Contribution Analysis of Scope of SRGAN in the Medical Field

Abstract

This paper focuses on the concept of generative adversarial networks (GANs) and their scope in the medical field. “Generative adversarial networks” has been one of the most prominent research areas in the domain of machine learning in the past few years. It consists of two neural networks competing against each other. One of them is a generator model which generates fake samples of data and the other is a discriminator model which receives both real data (from the training data) and fake data (from the generator model) and tries to identify them as real or fake. Use of generative adversarial networks has been done here for the purpose of “super-resolution” because GANs work on the concept of generative modeling. Since applying super-resolution to an image means adding more data to the image which was not previously there (2017), it would require generation of data which might not be actually real data but is so close to the real one that one cannot know the difference. Hence, when we apply super-resolution using generative adversarial networks it gives us way better results in comparison to many other approaches such as “SRCNN”.