Based On The Deep Learning Glaucoma Morphological Characteristics Of Automatic Identification Method Study

Glaucoma is one of the world’s second largest blinding disease.Fundus glaucoma patients often have big DVD than,choroid atrophy of arc and nerve fiber layer defect.Screening for glaucoma is a very complex and difficult task.At present the diagnosis of glaucoma basically rely on artificial observation.But the human visual system without its shortcomings,such as subjectivity,limitations,ambiguity and lack of persistence.In order to be able to implement the intelligent of the test automation and informatization,pressing for a can simulate human’s visual function and beyond its performance of computer image recognition and diagnosis system to identify and diagnose glaucoma pathological changes.Fundus image processing and analysis of related research abroad early,mainly in the retinal blood vessels segmentation,image registration,joining together,locating and segmenting the DVD and auxiliary diagnosis and treatment,etc.By fundus camera film type imaging medium for the development of the current digital imaging,digital fundus images can be obtained directly,saves the complex structure of the photographic film and film scanning digitizing process.Deep learning as one of the core research field of artificial intelligence,have efficient data analysis ability,and with the attention of many researchers and hard work,deep learning to today has become a very mature technology.And in the field of data mining,deep learning but also because of its good ability of data analysis and is widely used,in order to realize the goal of knowledge discovery.In this topic research,deep learning technology,to be used for the corresponding model is established in this work,in order to obtain the condition judgment problem has a high precision prediction performance of glaucoma model,in order to realize the auxiliary clinical goals.Deep learning in the network is an important application is the convolution of neural network for image classification.Convolution neural network is one of the deep learning network 2 d network model,and other deep learning network is the biggest difference between can and two-dimensional convolution layer data directly convolution operation.According to the above problem,this research subject in terms of automatic identification of glaucoma lesions will be exploratory attempt to in-depth study.Usingmultiple data preprocessing technology for appropriate treatment,in order to get more suitable for the data set of this study;Prediction model is established using deep learning algorithm,and on a comparison of their performance in order to build a more efficient forecasting model of glaucoma,automatic diagnosis of glaucoma has great realistic significance.