Analisis Sensitivitas Confidence Threshold pada Semi-Supervised FixMatch untuk Klasifikasi Multi-Kelas Citra Chest X-Ray
- Ahmad Kurniawan Universitas Islam Negeri Sultan Syarif Kasim Riau, Pekanbaru, Indonesia
- Muhammad Irsyad Universitas Islam Negeri Sultan Syarif Kasim Riau, Pekanbaru, Indonesia
-
Benny Sukma Negara
*
Universitas Islam Negeri Sultan Syarif Kasim Riau, Pekanbaru,
Indonesia
- Surya Agustian Universitas Islam Negeri Sultan Syarif Kasim Riau, Pekanbaru, Indonesia
- Nazruddin Safaat H Universitas Islam Negeri Sultan Syarif Kasim Riau, Pekanbaru, Indonesia (*) Corresponding Author
Abstract
Optimizing the confidence threshold in pseudo-labeling is a critical technical challenge in Semi-Supervised Learning (SSL) for multi-class medical image classification. A threshold that is too strict limits the utilization of unlabeled data, whereas a threshold that is too lenient introduces low-quality pseudo-labels into the training process. This study applies the FixMatch method with the DenseNet-169 architecture as the backbone network to classify three lung disease categories COVID-19, Pneumonia, and Normal under conditions of extremely limited labeled data. The dataset used is the COVID-19, Pneumonia, and Normal Chest X-Ray Images dataset from Mendeley Data, consisting of 5,218 chest X-ray images, divided into 70% training, 10% validation, and 20% testing sets. The experiments were systematically designed using three labeled-data proportions (5%, 10%, and 15%) and three confidence threshold values (τ = 0.90, 0.95, and 0.99), resulting in nine experimental scenarios. The results demonstrate that τ = 0.95 with 15% labeled data achieved the best performance, obtaining 97.41% accuracy, a 97.49% F1-score, and an AUC of 0.9963. This performance was achieved by balancing pseudo-label selectivity with a sufficient volume of effective training data. At a low labeled-data ratio (5%), the limited amount of labeled data meant that the lower mask rate at τ = 0.95 could not be adequately compensated, allowing τ = 0.99 to perform slightly better. In contrast, at a higher labeled-data ratio (15%), the selectivity of τ = 0.95 produced high-quality pseudo-labels while maintaining sufficient data volume, leading to improved generalization performance. This study contributes an empirical analysis of confidence threshold sensitivity in FixMatch for multi-class chest X-ray classification under limited labeled-data conditions. These findings reveal that the effectiveness of the confidence threshold is highly dependent on the availability of labeled data, and that determining an optimal threshold cannot be separated from the proportion of labeled data available.
Downloads
References
Alomar, K., Aysel, H. I., & Cai, X. (2023). Data Augmentation in Classification and Segmentation : A Survey and New Strategies. Journal of Imaging, 9(2), 46. https://doi.org/10.3390/jimaging9020046
Dalvi, P. P., Edla, D. R., & Purushothama, B. R. (2023). Diagnosis of Coronavirus Disease From Chest X-Ray Images Using DenseNet-169 Architecture. SN Computer Science, 4(3), 1–6. https://doi.org/10.1007/s42979-022-01627-7
Hmoud, M., Sheikh, A., Dandan, O. Al, Sami, A., Shamayleh, A., Jalab, H. A., & Ibrahim, R. W. (2023). Multi ‑ class deep learning architecture for classifying lung diseases from chest X ‑ Ray and CT images. Scientific Reports, 1–14. https://doi.org/10.1038/s41598-023-46147-3
Huang, S. C., Pareek, A., Jensen, M., Lungren, M. P., Yeung, S., & Chaudhari, A. S. (2023). Self-supervised learning for medical image classification: a systematic review and implementation guidelines. Npj Digital Medicine, 6(1). https://doi.org/10.1038/s41746-023-00811-0
