Despite a preponderance of evidence supporting lung cancer screening with low-dose CT (LDCT), including the U.S. National Lung Screening Trial (NLST) and other trials, uptake has been lower than many advocates would have liked.
Radiologists, primary care providers, data scientists, and screening advocates have been considering how to discern lung cancer screening (LCS) patterns, optimize follow-up CT screening, and reduce barriers, especially for those at highest risk.
For this special edition, AuntMinnie highlights the top stories in lung cancer screening this year in the medical imaging community and some of the AI systems noted for aiding doctors in their decision-making. We also highlight important topics in lung cancer screening at RSNA 2025.
Despite a preponderance of evidence supporting lung cancer screening with low-dose CT (LDCT), including the U.S. National Lung Screening Trial (NLST) and other trials, uptake has been lower than many advocates would have liked.
Radiologists, primary care providers, data scientists, and screening advocates have been considering how to discern lung cancer screening (LCS) patterns, optimize follow-up CT screening, and reduce barriers, especially for those at highest risk.
For this special edition, AuntMinnie highlights the top stories in lung cancer screening this year in the medical imaging community and some of the AI systems noted for aiding doctors in their decision-making. We also highlight important topics in lung cancer screening at RSNA 2025.
More eligible
Cancer screening advocates know that changes to the U.S. Preventive Services Task Force (USPSTF) lung cancer screening guidelines increased eligibility for individuals. However, they have also seen that greater eligibility alone does not translate into stronger participation in LCS screening programs.
For example, researchers at the David Geffen School of Medicine at the University of California, Los Angeles (UCLA), hypothesized that patients undergoing LCS after the release of the USPSTF's updated guidelines would be younger, more racially and ethnically diverse, and have less of a smoking history.
Medical imaging informatics researcher Yannan Lin, MD, PhD, and colleagues were correct about age and smoking history, and the guideline changes prompted a strong showing of newly eligible participants. They were surprised, though, to see less racial and ethnic diversity among participants, although some are now considered at higher risk for lung cancer.
LCS unknowns
Lung cancer is often diagnosed at later stages, when it is less likely to be curable. Unfortunately, screening participation patterns can be difficult to discern, a recent analysis of lung cancer screening data found.
Take the American Lung Association's (ALA) annual "State of Lung Cancer" reports. Researchers from Indiana University School of Medicine in Indianapolis suggested that a change in data-sharing requirements can make it "impossible" to gather true screening trends apart from methodological artifacts.
However, researchers were able to document a significant jump in lung cancer (high-risk) screening -- from 4.5% in 2023 to 16% in 2024. The Behavioral Risk Factor Surveillance System (BRFSS) played a key role in determining the better numbers, according to radiologist Peter Gunderman, MD, and colleagues. Since then, screening has ticked slightly upward again, as noted in the ALA 2025 report released November 4.
The American Lung Association's (ALA) annual State of Lung Cancer report released November 4 created tiers highlighting the rate of new lung cancer cases. The report includes state-specific measures of lung cancer incidence, adult smoking prevalence, estimated percent of radon tests at or above the U.S. EPA action level, five-year survival, early diagnosis, surgery as part of the first course of treatment, lack of treatment, and screening among those at high risk. American Lung Association
Looking at lung cancer screening and early diagnosis alone, the 2025 ALA report noted that screening rates were the best in Rhode Island (31% and 35.5%, respectively) and worst in Wyoming (9.7%). Early diagnosis rates were worst in Hawaii (21.8%).
Utah had the nation’s lowest lung cancer incidence rate, while Kentucky had the highest, according to the latest report.
LCS barriers
It is good news that lung cancer screening rates appear to be increasing, especially for those at high risk. At the American Roentgen Ray Society (ARRS) 2025 annual meeting, cardiothoracic radiologist Lauren Groner, DO, of Weill Cornell Medicine in New York City, offered insights into how to keep those numbers trending higher.
To get a better picture, researchers at Weill Cornell conducted focus groups with primary care providers (PCPs) and determined that several barriers -- at the provider, patient, and healthcare system levels -- perpetuate low lung cancer screening rates overall:
Insurance preauthorization
Patients' cognitive and psychosocial states
Provider-patient knowledge and communication
The culture of a busy primary care practice
Lack of patient follow-through on ordered exams
"[These] barriers affect providers' capability, opportunity, and motivation to screen patients for lung cancer," Groner said at ARRS. The most prominent barriers correspond to environmental context and resources, followed by beliefs about LCS consequences and knowledge, Groner added.
