Deep Learning-Based Early Detection of Lung Diseases Using Light Attention Visual Geometry Networks.
Gespeichert in:
| Titel: | Deep Learning-Based Early Detection of Lung Diseases Using Light Attention Visual Geometry Networks. |
|---|---|
| Autoren: | Christiyana, C. Callins, Senthilselvi, A., Elangovan, D., Dhanasekaran, S. |
| Quelle: | Cognitive Computation; Dec2025, Vol. 17 Issue 6, p1-32, 32p |
| Abstract: | Early detection of lung diseases is crucial for reducing the rate of mortality. However, traditional methods for lung disease detection are costly, time-consuming, and provide inaccurate disease diagnoses. Therefore, this study introduces the Light Attention Visual Geometry-based Oppositional Synergistic Swarm (LAVG-OSS) algorithm for the early detection of lung diseases through computed tomography scans. The LAVG-OSS algorithm combines Visual Geometry Group, Residual Neural Network, and Light Attention Connected Module to enhance feature extraction and accurately identify disease-affected regions. The proposed method applies several preprocessing techniques, such as image resizing, adaptive intensity adjustments, and noise elimination, followed by data augmentation to address overfitting issues. A modified deep autoencoder is employed for feature extraction, while the Light Attention Connected Module, together with Visual Geometry Group and Residual Neural Network focuses on relevant regions for disease detection. Synergistic Swarm Optimization with an Opposition Based Learning Strategy is implemented to avoid local optima and obtain faster convergence to improve computational efficiency. The model was tested on five datasets: DLCTlUNGDetectNet—Lung Tumor Dataset, Chest CT-Scan Images, Lung nodule infused images, Lung_cancer _images, and Medical Deepfakes: Lung_Cancer. Extensive experiments demonstrate that LAVG-OSS outperforms traditional methods in terms of accuracy (98.4%), precision (96.9%), recall (95%), F1-score (95.9%), specificity (97.5%), and area under the curve (0.97). On the whole, the proposed LAVG-OSS method has displayed 98.4% prediction accuracy for DLCTLUNGDetectNet, 98.8% prediction accuracy for Chest CT-Scan Images, 99.1% prediction accuracy for Lung Nodule-infused Images, 96.4% prediction accuracy for Lung Cancer Images, and 99.3% prediction accuracy for Medical Deepfakes. The results indicate the algorithm’s effectiveness in detecting lung diseases, providing a potential tool for improving early diagnosis and treatment outcomes. Future work will explore hybrid models integrating state-of-the-art techniques for enhanced accuracy and real-time application in clinical environments. [ABSTRACT FROM AUTHOR] |
| Copyright of Cognitive Computation is the property of Springer Nature and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.) | |
| Datenbank: | Complementary Index |
Full Text Finder 
Nájsť tento článok vo Web of Science | Abstract: | Early detection of lung diseases is crucial for reducing the rate of mortality. However, traditional methods for lung disease detection are costly, time-consuming, and provide inaccurate disease diagnoses. Therefore, this study introduces the Light Attention Visual Geometry-based Oppositional Synergistic Swarm (LAVG-OSS) algorithm for the early detection of lung diseases through computed tomography scans. The LAVG-OSS algorithm combines Visual Geometry Group, Residual Neural Network, and Light Attention Connected Module to enhance feature extraction and accurately identify disease-affected regions. The proposed method applies several preprocessing techniques, such as image resizing, adaptive intensity adjustments, and noise elimination, followed by data augmentation to address overfitting issues. A modified deep autoencoder is employed for feature extraction, while the Light Attention Connected Module, together with Visual Geometry Group and Residual Neural Network focuses on relevant regions for disease detection. Synergistic Swarm Optimization with an Opposition Based Learning Strategy is implemented to avoid local optima and obtain faster convergence to improve computational efficiency. The model was tested on five datasets: DLCTlUNGDetectNet—Lung Tumor Dataset, Chest CT-Scan Images, Lung nodule infused images, Lung_cancer _images, and Medical Deepfakes: Lung_Cancer. Extensive experiments demonstrate that LAVG-OSS outperforms traditional methods in terms of accuracy (98.4%), precision (96.9%), recall (95%), F1-score (95.9%), specificity (97.5%), and area under the curve (0.97). On the whole, the proposed LAVG-OSS method has displayed 98.4% prediction accuracy for DLCTLUNGDetectNet, 98.8% prediction accuracy for Chest CT-Scan Images, 99.1% prediction accuracy for Lung Nodule-infused Images, 96.4% prediction accuracy for Lung Cancer Images, and 99.3% prediction accuracy for Medical Deepfakes. The results indicate the algorithm’s effectiveness in detecting lung diseases, providing a potential tool for improving early diagnosis and treatment outcomes. Future work will explore hybrid models integrating state-of-the-art techniques for enhanced accuracy and real-time application in clinical environments. [ABSTRACT FROM AUTHOR] |
|---|---|
| ISSN: | 18669956 |
| DOI: | 10.1007/s12559-025-10522-1 |