHHS chose to utilize Rad AI Reporting, a program developed by the San Franciscobased developer Rad AI.
Sham Sokka, PhD, is the chief operating and technology officer of digital health for DeepHealth, which is based in Somerville, Massachusetts. DeepHealth is a wholly owned subsidiary of RadNet, a provider of outpatient diagnostic imaging services in the United States. Sokka says radiology departments should evaluate reporting as part of a full diagnostic workflow, not a standalone tool.
“The question is no longer only whether a platform can create a report, but whether it can help radiologists work more efficiently and consistently,” Sokka says. “[The platform should be] connecting image management, workflow orchestration, clinical AI findings, measurements, AI-generated impressions, quality assurance, and structured reporting in one seamless workflow.”
According to Sokka, radiologists are looking for technology that reduces reporting time and supports faster turnaround for patients and referring physicians. In addition, the health care organizations that employ them want reporting platforms that are in use in real-world clinical environments.
“Through DeepHealth’s connection with RadNet, our Reporting Pro software is being deployed across high-volume imaging settings. [This allows] the solution to be shaped by direct physician feedback and real-world workflow needs,” Sokka says.
Unexpected Shift
Microsoft’s March announcement jarred its long-time users. This resulted in health care systems feeling stranded and forced into a sped-up selection process. “[The change is] being imposed on organizations at the vendor’s timeline, not that of the client,” says Michael Finke, CEO of Jacobian, a Pittsburgh-based radiology reporting software developer.
Finke says the change that Jacobian staff witnessed firsthand at the mid-June 2026 Society for Imaging Informatics in Medicine meeting was palpable. “Beyond the operational disruption, these types of high-risk migrations lead to frustration among radiologists … When you layer in implementation, retraining, workflow reconfiguration, and productivity loss during transition, the true cost is far greater than the licensing delta alone,” Finke says. (The term “delta” means the difference between two types of systems.)
Health care organizations need an endto- end platform that can be deployed at enterprise scale and integrates deeply with existing systems and workflows, Finke says. At the same time, the platform should be supported by an experienced team, and its developers should be creative.
Finke says the team behind Jacobian’s reporting software, Fluency for Imaging, had many conversations about the fundamentals on which health care organizations depend. “It is clear that the risk tolerance of health care organizations in this forced migration is zero,” he says. “[Our] discussions are focused on trust, reliability, stability, and proven performance, not just innovation.”
Talukdar, whose practice adopted Rad AI Reporting in April 2026, says RAF looked at many programs before making a decision. “One of the things we liked about this program is the fact that you never have to break your visual concentration on the images to write the report,” he says. “With PS360, you had to tab over to the image, then tab over to the dictation system to report your findings about a specific structure like the liver, and then repeat, as opposed to keeping your eye on the images and discussing findings as you go.”
RAF’s radiologists had to do a fair amount of training to familiarize themselves with Rad AI. “We’re still undergoing training,” Talukdar says. “Rad AI has been involved in that process. But they’re open to feedback and adjustments. Both we and they are very committed on the continued partnership.”
RAF physicians particularly like that Rad AI developed its templates after talking to radiology practices. The practice also found common ground with Rad AI because Rad AI’s founder is a radiologist. RAF is composed of board certified and fellow-trained physicians.
AI’s Place
In radiology, AI is not considered an agent that has difficulty differentiating between good and bad data. According to Talukdar, most radiologists view AI as a necessary helper.
Elizabeth Bergey, MD, is the chief clinical officer of Rad AI. She says health care organizations should think about AI in radiology as a long-term strategy. There are two distinct categories of AI to implement into physician workflows.
“Generative AI supports reporting and documentation workflows,” Bergey says. “Pixel-based AI analyzes medical images to identify or prioritize potential findings or diagnoses …While they address different aspects of each radiologist’s workflow, they should not operate in isolation. The real opportunity is creating connected workflows where these technologies can communicate and work together.”
Cook thinks there are always risks associated with using any AI tool. She says, despite this, the potential benefits are worthwhile. “[This is] especially [true] since the radiologist will review the report before signing, and AI-generated content will not reach patients without review for accuracy,” Cook says.
