The global race in artificial intelligence for lung cancer screening is beginning to move beyond a familiar question: how accurately can an algorithm detect a pulmonary nodule?
As screening programs expand, a broader challenge is emerging. AI must work not only at the moment a CT scan is interpreted, but across the entire screening pathway — from identifying eligible individuals and reading scans to managing nodules, comparing prior images and ensuring long-term follow-up.
That shift was a central theme at the “Global Lung Cancer Screening with AI Symposium,” hosted by Coreline Soft in Seoul on Sept. 12 during the 2026 World Conference on Lung Cancer (WCLC).
The symposium brought together experts from Korea, Europe and North America, including Jin Mo Goo, professor of radiology at Seoul National University Hospital, who has been involved in research on Korea’s national lung cancer screening program, and Annemiek Snoeckx, professor of radiology at Antwerp University Hospital (UZA) in Belgium and a leading expert in European lung cancer screening.
Their perspectives came from different healthcare environments, but pointed toward a similar conclusion: in large-scale screening, algorithm performance alone is no longer enough.
Lung Cancer Screening Is Becoming a Digital Pathway
Snoeckx highlighted one of the central challenges facing Europe as lung cancer screening moves toward broader implementation.
Screening does not begin and end with radiology. It starts with identifying high-risk individuals and continues through scheduling, CT acquisition, image interpretation, reporting, recall when necessary and follow-up examinations.
Radiology remains at the center of the process, but it is only one component of a much larger system.
That becomes increasingly important as screening volumes rise. Hospitals often operate different IT systems that do not communicate smoothly with one another, creating what is often described as an “IT silo.”
Even a highly accurate AI algorithm may have limited clinical impact if radiologists must repeatedly move between separate screens, applications and workstations to use it.
The next phase of lung cancer screening AI, therefore, is likely to depend heavily on workflow integration and automation.
AI analysis, reporting, nodule management and follow-up need to operate within the radiologist’s existing environment with as little additional intervention as possible.
This also changes the way AI accuracy itself is viewed.
Rather than pursuing incremental gains in nodule detection alone, future development may increasingly focus on reducing unnecessary findings, integrating nodule management systems, identifying likely normal or low-risk examinations and digitizing the overall screening workflow.
That could be particularly important in population-based screening, where most scans do not contain immediately actionable disease. If AI can reliably help triage lower-risk examinations, radiologists could devote more attention to cases requiring specialist judgment.
Snoeckx also stressed that AI and radiologists should be viewed as complementary rather than competing forces, while algorithms must be validated again in the populations and clinical environments where they are actually deployed.
Korea Offers an Early Look at the Real-World Challenges
Korea provides a different perspective because it has already accumulated years of experience operating a national lung cancer screening program since 2019.
Goo focused on what happens after an AI algorithm leaves the laboratory and enters everyday clinical practice.
Strong technical performance does not automatically translate into widespread use by physicians. New user interfaces, separate workstations, integration with existing picture archiving and communication systems (PACS), and differences in hospital networks and IT infrastructure can all become barriers.
In practice, usability and reliability within the existing reading environment can become almost as important as the algorithm itself.
This experience may also carry implications for Korean medical AI companies seeking to expand overseas.
Coreline Soft accumulated real-world experience through Korea’s national lung cancer screening program before moving more aggressively into international markets. When competing algorithms deliver increasingly comparable baseline performance, experience in solving practical workflow problems may become another point of differentiation.
The same argument applies to the scope of the technology.
Rather than focusing solely on pulmonary nodule detection, Coreline Soft’s AVIEW platform has expanded into functions including nodule detection and quantification, comparison with prior CT examinations, emphysema assessment, coronary artery calcium analysis and interstitial lung disease-related analysis.
Research accumulated in Korea has also helped support international use of AVIEW, while experience from European screening and validation projects is creating another layer of clinical evidence.
In medical AI, particularly in screening, this creates a potentially important cycle: academic evidence supports real-world deployment, deployment generates new evidence, and those references can help open the next market.
Korea’s Experience Meets Europe’s Scaling Challenge
Korea and Europe are approaching lung cancer screening from different stages of implementation.
Korea has already encountered the practical question of what happens when AI is introduced into a national screening environment. Europe, meanwhile, is increasingly confronting the question of how screening can be expanded across institutions, regions and national healthcare systems.
Yet both are arriving at much the same problem.
In Korea, issues such as usability, reading workload, PACS integration and physician adoption have emerged through real-world operation. In Europe, the challenge increasingly involves connecting multiple institutions, integrating fragmented IT systems and automating follow-up as programs grow.
The direction of travel can be summarized as a continuous pathway:
Eligibility → CT acquisition → interpretation → prior-image comparison → nodule management → reporting → recall → long-term follow-up
AI that performs well at only one point in that chain may no longer be sufficient for large-scale screening.
That is also the context behind Coreline Soft’s message at WCLC 2026: “From Detecting Nodules to Managing Screening Pathways.”
Building on its experience in Korea’s national screening program, the company has entered screening environments in Germany, France, Italy and Australia, extending its focus from nodule detection toward prior-image comparison, longitudinal nodule analysis, structured reporting, multi-center reading and follow-up management.
The larger industry question is therefore shifting. The differentiator may no longer be simply which AI detects lung nodules most accurately, but which system can fit into different healthcare environments and connect interpretation, nodule management and longitudinal follow-up without adding complexity for clinicians.
“While lung cancer screening systems and healthcare environments differ from country to country, the challenge of reliably connecting interpretation and follow-up becomes common as screening populations and data volumes increase,” said Coreline Soft CEO Kim Jin-kook. “Based on the real-world experiences shared by medical professionals from Korea and abroad at this symposium, we will continue developing screening workflows that can be sustainably used in clinical practice.”
WCLC 2026 continues through Sept. 15 at COEX in Seoul. Coreline Soft is presenting AVIEW LCS Plus and other chest CT-based AI solutions at Booth No. 829, alongside its experience in national and multi-center lung cancer screening programs.
