Workflow Teleradiology Radiology

Right study, right radiologist, right time

Mike Moore October 6, 2026

Right study, right radiologist, right time

Most people picture radiology as the moment a doctor looks at an image and tells you what it means. Ginea Qualls spends her days on everything that has to happen before that moment is even possible. As the Senior Clinical System Engineer at Radiology Associates, Arkansas's largest independent practice, she keeps the enterprise imaging running for a group that reads close to a million studies a year across roughly two dozen hospitals and a hundred-plus clinics. Her whole job, as she describes it on episode four of Flight Path, is getting the right study, with the right history, to the right radiologist at the right time. When that works, nobody notices. When it does not, a read slows down and a patient waits.

It helps to see what a single study is actually carrying. When an exam is ordered, three things have to travel and then line up: the images, the order itself, and the patient's priors. Ginea routes that traffic through a DICOM router and an HL7 engine into the practice's archive and then onto the worklist, which is built from the orders coming in from each hospital. Those orders are what let her put the right study in front of the right reader.

 

Routing is the quiet skill

A large independent group does not read everything with everyone. It reads with subspecialists, and it reads across facilities that have no relationship to one another. So the worklist has to be smart about two things at once: who is allowed to read a study, and who should. Credentialing decides the first. Rather than wait for a new radiologist to be credentialed at every hospital, Ginea filters the worklist so a reader only sees the studies they are cleared for, which lets the practice bring on talent faster instead of stalling on paperwork.

Subspecialty decides the second. An MRI of the knee is tagged as a musculoskeletal study, and that tag is what routes it to a musculoskeletal reader. A suspected stroke is different again. Ginea uses the exam reason or a procedure code to identify it and lift it to the top of the list, because the clock on a stroke starts the moment the patient reaches the emergency department. The worklist itself is ordered by priority first, then by time remaining against the service-level agreement for that study type, then by class, whether the case is emergency, inpatient, or outpatient. The urgent study surfaces first, gets read, and the subspecialist goes back to their own queue.

The priors problem

Ask Ginea where image and data sharing actually breaks, and she does not hesitate: priors. Not because a prior study is complicated in itself, but because getting the right one, from a facility that may have nothing to do with yours, and making it line up with the current exam, is genuinely hard. In a state with many rural hospitals, a patient might begin care in southern Arkansas and end up at a larger hospital in Little Rock. The prior has to marry up to the new study, because it is the rest of the story. As she puts it, you would not open a book and start reading in the middle. A radiologist needs the earlier image to see whether the mass in front of them has grown or shrunk.

When a prior does not attach, the cause is often something small: a name that does not match, or a malformed piece of DICOM with an extra character where there should not be one. A lot of the work is detective work, tracing why two records that belong together will not connect. It sounds tedious. Ginea clearly loves it, and the reason is that every one of those puzzles is a read that gets back to a patient sooner.

Where AI earns its place, and where it does not

It would be easy to bolt an AI section onto a story like this and call it the future. Ginea is more disciplined than that. The clearest win she sees today is prioritization. AI tools can look at an image and flag something emergent, a stroke, a fracture, a bleed. The example she gives is an outpatient head CT that turns up a brain bleed. Because it was ordered as an outpatient study, it would normally sit in a lower-priority queue, potentially for hours. An AI flag lets her worklist elevate that case so a radiologist sees it quickly. The radiologist still makes the diagnosis. Every time. What the AI changed was the order things were looked at, not who decided what they meant.

She is also pointing AI at her own operation. To drive down how often radiologists have to stop and request a prior, she analyzes worklist data to find where those requests cluster, which hospital, which workflow, and closes the gap site by site. The goal is a prior-request rate down around one percent, and she can already see it trending. Two years ago, she notes, that kind of analysis would not have been a quick job. Now she uploads the data, gets it back fast, and works on something else while it runs.

The standard worth borrowing

What ties Ginea's approach together is not any single tool. It is a way of thinking about the work. Everything on the radiologist's plate that is not interpretation is a candidate to be removed, automated, or handed to someone else, so the scarcest person in the process spends their time reading. Every new capability is judged by whether it moves a number she can see. And every fix at the seam, a matched prior, a normalized feed, a cleaner study, is measured in the only unit that matters here: a patient getting an answer sooner.

Her read on the next few years follows from that. She expects agents to catch what is missing before a human has to, so studies arrive more complete and radiologists spend less time chasing and more time reading. Not the technology replacing the radiologist, but clearing the runway in front of them. In a field carrying real burnout, that is not a small thing. It is, in her words, not taking the role away. It is enhancing it.

Hear the full conversation with Ginea Qualls on Flight Path, available on YouTube, Spotify, and Apple Podcasts.