Related Experiment Video
Updated: Nov 5, 2025
Author Spotlight: A 3D Digital Model for the Diagnosis and Treatment of Pulmonary Nodules
Published on: May 19, 2023
Deep Learning for Malignancy Risk Estimation of Pulmonary Nodules Detected at Low-Dose Screening CT.
Kiran Vaidhya Venkadesh1, Arnaud A A Setio1, Anton Schreuder1
1From the Department of Medical Imaging, Radboud Institute for Health Sciences, Radboudumc, Nijmegen, the Netherlands (K.V.V., A.A.A.S., A.S., E.T.S., K.C., B.v.G., M.P., C.J.); Department of Digital Technology & Innovation, Siemens Healthineers, Erlangen, Germany (A.A.A.S.); Department of Radiology, Meander Medical Center, Amersfoort, the Netherlands (K.C.); Department of Diagnostic Imaging, Section of Radiology, Nordsjællands Hospital, Hillerød, Denmark (M.M.W.W.); Department of Medicine, Section of Pulmonary Medicine, Herlev-Gentofte Hospital, Hellerup, Denmark (Z.S.); and Department of Clinical Medicine, University of Copenhagen, Copenhagen, Denmark (Z.S.).
A new deep learning (DL) algorithm accurately estimates pulmonary nodule malignancy risk from CT scans. This AI tool shows performance comparable to thoracic radiologists, aiding lung cancer screening decisions.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Oncology
Background:
- Accurate malignancy risk estimation of pulmonary nodules on chest CT is vital for effective lung cancer screening.
- Current methods require careful interpretation, highlighting the need for advanced tools.
Purpose of the Study:
- To develop and validate a deep learning (DL) algorithm for estimating the malignancy risk of pulmonary nodules detected during screening CT scans.
Main Methods:
- A DL algorithm was trained on 16,077 nodules from the National Lung Screening Trial.
- External validation was conducted on three cohorts from the Danish Lung Cancer Screening Trial.
- Performance was assessed using the area under the receiver operating characteristic curve (AUC) and compared to the Pan-Canadian Early Detection of Lung Cancer (PanCan) model and clinicians.
Main Results:
- The DL algorithm significantly outperformed the PanCan model (AUC, 0.93 vs 0.90; P = .046).
- Algorithm performance was comparable to thoracic radiologists in cancer-enriched cohorts (AUC, 0.96 vs 0.90 and 0.86 vs 0.82).
Conclusions:
- The developed deep learning algorithm demonstrates excellent performance in estimating pulmonary nodule malignancy risk.
- This AI tool has the potential to provide reliable risk scores, assisting clinicians in optimizing lung cancer screening management.