Ihler, S., Kuhnke, F., Kuhlgatz, T., & Seel, T. (2024). Distribution-Aware Multi-Label FixMatch for Semi-Supervised Learning on CheXpert. IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops, 2295–2304. https://doi.org/10.1109/CVPRW63382.2024.00235
Ihongbe, I. E., Shereen, F., Mahmoud, T. F., & Rajasekaran, A. (2024). Evaluating Explainable Artificial Intelligence ( XAI ) techniques in chest radiology imaging through a human-centered Lens. PLoS ONE, 19, 1–27. https://doi.org/10.1371/journal.pone.0308758
Kumar, S., Shastri, S., Kansal, I., Singh, K., Popli, R., & Mansotra, V. (2022). LiteCovidNet : A lightweight deep neural network model for detection of COVID-19 using X-ray images. International Journal of Imaging Systems and Technology, 32(February), 1464–1480. https://doi.org/10.1002/ima.22770
Li, M., Jiang, Y., Zhang, Y., & Zhu, H. (2023). Medical image analysis using deep learning algorithms. Frontiers in Public Health, 11(November), 1–28. https://doi.org/10.3389/fpubh.2023.1273253
Liu, K., Liu, J., & Liu, S. (2024). Enhanced Semi-Supervised Medical Image Classification Based on Dynamic Sample Reweighting and Pseudo-Label Guided Contrastive Learning ( DSRPGC ). Mathematics, 12(22), 3572. https://doi.org/10.3390/math12223572
Liu, Z., Mao, H., Christoph, C. W., Trevor, F., Saining, D., & Berkeley, U. C. (2022). A ConvNet for the 2020s. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 11976–11986. https://doi.org/10.1109/CVPR52688.2022.01167
Nielsen, M., Wenderoth, L., Sentker, T., & Werner, R. (2023). Self-Supervision for Medical Image Classification : State-of-the-Art Performance with ~ 100 Labeled Training Samples per Class. Bioengineering, 10(8), 895. https://doi.org/10.3390/bioengineering10080895
Osapoetra, L. O., Moslemi, A., Moore-palhares, D., Halstead, S., Alberico, D., Hwang, A., Sannachi, L., & Curpen, B. (2025). End-to-end CNN-based deep learning enhances breast lesion characterization using quantitative ultrasound ( QUS ) spectral parametric images. 1–13.
Rainio, O., Jarmo, T., & Klen, R. (2024). Evaluation metrics and statistical tests for machine learning. Scientific Reports, 1–14. https://doi.org/10.1038/s41598-024-56706-x
Rajaraman, S., Ganesan, P., & Antani, S. (2022). Deep learning model calibration for improving performance in class-imbalanced medical image classification tasks. PLoS ONE, 17(1): e02, 1–23. https://doi.org/10.1371/journal.pone.0262838
Sahoo, P., Roy, I., Ahlawat, R., Irtiza, S., & Khan, L. (2022). Potential diagnosis of COVID ‑ 19 from chest X ‑ ray and CT findings using semi ‑ supervised learning. Physical and Engineering Sciences in Medicine, 45(1), 31–42. https://doi.org/10.1007/s13246-021-01075-2
Sajun, A. R., Zualkernan, I., & Sankalpa, D. (2022). Investigating the Performance of FixMatch for COVID-19 Detection in Chest X-rays. Applied Sciences (Switzerland), 12(9). https://doi.org/10.3390/app12094694
Selvaraju, R., Cogswell, M., Das, A., Vedantam, R., Parikh, D., & Batra, D. (2020). Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localization. International Journal of Computer Vision, 128(2), 336–359. https://doi.org/10.1007/s11263-019-01228-7
Shamrat, M., Javed, F. M., Azam, S., Karim, A., Ahmed, K., & Bui, F. M. (2023). High-precision multiclass classification of lung disease through customized MobileNetV2 from chest X-ray images. Computers in Biology and Medicine, 155(February), 106646. https://doi.org/10.1016/j.compbiomed.2023.106646
Shastri, S., Kansal, I., Kumar, S., Singh, K., Popli, R., & Mansotra, V. (2022). CheXImageNet : a novel architecture for accurate classification of Covid ‑ 19 with chest x ‑ ray digital images using deep convolutional neural networks. Health and Technology, 193–204. https://doi.org/10.1007/s12553-021-00630-x
Simon, G. J., & Aliferis, C. (2024). Artificial Intelligence and Machine Learning in Health Care and Medical Sciences. Springer.
