Solunum Sesi Sınıflandırması için Klasik ve Derin Öğrenme Modellerinin Karşılaştırılması

Yıl 2025, Cilt: 15 Sayı: 4, 1668 - 1695, 15.12.2025

Hayati Türe , Eren Aygün

https://doi.org/10.31466/kfbd.1715285

Öz

Bu çalışma, ICBHI 2017 solunum sesi veri seti üzerinde geliştirilmiş bir derin öğrenme yaklaşımıyla, normal ve patolojik solunum seslerini otomatik olarak sınıflandırmayı amaçlamaktadır. Çalışmada, segmentlenmiş solunum seslerinden MFCC, Mel spektrogram ve spektral öznitelikler elde edilmiş; ardından bu öznitelikler klasik makine öğrenmesi algoritmaları (XGBoost, SVM, KNN, Rastgele Orman) ve derin öğrenme modelleri (GhostNet v1-v4, EfficientNet-B0, ResNet50, MobileNetV3) ile eğitilmiştir. Veri artırma tekniklerinin (augmentasyon) katkısı da sistematik olarak incelenmiştir. Sonuçlar, GhostNet v4 modelinin %89 doğruluk ve 0.89 F1-skoru ile en iyi performansı gösterdiğini ortaya koymaktadır. Bu doğruluk oranı, ICBHI 2017 veri seti ile literatürde rapor edilen birçok yöntemi geride bırakmaktadır. Ayrıca, karışıklık matrisi analizleri modelin normal ve patolojik sınıfları yüksek tutarlılıkla ayırt edebildiğini göstermektedir. Elde edilen sonuçlar, akciğer seslerinin otomatik analizi için derin öğrenme temelli modellerin etkinliğini ortaya koymakta ve klinik karar destek sistemlerine entegre edilebilecek potansiyel çözümler sunmaktadır.

Anahtar Kelimeler

Solunum sesi analizi , Derin öğrenme , Makine Öğrenmesi , MFCC , Mel spektrogram , GhostNet

Comparison of Classical and Deep Learning Models for Respiratory Sound Classification

Öz

This study aims to automatically classify normal and pathological respiratory sounds using a deep learning-based approach developed on the ICBHI 2017 respiratory sound dataset. MFCCs, Mel spectrograms, and spectral features were extracted from segmented respiratory sound recordings. These features were then used to train classical machine learning algorithms (XGBoost, SVM, KNN, Random Forest) and deep learning models (GhostNet v1–v4, EfficientNet-B0, ResNet50, MobileNetV3). The impact of data augmentation techniques was also systematically examined. The results demonstrate that the GhostNet v4 model achieved the highest performance with 89% accuracy and an F1-score of 0.89. This accuracy outperforms many existing methods reported in the literature using the same dataset. Confusion matrix analyses further indicate that the model reliably distinguishes between normal and pathological classes. These findings highlight the effectiveness of deep learning-based models in the automatic analysis of respiratory sounds and suggest promising solutions for integration into clinical decision support systems.

Anahtar Kelimeler

Respiratory sound analysis , Deep learning , Machine learning , MFCC , Mel spectrogram , GhostNet

