Enhancing the resolution of Brain MRI images using Generative Adversarial Networks (GANs)
B. Ankitha, Ch. Srikanth, D. Venkatesh, D. Badrinath, G. Aditya, S. Akila Agnes · 2023
Generative Adversarial Networks (GANs) are an emerging technology in Artificial Intelligence (AI). They are an unsupervised learning framework that involves automatically learning patterns from input data to generate synthetic output that closely resembles the original input. GANs enable the creation of datasets by replicating copies of fake outputs based on given inputs. Using GANs, machines can generate unique content that is distinct from human-created data. GANs are widely used in various image-related areas, such as image-text conversion, photo-to-emoji conversion, face aging, new human face generation, super-resolution image conversion, and more. This paper focuses on enhancing the resolution of magnetic resonance images (MRI), which is a common method used in medical imaging. However, obtaining high-quality MRI scans often requires long waiting times and the use of lab equipment with restrictions. Additionally, there are health hazards associated with MRI radiation, resulting in lower-quality images. Super-resolution GANs (SRGANs) can potentially address this issue. Super-resolution is a method that can generate high-resolution images from lower-resolution ones. Previously, technologies such as bi-cubic interpolation, linear interpolation, and nearest neighbor interpolation were used to achieve higher resolution. However, the results obtained using the proposed SRGAN model outperform these conventional methods in terms of performance