Three ways AI can enhance radiology imaging workflows – right now
Introduction
While enhancing disease detection was the initial focus of AI development in radiology, it has become clear that AI can also significantly improve workflows and ease workloads for radiologists.
Workflow steps that involve radiology:
- Image Ordering
- Scheduling
- Acquisition
- Worklist Distribution
- Reading
- Reporting
- Patient Follow-Ups
These steps present an opportunity for AI-driven applications to increase efficiency and minimise errors for radiologists, technologists, and care teams. Now, it is a great opportunity to build a radiology ecosystem where every workflow step benefits from fresh AI-driven insights and efficiencies.
AI adoption in healthcare mirrors its progression in the auto industry — full automation still has a long way to go, but it is making steady progress with incremental advancements shaping the future.
For instance, the adoption of automated features in automobiles, like blind spot detection and collision alerts, have set the standard for further advancements and broader adoption. Similarly, in radiology, a series of targeted AI advancements can support the work of professionals. By building an intelligent imaging ecosystem with focused AI solutions, radiologists can increase efficiency and achieve more desirable patient outcomes — not in the distant future, but today.
In this blog, we explore how AI is currently being integrated to streamline radiology workflows.
1. Intelligent Worklist Distribution
Dr. Chopra recently conducted an informal survey of about 30 practicing radiologists and found that approximately 70% agreed that unequal work distribution and unfair compensation are among the major sources of job frustration.
Many PACS assign imaging studies based on self-selection or rules-based methods, contributing to inconsistent caseload distribution. At sites where compensation has a “relative value unit” (RVU) component, some radiologists may “cherry pick” studies, selecting higher paying or easier cases, leaving the remaining radiologists frustrated. Partly automated or manual caseload assignments further contribute to burnout and delays in delivery for patient care.
There is still room for improvement in optimising study assignments based on radiologists’ specialities and preferences. An AI-driven “intelligent worklist” approach can assist in balancing workloads and reducing variability by considering factors such as fair RVU distribution, radiologist specialty, urgency, peak times, etc.
A recent study presented at the SIIM (Society of Imaging Informatics Management) Conference on Machine Intelligence in Medical Imaging (CMIMI’24) in October provided strong evidence showing the intelligent worklist improving caseload management. The study compared radiologists’ performance with and without an intelligent worklist and found that an intelligent worklist approach resulted in a 34% more equitable distribution of studies. By dynamically balancing studies based on priority, value, radiologist preference and specialty, and turnaround time, an intelligent worklist ensures that studies are not simply assigned on a first come, first serve basis, but instead matched to the most qualified radiologists.
2. Generative AI for Reporting Workflows
AI has the potential to enhance reporting efficiency by integrating AI-generated results into radiology workflows and automating key steps in viewing and reporting activities. AI findings like measurements, can be displayed as DICOM image annotations within the viewer, allowing radiologists to quickly confirm or reject results and automatically populate structured reports with quantitative data.
This eliminates the manual work of measuring findings and inputting data into reports, which is useful for tracking disease progression over time. With reference to a study on AI-assisted lung nodule detection, we observed a 23% increase in efficiency in tracking patients with four or more lung nodules compared to manual methods.
Generative AI is also advancing with capabilities to accurately transcribe dictation into structured text that aligns with the radiologist’s reporting style. At RSNA 2023, Rad.ai became the first to demonstrate this unique capability, building on previous innovations in automated generation of impressions. Automated AI impressions are now widely available through vendors such as M*Modal, and major reporting vendors are integrating AI to populate dictated prescriptions to structured reports.
Moreover, generative AI excels at interpreting complex radiology and cardiology reporting guidelines. Our research shows that an AI agent for reporting can work alongside another guideline-trained AI agent to help radiologists create more accurate and compliant reports.
3. Automated Exam Follow-Up Detection
According to the American College of Radiology, up to 10% of radiology reports contain follow-up recommendations — yet about half of these follow-ups are never acted upon by patients. Alarmingly, lung nodules account for nearly half of these missed follow-ups, with over 60% of patients with incidentally detected pulmonary nodules, failing to seek appropriate care.
This presents an opportunity for improvement, and AI can help bridge this gap. Natural language processing tools can identify findings in previous patient imaging reports that require follow-ups. This information can then be automatically cross-referenced with the physician’s scheduling system to check if the patient has been contacted for follow-up care. If not, the system can automatically initiate a follow-up outreach and appointments scheduling. Although this is a relatively simple and low-touch AI application, it has the potential implications for patient care and mortality rates.
Building an Intelligent, Future-Proof Imaging Ecosystem
AI can be an important tool in delivering faster and more accurate patient care, always complementing the radiologist’s expertise rather than replacing it. Given that radiology is already a highly digital area of medicine, it doesn’t require radical rebuilding to fully leverage AI’s capabilities.
The examples discussed illustrate how targeted AI applications, which are not diagnostic in nature, can help reshape radiologist workflows to become more resilient and intelligent imaging ecosystems. We believe that the industry is only beginning to explore how AI can transform healthcare for both clinicians and patients.
Conclusion
The takeaway: AI adoption may be slow, but a growing body of proof may soon change that.
Healthcare leaders that we engage with express a keen interest in evaluating, justifying, and investing in AI. While there are opportunities to implement incremental and valuable changes, many leaders are uncertain about where to begin. It is also good to note that AI consulting is a fast-growing area within healthcare IT professional services.
Copyright 2025 Merative. All Rights Reserved.
Contact us via email for more information.