AGAN: An Anatomy Corrector Conditional Generative Adversarial Network

Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 12262)


The accurate segmentation of medical images has important consequences in clinical applications. Noisy and artefact-heavy images can result in erroneous image segmentation and often require expert understanding of the target anatomy by clinicians to interpret and compensate for missing and obfuscated data. This is especially true in ultrasound imaging where shadowing and speckle artefacts are common. We propose a novel approach to handle such artefacts using a conditional Generative Adversarial Network called Anatomical GAN (AGAN) that can correct anatomically-invalid pixel-wise segmentation and impose shape priors in carotid artery ultrasound images by learning the underlying structure of the arteries. These anatomically accurate outputs can then be used in the clinical work flow by clinicians or be further processed by other automated methods for assistance in clinical decision making. AGAN can be chained with any pixel-wise segmentation method and is generalisable for both anatomy and artefacts. Experimental results on a longitudinal ultrasound carotid artery dataset show that AGAN can correct anatomically-invalid segmentation masks obtained with different pixel-wise segmentation methods when other state-of-the-art methods fail.


Deep learning Convolutional neural networks Generative adversarial networks Segmentation Shape priors Ultrasound 


  1. 1.
    Abadi, M., et al.: TensorFlow: Large-scale machine learning on heterogeneous systems (2015). Software available from
  2. 2.
    Bouteldja, N., Merhof, D., Ehrhardt, J., Heinrich, M.P.: Deep multi-modal encoder-decoder networks for shape constrained segmentation and joint representation learning. In: Handels, H., Deserno, T.M., Maier, A., Maier-Hein, K.H., Palm, C., Tolxdorff, T. (eds.) Bildverarbeitung für die Medizin 2019, pp. 23–28. Springer Fachmedien Wiesbaden, Wiesbaden (2019)CrossRefGoogle Scholar
  3. 3.
    Dalca, A.V., Guttag, J.V., Sabuncu, M.R.: Anatomical priors in convolutional networks for unsupervised biomedical segmentation. CoRR abs/1903.03148 (2019).
  4. 4.
    Isola, P., Zhu, J.Y., Zhou, T., Efros, A.A.: Image-to-image translation with conditional adversarial networks. In: 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 5967–5976 (2016)Google Scholar
  5. 5.
    Kingma, D.P., Ba, J.: Adam: a method for stochastic optimization. CoRR abs/1412.6980 (2014)Google Scholar
  6. 6.
    Larrazabal, A.J., Martinez, C., Ferrante, E.: Anatomical priors for image segmentation via post-processing with denoising autoencoders. In: Shen, D., et al. (eds.) MICCAI 2019. LNCS, vol. 11769, pp. 585–593. Springer, Cham (2019). Scholar
  7. 7.
    Miyato, T., Kataoka, T., Koyama, M., Yoshida, Y.: Spectral normalization for generative adversarial networks. In: 6th International Conference on Learning Representations, ICLR 2018, Vancouver, BC, Canada, 30 April–3 May 2018, Conference Track Proceedings. (2018).
  8. 8.
    Oktay, O., et al.: Anatomically constrained neural networks (ACNN): application to cardiac image enhancement and segmentation. IEEE Trans. Med. Imaging (2017).
  9. 9.
    Ravishankar, H., Venkataramani, R., Thiruvenkadam, S., Sudhakar, P., Vaidya, V.: Learning and incorporating shape models for semantic segmentation. In: Descoteaux, M., Maier-Hein, L., Franz, A., Jannin, P., Collins, D.L., Duchesne, S. (eds.) MICCAI 2017. LNCS, vol. 10433, pp. 203–211. Springer, Cham (2017). Scholar
  10. 10.
    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). Scholar
  11. 11.
    Sekuboyina, A., Rempfler, M., Valentinitsch, A., Kirschke, J.S., Menze, B.H.: Adversarially learning a local anatomical prior: vertebrae labelling with 2D reformations. CoRR abs/1902.02205 (2019).
  12. 12.
    Shelhamer, E., Long, J., Darrell, T.: Fully convolutional networks for semantic segmentation. IEEE Trans. Pattern Anal. Mach. Intell. 39(4), 640–651 (2017).
  13. 13.
    Wang, T., Liu, M., Zhu, J., Tao, A., Kautz, J., Catanzaro, B.: High-resolution image synthesis and semantic manipulation with conditional GANs. CoRR abs/1711.11585 (2017).
  14. 14.
    Yi, Z., Zhang, H., Tan, P., Gong, M.: DualGAN: unsupervised dual learning for image-to-image translation. CoRR abs/1704.02510 (2017).
  15. 15.
    Zhang, H., Goodfellow, I.J., Metaxas, D.N., Odena, A.: Self-attention generative adversarial networks. In: Proceedings of the 36th International Conference on Machine Learning, ICML 2019, 9–15 June 2019, Long Beach, California, USA, pp. 7354–7363 (2019).
  16. 16.
    Zhou, Z., Rahman Siddiquee, M.M., Tajbakhsh, N., Liang, J.: UNet++: a nested U-net architecture for medical image segmentation. In: Stoyanov, D., et al. (eds.) DLMIA/ML-CDS -2018. LNCS, vol. 11045, pp. 3–11. Springer, Cham (2018). Scholar

Copyright information

© Springer Nature Switzerland AG 2020

Authors and Affiliations

  1. 1.See-Mode Technologies Pty. Ltd.MelbourneAustralia
  2. 2.Changi General HospitalSingaporeSingapore
  3. 3.National Neuroscience InstituteSingaporeSingapore
  4. 4.National University Health SystemSingaporeSingapore
  5. 5.National University of SingaporeSingaporeSingapore

Personalised recommendations