Walk into any radiology department in the country and you will likely find the same story: more studies to read, fewer radiologists to read them, and a worklist that grows faster than it shrinks. The radiologist shortage is not a future problem. It is a present one, and the data show it is not going away on its own anytime soon.
At the same time, artificial intelligence is making meaningful inroads into radiology workflows, offering a set of practical tools that help radiologists handle more volume without sacrificing accuracy or burning out. AI is not a silver bullet for the workforce crisis, and no honest observer would claim otherwise. But when combined with smart operational strategies like teleradiology, it represents one of the most viable paths toward closing the gap between imaging demand and radiologist capacity.
This article takes a clear-eyed look at the scope of the shortage, what is driving it, and how AI is helping practices and health systems manage the imbalance today.
The scale of the radiologist shortage in the United States comes into sharp focus when you look at the supply and demand data side by side. There are approximately 41,000 practicing radiologists in the U.S. today. Imaging volume, driven by an aging population, expanded preventive care guidelines, and the rising prevalence of chronic disease, is growing at 3 to 5% per year. The radiologist workforce is growing at roughly 1% annually. That gap between demand growth and supply growth is the core of the problem.
The Harvey L. Neiman Health Policy Institute published a landmark pair of companion studies in the Journal of the American College of Radiology in early 2025, projecting both radiologist supply and imaging demand through 2055. The findings were sobering. Even under optimistic scenarios where residency positions grow, imaging utilization is projected to rise by 16.9% to 26.9% by 2055 depending on modality, while the radiologist workforce is projected to grow by only 25.7% in that same period if residency slots do not expand. The conclusion from the study's lead author was direct: the shortage is not projected to get worse, but it is also not projected to improve without effective action.
The geographic picture is equally uneven. There are currently about 13 radiologists per 100,000 people nationally, but that figure masks wide regional disparities. States like Oklahoma, Mississippi, Nevada, and Wyoming have as few as 9 radiologists per 100,000 people. Some regions of Michigan have reported weeks-long waits for imaging results due to staffing shortages. For patients in those communities, the shortage is not a statistic. It is a delayed cancer diagnosis or a missed finding that should have been caught sooner.
Understanding why the shortage persists, despite strong interest in radiology as a specialty, requires looking at several converging factors.
Radiology is a highly competitive specialty that requires years of post-medical school training. Interest has not waned. In the 2025 National Resident Matching Program, 97.4% of diagnostic radiology positions were filled, and the number of applicants far exceeded available slots. Approximately 87% of diagnostic radiology applicants did not match into a PGY-1 position in 2025. The bottleneck is not a lack of qualified candidates. It is a lack of residency positions. Since 2021, only 29 new PGY-1 diagnostic radiology training positions have been added nationally. Growing the pipeline meaningfully will require years of sustained policy effort and investment in residency capacity.
The pandemic accelerated burnout and early departure from the field in ways that are still being felt. According to the Harvey L. Neiman Health Policy Institute, radiologist attrition rates have increased by 50% since 2020 compared to pre-pandemic levels. If post-COVID attrition rates persist rather than reverting to historical norms, the Neiman Institute projects there will be 3,116 fewer radiologists in the 2055 workforce compared to pre-COVID projections. That is a substantial and lasting effect on supply, driven not by a lack of new entrants but by experienced radiologists leaving the field earlier than they otherwise would have.
The average radiologist in a high-volume setting is expected to interpret an image every three to four seconds across an eight-hour shift, which amounts to approximately 6,300 images per day. At that pace, fatigue is not an occasional risk. It is a structural feature of the job. More than 45% of radiologists report experiencing burnout, and a global survey found that 47% believe working night shifts reduces their diagnostic accuracy. Burnout does not just harm individual radiologists. It degrades care quality, accelerates attrition, and ultimately makes the shortage worse. It is a self-reinforcing cycle that cannot be broken simply by asking radiologists to work harder or longer.
