%0期刊文章%@ 2291-9694 %I JMIR Publicat卡塔尔世界杯8强波胆分析ions %V 10 %N 11 %P 40878 %T儿童肾脏超声异常的自动筛查:深度学习和迁移学习方法%蔡阿，卢明钦，洪盛%张阿，岳川%黄永杰%付阿恩%+台中退伍军人总医院儿科，台湾大道四区1650号。，台湾台中，886 4 23592525 ext 5909, linshienfu@yahoo.com.tw %K迁移学习%K卷积神经网络%K小儿肾脏超声图像%K筛查%K小儿%K医学图像%K临床信息学%K深度学习%K超声图像%K人工智能%K诊断系统%D 2022 %7 2.11.2022 %9原论文%J JMIR Med Inform %G英文%X近年来，便携式超声探头的进展和推广使超声(US)成为医生诊断时的有用工具。随着机器学习和深度学习的出现，开发一种筛查肾US异常的计算机辅助诊断系统，可以帮助全科医生早期发现儿童肾脏疾病。目的:在本文中，我们试图评估深度学习技术的诊断性能，将肾脏图像分类为正常和异常。方法:选取330张正常和1269张异常儿童肾脏US图像，建立人工智能模型。异常图像包括结石、囊肿、高回声、占位性病变和肾积水。我们对原始图像进行预处理，用于后续的深度学习。我们重新定义了最终的连接层，用于从ResNet-50预训练模型中将提取的特征分类为异常或正常。模型的性能通过使用接受者工作特征曲线下的面积、准确性、特异性和敏感性的验证数据集进行测试。 Results: The deep learning model, 94 MB parameters in size, based on ResNet-50, was built for classifying normal and abnormal images. The accuracy, (%)/area under curve, of the validated images of stone, cyst, hyperechogenicity, space-occupying lesions, and hydronephrosis were 93.2/0.973, 91.6/0.940, 89.9/0.940, 91.3/0.934, and 94.1/0.996, respectively. The accuracy of normal image classification in the validation data set was 90.1%. Overall accuracy of (%)/area under curve was 92.9/0.959.. Conclusions: We established a useful, computer-aided model for automatic classification of pediatric renal US images in terms of normal and abnormal categories. %M 36322109 %R 10.2196/40878 %U https://medinform.www.mybigtv.com/2022/11/e40878 %U https://doi.org/10.2196/40878 %U http://www.ncbi.nlm.nih.gov/pubmed/36322109