OCT Image Recognition Of Retinopathy Based On Optimized Neural Network

Optical coherence tomography(OCT)technology has the advantages of high resolution,fast detection and no invasion to biological tissues.It is applied to the detection and diagnosis of human retinopathy.The accurate identification of retinopathy is the premise for doctors to treat the disease effectively.In view of the long time-consuming and difficult process for doctors to diagnose the lesions according to OCT images,this thesis proposes to use the optimization algorithm to optimize the BP neural network.The optimized BP neural network is used to effectively classify OCT data,Compared with BP neural network,it further improves the recognition accuracy of retinopathy types.so as to further improve the recognition accuracy of retinopathy types.Firstly,the experiment carries out image clipping,centering,enhancement and image horizontal correction on the original data set,and obtains the main research areas of different types of images.Then different feature extraction algorithms are used to extract the features of the processed image.Because the extracted image features contain a certain amount of irrelevant information,then different feature dimensionality reduction methods are used to remove the irrelevant features to obtain effective data for identifying retinopathy.Finally,80% of the data is used for the training of k-nearest neighbor algorithm,limit learning machine algorithm and BP neural network algorithm,and 20% of the data is used for the test of the algorithm.In the experiment,the recognition accuracy of the three algorithms is 94.58%,93.83%and 94.71%.Considering the disadvantage that BP neural network is easy to fall into local optimization,in the experiment,genetic algorithm,particle swarm optimization algorithm and ant colony optimization algorithm are used to optimize BP neural network.By further optimizing the weight and threshold of BP neural network,it is trained and recognized respectively in combination with the characteristics of dimensionality reduction.Compared with BP neural network,the recognition accuracy of Optimized BP neural network under different features is improved to varying degrees.Genetic algorithm,particle swarm optimization algorithm and ant colony optimization algorithm optimized BP neural network in turn.The recognition accuracy of retinopathy was 98.32%,96.68% and 96.95%.Finally,the optimization time of the three optimization algorithms is compared.The results show that genetic algorithm has higher recognition accuracy and less time cost than particle swarm optimization algorithm and ant colony optimization algorithm.In this thesis,combined with OCT technology,the optimization algorithm is used to optimize the BP neural network for the identification of retinopathy,which can effectively identify the type of retinopathy.At the same time,the optimization time cost of the three optimization algorithms is compared,and a retinopathy recognition scheme with less time cost and high accuracy is obtained.Figure [40] Table [8] Reference [85]...