Retinal Image Quality Classification Using Fine-Tuned CNN

  • Jing Sun
  • Cheng WanEmail author
  • Jun Cheng
  • Fengli Yu
  • Jiang Liu
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 10554)


Retinal image quality classification makes a great difference in automated diabetic retinopathy screening systems. With the increase of application of portable fundus cameras, we can get a large number of retinal images, but there are quite a number of images in poor quality because of uneven illumination, occlusion and patients movements. Using the dataset with poor quality training networks for DR screening system will lead to the decrease of accuracy. In this paper, we first explore four CNN architectures (AlexNet, GoogLeNet, VGG-16, and ResNet-50) from ImageNet image classification task to our Retinal fundus images quality classification, then we pick top two networks out and jointly fine-tune the two networks. The total loss of the network we proposed is equal to the sum of the losses of all channels. We demonstrate the super performance of our proposed algorithm on a large retinal fundus image dataset and achieve an optimal accuracy of 97.12%, outperforming the current methods in this area.


No-reference image quality assessment (NR-IQA) Convolutional neural networks (CNN) Retinal image Fine-tuning 


  1. 1.
    Ye, P.: Unsupervised feature learning framework for no-reference image quality assessment. In: IEEE Conference on Computer Vision and Pattern Recognition, vol. 157(10), pp. 1098–1105 (2012)Google Scholar
  2. 2.
    Wang, J.: A novel contourlet-based no-reference image quality assessment metric. In: International Research Association of Information and Computer Science 2014, pp. 3339–3352 (2014)Google Scholar
  3. 3.
    Lee, S.C., Wang, Y.: Automatic retinal image quality assessment and enhancement. In: Proceedings Spie, pp. 1581–1590 (1999)Google Scholar
  4. 4.
    Lalonde, M., Gagnon, L., Boucher, M.C.: Automatic visual quality assessment in optical fundus images. In: Proceedings of Vision Interface, vol. 18, pp. 437–450 (2001)Google Scholar
  5. 5.
    Yu, L., Tian, X., Li, T., Tian, J.: No-reference image quality assessment based on svm for video conferencing system. In: Lei, J., Wang, F.L., Li, M., Luo, Y. (eds.) Communications in Computer & Information Science 2016, vol. 345, pp. 555–560. Springer, Heidelberg (2012). doi: 10.1007/978-3-642-35211-9_70 Google Scholar
  6. 6.
    Suk, H.-I., Shen, D.: Deep learning-based feature representation for AD/MCI classification. In: Mori, K., Sakuma, I., Sato, Y., Barillot, C., Navab, N. (eds.) MICCAI 2013. LNCS, vol. 8150, pp. 583–590. Springer, Heidelberg (2013). doi: 10.1007/978-3-642-40763-5_72 CrossRefGoogle Scholar
  7. 7.
    Kang, L., Ye, P.: Convolutional neutral networks for no-reference image quality assessment. In: IEEE Conference on Computer Vision and Pattern Recognition 2014, pp. 1733–1740 (2014)Google Scholar
  8. 8.
    Tennakoon, R., Mahapatra, D., Roy, P.: Image quality classification for DR screening using convolutional neural networks. In: Chen, X., Garvin, K. (eds.) OMIA 2016, pp. 113–120 (2016)Google Scholar
  9. 9.
    Mahapatra, D.: Retinal image quality classification using neurobiological models of the human visual system. In: Chen, X., Garvin, K. (eds.) OMIA 2016, pp. 97–104 (2016)Google Scholar
  10. 10.
    Krizhevsky, A., Sutskever, I.: ImageNet classification with deep convolutional neural networks. In: International Conference on Neural Information Processing Systems 2012, vol. 25, pp. 1097–1105 (2012)Google Scholar
  11. 11.
    Szegedy, W., Liu, Y.: Going deeper with convolutions. In: Computer Vision and Pattern Recognition 2015, pp. 1–9 (2015)Google Scholar
  12. 12.
    Mohammadi, M., Das, S.: SNN: Stacked Neural Networks. arXiv:1605.08612 (2016)
  13. 13.
    Pratt, H., Coenen, F.: Convolutional neural networks for diabetic retinopathy. Proc. Comput. Sci. 90, 200–205 (2016)CrossRefGoogle Scholar

Copyright information

© Springer International Publishing AG 2017

Authors and Affiliations

  • Jing Sun
    • 1
  • Cheng Wan
    • 1
    Email author
  • Jun Cheng
    • 2
  • Fengli Yu
    • 1
  • Jiang Liu
    • 3
  1. 1.Nanjing University of Aeronautics and AstronauticsNanjingChina
  2. 2.Institute for Infocomm ResearchA*STARSingaporeSingapore
  3. 3.Ningbo Institute of Material Technology and EngineeringChinese Academy of SciencesNingboChina

Personalised recommendations