Advertisement

A Deep Learning Based Automatic Severity Detector for Diabetic Retinopathy

  • Rawan AlSaadEmail author
  • Somaya Al-maadeed
  • Md. Abdullah Al Mamun
  • Sabri Boughorbel
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 10934)

Abstract

Automated Diabetic Retinopathy (DR) screening methods with high accuracy have the strong potential to assist doctors in evaluating more patients and quickly routing those who need help to a specialist. In this work, we used Deep Convolutional Neural Network architecture to diagnosing DR from digital fundus images and accurately classifying its severity. We train this network using a graphics processor unit (GPU) on the publicly available Kaggle dataset. We used Theano, Lasagne, and cuDNN libraries on two Amazon EC2 p2.xlarge instances and demonstrated impressive results, particularly for a high-level classification task. On the dataset of 30,262 training images and 4864 testing images, our model achieves an accuracy of 72%. Our experimental results showed that increasing the batch size does not necessarily speed up the convergence of the gradient computations. Also, it demonstrated that the number and size of fully connected layers do not have a significant impact on the performance of the model.

Keywords

Deep learning Convolutional Neural Networks Medical imaging Diabetic retinopathy 

Notes

Acknowledgment

This work was supported by Sidra Medicine (authors RA and SB), as well as a grant from the Qatar National Research Fund through National Priority Research Program (NPRP) No. 6-249-1-053 (authors SA and MA).

References

  1. 1.
    Boyd, K.: What is Diabetic Retinopathy? https://www.aao.org/eye-health/diseases/what-is-diabetic-retinopathy (2017). Accessed 25 Feb 2017
  2. 2.
    Federation ID: Diabetes: Facts and Figures (2017). http://www.idf.org/about-diabetes/facts-figures. Accessed 25 Feb 2017
  3. 3.
    HMC Expert: Regular Checkups Can Help Prevent Diabetic Retinopathy (2017). https://www.hamad.qa/EN/news/2016/July/Pages/Regular-Checkups-can-help-prevent-Diabetic-Retinopathy-says-HMC-Expert.aspx. Accessed 25 Feb 2017
  4. 4.
    Elshafei, M., Gamra, H., Khandekar, R., Al Hashimi, M., Pai, A., Farouk Ahmed, M.: Prevalence and determinants of diabetic retinopathy among persons 40 years of age with diabetes in Qatar: a community-based survey. Eur. J. Ophthalmol. 21(1), 39 (2011)CrossRefGoogle Scholar
  5. 5.
    Kaggle: Diabetic Retinopathy Detection (2017). https://www.kaggle.com/c/diabetic-retinopathy-detection/data. Accessed 25 Feb 2017
  6. 6.
    Gulshan, V., Peng, L., Coram, M., Stumpe, M.C., Wu, D., Narayanaswamy, A., Venugopalan, S., Widner, K., Madams, T., Cuadros, J., et al.: Development and validation of a deep learning algorithm for detection of diabetic retinopathy in retinal fundus photographs. JAMA 316(22), 2402 (2016)CrossRefGoogle Scholar
  7. 7.
    Haloi, M.: arXiv preprint arXiv:1505.04424 (2015)
  8. 8.
    Wang, S., Yin, Y., Cao, G., Wei, B., Zheng, Y., Yang, G.: Hierarchical retinal blood vessel segmentation based on feature and ensemble learning. Neurocomputing 149, 708 (2015)CrossRefGoogle Scholar
  9. 9.
    Melinščak, M., Prentašić, P., Lončarić, S.: In: VISAPP 2015, 10th International Conference on Computer Vision Theory and Applications (2015)Google Scholar
  10. 10.
    Fraz, M.M., Remagnino, P., Hoppe, A., Uyyanonvara, B., Rudnicka, A.R., Owen, C.G., Barman, S.A.: Blood vessel segmentation methodologies in retinal images - a survey. Comput. Methods Programs Biomed. 108(1), 407 (2012)CrossRefGoogle Scholar
  11. 11.
    Pratt, H., Coenen, F., Broadbent, D.M., Harding, S.P., Zheng, Y.: Convolutional neural networks for diabetic retinopathy. Procedia Comput. Sci. 90, 200 (2016)CrossRefGoogle Scholar
  12. 12.
    Kaggle-dr: My Entry for the Kaggle Diabetic Retinopathy Competition for 20/661 Place (2017). https://github.com/ilyakava/kaggle-dr. Accessed 25 Feb 2017
  13. 13.
    Theano: Theano 0.8.2 Documentation (2017). http://deeplearning.net/software/theano/. Accessed 25 Feb 2017
  14. 14.
    Lasagne: Lightweight Library to Build and Train Neural Networks in Theano (2017). https://github.com/Lasagne/Lasagne. Accessed 25 Feb 2017
  15. 15.
    Pylearn2: Pylearn2 Documentation (2017). http://deeplearning.net/software/pylearn2/. Accessed 25 Feb 2017
  16. 16.
    NAVIDA: CUDA Toolkit Documentation (2017). http://docs.nvidia.com/cuda. Accessed 25 Feb 2017
  17. 17.
    GraphicsMagick: GraphicsMagick Image Processing System (2017). http://www.graphicsmagick.org. Accessed 25 Feb 2017
  18. 18.
    Wu, R., Yan, S., Shan, Y., Dang, Q., Sun, G.: arXiv preprint arXiv:1501.028767(8) (2015)
  19. 19.
    Simonyan, K., Zisserman, A.: arXiv preprint arXiv:1409.1556 (2014)
  20. 20.
    Cohen, J.: A coefficient of agreement for nominal scales. Educ. Psychol. Meas. 20(1), 37 (1960)CrossRefGoogle Scholar

Copyright information

© Springer International Publishing AG, part of Springer Nature 2018

Authors and Affiliations

  • Rawan AlSaad
    • 1
    • 2
    Email author
  • Somaya Al-maadeed
    • 2
  • Md. Abdullah Al Mamun
    • 2
  • Sabri Boughorbel
    • 1
  1. 1.Systems Biology DepartmentSidra MedicineDohaQatar
  2. 2.Department of Computer Science and EngineeringQatar UniversityDohaQatar

Personalised recommendations