Statistical and neural net methods for automatic glaucoma diagnosis determination

• Authors: František Pluháček , Jaroslav Pospíšil
• Languages of publication: EN

Abstracts

EN

Two new computer diagnostic methods for the automatic determination of glaucoma diagnosis are presented and compared in this paper: the statistical method and the neural net method. Both introduced methods evaluate colour glaucomatous changes within the optic disc area. The mentioned colour changes are numerically represented using a suitable image analysis process. Next, the investigated eye is classified to three defined glaucoma-risk classes with different reliability of the diagnosis determination. The verification and the reliability comparison of both studied methods are performed by virtue of the application of this methods to a set of normal healthy optic disc images and a set of glaucomatous optic disc images.

Keywords

EN

glaucoma; diagnostics method; pallor; image analysis; statistical evaluation; 87.57.Ra

• Publisher: De Gruyter Open
• Journal: Open Physics
• Year: 2004
• Volume: 2
• Issue: 1
• Pages: 12-24

Dates

published

1 - 3 - 2004

online

1 - 3 - 2004

Contributors

author

František Pluháček

• Department of Experimental Physics and Joint Laboratory of Optics of Palacký, University and Institute of Physics of Academy of Sciences, 17 listopadu 50a, 772 07, Olomouc, Czech Republic

author

Jaroslav Pospíšil

• Department of Experimental Physics and Joint Laboratory of Optics of Palacký, University and Institute of Physics of Academy of Sciences, 17 listopadu 50a, 772 07, Olomouc, Czech Republic

References

• [1] B. Shields: Textbook of Glaucoma, Williams & Wilkins, Baltimore, 1992.
• [2] K. Hanuš, J. Boguszaková, et al. Compendium of Eye Medicine, Grada, Praha, 1997.
• [3] A. P. Nesterov: Primary Glaucoma, Avicenum, Praha, 1991.
• [4] S. Řehák, R. Knobloch, et al. Medicine of the Eye, Avicenum, Praha, 1980.
• [5] P. Nagin, B. Schwartz: “Detection of increased pallor over time. Computerized image analysis in untreated ocular hypertension”, Ophthalmology, Vol. 92, (1985), pp. 252–261.
• [6] T.N. Cornsweet, S. Hersh, et al: “Quantification of the shape and color of the optic nerve head”, In: G.M. Breinin and I.M. Siegel (Ed.): Advances in Diagnostics Visual Optics, Springer-Verlag, New York, 1983, pp. 141–149.
• [7] R. Varma, G.L. Spaeth: “The PAR IS 2002: A new system for retinal digital image analysis”, Ophthalmic Surgery, Vol. 19, (1988), pp. 183–192.
• [8] A. Sugiyama, T. Moriyama, H. Ichiki: Retinal disease analyzer, U.S. Patent No. 5868134, U.S. Government, Washington, 1999.
• [9] F.S. Mikelberg, K. Wijsman, M. Schulzer: “Reproducibility of topographic parameters obtained with the heidelberg retina tomograph”, J. of Glaucoma, Vol. 2., (1993), pp. 101–103. [PubMed]
• [10] G. Zinser, U. Harbarth, H. Schröder: “Formation and analysis of three-dimensional data with the laser tomographic scanner (LST)”, In: J.E. Nasemann and R.O.W. Burk (Eds.): Scanning Laser Ophthalmoscopy and Tomography, Quintessenz Verlag, München, 1990, pp. 243–252.
• [11] K.U. Bartz-Schmidt, A. Sengersdorf et al.: “The cumulative normalised rim/disc area ratio curve”, Greafe's Achive for Clinical and Experimental Ophthalmology, Vol. 234, (1995), pp. 227–231. http://dx.doi.org/10.1007/BF00430414[Crossref]
• [12] F. Pluháček, J. Pospíšil: “Present methods of the computer-image analysis of a human retina with regard to diagnostics of glaucoma”, Fine Mechanics and Optics, Vol. 46, (2001), pp. 326–334.
• [13] L. Kubáček, L. Kubáčková: Statistics and Metrology, Palacký University Press, Olomouc, 2000.
• [14] J. Škrášek, Z. Tichý: Principles of Applied Mathematics III, SNTL, Praha, 1990.
• [15] B. Müller, J. Reinhardt, M.T. Strickland: Neural Networks: An Introduction, Springer, Berlin, 1995.
• [16] J.G. Taylor: Neural Networks and Their Applications, John Wiley and Sons, Chichester, 1996.
• [17] L. Machala, J. Pospíšil: “Proposal and verification of two methods for evaluation of the human iris video-camera images”, Optik, Vol. 112, (2001), pp. 335–340.

Identifiers

DOI

10.2478/BF02476270