Automated Lung Sound-Based Disease Classification Using a Multi-Model Pre-Trained CNN Framework
Emerging Science Innovation · 28.05.2026 · 7 · 01-13
DOI: 10.46604/emsi.2026.16374 ↗Kurz gefasst
Fachartikel in der Zeitschrift Emerging Science Innovation, online erschienen am 28.05.2026. Verfasst von Nisha Palanisamy und 1 weiteren Autor:innen. Die Arbeit nennt 21 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC 4.0).
Abstract (Originalsprache)
The early identification of pulmonary diseases is essential for timely treatment and better patient care. This study presents a deep learning framework for automated classification of lung sound based diseases using nine pre-trained convolutional neural networks (CNNs). The lung sounds are filtered using a sixth-order Butterworth bandpass filter and converted into log-Mel spectrograms before training. Evaluations of the candidate models include to ensure the reliability and clinical relevance of the diagnostic results. EfficientNet-B0 achieves the highest accuracy of 99.20%, precision of 99.20%, sensitivity of 99.20%, specificity of 99.73%, and F1-score of 99.22%, followed by DenseNet201 (99.00%) and ResNet101 (98.90%). These results demonstrate that pre-trained CNNs can learn effective, discriminative features from lung sounds, leading to precise and efficient classification of respiratory diseases.
Zitieren
APA
Palanisamy, N., & Jeganathan, V. (2026). Automated Lung Sound-Based Disease Classification Using a Multi-Model Pre-Trained CNN Framework. Emerging Science Innovation, 7, 01-13. https://doi.org/10.46604/emsi.2026.16374
Vancouver
Palanisamy N, Jeganathan V. Automated Lung Sound-Based Disease Classification Using a Multi-Model Pre-Trained CNN Framework. Emerging Science Innovation. 2026;7:01-13. doi:10.46604/emsi.2026.16374
BibTeX
@article{Palanisamy2026,
author = {Palanisamy, Nisha and Jeganathan, Vijayakumar},
title = {Automated Lung Sound-Based Disease Classification Using a Multi-Model Pre-Trained CNN Framework},
journal = {Emerging Science Innovation},
year = {2026},
volume = {7},
pages = {01--13},
doi = {10.46604/emsi.2026.16374}
}
EUVIMED