Journal of information and communication convergence engineering

The Korea Institute of Information and Commucation Engineering (한국정보통신학회)
권호 표지
DOI QR코드

DOI QR Code

Improving Chest X-ray Image Classification via Integration of Self-Supervised Learning and Machine Learning Algorithms

  • Tri-Thuc Vo (Department of Computer Science, Can Tho University) ;
  • Thanh-Nghi Do (Department of Computer Networks, Can Tho University)
  • Received : 2023.07.04
  • Accepted : 2024.03.04
  • Published : 2024.06.30
Download PDF
⟨ Previous Next ⟩

Abstract

In this study, we present a novel approach for enhancing chest X-ray image classification (normal, Covid-19, edema, mass nodules, and pneumothorax) by combining contrastive learning and machine learning algorithms. A vast amount of unlabeled data was leveraged to learn representations so that data efficiency is improved as a means of addressing the limited availability of labeled data in X-ray images. Our approach involves training classification algorithms using the extracted features from a linear fine-tuned Momentum Contrast (MoCo) model. The MoCo architecture with a Resnet34, Resnet50, or Resnet101 backbone is trained to learn features from unlabeled data. Instead of only fine-tuning the linear classifier layer on the MoCopretrained model, we propose training nonlinear classifiers as substitutes for softmax in deep networks. The empirical results show that while the linear fine-tuned ImageNet-pretrained models achieved the highest accuracy of only 82.9% and the linear fine-tuned MoCo-pretrained models an increased highest accuracy of 84.8%, our proposed method offered a significant improvement and achieved the highest accuracy of 87.9%.

