Deep learning (DL) is a subset of machine learning (ML) that utilizes artificial neural networks (ANN) which tries to mimic the human brain with the help of neural networks enabling it to uncover hidden patterns and make predictions. One of the algorithms of DL is convolutional neural network (CNN) which is made from multiple layers like convolutional layer, pooling layer, fully connected layer that helps in capturing various patterns in image recognition tasks. Through the integration of CNN architecture, the task of medical image processing is booming in today’s world. In this study, bone fracture (BF) prediction is made by detecting the region of interest (ROI) in X-ray images, the location where the probability of finding fracture is higher and then getting the desired part of image by cropping it. Then the X-ray images are passed through three pertained CNN models namely, ResNet50 (RN50), InceptionV3 (IV3), and DenseNet121(DN121) for BF prediction. This study shows the capability of CNN in fracture detection which is better than traditional methods as human interpretation may not be very accurate which can compromise the treatment. Hence, CNN holds a crucial role in the modern medical sector bridging the gap between technology and healthcare so as to enhance treatment quality resulting in improved lifestyle and avoiding human errors.