Detail publikace

Registration of the long-term autofluorescence retinal images taken from glaucomatic patients

Originální název

Registration of the long-term autofluorescence retinal images taken from glaucomatic patients

Anglický název

Registration of the long-term autofluorescence retinal images taken from glaucomatic patients

Jazyk

en

Originální abstrakt

This paper describes a new robust framework for registration of the autofluorescence images. The images were acquired from glaucomatous eyes between 2003 and 2012 by different laser scanning devices operating at wavelength 488 nm. Presented framework uses Fourier phase correlation theorem for robust estimate of the shift and rotation. This is followed by blood-vessels detection and standard intensity optimization framework using these blood-vessels images and correlation criteria. The results were evaluated subjectively and also by entropy correlation coefficient (ECC). We showed that our registration framework is comparable with manually registered images by the means of ECC statistical comparison.

Anglický abstrakt

This paper describes a new robust framework for registration of the autofluorescence images. The images were acquired from glaucomatous eyes between 2003 and 2012 by different laser scanning devices operating at wavelength 488 nm. Presented framework uses Fourier phase correlation theorem for robust estimate of the shift and rotation. This is followed by blood-vessels detection and standard intensity optimization framework using these blood-vessels images and correlation criteria. The results were evaluated subjectively and also by entropy correlation coefficient (ECC). We showed that our registration framework is comparable with manually registered images by the means of ECC statistical comparison.

BibTex


@inproceedings{BUT102065,
  author="Radim {Kolář} and Jan {Odstrčilík} and Pavlína {Macková} and Robert {Laemmer}",
  title="Registration of the long-term autofluorescence retinal images taken from glaucomatic patients",
  annote="This paper describes a new robust framework for registration of the autofluorescence images. The images were acquired from glaucomatous eyes between 2003 and 2012 by different laser scanning devices operating at wavelength 488 nm. Presented framework uses Fourier phase correlation theorem for robust estimate of the shift and rotation. This is followed by blood-vessels detection and standard intensity optimization framework using these blood-vessels images and correlation criteria. The results were evaluated subjectively and also by entropy correlation coefficient (ECC). We showed that our registration framework is comparable with manually registered images by the means of ECC statistical comparison.",
  address="Taylor & Francis",
  booktitle="Computational Vision and Medical Image Processing",
  chapter="102065",
  howpublished="print",
  institution="Taylor & Francis",
  year="2013",
  month="october",
  pages="115--119",
  publisher="Taylor & Francis",
  type="conference paper"
}