Image analysis and statistical evaluation of two-dimensional human eye retina images of healthy and glaucomatous eyes

摘要

In this paper, a new automatic glaucoma diagnostic method for determining the probability of glaucoma occurring in a eye under study is described. This method is based on computer analysis of two-dimensional images of the blind spot of the human eye retina and on successive statistical evaluation of the data obtained. First, the characteristic symptoms of glaucoma are shortly described. Next, a suitable numerical parameter of the retina blind spot is defined. The method of computer image analysis for automatic determination of the parameter in question is described and it is applied to a set of normal healthy eye images and to a set of glaucomatous eye images. The probability of glaucoma occurring for each value of the parameter introduced is suitably defined and computed by virtue of statistical evaluation of the results obtained.