Related Experiment Video
Updated: Aug 15, 2025
10:26
Author Spotlight: A 3D Digital Model for the Diagnosis and Treatment of Pulmonary Nodules
Published on: May 19, 2023
2.0K
An improved faster R-CNN algorithm for assisted detection of lung nodules
Jing Xu1, Haojie Ren1, Shenzhou Cai1
1School of Statistics and Mathematics, Zhejiang Gongshang University, Hangzhou, 310018, China; Collaborative Innovation Center of Statistical Data Engineering, Technology & Application, Zhejiang Gongshang University, Hangzhou, 310018, China.
Computers in Biology and Medicine
|January 1, 2023
Summary
Early detection of lung cancer is crucial. This study introduces an improved Faster R-CNN model for accurate pulmonary nodule detection in CT scans, enhancing early diagnosis and patient outcomes.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Computer-Aided Diagnosis
Background:
- Lung cancer morbidity and mortality are rising globally.
- Pulmonary nodules are early indicators of lung cancer, necessitating timely diagnosis.
- Computed Tomography (CT) is vital for screening, but manual review is time-consuming and prone to errors.
Purpose of the Study:
- To develop an automated system for early pulmonary nodule detection.
- To improve the accuracy and efficiency of lung nodule identification in CT scans.
- To reduce the diagnostic burden on radiologists.
Main Methods:
- An improved Faster R-CNN deep learning model was developed.
- Multi-scale training and path augmentation enhanced small object detection.
- Online Hard Example Mining (OHEM) adaptively adjusted training.
- Deformable convolution improved feature extraction and global context.
Main Results:
- The improved Faster R-CNN model achieved 90.7% precision (up from 76.4%) and 56.8% recall (up from 40.1%) on the LUNA16 dataset.
- The model demonstrated superior performance compared to YOLOv3 and Cascade R-CNN.
- Enhanced ability to detect small pulmonary nodules and improved feature extraction.
Conclusions:
- The proposed improved Faster R-CNN model significantly enhances pulmonary nodule detection accuracy and efficiency.
- This AI-driven approach can aid radiologists in early lung cancer diagnosis.
- The method shows promise for real-time batch processing of CT scans, improving clinical workflow.
