Related Experiment Video
Updated: Sep 8, 2025
Author Spotlight: Advancing 3D Modeling for Enhanced Diagnosis and Treatment of Pulmonary Nodules in Early-Stage Lung Cancer
Published on: October 13, 2023
Deep Learning to Optimize Candidate Selection for Lung Cancer CT Screening: Advancing the 2021 USPSTF Recommendations
Jong Hyuk Lee1, Dongheon Lee1, Michael T Lu1
1From the Department of Radiology, Seoul National University Hospital, 101 Daehak-ro, Jongno-gu, Seoul 03080, Korea (J.H.L., C.M.P., J.M.G., H.K.); Department of Biomedical Engineering, Chungnam National University College of Medicine, Chungnam National University Hospital, Daejeon, Korea (D.L.); Cardiovascular Imaging Research Center, Massachusetts General Hospital and Harvard Medical School, Boston, Mass (M.T.L., V.K.R.); Department of Radiology, Seoul National University College of Medicine, Seoul, Korea (C.M.P., J.M.G., H.K.); Institute of Radiation Medicine (C.M.P., J.M.G.) and Institute of Medical and Biological Engineering (C.M.P.), Seoul National University Medical Research Center, Seoul, Korea; Cancer Research Institute, Seoul National University, Seoul, Korea (J.M.G.); and Department of Internal Medicine, Healthcare Research Institute, Healthcare System Gangnam Center, Seoul National University Hospital, Seoul, Korea (S.H.C.).
A deep learning model effectively identifies lung cancer screening candidates, validating its use alongside U.S. Preventive Services Task Force guidelines. This AI tool reduces unnecessary screenings while maintaining high cancer detection rates.
Area of Science:
- Radiology
- Artificial Intelligence
- Oncology
Background:
- A deep learning (DL) model for lung cancer screening requires external validation using recent, real-world, non-U.S. data.
- The 2021 U.S. Preventive Services Task Force (USPSTF) provides guidelines for lung cancer screening.
Purpose of the Study:
- To externally validate a DL model for identifying lung cancer screening candidates.
- To assess the added benefits of the DL model to the 2021 USPSTF recommendations.
Main Methods:
- A single-center retrospective study included smokers aged 50-80 who underwent chest radiography between 2004 and 2018.
- The DL model's performance was evaluated using receiver operating characteristic curve analysis (AUC).
- Added value was assessed by comparing lung cancer inclusion rates, CT screening candidate proportions, and positive predictive values (PPV).
Main Results:
- The DL model achieved AUCs of 0.68 and 0.75 for incident lung cancers in the overall and USPSTF-eligible cohorts, respectively.
- Applying the DL model reduced the proportion of CT screening candidates from 45.1% to 35.8% (P < .001) within the USPSTF-eligible sample.
- This reduction in screening candidates did not significantly impact the cancer inclusion rate (0.3% vs 0.3%, P = .85) or PPV (0.9% vs 0.7%, P = .42).
Conclusions:
- External validation confirmed the DL model's efficacy in identifying lung cancer screening candidates.
- The DL model offers added value to the 2021 USPSTF recommendations by reducing the number of individuals recommended for low-dose CT screening.
- The model maintains high cancer detection rates and positive predictive value, optimizing lung cancer screening efficiency.