The RSNA estimates that people 65 and older account for 30% of all imaging each year, and the U.S. Census Bureau projects that 77 million Americans will be 65 or older by 2034. Advanced imaging modalities, including CT, MRI, and PET scans, are being used more frequently not just for initial diagnosis but to monitor treatment response and disease progression. More imaging is being ordered, studies are growing in complexity, and the case mix is shifting toward higher-acuity work. The demand side of the equation is not slowing down.
A peer-reviewed study published in npj Health Systems in 2025 identified three primary ways AI can help address the radiologist shortage: demand management, workflow efficiency, and capacity building. Each addresses a different dimension of the problem, and together they offer a practical framework for how AI can support the workforce without replacing it.
One often overlooked contributor to radiologist overload is unnecessary imaging. Studies consistently show that a meaningful share of imaging studies ordered in emergency and outpatient settings does not change clinical management. AI-powered clinical decision support tools can help referring clinicians order the right study for the right indication from the outset, reducing the volume of low-value imaging that flows into the radiology pipeline. Fewer unnecessary studies means a smaller total workload for the same number of radiologists, which creates real capacity without adding a single person to the workforce.
One of the most immediate applications of AI in radiology is worklist prioritization. Traditional worklists move on a first-in, first-out basis, meaning a routine knee MRI might sit ahead of a head CT showing signs of stroke simply because it arrived earlier. AI-powered triage changes this by continuously analyzing incoming studies and flagging potential critical findings in real time, automatically moving urgent cases to the front of the queue.
This matters for patient outcomes directly. Faster identification of time-sensitive findings like intracranial hemorrhage, large vessel occlusion, or pulmonary embolism shortens the window between imaging and treatment. It also matters for radiologist efficiency. Radiologists working from intelligently sorted worklists spend less cognitive energy on sequencing decisions and more on the clinical work that requires their expertise.
Report generation is one of the most time-intensive parts of a radiologist's day. A pilot study published in 2024 found that AI-generated draft reports reduced average reporting time from 573 seconds to 435 seconds per study, a 24% reduction, with no loss in diagnostic accuracy. A study from Northwestern Medicine found a 15.5% overall efficiency gain in radiograph reporting, with some radiologists achieving up to 40% faster completion times when using AI assistance.
These are not trivial gains. Applied across hundreds of studies per day in a high-volume practice, a 15 to 24% reduction in reporting time meaningfully increases the number of patients a radiologist can serve in a given shift without extending hours or cutting corners. That is the kind of capacity building the field needs, and it is being delivered through tools that work within the radiologist's existing workflow rather than demanding new ones.
For high-volume screening programs, AI is increasingly being used as a second reader that reviews studies independently and flags cases for closer attention. The MASAI trial in Sweden, published in Lancet Digital Health in 2025, found that AI-supported mammography screening reduced radiologist workload by 44.2% while improving cancer detection rates. For a specialty that is already stretched thin, the ability to safely offload a portion of high-volume screening reads to AI review, with radiologist oversight and sign-off, represents a genuine capacity multiplier.
It is important to be precise about what this means in practice. AI does not finalize reports, make independent clinical decisions, or replace the radiologist's professional judgment and accountability. What it does is handle the initial pass on a large volume of routine studies, surfacing the ones that warrant closer attention and allowing the radiologist to focus their time where it is most needed.
AI's ability to address the shortage does not operate in isolation. It is most effective when combined with structural changes in how radiology services are delivered. Teleradiology is the single most scalable workforce strategy available to health systems managing the shortage today, and AI makes it substantially more effective.
Teleradiology expands the geographic reach of qualified radiologists, allowing subspecialists to provide coverage for facilities that could not otherwise attract or retain them. A rural hospital in Wyoming with 9 radiologists per 100,000 people in its region can access MSK subspecialty reads, overnight coverage, or PET imaging expertise through a teleradiology partner without competing in a labor market that is already deeply constrained. That access to subspecialty depth is what transforms teleradiology from a coverage stopgap into a genuine quality improvement strategy.
