Related Experiment Video
Updated: Jan 9, 2026
10:26
Author Spotlight: A 3D Digital Model for the Diagnosis and Treatment of Pulmonary Nodules
Published on: May 19, 2023
2.5K
Risk-stratified classification of pulmonary nodule malignancy via a machine learning model integrating imaging and
Huiting Wang1, Hairong Huang2, Feng Li1
1Department of Thoracic Surgery and Oncology, The First Affiliated Hospital of Guangzhou Medical University, China State Key Laboratory of Respiratory Disease & National Clinical Research Center for Respiratory Disease, Guangzhou, 510165, China.
The Lancet Regional Health. Western Pacific
|December 1, 2025
Summary
This study developed an integrated machine learning model combining low-dose computed tomography (LDCT) radiomics and plasma cell-free DNA (cfDNA) fragmentomics for improved lung nodule malignancy classification and invasiveness prediction.
Area of Science:
- Oncology
- Radiology
- Genomics
Background:
- Accurate risk stratification of pulmonary nodules is crucial for early lung cancer detection.
- Machine learning models offer potential for improved classification and prediction.
- Integrating imaging and molecular data may enhance diagnostic accuracy.
Purpose of the Study:
- To develop and validate a machine learning model integrating LDCT radiomics and plasma cfDNA fragmentomics for pulmonary nodule malignancy classification and invasiveness prediction.
- To assess the performance of the integrated model compared to individual imaging and cfDNA models.
- To evaluate the model's ability to predict tumor aggressiveness.
Main Methods:
- A multicenter study with 1356 participants (discovery and validation cohorts).
- A deep learning imaging model for nodule detection and classification from LDCT scans.
- A cfDNA model analyzing four whole-genome fragmentation features.
- Integration of imaging and cfDNA models using a stacked ensemble algorithm.
- Development of a separate invasion prediction model.
Main Results:
- The integrated imaging-cfDNA model achieved high AUCs (0.950 internal, 0.966 external) for malignancy classification.
- The combined model demonstrated improved specificity (0.60) while maintaining 95% sensitivity.
- The invasion prediction model stratified lung cancers with AUCs of 0.884 (internal) and 0.880 (external).
Conclusions:
- This multimodal approach enhances pulmonary nodule risk stratification by integrating radiomic and molecular biomarkers.
- The model significantly improves diagnostic accuracy, potentially reducing unnecessary procedures and missed diagnoses.
- The integrated model shows clinical utility in lung cancer screening.