Keywords

References

  1. Z. Akkus et al. "Deep learning for brain MRI segmentation: state of the art and future directions," Journal of Digital Imaging, vol. 30, no. 4, pp. 449-459, Jun. 2017. DOI: 10.1007/s10278-017-9983-4.
  2. C. H. Liang et al., "Identifying pulmonary nodules or masses on chest radiography using deep learning: external validation and strategies to improve clinical practice," Clinical Radiology, vol. 75, no. 1, pp. 38-45, Jan. 2020. DOI: 10.1016/j.crad.2019.08.005.
  3. J. Ker et al., "Deep learning applications in medical image analysis," IEEE Access, vol. 6, pp. 9375-9389, 2018. DOI: 10.1109/ACCESS.2017.2788044.
  4. K. He et al., "Momentum contrast for unsupervised visual representation learning," in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, Online, pp. 9729-9738, 2020.
  5. X. Chen et al., "Improved baselines with momentum contrastive learning," arXiv preprint arXiv:2003.04297, Mar. 2020. DOI: 10.48550/arXiv.2003.04297.
  6. T. Chen et al., "A simple framework for contrastive learning of visual representations," arXiv preprint arXiv:2002.05709, 2020. DOI: 10.48550/arXiv.2002.05709.
  7. T. Chen et al., Big self-supervised models are strong semi-supervised learners," arXiv preprint arXiv:2006.10029, 2020. DOI: 10.48550/arXiv.2006.10029.
  8. H. Sowrirajan et al., "Moco pretraining improves representation and transferability of chest x-ray models," in Medical Imaging with Deep Learning, vol. 143, pp. 728-744. PMLR, 2021. DOI: 10.48550/arXiv.2010.05352v1.
  9. V. N. Vapnik, The Nature of Statistical Learning Theory, 2nd ed. Springer-Verlag New York, 2000.
  10. G. Ke et al., "Lightgbm: A highly efficient gradient boosting decision tree," Advances in Neural Information Processing Systems, vol. 30, 2017.
  11. T. Chen and C. Guestrin, "Xgboost: A scalable tree boosting system," in Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data mining, San Francisco, USA, pp. 785-794, Aug. 2016. DOI: 10.1145/2939672.2939785.
  12. A. V. Dorogush et al., CatBoost: gradient boosting with categorical features support," arXiv preprint arXiv:1810.11363, Oct. 2018. DOI: 10.48550/arXiv.1810.11363.
  13. Ege Kesim et al., "X-ray chest image classification by a small-sized convolutional neural network," in 2019 Scientific Meeting on Electrical-Electronics & Biomedical Engineering and Computer Science (EBBT), Istanbul, Turkey, pp. 1-5, 2019. DOI: 10.1109/EBBT.2019.8742050.
  14. V. Chouhan et al., "A novel transfer learning based approach for pneumonia detection in chest x-ray images," Applied Sciences, vol. 10, no. 2, pp. 559, Jan. 2020. DOI: 10.3390/app10020559.
  15. P. Rajpurkar et al., "Deep learning for chest radiograph diagnosis: A retrospective comparison of the chexnext algorithm to practicing radiologists," PLoS medicine, vol. 15, no. 11, pp. e1002686, Nov. 2018. DOI: 10.1371/journal.pmed.1002686.
  16. A. Bhandary et al., "Deep-learning framework to detect lung abnormality-a study with chest x-ray and lung CT scan images," Pattern Recognition Letters, vol. 129, pp. 271-278, Jan. 2020. DOI: 10.1016/j.patrec.2019.11.013.
  17. M. Wozniak et al., "Small lung nodules detection based on local variance analysis and probabilistic neural network," Computer Methods and Programs in Biomedicine, vol. 161, pp. 173-180, Jul. 2018. DOI: 10.1016/j.cmpb.2018.04.025.
  18. S. S. Yadav and S. M. Jadhav. "Deep convolutional neural network based medical image classification for disease diagnosis," Journal of Big data, vol. 6, no. 1, pp. 1-18, Dec. 2019. DOI: 10.1186/s40537-019-0276-2.
  19. T-N. Do et al., "SVM on Top of Deep Networks for Covid-19 Detection from Chest X-ray Images," Journal of Information and Communication Convergence Engineering, vol. 20, no. 3, pp. 219-225, Sep. 2022. DOI: 10.56977/jicce.2022.20.3.219.
  20. P. Afshar et al., "Covid-caps: A capsule network-based framework for identification of covid-19 cases from x-ray images," Pattern Recognition Letters, vol. 138, pp. 638-643, Oct. 2020. DOI: 10.1016/j.patrec.2020.09.010.
  21. E. E-D. Hemdan et al., "Covidx-net: A framework of deep learning classifiers to diagnose covid-19 in x-ray images," arXiv preprint arXiv:2003.11055, Mar. 2020. DOI: 10.48550/arXiv.2003.11055.
  22. T. Ozturk et al., "Automated detection of covid-19 cases using deep neural networks with x-ray images," Computers in Biology and Medicine, vol. 121, p. 103792, Jun. 2020. DOI: 10.1016/j.compbiomed.2020.103792.
  23. K. He et al., "Deep residual learning for image recognition," in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, USA, pp. 770-778, 2016.
  24. J. Irvin et al., "Chexpert: A large chest radiograph dataset with uncertainty labels and expert comparison," in Proceedings of the AAAI conference on artificial intelligence, Honolulu, USA, vol. 33, no. 1, pp. 590-597, Jul. 2019. DOI: 10.1609/aaai.v33i01.3301590.
  25. H. Q. Nguyen et al., "Vindr-cxr: An open dataset of chest x-rays with radiologist's annotations," Scientific Data, vol. 9, no. 1, pp. 429, Jul. 2022. DOI: 10.1038/s41597-022-01498-w.
  26. D. S. Kermany et al., "Identifying medical diagnoses and treatable diseases by image-based deep learning," Cell, vol. 172, no. 5, pp. 1122-1131, Feb. 2018. DOI: 10.1016/j.cell.2018.02.010.
  27. J. Shiraishi et al., "Development of a digital image database for chest radiographs with and without a lung nodule: receiver operating characteristic analysis of radiologists' detection of pulmonary nodules," American Journal of Roentgenology, vol. 174, no. 1, pp. 71-74, Jan. 2000. DOI: 10.2214/ajr.174.1.1740071.
  28. J. Saltz et al., Stony Brook University Covid-19 Positive Cases, 2020, [online] Available: https://doi.org/10.7937/TCIA.BBAG-2923
  29. A. Haghanifar et al., Covid-19 chest x-ray image repository," 2021, [online] Available: https://figshare.com/articles/dataset/COVID-19Chest XRay Image Repository/12580328.
  30. J. P. Cohen et al., Covid-19 image data collection: Prospective predictions are the future, arXiv preprint arXiv:2006.11988, Jun. 2020. DOI: /10.48550/arXiv.2006.11988.
  31. M. D. L. I. Vaya et al., Bimcv covid-19+: a large annotated dataset of rx and ct images from covid-19 patients," arXiv preprint arXiv: 2006.01174, Jun. 2020. DOI: 10.48550/arXiv.2006.01174.
  32. H. B. Winther et al., Covid-19 image repository, 2020, [online], Available: https://doi.org/10.6084/m9.figshare.12275009.v1
  33. SIIM-ACR Pneumothorax Segmentation, [online], Available: https://www.kaggle.com/c/siim-acr-pneumothorax-segmentation
  34. X. Wang et al., "Chestx-ray8: Hospital-scale chest x-ray database and benchmarks on weakly-supervised classification and localization of common thorax diseases," in Proceedings of the IEEE conference on computer vision and pattern recognition, Honolulu, USA, pp. 2097-2106, 2017. DOI: 10.1109/cvpr.2017.369.
  35. M. Abadi et al., Tensorflow: Large-scale machine learning on heterogeneous distributed systems," arXiv preprint arXiv:1603.04467, Mar. 2016. DOI: 10.48550/arXiv.1603.04467.
  36. F. Pedregosa et al., "Scikitlearn: Machine learning in python," The Journal of Machine Learning Research, vol. 12, pp. 2825-2830, Oct. 2011.