Radiologist shortage in the U.S. spurs AI-powered overhaul of imaging workflows

A major new study published in npj Health Systems titled “AI solutions to the radiology workforce shortage” outlines how artificial intelligence (AI) could be harnessed to tackle the growing strain on the U.S. radiology sector.

Conducted by researchers from the University of Texas MD Anderson Cancer Center and the University of Texas Health Science Center, the study dissects the causes and consequences of the radiologist shortfall, while presenting AI-powered interventions in demand management, workflow efficiency, and capacity building as an integrated solution.

What is causing the radiology workforce shortage?

The U.S. radiology sector is buckling under dual pressures: rising demand and stagnating supply. An aging population, greater prevalence of chronic illnesses, and technological advances in imaging are driving up demand for diagnostic services. Between 2010 and 2020, the number of Americans aged 65 or older surged by nearly 39%, and cancer cases are projected to grow by 42% by 2045. This has dramatically escalated the volume of imaging studies.

Yet the number of qualified radiologists has not kept pace. As of the latest estimates, the U.S. has around 34,000 practicing radiologists, of whom a significant portion are nearing retirement. Only about 1,400 new radiology residents enter the field each year, far short of the over 1,900 open positions. Compounding this issue is a cap on Medicare-funded residency slots that dates back to 1996, limiting training expansion unless alternative funding is found.

The consequences of this imbalance are already being felt: longer wait times, delayed diagnoses, and increasing rates of burnout among radiologists. In private practice, burnout rates exceed 45%, and radiology departments nationwide report difficulty in retaining staff and maintaining service standards.

How can AI help manage imaging demand and streamline workflow?

The authors classify AI applications into three impact zones: managing demand, improving workflow, and building system capacity.

In terms of demand management, AI tools can mitigate unnecessary imaging through predictive analytics, decision-support systems, and risk-based screening protocols. Studies cited in the paper estimate that 20–50% of imaging may be medically unnecessary. AI-powered platforms using American College of Radiology Appropriateness Criteria can assist referring physicians in selecting suitable tests, thus conserving resources and reducing clinician burden.

AI also plays a crucial role in triaging and scheduling. Algorithms can identify non-urgent cases and reschedule them to off-peak hours, helping smooth demand fluctuations. Predictive models can even anticipate peak loads and adjust workflows proactively. For patients with chronic diseases, AI systems offer ongoing monitoring, reducing the frequency of imaging through early interventions, behavior modification prompts, and predictive risk scores.

To optimize workflow, AI enhances image acquisition, processing, and interpretation. Deep learning models now enable faster MRI scans and automatic quality checks that detect errors like motion artifacts before images reach radiologists. Automated protocoling of CT and MRI exams ensures scans align with clinical needs, reducing redundancy and exposure risks.

Another breakthrough is the automation of report generation. Natural language generation tools can draft imaging reports based on AI-assisted image interpretation, standardizing output and minimizing turnaround time. Clinical history summarization, often a time-consuming burden, is being accelerated through large language models capable of distilling electronic health record data into concise summaries, improving accuracy and saving time.

AI also integrates into legacy systems such as PACS and EHRs, ensuring interoperability and reducing workflow disruption. These digital efficiencies allow radiologists to focus on high-value tasks, such as complex diagnoses and direct patient care.

Can AI solve the supply gap by expanding radiologist capacity?

Beyond streamlining processes, the study emphasizes capacity building as AI’s most transformative promise. By alleviating repetitive tasks, AI enables radiologists to handle larger caseloads without sacrificing quality. AI assistance in image interpretation, such as for chest radiographs or mammography, has demonstrated sensitivity gains of up to 26% across skill levels, enhancing both efficiency and diagnostic accuracy.

The report notes that AI can reduce burnout by minimizing cognitive fatigue and enabling more flexible work arrangements, including remote teleradiology. These configurations are critical to retaining skilled professionals and preventing attrition.

On the educational front, AI supports radiology trainees through simulated learning environments, personalized case assignments, and the generation of synthetic training cases. Tools leveraging natural language processing help monitor trainee performance, identify gaps, and deliver competency-based instruction, even in programs with limited case diversity.

AI also facilitates improved patient communication, which is often undervalued in radiology. Natural language tools and visualization software can translate complex findings into patient-friendly summaries or 3D models. This not only enhances understanding but also frees up radiologists for more meaningful interactions.

Finally, AI enables equitable access to imaging services through advanced resource allocation and remote diagnostics. From predicting seasonal demand spikes to optimizing population-wide screening efforts, AI helps ensure that healthcare systems can deliver timely imaging to those who need it most.

By integrating demand management, workflow enhancements, and capacity expansion, AI can create a self-reinforcing loop: reducing overutilization, improving efficiency, and freeing up human capital for higher-value tasks. In doing so, it supports the long-term sustainability of radiology as both a profession and a vital healthcare function.

The researchers warn that the benefits of AI are not automatic. Poor implementation or lack of training can backfire, potentially worsening burnout. Thus, careful planning, stakeholder engagement, and robust clinical validation are essential to ensure these technologies are effective, ethical, and broadly accessible.