Kaynakça

• Bardou, D., Zhang, K., & Ahmad, S. M. (2018). Lung sounds classification using convolutional neural networks. Artificial Intelligence in Medicine, 88, 58-69.
• Chen, H., Yuan, X., Pei, Z., Li, M., & Li, J. (2019). Triple-classification of respiratory sounds using optimized s-transform and deep residual networks. IEEE Access, 7, 32845-32852.
• Choi, Y., & Lee, H. (2023). Interpretation of lung disease classification with light attention connected module. Biomedical Signal Processing and Control, 84, 104695.
• Demir, F., Ismael, A. M., & Sengur, A. (2020). Classification of lung sounds with CNN model using parallel pooling structure. IEEE Access, 8, 105376-105383.
• Demir, F., Sengur, A., & Bajaj, V. (2019). Convolutional neural networks based efficient approach for classification of lung diseases. Health Information Science and Systems, 8(1), 4.
• Fraiwan, L., Hassanin, O., Fraiwan, M., Khassawneh, B., Ibnian, A. M., & Alkhodari, M. (2021). Automatic identification of respiratory diseases from stethoscopic lung sound signals using ensemble classifiers. Biocybernetics and Biomedical Engineering, 41(1), 1-14.
• Gairola, S., Tom, F., Kwatra, N., & Jain, M. (2021, November). Respirenet: A deep neural network for accurately detecting abnormal lung sounds in limited data setting. In 2021 43rd Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC) (pp. 527-530). IEEE.
• Han, K., Wang, Y., Tian, Q., Guo, J., Xu, C., & Xu, C. (2020). Ghostnet: More features from cheap operations. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (pp. 1580-1589).
• He, K., Zhang, X., Ren, S., & Sun, J. (2016). Deep residual learning for image recognition. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (pp. 770-778).
• Huang, D. M., Huang, J., Qiao, K., Zhong, N. S., Lu, H. Z., & Wang, W. J. (2023). Deep learning-based lung sound analysis for intelligent stethoscope. Military Medical Research, 10(1), 44.
• Kansal, K., Chandra, T. B., & Singh, A. (2024). ResNet-50 vs. EfficientNet-B0: Multi-Centric Classification of Various Lung Abnormalities Using Deep Learning" Session id: ICMLDsE. 004". Procedia Computer Science, 235, 70-80.
• Khan, R., Khan, S. U., Saeed, U., & Koo, I. S. (2024). Auscultation-Based Pulmonary Disease Detection through Parallel Transformation and Deep Learning. Bioengineering, 11(6), 586.
• Kim, J. W., Toikkanen, M., Bae, S., Kim, M., & Jung, H. Y. (2024, July). Repaugment: Input-agnostic representation-level augmentation for respiratory sound classification. In 2024 46th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC) (pp. 1-6). IEEE.
• Kim, Y., Hyon, Y., Jung, S. S., Lee, S., Yoo, G., Chung, C., & Ha, T. (2021). Respiratory sound classification for crackles, wheezes, and rhonchi in the clinical field using deep learning. Scientific Reports, 11(1), 1-11.
• Liu, R., Cai, S., Zhang, K., & Hu, N. (2019, November). Detection of adventitious respiratory sounds based on convolutional neural network. In 2019 International Conference on Intelligent Informatics and Biomedical Sciences (ICIIBMS) (pp. 298-303). IEEE.
• Mang, L. D., González Martínez, F. D., Martinez Muñoz, D., García Galán, S., & Cortina, R. (2024). Classification of adventitious sounds combining cochleogram and vision transformers. Sensors, 24(2), 682.
• Nguyen, T., & Pernkopf, F. (2020, July). Lung sound classification using snapshot ensemble of convolutional neural networks. In 2020 42nd Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC) (pp. 760-763). IEEE.
• Nguyen, T., & Pernkopf, F. (2022). Lung sound classification using co-tuning and stochastic normalization. IEEE Transactions on Biomedical Engineering, 69(9), 2872-2882.
• Petmezas, G., Cheimariotis, G. A., Stefanopoulos, L., Rocha, B., Paiva, R. P., Katsaggelos, A. K., & Maglaveras, N. (2022). Automated lung sound classification using a hybrid CNN-LSTM network and focal loss function. Sensors, 22(3), 1232.
• Poirè, A. M., Simonetta, F., & Ntalampiras, S. (2022, September). Deep feature learning for medical acoustics. In International Conference on Artificial Neural Networks (pp. 39-50). Cham: Springer Nature Switzerland.
• Prabhakar, S. K., & Won, D. O. (2023). HISET: Hybrid interpretable strategies with ensemble techniques for respiratory sound classification. Heliyon, 9(8).
• Sabry, A. H., Bashi, O. I. D., Ali, N. N., & Al Kubaisi, Y. M. (2024). Lung disease recognition methods using audio-based analysis with machine learning. Heliyon, 10(4).
• Serbes, G., Sakar, C. O., Kahya, Y. P., & Aydin, N. (2011, August). Feature extraction using time-frequency/scale analysis and ensemble of feature sets for crackle detection. In 2011 Annual International Conference of the IEEE Engineering in Medicine and Biology Society (pp. 3314-3317). IEEE.
• Shuvo, S. B., Ali, S. N., Swapnil, S. I., Hasan, T., & Bhuiyan, M. I. H. (2020). A lightweight CNN model for detecting respiratory diseases from lung auscultation sounds using EMD-CWT-based hybrid scalogram. IEEE Journal of Biomedical and Health Informatics, 25(7), 2595-2603.
• Tan, M., & Le, Q. (2019, May). Efficientnet: Rethinking model scaling for convolutional neural networks. In International Conference on Machine Learning (pp. 6105-6114). PMLR.
• Tariq, Z., Shah, S. K., & Lee, Y. (2022). Feature-based fusion using CNN for lung and heart sound classification. Sensors, 22(4), 1521.
• Türkiye Solunum Araştırmaları Derneği. (t.y.). Solunum hastalıkları önemli bir mortalite nedeni. Medikal Akademi. https://www.medikalakademi.com.tr/solunum-hastalklar-oenemli-bir-mortalite-nedeni/
• Tzeng, J. T., Li, J. L., Chen, H. Y., Huang, C. H., Chen, C. H., Fan, C. Y., ... & Lee, C. C. (2025). Improving the Robustness and Clinical Applicability of Automatic Respiratory Sound Classification Using Deep Learning–Based Audio Enhancement: Algorithm Development and Validation. JMIR AI, 4(1), e67239.
• Wanasinghe, T., Bandara, S., Madusanka, S., Meedeniya, D., Bandara, M., & Díez, I. D. L. T. (2024). Lung sound classification with multi-feature integration utilizing lightweight CNN model. IEEE Access, 12, 21262-21276.
• Wang, Z., & Sun, Z. (2024). Performance evaluation of lung sounds classification using deep learning under variable parameters. EURASIP Journal on Advances in Signal Processing, 2024(1), 51.
• Wu, C., Ye, N., & Jiang, J. (2024). Classification and recognition of lung sounds based on improved Bi-ResNet model. IEEE Access.
• Xu, M., & Wiese, L. (2023, September). Application and performance improvement of transfer learning on ICBHI lung sound dataset. In Proceedings of SAI Intelligent Systems Conference (pp. 156-173). Cham: Springer Nature Switzerland.
• Yu, S., Ding, Y., Qian, K., Hu, B., Li, W., & Schuller, B. W. (2022, May). A glance-and-gaze network for respiratory sound classification. In ICASSP 2022-2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) (pp. 9007-9011). IEEE.
• Zhang, W., & Liu, X. (2024). DECODING BREATH: Machine learning advancements in diagnosing pulmonary diseases via lung sound analysis. Scientific and Engineering Letters, 12(1), 1-11.