PUV

Article DOIhttps://doi.org/10.1016/j.urology.2020.05.019
ObjectiveTo diagnose children with PUV
AI ApproachCNN with transfer learning
Data Source(s)Institutional series (157 children: 3504
sagittal ultrasounds, 2558 transverse ultrasounds)
Model InputSagittal and/or transverse features of renal ultrasounds
Model OutcomePUV
Model MetricsAUROC = 0.96, accuracy = 93%
Model Usabilityhttps://github.com/YS181818/CAKUT_diagnosis/tree/master
AI = Artificial intelligence, AUROC = Area under the receiver operator characteristic, CNN = Convolutional neural network, PUV = Posterior urethral valves

Leave a comment