Skip to main content
placeholder image

Fundus blood vessels detection based on pulse coupled neural network

Journal Article


Abstract


  • Precise fundus image blood vessels detection is an important step in diabetic retinopathy screening. In this paper, based on pulse coupled neural network (PCNN), a new fundus blood vessel detection method is proposed. In preprocessing, the contrast limited adaptive histogram equalization (CLAHE) and the two-dimensional Gaussian matched filter are used to enhance the contrast of the fundus image. Then based on PCNN model and maximal entropy criterion, the preprocessed fundus image is segmented. Finally, area filtering and breakpoint connection are performed, and the final vessel detection result is obtained. Experiment results show that not only main blood vessels but also tiny vessels are detected in normal and abnormal fundus images, and it is superior to Hoover method and very similar with hand-labeled result.

Authors


  •   Wang, Shuqin (external author)
  •   Xiao, Zhitao (external author)
  •   Wu, Jun (external author)
  •   Geng, Lei (external author)
  •   Zhang, Fang (external author)
  •   Xi, Jiangtao

Publication Date


  • 2012

Citation


  • S. Wang, Z. Xiao, J. Wu, L. Geng, F. Zhang & J. Xi, "Fundus blood vessels detection based on pulse coupled neural network," International Journal of Digital Content Technology and its Applications, vol. 6, (15) pp. 467-474, 2012.

Scopus Eid


  • 2-s2.0-84866433805

Ro Metadata Url


  • http://ro.uow.edu.au/infopapers/2299

Has Global Citation Frequency


Number Of Pages


  • 7

Start Page


  • 467

End Page


  • 474

Volume


  • 6

Issue


  • 15

Place Of Publication


  • Korea

Abstract


  • Precise fundus image blood vessels detection is an important step in diabetic retinopathy screening. In this paper, based on pulse coupled neural network (PCNN), a new fundus blood vessel detection method is proposed. In preprocessing, the contrast limited adaptive histogram equalization (CLAHE) and the two-dimensional Gaussian matched filter are used to enhance the contrast of the fundus image. Then based on PCNN model and maximal entropy criterion, the preprocessed fundus image is segmented. Finally, area filtering and breakpoint connection are performed, and the final vessel detection result is obtained. Experiment results show that not only main blood vessels but also tiny vessels are detected in normal and abnormal fundus images, and it is superior to Hoover method and very similar with hand-labeled result.

Authors


  •   Wang, Shuqin (external author)
  •   Xiao, Zhitao (external author)
  •   Wu, Jun (external author)
  •   Geng, Lei (external author)
  •   Zhang, Fang (external author)
  •   Xi, Jiangtao

Publication Date


  • 2012

Citation


  • S. Wang, Z. Xiao, J. Wu, L. Geng, F. Zhang & J. Xi, "Fundus blood vessels detection based on pulse coupled neural network," International Journal of Digital Content Technology and its Applications, vol. 6, (15) pp. 467-474, 2012.

Scopus Eid


  • 2-s2.0-84866433805

Ro Metadata Url


  • http://ro.uow.edu.au/infopapers/2299

Has Global Citation Frequency


Number Of Pages


  • 7

Start Page


  • 467

End Page


  • 474

Volume


  • 6

Issue


  • 15

Place Of Publication


  • Korea