首页> 外国专利> COMPUTER-IMPLEMENTED METHOD, SYSTEM AND COMPUTER PROGRAM PRODUCT FOR DETECTING A RETINAL CONDITION FROM EYE FUNDUS IMAGES

COMPUTER-IMPLEMENTED METHOD, SYSTEM AND COMPUTER PROGRAM PRODUCT FOR DETECTING A RETINAL CONDITION FROM EYE FUNDUS IMAGES

机译:用于检测来自眼底图像的视网膜条件的计算机实现的方法,系统和计算机程序产品

摘要

Computer-implemented method, system and computer program product for detecting a retinal condition from eye fundus images. The method (100) comprises:receiving (110), by an image classifier (506), one or more eye fundus images Ii (102) acquired from a same eye;applying (120) a same convolutional neuronal network (302) on each eye fundus image Ii (102) to obtain a corresponding prediction yi (122);applying (130) a pooling function (304) on the predictions yi (122) to obtain a classification Y (132) of the one or more eye fundus images Ii (102) defined as: , wherein ε is a constant and N is the number of eye fundus images Ii received; anddetecting (140) a retinal condition (142) based on the classification Y (132). The image classifier is advantageously trained using only eye examination labels assigned by ophthalmologists, without requiring low-level image labels.
机译:用于从眼底图像检测视网膜条件的计算机实现的方法,系统和计算机程序产品。 方法(100)包括:通过图像分类器(506),从同一眼睛获取的图像分类器(506),通过图像分类器(506),每个眼底图像II(102);每个相同的卷积神经元网络(302)施加相同的卷积神经网络(302) 眼底图像II(102)获得相应的预测yi(122);在预测yi(122)上施加(130)池函数(304)以获得一个或多个眼底图像的分类Y(132) II(102)定义为:<![cdata [y =ΣInyiyiσInyi+ε]>,其中ε是常数,n是收到的眼底图像的数量; 基于分类Y(132)的视网膜条件(142)。 图像分类器有利地使用眼科医生分配的眼睛检查标签,而不需要低级图像标签。

著录项

相似文献

  • 专利
  • 外文文献
  • 中文文献
获取专利

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号