"Screening requires access to a center that offers LDCT, insurance coverage, a primary care office that can navigate the insurance-approval process, and further time and transportation to follow through with the scan,” wrote Madison Wulfeck, MD, a cardiothoracic radiologist with Radiology Partners-Florida, for an American College of Radiology (ACR) November 10 blog.
Lung cancer prevalence
American Cancer Society (ACS) data indicate that by January 2035, lung cancer prevalence for men and women is projected to increase by 27% and 29%, respectively.
LDCT is the gold standard for lung cancer screening and has been shown to reduce lung cancer mortality -- by up to 20%, according to ALA's new report.
The problem is that some screening trials have reported high false-positive rates -- which can lead to unnecessary follow-up procedures, patient anxiety, and higher healthcare costs. To make things trickier, pulmonary nodules found on LDCT are common, but determining which are malignant can be a challenge. Data scientists are working on that.
Nodule findings
"AI accounts for factors that we might not even see on the CT scan to further assess a nodule as likely to be malignant," said doctoral candidate Noa Antonissen, MD, of Radboud University Medical Center in Nijmegen, the Netherlands, in a RSNA statement in September.
Scientists at Radboud have developed an AI deep-learning tool that estimates the malignancy risk of lung nodules. They found that the algorithm reduced false-positive results by almost 40%.
"Accurate risk classification of these nodules could reduce unnecessary procedures," Antonissen and colleagues wrote in a paper published in Radiology.
"Deep-learning algorithms can assist radiologists in deciding whether follow-up imaging is needed, but prospective validation is required to determine the clinical applicability of these tools and to guide their implementation in practice," the group noted. "Reducing false-positive results will make lung cancer screening more feasible."
Only a minority of lung nodules represent cancer, according to Mario Silva, MD, PhD, a chest radiologist at the University of Parma in Italy. Overdiagnosis risk warrants careful consideration of nodule assessment, Silva said in connection with ECR 2025 in Europe.
Silva advised that after a baseline screening, active surveillance of low-to-intermediate risk nodules should be based on the balance of the risk of stage shift. He cautioned that subsolid nodules are prototype examples of overdiagnosed findings. They are indicative of lung cancer, but they are also considered a biomarker of having a comorbidity from something else that causes death.
Overinvestigation
Because there is considerable variation in the progression of lung cancer in different people, a one-size-fits-all strategy for follow-up screening may not be optimal for many individuals. Minimizing harm and maximizing benefits from CT lung cancer screening reporting is a major objective and urgent priority, Silva emphasized.
"Avoid overinvestigation," Silva added. "It takes six to 12 months to see measurable accurate changes. We typically wait one to two years to see the clinical changes."
To that end, cardiothoracic radiologist Pedro Staziaki, MD, at the University of Vermont Medical Center in Burlington, clarified strides made to improve lung cancer risk prediction.
Another study found AI to be useful for three-year lung cancer risk prediction, according to the authors of a study published October 21 in Radiology. Lung cancer screening with LDCT relies on characterizing nodules using the Lung-RADS framework.
The study compared a three-year lung cancer prediction model enhanced by the incorporation of "global lung features," called ScreenLungNet, to Lung-RADs. Ultimately, the "lung milieu" raises questions about patient management, Staziaki penned in his accompanying editorial.
Risk prediction
"Even though the current way of assessing lung cancer screening CT examinations is efficient, pragmatic, well-researched, and well-developed, it can seem quite primitive," he wrote. "What I mean is that the Lung-RADS v2022 assessment doesn’t consider the lung parenchymal milieu of each patient."
In other words, factors such as fine particulates, chronic inflammation, emphysema, or pulmonary fibrosis are not accounted for in the lung cancer risk scoring system. This is where a global lung assessment with deep learning could be useful, he added.
"As radiologists concentrate on the task at hand (scrolling through images and identifying the most suspicious nodules), we often miss the bigger picture: each lung is a complex biologic system, composed of intricate tissues forming the parenchyma and stroma," Staziaki wrote. "These tissues have spent years cycling through large volumes of air and have developed specific characteristics, including injury, inflammation, and scarring."
