Oftalʹmologiâ (Oct 2018)

The Morphological Description Automation of the Optic Nerve Head Boundary in Digital Fundus Images

  • E. G. Tanaeva,
  • R. G. Khafizov

DOI
https://doi.org/10.18008/1816-5095-2018-3-325-329
Journal volume & issue
Vol. 15, no. 3
pp. 325 – 329

Abstract

Read online

Purpose. To develop the automating method for the description of the optic nerve head condition. The method allows to differentiate the optic nerve head with clear boundaries typical for the normal fundus and optic nerve head with blurred boundaries typical for different types of ophthalmopathology. Methods. The proposed method is based on integration algorithm of luminance samples along the diagonal. It shows that optic nerve head images with blurred boundaries are characterized by a more linear increase in brightness in the frontal region compared to the accumulation result for the case of normal optic nerve head. To identify this feature the firstorder derivative can be used. To reduce the effect of blood vessels on the result of processing luminance samples along the diagonal is performed in the red channel since the blood vessels effect on the optic nerve head border is minimized in the red color component. Results. The application of this method to images of optic nerve head with diffuse boundaries shows that the accumulation result is characterized by a slower brightness variation. Тhe analysis of 20 fundus images revealed that the value of the first-order derivative of the result of accumulation of luminance readings diagonally for images of optic nerve head with blurred boundaries is 2 times smaller than for images of optic nerve head with clear boundaries. Conclusions. Тhe presented method can be used to create expert systems which allow to automate the process of morphological description of the optic nerve head border from the fundus images and also can partially solve the diagnostic problem. Since the disappearance of the clarity of the optic nerve head borders and the appearance of their blurring are the main diagnostic criterions for various types of pathology this method can be used to create medical expert systems and software for the fundus images processing.

Keywords