Related Experiment Video
Updated: May 20, 2025
05:22
Author Spotlight: Demonstrating Systematic Endobronchial Ultrasound to New Endoscopists
Published on: August 11, 2023
1.7K
Diagnosis of Lung Cancer Using Endobronchial Ultrasonography Image Based on Multi-Scale Image and Multi-Feature
Huitao Wang1, Takahiro Nakajima2, Kohei Shikano3
1Department of Medical Engineering, Graduate School of Science and Engineering, Chiba University, Chiba 263-8522, Japan.
Tomography (Ann Arbor, Mich.)
|March 26, 2025
Summary
This study introduces M3-Net, an AI system using deep learning on EBUS images for early lung cancer detection. The AI model shows promise in improving diagnosis accuracy and patient survival rates.
Area of Science:
- Medical Imaging
- Artificial Intelligence in Oncology
- Pulmonology
Background:
- Lung cancer remains a leading cause of cancer mortality worldwide.
- Early diagnosis is critical for improving lung cancer survival rates.
- Advancements in AI and imaging offer new avenues for lung cancer detection.
Purpose of the Study:
- To develop a computer-aided diagnosis (CAD) system for lung cancer detection.
- To utilize endobronchial ultrasonography (EBUS) images and deep learning algorithms.
- To enhance early lung cancer detection and improve patient survival.
Main Methods:
- Proposed M3-Net, a multi-branch deep learning framework with an attention-based mechanism.
- Integrated multiple features for comprehensive lung cancer assessment.
- Validated the framework on a dataset of 1140 EBUS images from 95 patients (13 benign, 82 malignant).
Main Results:
- Achieved an accuracy of 0.76 and an F1-score of 0.75.
- Reported an Area Under the Curve (AUC) of 0.83.
- Demonstrated sensitivity of 0.72, specificity of 0.80, PPV of 0.80, and NPV of 0.75.
Conclusions:
- The proposed attention-based multi-feature fusion framework shows significant potential for assisting lung cancer diagnosis.
- M3-Net can enhance diagnostic performance by providing comprehensive information from EBUS images.
- This AI-driven approach could aid in earlier and more accurate lung cancer detection.
