The application of artificial intelligence to pattern recognition in medicine has been described in recent studies in machine learning.1 In particular, the use of machine learning in the form of the convolutional neural network has underpinned the current popularity of so-called deep learning algorithms, which are essentially convolutional neural network models distinguished by having a large number of hidden layers with many adjustable parameters requiring a large quantity of data for training. A deep learning algorithm can be applied to raw data without the need for data normalization and can be used to extract the inherent nonobvious features from training images that are necessary for accurate discrimination.1,2