The Application of Neural Network Technology in the Fundus Angiography Image Segmentation

Abstract

The digital images as well as the data obtained by the fund us fluorescence angiography (FFA) can reflect the structure of retinal vessels, the hemodynamic changes, the vascular pathological physical changes and the pathological changes of other related structures, which have been widely applied in the differential diagnosis of the retina, the choroid and the optic nerve disease. According to the characteristics of FFA images, the BP neural network algorithm and the genetic neural network algorithm have been respectively employed to segment and contrast the lesion areas in fund us angiography vascular images as well as the fund us angiography images. Then the clinicians can get the more accurate measurement data of lesion areas and observe the more subtle vascular changes, which can provide an important basis for the treatment of the heart, the brain vascular system and the diabetes.