Variable generalization performance of a deep learning model to detect pneumonia in chest radiographs: A cross-sectional study

There is interest in using convolutional neural networks (CNNs) to analyze medical imaging to provide computer-aided diagnosis (CAD). Recent work has suggested that image classification CNNs may not generalize to new data as well as previously believed. We assessed how well CNNs generalized across t...

Ausführliche Beschreibung