Deep learning is a subset of machine learning based on multi-layered (a.k.a. “deep“) artificial neural networks. Their highly flexible architectures can learn directly from data (such as images, video or text) without the need of hand-coded rules and can increase their predictive accuracy when provided with more data.

Deep learning in Radiology

Deep learning has done remarkably well in image classification and processing tasks, mainly owing to convolutional neural networks (CNN), a subcategory of artificial neural network that makes the explicit assumption that the inputs are images. In the past few years, CNNs have been the basis for some of the most influential innovations in the field of computer vision ^2,3. Radiology, being a heavily image-based specialty, has naturally become a prominent field of application for deep learning methods with very good results.

History and etymology

A major breakthrough in the field of deep learning was in 1998, when Lecun and colleagues applied successfully their novel convolutional neural network, LeNet, to handwritten digit classification ⁴. However, deep learning methods  did not receive wide attention until 2012. That year, at the ImageNet challenge, Krizhevsky and Hinton⁵ developed a CNN named AlexNet that surpassed all the other competing classic machine learning techniques and won the competition. Today, deep learning and CNNs are considered to represent the state of the art in image analysis ^2,3.