Transfer Learning for Segmentation of Injured Lungs Using Coarse-to-Fine Convolutional Neural Networks
Deep learning using convolutional neural networks (ConvNets) achieves high accuracy across many computer vision tasks, with the ability to learn multi-scale features and generalize across a variety of input data. In this work, we propose a deep learning framework that utilizes a coarse-to-fine cascade of 3D ConvNet models for segmentation of lung structures obtained from computed tomographic (CT) images. Deep learning requires a large number of training datasets, which may be challenging in medical imaging, especially for rare diseases. In the present study, transfer learning is utilized for lung segmentation of CT scans in large animal models of the acute respiratory distress syndrome (ARDS) using only 13 subjects. The method was quantitatively evaluated on a human dataset, consisting of 395 3D CT scans from 153 subjects, and an animal dataset consisting of 148 3D CT images from 5 porcine subjects. The human dataset achieved an average Jacaard index of 0.99, and an average symmetric surface distance (ASSD) of 0.29 mm. The animal dataset had an average Jacaard index of 0.94, and an ASSD of 0.99 mm.
KeywordsSegmentation Computed tomography Deep learning
This work was supported in part by NIH grant CA166703 (J. M. Reinhardt and S. E. Gerard), the National Heart, Lung, and Blood Institute Grants R01-HL-112986, and R01-HL-126838 (D. W. Kaczka), the University of Iowa, Department of Anesthesia (D. W. Kaczka and J. Herrmann), and by the Office of the Assistant Secretary of Defense for Health Affairs through the Peer Reviewed Medical Research Program under Award no. W81XWH-16-1-0434 (D. W. Kaczka).
S.E. Gerard received support from a Presidential Fellowship through the University of Iowa Graduate College and from a NASA Iowa Space Grant Consortium Fellowship.
The authors thank Dr. Eric Hoffman and Dr. Junfeng Guo from the University of Iowa for providing the CT image data and ground truth segmentations for the IPF cases.
Disclosure: Opinions, interpretations, conclusions, and recommendations are those of the authors and are not necessarily endorsed by the U.S. Department of Defense. J. M. Reinhardt is a shareholder in VIDA Diagnostics, Inc.
- 2.Guo, J., Fuld, M.K., Alford, S.K., Reinhardt, J.M., Hoffman, E.A.: Pulmonary analysis software suite 9.0: integrating quantitative measures of function with structural analyses. In: Brown, M., et al., (eds.) First International Workshop on Pulmonary Image Analysis, pp. 283–292 (2008)Google Scholar
- 7.Krizhevsky, A., Sutskever, I., Hinton, G.E.: Imagenet classification with deep convolutional neural networks. In: Advances in Neural Information Processing Systems, pp. 1097–1105 (2012)Google Scholar
- 8.Harrison, A.P., Xu, Z., George, K., Lu, L., Summers, R.M., Mollura, D.J.: Progressive and multi-path holistically nested neural networks for pathological lung segmentation from CT images. In: Descoteaux, M., Maier-Hein, L., Franz, A., Jannin, P., Collins, D.L., Duchesne, S. (eds.) MICCAI 2017. LNCS, vol. 10435, pp. 621–629. Springer, Cham (2017). https://doi.org/10.1007/978-3-319-66179-7_71CrossRefGoogle Scholar
- 9.Anthimopoulos, M., Christodoulidis, S., Ebner, L., Geiser, T., Christe, A., Mougiakakou, S.: Semantic segmentation of pathological lung tissue with dilated fully convolutional networks. arXiv preprint arXiv:1803.06167 (2018)
- 16.Gerard, S.E., Patton, T.J., Christensen, G.E., Bayouth, J.E., Reinhardt, J.M.: FissureNet: A deep learning approach for pulmonary fissure detection in CT images. IEEE Trans. Med. Imaging (2018)Google Scholar
- 17.Ronneberger, O., Fischer, P., Brox, T.: U-Net: convolutional networks for biomedical image segmentation. In: Navab, N., Hornegger, J., Wells, W.M., Frangi, A.F. (eds.) MICCAI 2015. LNCS, vol. 9351, pp. 234–241. Springer, Cham (2015). https://doi.org/10.1007/978-3-319-24574-4_28CrossRefGoogle Scholar
- 18.Glorot, X., Bengio, Y.: Understanding the difficulty of training deep feedforward neural networks. In: Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics, pp. 249–256 (2010)Google Scholar
- 19.Kingma, D., Ba, J.: Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980 (2014)
- 20.Theano Development Team: Theano: A Python framework for fast computation of mathematical expressions. arXiv e-prints abs/1605.02688, May 2016
- 21.Dieleman, S., et al.: Lasagne: First Release, vol. 3. Zenodo, Geneva (2015)Google Scholar