CP-MCNN: Multi-label Chest X-ray Diagnostic Based on Confidence Predictor and CNN
- 373 Downloads
Chest X-ray as a sensing mode is worthwhile to be paid attention in terms of its conversation and prevalence, and it as a typical multi-label problem where each example is represented by a single instance while associated with a set of labels simultaneously. Early researches on chest X-ray mainly using Convolutional Neural Network (CNN), although it has outperformance in experiment, diagnosis of chest x-ray as a typical high-risk problem, CNN lacks confidence evaluation its output to make a judgment. To solve this problem, we propose a new framework of Confidence Prediction-Multi-label Convolutional Neural Network (CP-MCNN) that plugs MCNN into Confidence Predictor. It can provide calibrated confidential evaluation for MCNN. On chestx-ray14 dataset, the experimental results show that CP-MCNN performs better than MCNN in terms of Sub-accuracy, Hamming-loss, Ranking-loss and Average Precision. Moreover, CP-MCNN can provide well-calibrated confidence prediction on chest X-ray sensor picture in order to enhance its reliability and interpretability.
KeywordsSensor Chest X-ray Conformal predictor CNN Multi-label
This work is supported by National Natural Science Foundation of China under Grant No. 61673186, the Natural Science Foundation of Fujian Province in China under Grant No. 2012J01274.
- 1.Balasubramanian, V., Ho, S.S., Vovk, V.: Conformal prediction for reliable machine learning: theory, adaptations and applications (2014)Google Scholar
- 4.Li, Z., Zheng, Y., Zhang, C., Shi, Z.: Combining deep feature and multi-label classification for semantic image annotation. J. Comput.-Aided Des. Comput. Graph. 30(2), 318 (2018)Google Scholar
- 5.Dewey, M., Kachelrieß, M.: Fundamentals of X-ray computed tomography: acquisition and reconstruction. In: Sack, I., Schaeffter, T. (eds.) Quantification of Biophysical Parameters in Medical Imaging, pp. 325–339. Springer, Cham (2018). https://doi.org/10.1007/978-3-319-65924-4_14CrossRefGoogle Scholar
- 8.Rajpurkar, P., et al.: CheXNet: radiologist-level pneumonia detection on chest x-rays with deep learning 1711, 1 November 2017. http://adsabs.harvard.edu/abs/2017arXiv171105225R
- 9.Schmidhuber, J.: Deep learning in neural networks: an overview (2014)Google Scholar
- 10.Shafer, G., Vovk, V.: A tutorial on conformal prediction. J. Mach. Learn. Res. 9, 371–421 (2008). \(<\)Go to ISI\(>\)://000256642000002Google Scholar
- 11.Iwata, T., Ghahramani, Z.: Improving output uncertainty estimation and generalization in deep learning via neural network Gaussian processes (2017)Google Scholar
- 14.Wang, X., Peng, Y., Lu, L., Lu, Z., Bagheri, M., Summers, R.M.: Chestx-ray8: hospital-scale chest x-ray database and benchmarks on weakly-supervised classification and localization of common thorax diseases. In: IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 3462–3471 (2017)Google Scholar
- 15.Zhang, M., Zhou, Z.: A review on multi-label learning algorithms (2013)Google Scholar
- 16.Zhao, H.J.W.C.C.: Pulmonary tuberculosis detection model of chest x-ray images using convolutional neural network, 8 July 2018Google Scholar
- 17.Zhu, J., Liao, S., Yi, D., Lei, Z., Li, S.Z.: Multi-label CNN based pedestrian attribute learning for soft biometrics. In: International Conference on Biometrics, pp. 535–540 (2015)Google Scholar