首页>
外国专利>
CLASSIFICATION ALGORITHM FOR RETINAL OCT IMAGE BASED ON THREE-DIMENSIONAL CONVOLUTIONAL NEURAL NETWORK
CLASSIFICATION ALGORITHM FOR RETINAL OCT IMAGE BASED ON THREE-DIMENSIONAL CONVOLUTIONAL NEURAL NETWORK
展开▼
机译:基于三维卷积神经网络的视网膜OCT图像分类算法
展开▼
页面导航
摘要
著录项
相似文献
摘要
Disclosed in the present invention is a classification algorithm for a retinal OCT image based on a three-dimensional convolutional neural network, comprising the following steps: S01: collecting three types of retinal OCT images, and classifying and marking the three types of retinas; S02: preprocessing data, down-sampling the three-dimensional OCT image data to obtain three-dimensional images with a uniform size which are to be inputted into a three-dimensional convolutional neural network; S03: according to a migration learning idea, pre-training a three-dimensional convolutional neural network model by using a large number of marked natural images; S04: slightly adjusting the well-trained three-dimensional convolutional neural network model by using the preprocessed retinal OCT images, adding a branch network after a convolutional layer in the middle of a main stream network, and fusing output layers of the main stream network and the branch network; S05: preprocessing test images according to step S02, performing testing using the three-dimensional convolutional neural network model which has been slightly adjusted in S04, and outputting a classification result. The present invention can classify three-dimensional retinal OCT images and improve classification accuracy.
展开▼