Streamlining AI Consumption for Radiologists

There are nearly 1,000 FDA-cleared artificial intelligence (AI) devices for radiology on the U.S. market. Choice is a good thing. But so many options can also create a lot of confusion. Add to this the fact that many practices and hospitals undergo mergers or acquisitions, and those institutions often bring with them their own native PACS and AI systems.  

What’s more, AI delivery varies from vendor to vendor. Some AI tech comes as software as a service (SaaS), some are in the cloud, and others are integrated into the hardware of a modality, like a CT or MRI machine. For busy radiologists, the unfortunate and all too common result is fragmented and complex workflows that often lead to AI avoidance rather than acceptance. 

Yet, we know AI is helping radiologists by acting as a key component in day to day workflows. AI is assisting with automating routine tasks, improving diagnostic speed and accuracy, and prioritizing critical cases for review. So, how can healthcare enterprises and imaging centers  ensure radiologists are making the most of AI?  

Some, like University Radiology Group (URG), the largest provider of subspecialty radiology and teleradiology services in New Jersey, have partnered with Fujifilm to streamline the consumption of AI for radiologists.  

Approximately five years ago URG first explored Fujifilm’s Synapse® AI Orchestrator for use in their practice, an open imaging workflow orchestrator that uses an advanced rules engine to seamlessly bring the radiologist’s preferred imaging algorithms directly to Synapse PACS workflows. Designed for true interoperability, Synapse AI Orchestrator allows healthcare IT staff to deploy and maintain a multitude of AI applications for diverse clinical workflows across the enterprise.  

At RSNA 2025, URG President Dr. Roger Yang, MD, FACR, presented at the Innovation Theater on his teams’ experience, with Fujifilm and our Synapse AI Orchestrator that has been in use clinically across the organization for a number of years.  

URG provides preliminary and final interpretations along with consultative support for 14 affiliated regional and academic hospitals. It also operates and reads exams for its own 24 diagnostic imaging centers. URG radiologists read more than 2.1 million exams annually, including in New Jersey and Israel —so simplifying the AI consumption process was a top priority. URG turned to Fujifilm to develop a system that smartly deployed AI across a variety of modalities, so that it became functional for all radiologists. In addition to enhancing AI acceptance and diagnostic accuracy, a longer secondary goal was to boost provider productivity,  once their users were content and comfortable with their AI-driven workflow.  

In his compelling presentation, Dr. Yang discussed: 

• URG’s goal: For URG, the primary objective for implementing AI, specifically mammography in his presentation, across it’s network wasn’t to strictly increase radiologist productiivity. Rather, it was to make them more comfortable and content with using AI in their day-to-day workflow.  To acheive this, URG worked with Fujifilm to streamline all of its AI data across imaging modalities, in and out of PACS, and present it to radiologists in a way that was most usable for them. By building a more seamless and intuitive technology,  radiologists were given an easy way to look at all AI and leverage the insights through a seamless and straightforward workflow. 

• How Synapse AI Orchestrator works: The server is aware of all orders coming in and commands the PACS to move specific scans to the appropriate AI engine located either on-premises or in the cloud. The data is immediately processed and  results are completed, it then goes back to Synapse AI Orchestrator. Next, Synapse AI Orchestrator sends the results to Synapse PACS and displays the notifications directly in the worklist, all in realtime and wihtout impact to the radiologist. This process happens at the speed of light – all within 1-2 seconds!  

• URG’s first test case: Synapse AI Orchestrator supported URG’s mammographers by providing AI results prior to the start of radiologist’s interpretation, and via a triage tool that flagged abnormal cases requiring more urgent attention.  

• Results of test case: Overall, URG saw a 29% average reduction in interpretation time, saving about  1 minute and 26 seconds per case. They also saw a 30% improvement in the number of mammograms read. Radiologists’ RVU productivity increased by 16% in the one-year that was examined. This was measured for all exams, not just screening mammograms.  

Perhaps most importantly, Dr. Yang states that there have been significant improvements to his team’s openness and comfort with incorporating AI findings into their reading workflow, while also driving impressive efficiency gains. That comfort factor is key, says Yang, as anecdotal industry research indicates that some 80% of AI installs are unsuccessful—with only 1 in 5 radiologists utilizing the AI tools their healthcare facility invests in.  

That’s why Dr. Yang measures the success of AI Orchestrator in “radiologist satisfaction.” It’s all about the “human factor.” Once you clear up the confusion and streamline the consumption of AI, radiologists embrace those algorithms. And that’s a good thing for your practice and your patients.  

To view Dr. Yang’s full presentation, please click here or click the embedded video below.