Configuração de hyper-parâmetros de modelos deep learning para auxílio no pós-diagnóstico de Tuberculose
Abstract
Tuberculosis (TB) is recognised as the deadliest disease in the world, according to the World Health Organization (WHO), being one of the ten leading causes of death in the world, as well as being the leading cause of death for people with HIV. Brazil is one of the countries with a high TB burden, and one of the highest mortality rates in the country is in the state of Amazonas. The aim of this paper is to analyse deep learning (DL) models to assist in the post-diagnosis of TB, predicting the severity of the disease in the patient. Two DL models are proposed and the Grid-search technique is applied to define configurations with the best performance. DL models yield interesting results, with a fully connected DL configuration reaching specificity of 83.4%.
References
Cao, Y., Liu, C., Liu, B., Brunette, M. J., Zhang, N., Sun, T., Zhang, P., Peinado, J., Garavito, E. S., Garcia, L. L., et al. (2016). Improving tuberculosis diagnostics using deep learning and mobile health technologies among resource-poor and marginalized communities. In IEEE International Conference on Connected Health (CHASE), pages 274–281. IEEE.
Goldman, L. and Schafer, A. I. (2015). Goldman-Cecil Medicine E-Book. Elsevier Health Sciences.
Han, H., Zhu, X., and Li, Y. (2018). Edlt: Enabling deep learning for generic data classification. In 2018 IEEE International Conference on Data Mining (ICDM), pages 147–156.
Lakhani, P. and Sundaram, B. (2017). Deep learning at chest radiography: automated classification of pulmonary tuberculosis by using convolutional neural networks. Radiology, 284(2):574–582.
LeCun, Y., Bengio, Y., and Hinton, G. (2015). Deep learning. nature 521.
Liu, X.-Y., Wu, J., and Zhou, Z.-H. (2008). Exploratory undersampling for classimbalance learning. IEEE Transactions on Systems, Man, and Cybernetics, Part B (Cybernetics), 39(2):539–550.
Rajaraman, S., Candemir, S., Xue, Z., Alderson, P. O., Kohli, M., Abuya, J., Thoma, G. R., and Antani, S. (2018). A novel stacked generalization of models for improved tb detection in chest radiographs. In IEEE Engineering in Medicine and Biology Society (EMBC), pages 718–721. IEEE.
Ryndak, M. B. and Laal, S. (2019). Mycobacterium tuberculosis primary infection and dissemination: A critical role for alveolar epithelial cells. Frontiers in Cellular and Infection Microbiology, 9:299.
Shin, H., Roth, H. R., Gao, M., Lu, L., Xu, Z., Nogues, I., Yao, J., Mollura, D., and Summers, R. M. (2016). Deep convolutional neural networks for computer-aided detection: Cnn architectures, dataset characteristics and transfer learning. IEEE Transactions on Medical Imaging, 35(5):1285–1298.