When AI is layered on top of a teleradiology model, the operational benefits compound. Intelligent triage ensures teleradiologists are never sorting through undifferentiated worklists to find the urgent cases. Automated case routing directs studies to the right subspecialist based on modality, body part, and clinical indication. AI-assisted reporting reduces the time each study takes, increasing the number of reads a teleradiologist can complete without sacrificing accuracy or attention. Together, these tools allow a well-structured teleradiology team to serve a significantly larger geographic footprint than would be possible without them.
This is the model Transparent Imaging has been built around since its founding in 2019. Created by radiologists for radiologists, the practice exists because co-founders David Zelman, D.O., specializing in PET and Body Imaging, and Eric Ledermann, D.O., M.B.A., specializing in MSK Radiology, saw firsthand that the shortage of subspecialty expertise was a problem with a solvable structure. By building a team of 100+ radiologists across subspecialties, Transparent Imaging gives imaging centers and hospital systems access to highly accurate, peer-reviewed reads and subspecialty expertise on complex cases, with the turnaround times and consultation support that referring physicians depend on. The goal has always been to make the kind of radiology care available at a major academic center accessible to any practice that needs it, regardless of geography or size.
An honest assessment of AI's role in addressing the shortage also requires acknowledging its limitations. A survey of 185 radiologists published in 2022 found that only 22.7% reported a workload reduction from AI tools, while 69.8% reported no workload reduction at all. This does not mean AI is failing. It means that the benefits of AI are highly dependent on implementation quality, workflow integration, and the specific tools being used. Poorly integrated AI that adds steps to a workflow rather than removing them can increase burden rather than reduce it.
AI also does not solve the structural constraints on the training pipeline. More residency positions and sustained policy investment are necessary for any meaningful long-term improvement in workforce supply. AI buys time and creates capacity. It does not replace the fundamental need to train more radiologists.
And AI does not replace the clinical accountability, judgment, and expertise that a trained radiologist brings to a complex case. A subspecialist reading a difficult MSK study, a PET imaging expert interpreting an ambiguous oncology scan, or a neuroradiologist working through a challenging brain MRI is doing something that current AI tools are not designed to replicate. The value of genuine subspecialty expertise remains high precisely because the hardest cases are the ones where it matters most.
For radiology directors and hospital leaders managing the shortage today, the practical path forward involves a combination of strategies rather than a single solution.
The first priority is understanding where the bottlenecks actually are. Is the problem overnight and weekend coverage? Subspecialty depth for complex cases? Turnaround times on high-volume routine studies? The answer shapes the solution. A facility struggling with overnight reads needs a different approach than one facing a backlog in MSK interpretation or a gap in PET imaging expertise.
The second priority is evaluating teleradiology partnerships that already have AI-enhanced workflows in place. Building AI infrastructure from scratch is a significant undertaking for most imaging centers. Partnering with a teleradiology group that has already integrated intelligent triage, AI-assisted reporting, and subspecialty routing into its workflows allows facilities to capture the operational benefits without the implementation burden.
When evaluating partners, look for radiologist-led practices where clinicians, not software vendors, are driving decisions about which tools to use and how to deploy them. Look for peer-reviewed quality processes, measurable accuracy standards, and demonstrated subspecialty depth. Ask specifically about consultation support for study ordering, because getting the right scan ordered in the first place reduces unnecessary work downstream. And look for onboarding and licensing assistance that allows the partnership to get off the ground quickly rather than stalling in a prolonged credentialing process.
These are not hypothetical criteria. They reflect what the research shows about what actually works in practice, and they reflect the standard that Transparent Imaging was founded to deliver.
