Related Experiment Video
Updated: Dec 14, 2025
04:23
A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images
Published on: April 21, 2023
2.2K
Lung Nodule Detection using Convolutional Neural Networks with Transfer Learning on CT Images
Jun Gao1, Qian Jiang1, Bo Zhou2
1College of Information Engineering, Shanghai Maritime University, Shanghai 201306, China.
Combinatorial Chemistry & High Throughput Screening
|July 16, 2020
Summary
This study presents an advanced Convolutional Neural Network (CNN) method for lung nodule detection in CT scans. The developed technique achieves high accuracy and specificity, improving early lung cancer diagnosis.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Oncology
Background:
- Lung nodule detection is crucial for early lung cancer diagnosis and improving patient survival rates.
- Convolutional Neural Networks (CNNs) show promise for automated lung nodule detection in Computed Tomography (CT) images.
- Further improvements in accuracy and efficiency are needed for clinical application.
Purpose of the Study:
- To develop and evaluate an automatic lung nodule detection method using CNNs with transfer learning.
- To compare the performance of VGG16, VGG19, and ResNet50 models for lung nodule detection.
- To optimize CNN hyperparameters and training strategies for enhanced performance.
Main Methods:
- Compared three state-of-the-art CNN models (VGG16, VGG19, ResNet50).
- Utilized transfer learning with layer freezing and fine-tuning strategies.
- Optimized hyperparameters including optimizer, batch size, and epoch.
Main Results:
- Achieved high accuracy (96.86%), precision (91.10%), sensitivity (90.78%), specificity (98.13%), and AUC (99.37%) on the LUNA16 dataset.
- Demonstrated the effectiveness of transfer learning in lung nodule detection.
- Obtained state-of-the-art specificity compared to existing methods.
Conclusions:
- The proposed CNN-based method with transfer learning is effective for lung nodule detection.
- The approach achieves high diagnostic performance, particularly in specificity.
- This method shows significant potential for clinical application in lung cancer screening.
