Classification of deep features obtained from mpox images using Xception by selecting them with PCA and mRMR and using SVM

Öz

Mpox is a dsDNA virus that shows lesions on the skin similar to those of chickenpox, measles, and smallpox. Clinical examination is based on traditional methods such as Polymerase Chain Reaction (PCR) test kits and skin lesion testing by electron microscopy. The costly and time-consuming nature of PCR tests and the similarity of Mpox to chickenpox, measles, and smallpox make electron microscopy tests difficult for early diagnosis. In this study, we attempted to solve these challenges by using transfer learning-based convolutional neural network (CNN) models to diagnose Mpox disease. The experiments were conducted on the publicly available Mpox Close Skin Images (MCSI) dataset, which is preprocessed, homogeneous, and has a balanced distribution. DenseNet-121, DenseNet-169, DenseNet-201, Inception-V3, MobileNet, MobileNetV2, NasNetMobile, and Xception CNN models were compared, and it was analyzed that the Xception model was more successful than other models in the Mpox classification task. The 204800 deep feature maps obtained from the layer just before the fully connected layer of the Xception model were reduced to 400 by Principal component analysis (PCA). Then, the deep feature maps were filtered with the Minimum Redundancy Maximum Relevance (mRMR) algorithm and passed through the feature selection process. After feature selection, the 100 feature maps obtained after the classification process of the Support Vector Machine (SVM) algorithm yielded an accuracy of 89.70%, precision of 89.69%, sensitivity of 89.70%, F1 of 89.66%, and specificity of 96.57%. The GridCVSearch method to optimize hyper-parameters was used, and a Repeated 5-fold cross-validation technique was used in all experimental studies. As a result of these results, our approach showed that it could increase the diagnostic accuracy rate of the disease, reduce the overall misdiagnosis rate, and be a potential alternative decision support system to traditional methods such as PCR and electron microscopy.

Anahtar Kelimeler

• Deep Learning
• Convolutional neural network
• Mpox
• Monkeypox
• Transfer learning
• Classification

Kaynakça

1. Akram, A., Jamjoom, A. A., Innab, N., Almujally, N. A., Umer, M., Alsubai, S., & Fimiani, G. (2024). SkinMarkNet: an automated approach for prediction of monkeyPox using image data augmentation with deep ensemble learning models. Multimedia Tools and Applications. doi:10.1007/s11042-024-19862-w google scholar
2. Ali, S. N., Ahmed, M. T., Paul, J., Jahan, T., Sani, S. M., Noor, N., & Hasan, T. (2022). Mpox skin lesion detection using deep learning models: A feasibility study. ArXiv Preprint ArXiv:2207.03342. https://doi.org/10.48550/arXiv.2207.03342 google scholar
3. Ali, S. N., Ahmed, Md. T., Jahan, T., Paul, J., Sakeef Sani, S. M., Noor, N., Asma, A. N., & Hasan, T. (2024). A web-based mpox skin lesion detection system using state-of-the-art deep learning models considering racial diversity. Biomedical Signal Processing and Control, 98, 106742. doi:10.1016/j.bspc.2024.106742 google scholar
4. Altun, M., Gürüler, H., Özkaraca, O., Khan, F., Khan, J., & Lee, Y. (2023). Mpox detection using CNN with transfer learning. Sensors, 23(4), 1783. doi:10.3390/s23041783 google scholar
5. Bala, D., Hossain, M. S., Hossain, M. A., Abdullah, M. I., Rahman, M. M., Manavalan, B., Gu, N., Islam, M. S., & Huang, Z. (2023). MonkeyNet: A robust deep convolutional neural network for monkeypox disease detection and classification. Neural Networks, 161, 757–775. https://doi.org/10.1016/j.neunet.2023.02.022 google scholar
6. Başaran, E. (2022). Classification of leukocytes with SVM by selecting SqueezeNet and LIME properties using the mRMR method. Signal, Image and Video Processing, 16(7), 1821–1829. doi:10.1007/s11760-022-02141-2 google scholar
7. Başaran, E., & Çelik, Y. (2024). Skin cancer diagnosis using CNN features with the genetic algorithm and particle swarm optimization methods. Transactions of the Institute of Measurement and Control, 46(14), 2706–2713. doi: 10.1177/01423312241253926 google scholar
8. Bassma, G., & Tayeb, S. (2018). Support vector machines for improving vehicle localization in urban canyons. MATEC Web of Conferences, 200, 00004. google scholar

