Zheng C, Xie X, Huang L, et al. - Based on spectral domain optical coherence tomography (SD-OCT) peripapillary retinal nerve fiber layer (pRNFL) images, researchers created a deep learning (DL) model for automated detection of glaucoma and contrasted diagnostic capability against hand-craft features (HCFs). A DL model based on a pretrained convolutional neural network (CNN) was trained using a retrospective training set of 1,501 pRNFL OCT images containing 690 images of 153 patients with glaucoma and 811 images of 394 healthy individuals. In addition, it was tested in an independent test set of 50 images from 50 glaucoma patients and 52 images from 52 normal individuals. Findings revealed that the DL model achieved an AROC of 0.99 which was significantly larger than the AROC values of all other HCFs. Overall, the authors concluded that DL models based on pretrained CNN are able to identify glaucoma based on SD-OCT pRNFL images with high sensitivity and specificity.