Sokka’s thoughts on AI relate to the fact that this moment is an inflection point for radiology. “Reporting is moving from a documentation function to a more intelligent, connected part of the diagnostic workflow and infrastructure,” he says. “AI should not sit outside the report or require radiologists to move between disconnected tools. It should be integrated into the reporting process so that measurements, clinical AI findings, and relevant results can flow directly into structured report creation.”
Sokka adds that workflow fit is a key consideration for radiologists. He says the real value for AI comes when AI is embedded into how radiologists already work.
“Reporting tools should reduce manual documentation, support review and refinement, and help radiologists finalize reports within a connected reporting environment. Ultimately, organizations should also consider whether the reporting environment can scale across modalities, sites, and clinical use cases without adding complexity,” Sokka says.
Finke says AI presents an opportunity for health care organizations to determine which software developer they can trust to be at “the center of everything. The organizations best positioned for the future are those that make a deliberate architectural decision now, to establish their platform as the control plane of diagnostic imaging,” he says. “[That’s] where every data source, clinical insight, AI model, and workflow automation flows through a single trusted environment.” Finke adds that EHR data, prior studies, clinical guidelines, and AI-generated insights should converge at the point where the radiologist works.
Workforce Shortage
For years, the United States has suffered a workforce shortage in radiology that has resulted in personal and professional stress for practicing radiologists. This phenomenon has led to high rates of attrition, particularly after the COVID-19 pandemic. If attrition rates continue to be as high as they have been post-COVID, the 2055 radiologist workforce will be substantially smaller. That finding comes from a 2025 study published in the Journal of the American College of Radiology.
One solution could be hiring international medical graduate radiologists who complete the American Board of Radiology’s alternate pathway. In 2022, the American Board of Radiology changed its alternate pathway requirements for foreign-trained radiologists. This made the process of board certification easier and more flexible. Yet changes to immigration policies have made it harder for foreign-trained radiologists to work in the United States. As a result, for now, the workforce shortage is almost certain to continue.
Present and future workforce shortages are a factor that health care systems consider when choosing reporting software. They are trying to offset the problems they face by utilizing technology with new capabilities.
Sokka says, at this point, radiology is undergoing a transformation. “Radiologists are under growing pressure from rising imaging volumes, workforce constraints, fragmented technology, and increasing clinical complexity,” he says. “Replacing a reporting system is not just an IT project. It is a chance to reduce fragmentation and build a more connected imaging environment that supports radiologists and the care experience.”
Talukdar says having tools that preserve a practice’s human capital is beneficial. “Anything that helps us become more efficient without increasing burnout is welcome,” he says. He adds that there are some small hiccups with AI, but nothing outside the norm.
“What we see is what’s to be expected with a normal technological implementation,” Talukdar says.
Changes Ahead
Since new software means many changes, some health care organizations are considering how they will explain their tools to patients and their loved ones. This is especially true when it comes to AI. There is widespread public distrust of AI and concerns about the water and electricity needs of data centers that contain AI infrastructure. Currently, most patients do not ask how a radiologist’s report is generated.
“Most patients are focused on themselves and what they’re going through. They ask, ‘What does this cancer mean for me? How will it affect my life?’ As AI products become more patient facing, they may begin to understand and ask more about AI’s role in radiology reporting,” Talukdar says.
Health care organizations are also looking at how reporting works for radiologists today. “That means looking at where time is lost,” Bergey says. “Are radiologists waiting on the system? Are they correcting speech recognition errors? Are templates helping or slowing them down? Are they moving between too many systems to complete a report? Are they relying on workarounds that have become normalized over time?”
Although such issues may seem small in isolation, “Across thousands of exams, they can have a meaningful impact on efficiency, satisfaction, and capacity,” Bergey says.
Davis says features like AI are only part of an evaluation. Radiology is a complex environment, and radiologists must move across PACS, the EHR, worklists, templates, speech recognition platforms, and other systems throughout the day.
“What matters is whether a solution performs reliably under real clinical volume, reduces friction in workflows, and helps radiologists work more efficiently without adding complexity,” Davis says. “For us, the right [software development] partner was one that could support our current needs, innovate with us, and continue advancing the reporting experience as our workflows and technology environment evolve.”
Jessica Zimmer is a freelance writer living in northern California. She specializes in covering AI and legal matters.