The radiologist shortage will not be resolved quickly. The Neiman Institute's projections make clear that even under favorable scenarios, the gap between imaging demand and radiologist supply will persist for decades without structural changes to residency capacity and workforce policy. AI accelerates the efficiency of the existing workforce. Teleradiology extends its geographic reach. Neither eliminates the fundamental supply constraint.
What they do, together, is allow the radiologists who are practicing today to do more, with less fatigue, with better tools, and with the flexibility to work in ways that reduce rather than accelerate burnout. That matters enormously for retention. Every experienced radiologist who stays in the field because the workload is manageable and the flexibility is real represents a meaningful contribution to workforce capacity. Every practice that creates a sustainable environment for its radiologists is investing in the long-term supply of the specialty.
That is ultimately why the radiologist-first philosophy matters in this context. The shortage is not just a logistics problem. It is a human one. Solving it requires treating radiologists as professionals whose experience, wellbeing, and clinical judgment are worth protecting, not just as reading units to be optimized. AI and teleradiology, done well, serve that goal. Done poorly, they make it worse.
It is significant and well-documented. There are approximately 41,000 practicing radiologists in the U.S. serving a population that is ordering imaging at a rate that is growing 3 to 5% per year. The radiologist workforce is growing at roughly 1% annually. The Harvey L. Neiman Health Policy Institute projects this imbalance will persist through at least 2055 without meaningful changes to residency training capacity. In some states and rural regions, the shortage is acute, with as few as 9 radiologists per 100,000 people and patients waiting weeks for imaging results. The ACR Career Center listed over 1,700 open radiology positions as of 2024, and industry data suggest roughly 50% of radiology job searches in recent years have gone unfilled.
No, and the professional consensus is clear on this point. AI is not designed or approved to practice medicine independently. Every AI-flagged finding requires radiologist review and sign-off. What AI can do is increase the effective capacity of the radiologists who are practicing by reducing the time spent on routine tasks, improving worklist prioritization, and assisting with report generation. The most credible studies show that AI works best as a tool that augments radiologist performance, not one that substitutes for it. The shortage requires growing the workforce through training pipeline investment. AI helps manage the gap in the meantime.
The tools with the most documented real-world impact fall into three categories. First, AI triage and prioritization tools that automatically flag critical findings and reorder worklists by clinical urgency, getting stroke, hemorrhage, and PE cases to radiologists faster. Second, AI-assisted reporting tools that use speech recognition and natural language processing to generate structured draft reports, with studies showing 15 to 24% reductions in reporting time in real clinical settings. Third, AI second-reader tools for high-volume screening programs, most notably mammography, where the MASAI trial in Sweden demonstrated a 44.2% workload reduction while maintaining or improving detection rates. These are the areas where evidence is strongest and adoption is most advanced.
Teleradiology addresses the geographic dimension of the shortage by allowing qualified radiologists to provide coverage for facilities that cannot recruit or retain on-site staff. A small hospital in a rural state can access subspecialty reads, overnight coverage, and complex case consultation through a teleradiology partner regardless of local workforce constraints. AI makes teleradiology more effective by handling the operational work that would otherwise consume radiologist time: sorting worklists, routing cases to the right subspecialist, assisting with report drafting, and flagging urgent findings so they are never buried. Together, teleradiology and AI allow a smaller number of well-supported radiologists to serve a larger geographic footprint at a higher level of quality than either strategy could achieve alone.
The most important qualities are subspecialty depth, quality standards, and operational reliability. Look for a partner with genuine subspecialty coverage across the modalities and body parts your facility handles, not just general diagnostic radiology. Ask about peer-reviewed quality processes and how accuracy is measured and monitored. Evaluate turnaround time commitments for both routine and urgent studies. Ask whether the group offers consultation support for study ordering, which reduces unnecessary imaging and improves downstream efficiency. And look for onboarding and licensing support that allows the partnership to become operational quickly. In a shortage environment, the time it takes to get a new partner up and running has real clinical cost. A well-structured teleradiology partner should be able to address that directly.