The point is important because lung cancer deaths related to smoking are projected to decrease over time. But, as smoking habits have declined, the share of lung cancers in people who have never smoked is increasing, according to Staziaki.
In fact, some researchers are trying to determine the value of lung cancer screening in the general population, those not necessarily at risk. A group at Guangzhou Institute of Respiratory Disease in China suggests the need for more studies to evaluate the importance of identifying high-risk factors or prescreen enriching biomarkers.
A key question is LCS timing: annual or biannual. Canadian researchers have investigated using risk assessment scores to determine biannual screening eligibility. Their talk at RSNA 2025 will share results of a retrospective analysis of nearly 5,000 patients that determined how timing affected malignancy missed.
Radiologists or PCPs
For now, radiologists and PCPs have an opportunity to collaborate, according to the Weill Cornell group. They say radiologists are poised to lead efforts to improve lung cancer screening uptake and can help boost LCS uptake by educating and supporting their PCP peers. Looking forward, five strategies may help to guide their discourse:
Conduct educational meetings and training with primary care providers and clinic staff.
Develop and distribute educational materials (for example, a website Groner's group developed called LungCheck.org).
Develop an "implementation blueprint."
Train LCS navigators.
Modify the electronic health record system. One example would be embedding a link to LungCheck.org, a best practice advisory for LCS, and optimized smoking history tracking.
Also, look for a broadened American College of Radiology (ACR)'s Lung Cancer Screening Registry (LCSR). Established in 2015, the LCSR has been collecting detailed clinical data and reporting through Lung-RADS. However, this year, the LCSR is evolving into an Early Lung Cancer Detection Registry (ELCDR) for radiologists and other physicians.
"Along the way we realized that, while we've been managing incidental pulmonary nodules (IPNs) for a very long time, this issue hasn't received the same attention and performance improvement measures that lung cancer screening has had," explained Ella Kazerooni, MD, professor of radiology at University of Michigan Health, for the ACR. "Today, we recognize that early detection is a combination of both screening and incidental pulmonary nodule management."
ACR registry changes
Kazerooni, who also serves as vice chair of the National Comprehensive Cancer Network (NCCN) Lung Cancer Screening panel and chairs the ACR Committee on Lung Cancer Screening, including the Lung-RADS committee, confirmed that an estimated 60% of patients with actionable incidental pulmonary nodules (IPNs) don’t receive guideline-concordant follow-up imaging.
"Lack of appropriate recommendations for management, lack of communication about that follow-up to patients and ordering providers, and lack of tracking that the recommendations are completed all lead to lack of adherence to follow-up recommendations," Kazerooni continued.
Expected to be available late this year, the ELCDR will provide diagnostic performance feedback on management of actionable IPNs that focus both on the quality of radiology report recommendations and timely follow-up completion rates, added Ben Wandtke, MD, chief of diagnostic imaging at FF Thompson Hospital and professor of imaging sciences at the University of Rochester Medical Center in New York. Wandtke also serves as vice chair of the ACR Quality and Safety Commission.
AI tools are getting better at nodule detection, risk stratification, and even helping with incidental findings and follow-up, Wulfeck noted for the ACR. The challenge is facilitating seamless integration into clinical practice.
“Encourage your institution to make lung cancer screening a part of the routine care pathways," Wulfeck suggested. “Nudge primary care colleagues who don’t consistently refer patients and provide them with the resources they need to navigate the nuances. Advocacy doesn’t have a relative value unit conversion, but it has clinical value and can keep moving the needle in the right direction.”
RSNA 2025
RSNA 2025 includes many sessions dedicated to important topics in lung cancer screening. Please refer to the RSNA catalog. In the meantime, a small sampling is below.
ScreenLungNet: S4-SSCH02-1, Personalized long-term prediction of lung cancer risk from a single LDCT screening, Sunday, November 30, 1 p.m.-1:10 p.m., Room E353C.
Indeterminate lung nodule malignancy risk estimation: S4-SSCH02-3, Benchmarking radiologists and AI for estimation on screening CT, the LUNA25 Challenge, Sunday, November 30, 1:20 p.m.-1:30 p.m., Room E353C.
Biannual lung cancer screening: S4-SSCH02-4, using risk assessment scores, Sunday, November 30, 1:30 p.m.-1:40 p.m., Room E353C.