ABSTRACT
The authors test whether for evaluation of the severity of diabetic retinopathy it is possible to use the method of computer assisted mathematical processing of the retinal image. This method is based on the identification of vascular endings in the visualized part of the retina. It is assumed that with the severity of the finding the number of vascular endings increases in conjunction with the process of neogenesis. For assessment of the number of vascular endings the authors use the Adaptive Contrast Control (ACC) method. By a special procedure the vascular endings are identified and their number is assessed in the investigated area of the retina. The method makes it possible to assess in addition to the number of vascular endings also the total length of the vessels, their volume, surface, area they cover and the histogram of the diameter of the vessels. The authors examined 19 patients (38 eyes) with a quite normal finding on the retina. The mean age of the group was 37.3 years. Moreover they examined 10 patients (20 eyes) with middle advanced and advanced non-proliferative diabetic retinopathy and 10 patients (20 eyes) with risk proliferative diabetic retinopathy. In every patient a standard digital photograph of the fundus of both eyes size "tif" was made and each picture was subjected to mathematical analysis. The group of patients was divided into three sub-groups (patients with a normal finding, a non-proliferative finding and a finding of proliferative diabetic retinopathy). In the conclusion the authors provide evidence that with increasing severity of the finding on the retina also the number of vascular endings increases.
Subject(s)
Data Interpretation, Statistical , Diabetic Retinopathy/pathology , Fluorescein Angiography , Retina/pathology , Adolescent , Adult , Aged , Humans , Image Processing, Computer-Assisted , Middle AgedABSTRACT
A new shearing-interferometry method is used to characterize the basic statistical properties of very rough surfaces of transparent solids. The mean level of these rough surfaces can be curved or flat. Using this method, one evaluates the root-mean-square values (standard deviations) of the heights and slopes of the surface irregularities. Moreover the values of autocorrelation lengths, autocorrelation functions, and one-dimensional height-distribution functions are also determined. The method is applied to a statistical analysis of rough flat or curved glass surfaces. The results from this method agree very well with those obtained with a stylus-type surface instrument and speckle contrast method.
