Abstract: The excellent level of healthcare services is contributed by deep learning. Due to the exponential growth of digital medical data throughout time early diagnosis and disease prediction become more efficient because of in-depth learning techniques which greatly minimise fatality. This paper concentrates mostly on providing an overview of profound knowledge in the area of processing and analysis of medical images. We showed the application of novel deep learning oncological architectures to forecast various types of cancers such as the brain, lung, skin, etc. The state-of-the-art architectures do analyses of 2D and 3D medical images to improve and improve patient diagnosis. The application of prominent methodologies in the field of machine learning, like set and transport learning, improved the performance of profound neural networks in the analysis and analysis of medical images. The current deeper networks are urging the new Capsule Network (CapsNet) image classification network to improve categorization and detection. It becomes increasingly important because the CapsNet equivalence features disincentivate the influence of any structural invariance on the network of an input image.