Follow-up to ScreenTrustCAD presented at RSNA 2024 highlights 1-year post-AI adoption success in improving cancer detection, radiologist workload, and screening efficiency
SEOUL, South Korea, December 3, 2024 -- Lunit (KRX:328130.KQ), a leading provider of AI-powered solutions for cancer diagnostics and therapeutics, unveiled follow-up findings from the groundbreaking ScreenTrustCAD trial at the Radiological Society of North America (RSNA) 2024 Annual Meeting. This study, conducted at Capio S:t Göran Hospital in Sweden, reinforces the real-world impact of Lunit INSIGHT MMG, which has successfully replaced one of the two human readers in the hospital’s double-reading breast cancer screening protocol.
The original ScreenTrustCAD trial (2021–2022), published in The Lancet Digital Health, was the first prospective interventional study to demonstrate that AI could match or even surpass traditional two-radiologist systems. Building on these findings, the hospital fully integrated AI in mid-2023 and became the first in the world to implement AI as an independent reader, fully replacing one radiologist. The new analysis compares screening performance before and after AI adoption, offering real-world validation of Lunit INSIGHT MMG’s clinical and operational benefits.
Real-World Results: Efficiency Gains and Enhanced Screening Metrics
The follow-up study analyzed screening metrics from two nine-month periods: July 2018–March 2019 (pre-AI / pre-COVID) and July 2023–March 2024 (post-AI implementation).
In brief
Lunit reported follow-up results from the ScreenTrustCAD breast-screening deployment at Capio S:t Göran Hospital in Sweden, presented at RSNA 2024. The analysis compared screening periods before and after the hospital adopted Lunit INSIGHT MMG as an independent reader, replacing one of two radiologists in its double-reading protocol. Lunit says the post-adoption period showed lower recall and false-positive rates, higher positive predictive value and cancer detection, and reduced reading time. The company and a study presenter said the change also eased radiologist workloads and shortened waits for clinical mammograms. The release describes the findings as real-world evidence from the hospital; it does not provide details about other sites or broader deployment.
Key takeaways
- Lunit says the hospital used its AI system in place of one reader in a two-reader breast-screening process.
- Recall rates fell from 2.8% to 2.5% in the compared screening periods.
- Cancer detection rose from 4.8 to 5.5 per 1,000 screened, according to the company.
- The company reported reading time decreased by over 36% after adoption.
Key findings include:
- 11% Decrease in Recall Rates: Recall rates dropped from 2.8% to 2.5%, minimizing unnecessary follow-ups for patients.
- Fewer False Positives: The rate of false positives decreased significantly from 89.6% to 78.0%, reflecting a notable improvement in screening accuracy.
- Improved Positive Predictive Values (PPV): The PPV for recalls increased to 22.1% with AI, compared to 16.9% without AI, indicating fewer unnecessary recalls and a more efficient screening process.
- Increased Cancer Detection Rates: Cancer detection rates increased by 15%, rising from 4.8 (traditional double reading) to 5.5 per 1,000 screened (with AI).
- Reduced Overall Reading Time: Reading time decreased by over 36%, enabling radiologists to allocate more time to complex cases and other critical tasks.
These findings demonstrate AI’s ability to enhance diagnostic accuracy and significantly improve screening efficiency.
Pioneering AI Integration in Breast Screening
The adoption of Lunit INSIGHT MMG at Capio S:t Göran Hospital marks a global milestone in AI-driven healthcare. By replacing one human reader, the AI system addresses the growing radiologist shortage while maintaining the rigor of double-reading protocols. Additionally, it reduces recall rates, sparing more women the stress and anxiety of a potential cancer diagnosis. These results align with Lunit’s mission to advance equitable, efficient, and high-quality cancer detection while improving patient experiences and outcomes.
“AI has become an indispensable part of our screening program, and these results showcase its potential for broader adoption," said Dr. Karin Dembrower, senior breast radiologist at Capio S:t Göran Hospital and study presenter. "By replacing one radiologist, we’ve reduced workloads, improved cancer detection rates, and decreased unnecessary recalls. This shift has eliminated overtime on evenings and weekends, allowing radiologists to focus on complex cases—the work we are trained for. For patients, it means faster and more accurate care. Previously, the waiting time for a clinical mammogram was five to six weeks. Now, with AI, it’s down to zero.”
“This follow-up to the ScreenTrustCAD trial demonstrates the transformative potential of Lunit INSIGHT MMG in real-world clinical practice," said Brandon Suh, CEO of Lunit. "By reducing reading time by 36% while improving cancer detection rates and screening efficiency, we are proving that AI is not just a tool but a critical partner in addressing global challenges like radiologist shortages and the growing demand for efficient cancer care. As AI continues to reshape radiology, we remain committed to driving innovations that empower clinicians, optimize patient outcomes, and redefine the future of breast cancer screening.”
Explore Lunit’s innovations at booth #4929 in the AI Showcase during RSNA 2024, December 1–6 in Chicago, Illinois.
If you detect any issues, problems, or errors in this press release content, kindly contact [email protected] to notify us (it is important to note that this email is the authorized channel for such matters, sending multiple emails to multiple addresses does not necessarily help expedite your request). We will respond and rectify the situation in the next 8 hours.
Questions around this news
What did the follow-up analysis of the ScreenTrustCAD deployment report?
The analysis compared screening metrics before and after AI adoption at Capio S:t Göran Hospital. It reported lower recall and false-positive rates, a higher positive predictive value, and an increase in cancer detection from 4.8 to 5.5 per 1,000 screened. Reading time also fell by over 36%, according to the release.
How was AI used in the hospital’s breast cancer screening program?
The hospital incorporated an AI system as an independent reader, replacing one of the two human readers in its double-reading protocol. The release says the hospital began full integration in mid-2023, following the earlier ScreenTrustCAD trial, and describes it as the first implementation of this kind.
What does positive predictive value mean in breast screening?
Positive predictive value (PPV) is the proportion of positive screening results that are confirmed to indicate the condition being screened for. In breast screening, it can help describe how often a recall or other positive finding corresponds to cancer. PPV is influenced by the population screened and how positive results are defined.
What is double reading in mammography screening?
Double reading is a screening approach in which two readers independently review the same mammograms. If their assessments differ, a program may use a defined process to resolve the disagreement. The approach provides a second review, though its staffing and workflow requirements vary across screening programs.
Earlier from Lunit
- Lunit Inks Partnership with Salud Digna to Advance AI in Medical Imaging
- Lunit Announces Collaboration with AstraZeneca to Develop AI-Powered Digital Pathology Risk Assessment Tools for NSCLC
- Lunit Shows Promise of AI in Predicting Immunotherapy Response for Rare Cancer Patients at SITC 2024
- Lunit AI Predicts Breast Cancer Risk Up to 6 Years in Advance
- Lunit Joins Roche's Digital Pathology Open Environment to Advance Cancer Biomarker Testing
