Han SS, Moon IJ, Lim W, et al. - In order to analyze whether an algorithm can automatically locate suspected areas and predict the probability of a lesion being malignant, researchers examined a sum of 924,538 training image-crops involving various benign lesions using a region-based convolutional neural network. They applied the area under the receiver operating characteristic curve, F1 score, and Youden index score to correlate the performance of the algorithm with that of the participants. Convolutional neural networks were trained with 1106,886 image crops to locate and diagnose cancer, after manually or automatically annotating these possible lesions based on image findings. They obtained validation data sets from 3 hospitals between January 1, 2010, and September 30, 2018. The outcomes indicated that the algorithm could localize and diagnose skin cancer without preselection of suspicious lesions by dermatologists.