Related Experiment Video
Updated: Jun 29, 2025
Author Spotlight: A 3D Digital Model for the Diagnosis and Treatment of Pulmonary Nodules
Published on: May 19, 2023
Predicting Invasiveness of Lung Adenocarcinoma at Chest CT with Deep Learning Ternary Classification Models
Zhengsong Pan1, Ge Hu1, Zhenchen Zhu1
1From the Department of Radiology (Z.P., Z. Zhu, W.S., L.S., Z.J.), Medical Research Center (G.H.), State Key Laboratory of Complex Severe and Rare Disease (G.H.), Peking Union Medical College Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, No. 1 Shuaifuyuan, Dongcheng District, Beijing 100730, China; 4 + 4 Medical Doctor Program (Z.P., Z. Zhu), Department of Epidemiology and Health Statistics (W.H.), Institute of Basic Medicine Sciences (W.H.), Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing, China; Deepwise AI Laboratory, Beijing Deepwise & League of PhD Technology, Beijing, China (W.T., Z. Zhou, Y.Y.); and Department of Computer Science, The University of Hong Kong, Hong Kong, China (Y.Y.).
Deep learning models with an adjudication strategy significantly improved the classification of lung adenocarcinoma invasiveness on CT scans. This approach enhances diagnostic accuracy for pure ground-glass nodules, aiding clinical management.
Area of Science:
- Radiology
- Artificial Intelligence
- Oncology
Background:
- Preoperative classification of lung adenocarcinoma invasiveness (preinvasive, minimally invasive, invasive) is crucial for patient management but challenging for pure ground-glass nodules (pGGNs).
- Deep learning (DL) offers potential for improving this ternary classification task.
Purpose of the Study:
- To evaluate a strategy combining DL models with an adjudication approach to enhance the prediction of lung adenocarcinoma invasiveness at chest CT.
- To assess the performance of this strategy in classifying pGGNs.
Main Methods:
- Developed six DL ternary classification models using a multicenter dataset of lung nodules.
- Incorporated framework optimization, joint learning, and an adjudication strategy (simulating multireader consensus) to refine DL models.
- Tested models on an external dataset of pGGNs.
Main Results:
- The proposed adjudication strategy significantly improved DL model performance in external testing (P < .001).
- For minimally invasive adenocarcinoma, the adjudicated model (model 6) achieved 85% accuracy, 75% sensitivity, and 89% specificity, outperforming the non-adjudicated model (model 3).
- The adjudicated model demonstrated superior consistency across diagnostic indexes compared to other models.
Conclusions:
- Combining framework optimization, joint learning, and an adjudication approach enhances DL-based classification of lung adenocarcinoma invasiveness on chest CT.
- This strategy improves diagnostic performance, particularly for challenging cases involving pure ground-glass nodules.
More Related Videos
07:53Author Spotlight: Advancing 3D Modeling for Enhanced Diagnosis and Treatment of Pulmonary Nodules in Early-Stage Lung Cancer
Published on: October 13, 2023
08:05Lung CT Segmentation to Identify Consolidations and Ground Glass Areas for Quantitative Assesment of SARS-CoV Pneumonia
Published on: December 19, 2020