An Architecture based on CNNs and BiLSTMs for Slice-Level and Series-Level Intracranial Hemorrhage Identification in CT Scans

  • Daniel Henrique Comério IFES
  • Karin Satie Komati IFES
  • Thiago Oliveira-Santos UFES
  • Filipe Mutz UFES

Resumo


This work proposes an architecture of neural networks for estimating the probability of intracranial hemorrhage and its subtypes in computed tomography images and series. The architecture consists of three stages, with the first being a CNN and the other two being BiLSTM recurrent networks. The first stage receives as input a CT image and returns the hemorrhages probabilities. The second stage improves these estimates using contextual information from neighboring images. The final stage integrates the predictions from all slices in order to provide an unified output for the series. Extensive experiments were performed using the datasets RNSA, CQ500, and PhysioNet for evaluating configurations of the architecture, improvements produced by each component, and generalization to new data. The best configuration uses the DenseNet-121 as backbone and achieved average accuracy, precision, recall and f1-score over datasets of 91%, 91%, 90% and 90% confirming the model’s robustness and generalization.

Palavras-chave: Convolutional neural networks, BiLSTM, Computed tomography, Intracranial hemorrhage

Referências

H. Salehinejad, J. Kitamura, N. Ditkofsky, A. Lin, A. Bharatha, S. Suthiphosuwan, H.-M. Lin, J. Wilson, M. Mamdani, and E. Colak, “A real-world demonstration of machine learning generalizability in the detection of intracranial hemorrhage on head computerized tomography,” Scientific Reports, vol. 11, 08 2021.

X. Wang, T. Shen, S. Yang, J. Lan, Y. Xu, M. Wang, J. Zhang, and X. Han, “A deep learning algorithm for automatic detection and classification of acute intracranial hemorrhages in head ct scans,” NeuroImage: Clinical, 08 2021.

Brasil and M. da Saúde, “Linha de cuidados em acidente vascular cerebral (avc) na rede de atenção às urgências e emergências,” 2012.

M. R. Arbabshirani, B. K. Fornwalt, G. J. Mongelluzzo, J. D. Suever, B. D. Geise, A. A. Patel, and G. J. Moore, “Advanced machine learning in action: identification of intracranial hemorrhage on computed tomography scans of the head with clinical workflow integration,” NPJ digital medicine, vol. 1, no. 1, pp. 1–7, 2018.

A. Flanders, L. Prevedello, G. Shih, S. Halabi, J. Kalpathy-Cramer, R. Ball, J. Mongan, A. Stein, F. Kitamura, M. Lungren, G. Choudhary, L. Cala, L. Coelho, M. Mogensen, F. Morón, E. Miller, I. Ikuta, V. Zohrabian, O. McDonnell, and J. Nath, “Construction of a machine learning dataset through collaboration: The rsna 2019 brain ct hemorrhage challenge,” Radiology: Artificial Intelligence, vol. 2, p. e190211, 05 2020.

A. Graves, S. Fernández, and J. Schmidhuber, “Bidirectional lstm networks for improved phoneme classification and recognition,” in International conference on artificial neural networks, pp. 799–804, Springer, 2005.

V. Feigin, M. Brainin, B. Norrving, S. Martins, R. Sacco, W. Hacke, M. Fisher, J. Pandian, and P. Lindsay, “World stroke organization (WSO): Global stroke fact sheet 2022,” International Journal of Stroke, vol. 17, pp. 18–29, 01 2022.

M. Grewal, M. M. Srivastava, P. Kumar, and S. Varadarajan, “Radnet: Radiologist level accuracy using deep learning for hemorrhage detection in ct scans,” in 2018 IEEE 15th International Symposium on Biomedical Imaging (ISBI 2018), pp. 281–284, IEEE, 2018.

S. Chilamkurthy, R. Ghosh, S. Tanamala, M. Biviji, N. G. Campeau, V. K. Venugopal, V. Mahajan, P. Rao, and P. Warier, “Deep learning algorithms for detection of critical findings in head ct scans: a retrospective study,” The Lancet, vol. 392, no. 10162, pp. 2388–2396, 2018.

E. P. Reis, F. Nascimento, M. Aranha, F. M. Secol, B. Machado, M. Felix, A. Stein, and E. Amaro, “Brain hemorrhage extended (bhx): Bounding box extrapolation from thick to thin slice ct images,” PhysioNet, vol. 101, no. 23, pp. e215–20, 2020.

M. Hssayeni, “Computed tomography images for intracranial hemorrhage detection and segmentation (version 1.3.1). physionet,” 2020.

T. M. Buzug, “Computed tomography,” in Springer handbook of medical technology, pp. 311–342, Springer, 2011.

G. Huang, Z. Liu, and K. Weinberger, “Densely connected convolutional networks,” p. 12, 08 2016.

S. Xie, R. Girshick, P. Dollar, Z. Tu, and K. He, “Aggregated residual transformations for deep neural networks,” pp. 5987–5995, 07 2017.

F. Iandola, S. Han, M. Moskewicz, K. Ashraf, W. Dally, and K. Keutzer, “Squeezenet: Alexnet-level accuracy with 50x fewer parameters and ¡0.5mb model size,” 02 2016.

J. Deng, W. Dong, R. Socher, L.-J. Li, K. Li, and L. Fei-Fei, “Imagenet: A large-scale hierarchical image database,” in 2009 IEEE conference on computer vision and pattern recognition, pp. 248–255, Ieee, 2009.
Publicado
13/11/2023
COMÉRIO, Daniel Henrique; KOMATI, Karin Satie; OLIVEIRA-SANTOS, Thiago; MUTZ, Filipe. An Architecture based on CNNs and BiLSTMs for Slice-Level and Series-Level Intracranial Hemorrhage Identification in CT Scans. In: WORKSHOP DE VISÃO COMPUTACIONAL (WVC), 18. , 2023, São Bernardo do Campo/SP. Anais [...]. Porto Alegre: Sociedade Brasileira de Computação, 2023 . p. 12-17. DOI: https://doi.org/10.5753/wvc.2023.27525.

Artigos mais lidos do(s) mesmo(s) autor(es)

Obs.: Esse plugin requer que pelo menos um plugin de estatísticas/relatórios esteja habilitado. Se o seu plugins de estatísticas oferece mais que uma métrica, então, por favor, também selecione uma métrica principal na página de configurações administrativas do site e/ou da revista.