Convolutional neural networks: an overview and application in radiology

This review article introduces convolutional neural networks, a class of artificial neural networks that has become dominant in computer vision and is increasingly of interest in radiology. It explains how a CNN automatically and adaptively learns spatial hierarchies of features through backpropagation, using building blocks such as convolution layers, pooling layers, and fully connected layers. The goal is to give radiology practitioners a working understanding of the concepts, advantages, and limitations of CNNs.

Beyond the fundamentals, the article surveys applications of CNNs to various radiological tasks and discusses the field's challenges and future directions, with particular attention to two practical obstacles—small datasets and overfitting—and techniques for minimizing them. The authors emphasize that familiarity with CNNs is essential to leverage their potential in diagnostic radiology, with the ultimate aim of augmenting radiologist performance and improving patient care.