|
|
|
|
|
Abstract
Generative Adversarial Networks (GANs) have been developed for applications in the medical imaging field, such as reducing image noise, providing higher spatial resolution, and generating synthetic images start some years ago. In the case of synthetic image generation, GANs have been developed and modified to give a better synthetic image. It is necessary to conduct an in-depth study related to variation layers in the Generator and Discriminator of GANs architecture. The difference in the layer arrangement certainly impacts the generated synthetic image differently. This study is carried out the impact of changing the Generator and Discriminator 3 layers, 5 layers, and 7 layers to make synthetic X-ray image. A quantitative comparison of synthetic images and original X-ray images was also carried out to determine the performance of GANs objectively.
Keywords: X-ray image, GANs, Synthetic Image, Generator.