Advertisement

Some Investigations on Robustness of Deep Learning in Limited Angle Tomography

  • Yixing Huang
  • Tobias Würfl
  • Katharina Breininger
  • Ling Liu
  • Günter Lauritsch
  • Andreas Maier
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 11070)

Abstract

In computed tomography, image reconstruction from an insufficient angular range of projection data is called limited angle tomography. Due to missing data, reconstructed images suffer from artifacts, which cause boundary distortion, edge blurring, and intensity biases. Recently, deep learning methods have been applied very successfully to this problem in simulation studies. However, the robustness of neural networks for clinical applications is still a concern. It is reported that most neural networks are vulnerable to adversarial examples. In this paper, we aim to investigate whether some perturbations or noise will mislead a neural network to fail to detect an existing lesion. Our experiments demonstrate that the trained neural network, specifically the U-Net, is sensitive to Poisson noise. While the observed images appear artifact-free, anatomical structures may be located at wrong positions, e.g. the skin shifted by up to 1 cm. This kind of behavior can be reduced by retraining on data with simulated Poisson noise. However, we demonstrate that the retrained U-Net model is still susceptible to adversarial examples. We conclude the paper with suggestions towards robust deep-learning-based reconstruction.

Keywords

Deep learning Limited angle tomography Adversarial example 

References

  1. 1.
    Chen, Z., Jin, X., Li, L., Wang, G.: A limited-angle \(\text{ CT }\) reconstruction method based on anisotropic \(\text{ TV }\) minimization. Phys. Med. Biol. 58(7), 2119–2141 (2013)CrossRefGoogle Scholar
  2. 2.
    Evtimov, I., et al.: Robust physical-world attacks on deep learning models. arXiv preprint 1 (2017)Google Scholar
  3. 3.
    Goodfellow, I.J., Shlens, J., Szegedy, C.: Explaining and harnessing adversarial examples. arXiv preprint (2014)Google Scholar
  4. 4.
    Gu, J., Ye, J.C.: Multi-scale wavelet domain residual learning for limited-angle CT reconstruction. In: Proceedings of Fully3D, pp. 443–447 (2017)Google Scholar
  5. 5.
    Hammernik, K., Würfl, T., Pock, T., Maier, A.: A deep learning architecture for limited-angle computed tomography reconstruction. In: Maier-Hein, K.H., Deserno, T.M., Handels, H., Tolxdorff, T. (eds.) Bildverarbeitung für die Medizin 2017. Informatik aktuell, pp. 92–97. Springer, Heidelberg (2017).  https://doi.org/10.1007/978-3-662-54345-0_25CrossRefGoogle Scholar
  6. 6.
    Huang, Y., et al.: Restoration of missing data in limited angle tomography based on Helgason-Ludwig consistency conditions. Biomed. Phys. Eng. Express 3(3), 035015 (2017)CrossRefGoogle Scholar
  7. 7.
    Kurakin, A., Goodfellow, I., Bengio, S.: Adversarial examples in the physical world. arXiv preprint (2016)Google Scholar
  8. 8.
    Maier, A., et al.: Precision learning: towards use of known operators in neural networks. In: International Conference on Pattern Recognition (2018, to appear). https://arxiv.org/abs/1712.00374
  9. 9.
    Odena, A., Dumoulin, V., Olah, C.: Deconvolution and checkerboard artifacts. Distill 1(10), e3 (2016)CrossRefGoogle Scholar
  10. 10.
    Qu, G.R., Lan, Y.S., Jiang, M.: An iterative algorithm for angle-limited three-dimensional image reconstruction. Acta Math. Appl. Sin. Engl. Ser. 24(1), 157–166 (2008)MathSciNetCrossRefGoogle Scholar
  11. 11.
    Riess, C., Berger, M., Wu, H., Manhart, M., Fahrig, R., Maier, A.: TV or not TV? That is the question. In: Proceedings of Fully3D, pp. 341–344 (2013)Google Scholar
  12. 12.
    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
  13. 13.
    Sidky, E., Pan, X.: Image reconstruction in circular cone-beam computed tomography by constrained, total-variation minimization. Phys. Med. Biol. 53(17), 4777–4807 (2008)CrossRefGoogle Scholar
  14. 14.
    Szegedy, C., et al.: Intriguing properties of neural networks. arXiv preprint (2013)Google Scholar
  15. 15.
    Würfl, T., et al.: Deep learning computed tomography: learning projection-domain weights from image domain in limited angle problems. IEEE Trans. Med. Imaging 37(6), 1454–1463 (2018)CrossRefGoogle Scholar
  16. 16.
    Yuan, X., He, P., Zhu, Q., Bhat, R.R., Li, X.: Adversarial examples: attacks and defenses for deep learning. arXiv preprint (2017)Google Scholar

Copyright information

© Springer Nature Switzerland AG 2018

Authors and Affiliations

  • Yixing Huang
    • 1
  • Tobias Würfl
    • 1
  • Katharina Breininger
    • 1
  • Ling Liu
    • 1
  • Günter Lauritsch
    • 2
  • Andreas Maier
    • 1
    • 3
  1. 1.Friedrich-Alexander Universität Erlangen-NürnbergErlangenGermany
  2. 2.Siemens Healthcare GmbHForchheimGermany
  3. 3.Erlangen Graduate School in Advanced Optical Technologies (SAOT)ErlangenGermany

Personalised recommendations