Related Experiment Video
Updated: Sep 4, 2025
07:53
Author Spotlight: Advancing 3D Modeling for Enhanced Diagnosis and Treatment of Pulmonary Nodules in Early-Stage Lung Cancer
Published on: October 13, 2023
1.6K
Attention Layer-Based Multidimensional Feature Extraction for Diagnosis of Lung Cancer.
Manisha Bhende1, Anuradha Thakare2, V Saravanan3
1Marathwada Mitra Mandal's Institute of Technology, Pune, India.
Biomed Research International
|July 14, 2022
Summary
This study introduces an AI algorithm for diagnosing invasive adenocarcinoma nodules in lung CT scans, improving accuracy over traditional methods. The new approach enhances early detection of lung cancer, aiding in better patient outcomes.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Oncology
Background:
- Current lung cancer screening relies on radiologist experience with CT images, which has limitations.
- Intraoperative pathology has low accuracy for small nodules and is invasive.
- Accurate diagnosis of invasive adenocarcinoma nodules is crucial for patient prognosis.
Purpose of the Study:
- To develop and validate an algorithm for diagnosing invasive adenocarcinoma nodules in ground-glass pulmonary nodules using CT images.
- To improve diagnostic accuracy and efficiency compared to existing methods.
Main Methods:
- An algorithm was developed using CT image data, incorporating nodule space and plane features.
- Sample data were designed in 2D and 3D dimensions.
- A neural network structure based on attention mechanism and residual learning was employed, fusing 2D and 3D network features.
Main Results:
- The algorithm was tested on 1760 ground-glass nodules (5-20mm diameter).
- Cross-validation on 1420 invasive nodule samples showed 82.7% classification accuracy.
- Sensitivity was 82.9% and specificity was 82.6%.
Conclusions:
- The developed algorithm demonstrates promising performance in diagnosing invasive adenocarcinoma nodules.
- This AI-based approach offers a potential improvement for early lung cancer detection and diagnosis.