9. Çalışkan, A. (2023). In brain tumor detection, training of mr images created by heat map technique with CNN models-extraction of type-based activation sets and selection of best features using mRMR method. Int Res Eng Sci, 5, 7. google scholar
10. Campana, M. G., Colussi, M., Delmastro, F., Mascetti, S., & Pagani, E. (2023). Mpox Close Skin Images dataset (MCSI). Available at https://zenodo.org/records/8360076 (accessed 14 Dec 2024). google scholar
11. Campana, M. G., Colussi, M., Delmastro, F., Mascetti, S., & Pagani, E. (2024). A transfer learning and explainable solution to detect mpox from smartphone images. Pervasive and Mobile Computing, 98, 101874. doi:10.1016/j.pmcj.2023.101874 google scholar
12. Chakroborty, S. (2024). H-MpoxNet: A hybrid deep learning framework for mpox detection from image Data. MedRxiv, 2024.11.26.24318006. https://doi.org/10.1101/2024.11.26.24318006 google scholar
13. Chollet, F. (2017). Xception: Deep learning with depthwise separable convolutions. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 21–26 July 2017, pp. 1251–1258, Honolulu, HI, USA. google scholar
14. Fushiki, T. (2011). Estimation of the prediction error using K-fold cross-validation. Statistics and Computing, 21, 137–146. doi:10.1007/s11222-009-9153-8 google scholar
15. Göl, M., Aktürk, C., Talan, T., Vural, M. S., & Türkbeyler, İ. H. (2025). Predicting malnutrition‐based anemia in geriatric patients using machine learning methods. Journal of Evaluation in Clinical Practice, 31(2), e14142. https://doi.org/10.1111/jep.14142 google scholar
16. Greenacre, M., Groenen, P. J.F., Hastie, T., d’Enza, A. I., Markos, A., & Tuzhilina, E. (2022). Principal component analysis. Nat Rev Methods Primers 2, 100. doi:10.1038/s43586-022-00184-w google scholar
17. Haque, M. E., Ahmed, M. R., Nila, R. S., & Islam, S. (2022). Classification of human mpox disease using deep learning models and attention mechanisms. ArXiv Preprint ArXiv:2211.15459. doi:10.48550/arXiv.2211.15459 google scholar
18. Hu, Q., Pan, W., An, S., Ma, P., & Wei, J. (2010). An efficient gene selection technique for cancer recognition based on neighborhood mutual information. International Journal of Machine Learning and Cybernetics, 1(1), 63–74. doi:10.1007/s13042-010-0008-6 google scholar
19. Jaradat, A. S., Al Mamlook, R. E., Almakayeel, N., Alharbe, N., Almuflih, A. S., Nasayreh, A., Gharaibeh, H., Gharaibeh, M., Gharaibeh, A., & Bzizi, H. (2023). Automated mpox skin lesion detection using deep learning and transfer learning techniques. International Journal of Environmental Research and Public Health, 20(5). doi:10.3390/ijerph20054422 google scholar
20. Khan, S. U. R., Asif, S., Bilal, O., & Ali, S. (2024). Deep hybrid model for Mpox disease diagnosis from skin lesion images. International Journal of Imaging Systems and Technology, 34(2), e23044. doi:10.1002/ima.23044 google scholar
21. Kherif, F., & Latypova, A. (2020). Principal component analysis. In A. Mechelli & S. Vieira (Eds.), Machine Learning (pp. 209–225). Academic Press. doi:10.1016/B978-0-12-815739-8.00012-2 google scholar
22. Lasisi, A., & Attoh-Okine, N. (2018). Principal component analysis and track quality index: A machine learning approach. Transportation Research Part C: Emerging Technologies, 91, 230–248. doi:10.1016/j.trc.2018.04.001 google scholar
23. Lu, S., Lu, Z., & Zhang, Y.-D. (2019). Pathological brain detection based on AlexNet and transfer learning. Journal of Computational Science, 30, 41–47. google scholar
24. Mercier, G., & Lennon, M. (2003). Support vector machines for hyperspectral image classification with spectral-based kernels. IGARSS 2003. 2003 IEEE International Geoscience and Remote Sensing Symposium. Proceedings (IEEE Cat. No. 03CH37477), 1, 288–290. google scholar
25. Peng, H., Long, F., & Ding, C. (2005). Feature selection based on mutual information criteria of max-dependency, max-relevance, and min-redundancy. IEEE Transactions on Pattern Analysis and Machine Intelligence, 27(8), 1226–1238. google scholar
26. Raha, A. D., Gain, M., Debnath, R., Adhikary, A., Qiao, Y., Hassan, M. M., Bairagi, A. K., & Islam, S. M. S. (2024). Attention to monkeypox: An interpretable mpox detection technique using attention mechanism. IEEE Access, 12, 51942–51965. doi: 10.1109/ACCESS.2024.3385099 google scholar
27. Rodriguez, J. D., Perez, A., & Lozano, J. A. (2009). Sensitivity analysis of k-fold cross validation in prediction error estimation. IEEE Transactions on Pattern Analysis and Machine Intelligence, 32(3), 569–575. doi: 10.1109/TPAMI.2009.187 google scholar
28. Sitaula, C., & Shahi, T. B. (2022). Mpox virus detection using pre-trained deep learning-based approaches. Journal of Medical Systems, 46(11), 78. doi:10.1007/s10916-022-01868-2 google scholar
29. Şenol, A., Talan, T. & Aktürk, C. (2024). A new hybrid feature reduction method using the MCMSTClustering algorithm with various feature projection methods: a case study on sleep disorder diagnosis. SIViP, 18, 4589–4603. doi:10.1007/s11760-024-03097-1 google scholar
30. Thorat, R., & Gupta, A. (2024). Transfer learning-enabled skin disease classification: the case of mpox detection. Multimedia Tools and Applications, 83(35), 82925–82943. DOI: 10.1007/s11042-024-18750-7 google scholar
31. Toğaçar, M., Ateş, F. F., & Çalışkan, A. (2022). Ultrason tabanlı meme kayseri görüntülerinin erin öğrenme yaklaşımları ile sınıflandırılması [Classification of ultrasound-based breast cancer images with deep learning approaches]. Fırat Üniversitesi Fen Bilimleri Dergisi, 34(2), 179–187. https://dergipark.org.tr/tr/pub/fufbd/issue/72741/1170793 google scholar
32. Uysal, F. (2023). Detection of mpox from human skin images with a hybrid deep learning model. Diagnostics, 13(10). doi:10.3390/diagnostics13101772 google scholar
33. Wong, T.-T., & Yeh, P.-Y. (2019). Reliable accuracy estimates from k-fold cross validation. IEEE Transactions on Knowledge and Data Engineering, 32(8), 1586–1594. doi: 10.1109/TKDE.2019.2912815 google scholar