A Study Of Machine Learning Classification Methods For Ophthalmic Diseases

With the continuous improvement of living standards,scientific and technological progress has brought a lot of convenience to people,but it also has a negative impact on human health,especially in the eyes,people’s excessive use of eyes every day leads to a variety of eye diseases,the whole society presents a trend of younger patients with eye diseases.Nowadays,with the increasing number of patients with eye diseases,the number of ophthalmologist is obviously not able to meet the needs of people to see a doctor in time,and the efficiency of artificial examination of the disease is not high,it is likely to miss the critical period of the best treatment of the disease.With the continuous development of artificial intelligence technology,with the advantage of high computing power of computer,artificial intelligence has changed the medical status quo to a large extent.As the main technology of artificial intelligence,machine learning has become increasingly mature in image recognition after continuous iterative development,and medical image processing based on machine learning has become a hot emerging research direction.This paper mainly studies the machine learning method of image classification of eye diseases,and continuously improves it to improve the effect of image classification of eye diseases.The main research contents are as follows:Machine learning requires a lot of image data for training.In order to build the data set required by the experiment,the ophthalmic image data of the Fourth Affiliated Hospital of China Medical University is adopted as the research object in this project.I follow the doctors to learn the knowledge related to eye diseases,and screen and preprocess the ophthalmic image data according to my own needs to make a small sample ophthalmic disease data set.Prepare for subsequent model training.In this paper,the deep learning network was used for feature extraction,the support vector machine was used for classification,and the data set of eye diseases was put into the model for training.Considering the features of medical images and the feature space extracted from the data,a machine learning classification framework integrating multiple feature selection and optimization methods was proposed.Firstly,in view of the different features extracted from different deep learning networks,feature fusion is used to combine the features of different networks to add more features with judgment ability.Then,in the complex case of multi-classification,it is not conducive to the support vector machine algorithm to find the optimal classification hyperplane for classification,and the principal component analysis is used to reduce the dimension of high-dimensional feature space.Finally,some features extracted from the deep learning network are used as noise,which will affect the classification effect.In order to eliminate these redundant and misleading features,the hybrid gray Wolf predator optimization algorithm is used to leave the best feature subset from the feature space,and the optimization algorithm is embedded in the support vector machine for classification.By comparing the original model,the improved model and the current mainstream image classification algorithms in many aspects,the classification model adopted in this topic can achieve better results.