Opportunistic CT Screening Enters a New Era
Technological advances are expanding the potential for using CT screening for preventive care
In concert with the RadioGraphics monograph, RSNA News shares the latest on the role of medical imaging in disease screening.
Opportunistic CT screening—in which radiologists extract clinically relevant imaging biomarkers from CT scans being performed for unrelated primary indications—has reached an inflection point, according to Siddhant Dogra, MD, of the Department of Radiology at NYU Grossman School of Medicine in New York City.
“The goal of opportunistic CT screening is to identify patients who already have an underrecognized condition, or who are at elevated risk for a condition, early enough to intervene before they experience an adverse event,” said Dr. Dogra, who co-authored a RadioGraphics article on the subject.
According to Dr. Dogra, AI and other technologies involved in opportunistic screening have rapidly improved and have become more readily accessible in recent years. In addition, a growing number of studies have demonstrated potential clinical and financial value, supporting clinical implementation.
“This is a great moment to introduce readers who are new to the concept of opportunistic screening to the most established applications with the greatest evidence base, as well as to briefly highlight emerging areas of research,” he said.
“Just as importantly, we wanted to help move the conversation beyond technical feasibility and focus on what is required for successful clinical deployment,” Dr. Dogra added.
“Advances in image segmentation and automated analysis now allow algorithms to identify the relevant anatomy accurately and perform standardized measurements across large numbers of routine CT examinations.”
— SIDDHANT DOGRA, MD
Applications for Opportunistic CT Screening
Technological advances have made it easier to do things such as measure vertebral body trabecular attenuation for low bone mineral density, something that has traditionally been done manually, Dr. Dogra noted.
“The problem is that manually performing these measurements is too time-consuming and potentially too variable to be practical at scale,” he said. “Advances in image segmentation and automated analysis now allow algorithms to identify the relevant anatomy accurately and perform standardized measurements across large numbers of routine CT examinations.”
Moreover, because the segmented structures can be displayed visually, radiologists can also review the outputs relatively quickly for quality control, he noted.
“Some of the underlying segmentation tools are even available as open-source software, such as TotalSegmentator,” Dr. Dogra said.
Similar approaches can segment skeletal muscle to calculate muscle area and attenuation as markers of sarcopenia and myosteatosis.
“Algorithms can also quantify hepatic attenuation to detect hepatic steatosis and segment visceral and subcutaneous adipose tissue to provide measures of body composition and cardiometabolic risk,” he said.
According to Dr. Dogra, newer deep learning models may expand opportunistic screening beyond these relatively transparent, measurement-based applications.
“For example, models are being explored that can predict diseases and future health risks from imaging patterns that radiologists may not be able to perceive reliably,” he said. “These approaches are promising and potentially very powerful, but they can also be more difficult to interpret and lack quality-control, so many remain areas of active research.”
Opportunistic assessment of vertebral bone mineral density (BMD). Sagittal noncontrast CT images of the abdomen and pelvis before opportunistic assessment (A), with automated segmentation (color overlay in B) of vertebral bodies across multiple levels (B), and with regions of interest (green markers in C) placed on trabecular bone (C), enabling attenuation-based screening for low BMD.
https://doi.org/10.1148/rg.260009 ©RSNA 2026
Navigating Challenges
Establishing an opportunistic screening program can be a tremendous undertaking, regardless of whether an institution develops its own algorithm or integrates a commercial product, according to Dr. Dogra.
“The model output isn't necessarily useful by itself; institutions must validate any algorithm on their own patient population and imaging protocols, integrate it into existing clinical workflows, establish quality control and ongoing monitoring processes and determine how results will be communicated and acted upon,” he said. “These efforts are on top of the demands of an already very busy clinical practice.”
In addition, he noted that these tasks require financial and institutional investment, while reimbursement pathways and prospective evidence of improved outcomes have historically been limited.
“Recent modeling studies demonstrating potential clinical and financial value, prospective implementation and outcomes data, a growing number of commercial tools, and new potential reimbursement pathways will hopefully help reduce these barriers and provide stronger incentives for adoption,” he said.
Dr. Dogra offered advice to radiologists interested in doing more opportunistic CT screening.
“Take an honest look at the needs and capabilities of your own group or institution. Instead of immediately jumping in with an algorithm, start by identifying a meaningful clinical gap,” he said.
Some appropriate questions include: Which condition is commonly underrecognized in your patient population? Is there strong evidence that detecting it will change management and potentially change outcomes? Is there an established pathway for evaluation and treatment?
“Applications such as osteoporosis might be reasonable starting points because the evidence base and downstream clinical pathways are well established,” Dr. Dogra said.
The next key step is to bring the relevant stakeholders together early.
“A truly successful opportunistic screening program requires collaboration among radiologists, informaticians, referring clinicians and operational leadership,” he said.
Before deployment, stakeholders should agree on the care pathway, including who will receive the results, how they will be communicated, who will assume responsibility for follow-up and what action an abnormal result should prompt, he added.
“Once the algorithm is developed or licensed, it should be validated locally across your patient population, scanners and imaging protocols,” Dr. Dogra continued. “Of course, once deployed, radiologists should monitor algorithm performance but also track operational and clinical outcomes—for example, how often results are acted upon, whether the patients complete follow-up and whether the program ultimately changes management or improves outcomes.”
Opportunistic screening lets radiologists potentially contribute to preventive care, risk stratification and population health, and overall engage more deeply in interdisciplinary work, he noted.
“In that sense, I think opportunistic screening is an example of how AI can help radiologists perform new types of clinically meaningful work, rather than just interpret existing examinations more quickly or more accurately,” Dr. Dogra said.
For More Information
Access the RadioGraphics article, “Opportunistic CT Screening: Clinical Applications, Technical Foundations, and Best Practices.”
Read previous RSNA News stories on opportunistic screening: