摘要
胆道闭锁是婴儿期最严重的肝脏疾病之一,目前的早期筛查方法主要是大便比色卡。由于大便比色卡这种方法存在的种种问题,为了更准确地筛查早期婴儿胆道闭锁,本文提出了基于卷积神经网络的早期婴儿胆道闭锁的筛查方法。首先,通过相机获得早期婴儿的大便图像,然后对获得的图像进行一定的预处理,最后通过不断的训练卷积神经网络的模型和调参数,对获得的图像进行分类识别,并获得高的准确率。实验结果表明本文提出的方法较传统识别的方法有效地提高了识别的准确率。
Biliary atresia is one of the most serious liver diseases in infancy,and its early screening method is mainly stool colorimetric card.In order to screen early infant biliary atresia more accurately,a convolutional neural network based screening method for early infant biliary atresia is proposed.Firstly,get the early infant’s stool image by the camera,then do some preprocessing on the acquired image,finally,classify and recognize the acquired image by training convolution neural network model and adjusting the parameters,and obtain high accuracy.Experimental results show that the proposed method is more effective than the traditional recognition method to improve the recognition accuracy.
作者
万志勇
周小安
WAN Zhiyong;ZHOU Xiao’an(College of Information Engineering,Shenzhen University,Shenzhen Guangdong 518000,China)
出处
《智能计算机与应用》
2019年第1期77-79,83,共4页
Intelligent Computer and Applications
关键词
卷积神经网络
胆道闭锁
深度学习
convolutional neural network
biliary atresia
deep learning