Sohn, K., Berthelot, D., Li, Chun, L., Zhang, Z., Carlini, N., Cubuk, E. D., Kurakin, A., Zhang, H. H., & Raffel, C. (2020). Semi-Supervised : FixMatch. Advances in Neural Information Processing Systems, 37(10), 1575–1585. https://arxiv.org/abs/2001.07685
Takahashi, S., Sakaguchi, Y., Kouno, N., Takasawa, K., Ishizu, K., & Akagi, Y. (2024). Comparison of Vision Transformers and Convolutional Neural Networks in Medical Image Analysis : A Systematic Review. Journal of Medical Systems, 48(1), 1–22. https://doi.org/10.1007/s10916-024-02105-8
Umair, M., Khan, M. S., Ahmed, F., Baothman, F., Alqahtani, F., Alian, M., & Ahmad, J. (2021). Detection of COVID-19 Using Transfer Learning and Grad-CAM Visualization on Indigenously Collected. June. https://doi.org/https://doi.org/10.3390/s21175813
Wang, L., Guo, D., Wang, G., & Zhang, S. (2021). Annotation-Efficient Learning for Medical Image Segmentation based on Noisy Pseudo Labels and Adversarial Learning. IEEE Transactions on Medical Imaging, 40(8), 2235–2246. https://doi.org/10.1109/TMI.2020.3047807
Wang, Y., Chen, H., Heng, Q., Hou, W., Fan, Y., Savvides, M., & Shinozaki, T. (2023). FreeMatch: Self-adaptive thresholding in semi-supervised learning. International Conference on Learning Representations, 1–20.
Yang, L., Feng, L., Shi, Y., Qi, L., & Zhang, W. (2023). Revisiting Weak-to-Strong Consistency. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 7236–7246. https://doi.org/10.1109/CVPR52729.2023.00699
Zhang, B., Yang, W., Hou, W., Wu, H., Wang, J., Okumura, M., & Shinozaki, T. (2021). FlexMatch : Boosting Semi-Supervised Learning with Curriculum Pseudo Labeling. Advances in Neural Information Processing Systems, 34(NeurIPS), 18408–18419.
Zhang, W., Zhu, L., Hallinan, J., Makmur, A., Zhang, S., Cai, Q., & Ooi, Chin, B. (2022). BoostMIS : Boosting Medical Image Semi-supervised Learning with Adaptive Pseudo Labeling and Informative Active Annotation. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 20666–20676. https://doi.org/10.1109/CVPR52729.2022.02007
World Health Organization. (2024). Pneumonia. Retrieved from https://www.who.int/health-topics/pneumonia
Bila bermanfaat silahkan share artikel ini
Berikan Komentar Anda terhadap artikel Analisis Sensitivitas Confidence Threshold pada Semi-Supervised FixMatch untuk Klasifikasi Multi-Kelas Citra Chest X-Ray
Pages: 228-241
Copyright (c) 2026 Ahmad Kurniawan, Muhammad Irsyad, Benny Sukma Negara, Surya Agustian, Nazruddin Safaat H
This work is licensed under a Creative Commons Attribution 4.0 International License.
Authors who publish with this journal agree to the following terms:
- Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under Creative Commons Attribution 4.0 International License that allows others to share the work with an acknowledgment of the work's authorship and initial publication in this journal.
- Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgment of its initial publication in this journal.
- Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) prior to and during the submission process, as it can lead to productive exchanges, as well as earlier and greater citation of published work (Refer to The Effect of Open Access).
