Otimização de Hiperparâmetros de Redes Neurais Profundas para Detecção de Cardiomegalia em radiografias do tórax

• Saulo Enock Rodrigues Fernandes UFMA
• Ricardo Costa da Silva Marques UFMA
• João Dallyson Sousa de Almeida UFMA
• Anselmo Cardoso de Paiva UFMA
• Geraldo Braz Júnior UFMA

DOI: https://doi.org/10.5753/sbcas.2022.222560

Abstract

Chest X-Ray is one of the most commonly used exams for diagnosing thoracic diseases. However, due to the complexity of chest diseases, there is a demand for skilled and experienced physicians to mitigate the error chances in diagnosing such pathology. Multiple approaches using Neural Networks have been devised to aid specialists in detecting chest diseases such as lung opacity and pneumonia. This paper proposes a methodology striving to help health professionals diagnose thoracic diseases. Therefore, we present an image quantity balancing and hyperparameter optimization approach. The tests were focused on cardiomegaly classification and the results obtained proved to be promising with values of 0.919 (AUC), 0.873 (Accuracy), 0.842 (Precision), 0.876 (F1Score), 0.913 (Sensitivity), 0.790 (Specificity), approaching the best metrics present in the literature.

Keywords: Deep Neural Networks, Diagnosis of Cardiomegaly, Hyperparameters Optimization

References

Agostoni, P., Cattadori, G., Guazzi, M., Palermo, P., Bussotti, M., and Marenzi, G. (2000). Cardiomegaly as a possible cause of lung dysfunction in patients with heart failure. American Heart Journal, 140(5):A17-A21.

Allaouzi, I. and Ahmed, M. B. (2019). A novel approach for multi-label chest x-ray classification of common thorax diseases. IEEE Access, 7:64279-64288.

Anderson, L., Dean, A., Falzon, D., Floyd, K., Baena, I., Gilpin, C., Glaziou, P., Hamada, Y., Hiatt, T., and A. Char, e. a. (2015). Global tuberculosis report 2015. World Health Organization, 20th edition.

Bellaviti, N., Bini, F., Pennacchi, L., Pepe, G., Bodini, B., Ceriani, R., D'Urbano, C., and Vaghi, A. (2016). Increased incidence of spontaneous pneumothorax in very young people: Observations and treatment. Chest Journal, 150(4):560A.

Bergstra, J., Komer, B., Eliasmith, C., Yamins, D., and Cox, D. D. (2015). Hyperopt: a python library for model selection and hyperparameter optimization. Computational Science & Discovery, 8(1):014008.

Bui, A. L., Horwich, T. B., and Fonarow, G. C. (2011). Epidemiology and risk profile of heart failure. Nature Reviews Cardiology, 8(1):30-41.

Cardenas, D., Junior, J. F., Moreno, R., Rebelo, M., Krieger, J., and Gutierrez, M. (2020). Multicenter validation of convolutional neural networks for automated detection of cardiomegaly on chest radiographs. In Anais do XX Simpósio Brasileiro de Computação Aplicada à Saúde, pages 179-190, Porto Alegre, RS, Brasil. SBC.

Chollet, F. (2017). Xception: Deep learning with depthwise separable convolutions. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 1251-1258.

Fernandes, V., Junior, G. B., de Paiva, A. C., Silva, A. C., and Gattass, M. (2021). Bayesian convolutional neural network estimation for pediatric pneumonia detection and diagnosis. Computer Methods and Programs in Biomedicine, 208:106259.

He, K., Zhang, X., Ren, S., and Sun, J. (2015). Deep residual learning for image recognition. CoRR, abs/1512.03385.

Huang, G., Liu, Z., and Weinberger, K. Q. (2016). Densely connected convolutional networks. CoRR, abs/1608.06993.

Irvin, J., Rajpurkar, P., Ko, M., Yu, Y., Ciurea-Ilcus, S., Chute, C., Marklund, H., Haghgoo, B., Ball, R., Shpanskaya, K., Seekins, J., Mong, D. A., Halabi, S. S., Sandberg, J. K., Jones, R., Larson, D. B., Langlotz, C. P., Patel, B. N., Lungren, M. P., and Ng, A. Y. (2019). Chexpert: A large chest radiograph dataset with uncertainty labels and expert comparison. AAAI-19/IAAI-19/EAAI-20 Proceedings, 33(1):590-597.

Pham, H. H., Le, T. T., Tran, D. Q., Ngo, D. T., and Nguyen, H. Q. (2020). Interpreting chest x-rays via cnns that exploit hierarchical disease dependencies and uncertainty labels.

Randtke, M. A., Andrews, B. P., and Mach, W. J. (2015). Pathophysiology and prevention of intraoperative atelectasis: A review of the literature. Journal of PeriAnesthesia Nursing, 30(6):516-527.

Silva, J. D. D. d. (2011). Repercussões da cardiomegalia na distribuição regional da ventilação na caixa torácica em indivíduos com cardiopatia chagásica crônica. Programa de Pós-Graduação em Fisioterapia, Universidade Federal de Pernambuco, Dissertação (Mestrado).

Simonyan, K. and Zisserman, A. (2014). Very deep convolutional networks for large-scale image recognition. arXiv preprint arXiv:1409.1556.

Szegedy, C., Vanhoucke, V., Ioffe, S., Shlens, J., and Wojna, Z. (2015). Rethinking the inception architecture for computer vision. CoRR, abs/1512.00567.

Yang, H.-Y., Yang, J., Pan, Y., Cao, K., Song, Q., Gao, F., and Yin, Y. (2019). Learn to be uncertain: Leveraging uncertain labels in chest x-rays with bayesian neural networks. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops.