Convolutional neural networks (CNNs), form of artificial neural network (ANN) prominent in computer vision, are finding traction in diversity of sectors, comprising radiology. CNN employs a variety of building pieces, including as convolution, pooling layers, & fully linked layers, for acquiring spatial data hierarchy autonomously & adaptively via backpropagation. This review paper investigates core concepts of CNN & how se are used to numerous radiological jobs, as well as issues & future prospects in radiology. In addition, this work will explore two issues that arise when using CNN to radiological tasks: restricted datasets & overfitting, as well as approaches for mitigating m. Conceptual underst&ing, advantages, & limitations of CNN is crucial for realising its full potential in diagnostic radiology & improving radiologists' performance & patient care.