Realistically improvement of chest X-ray resolution using generative adversarial network

X-ray is one of the medical imaging methods, which helps physicians correctly diagnose diseases. Improper adjustment of X-ray tube parameters, different types of artifacts, and noise are factors affecting the quality of radiographic images. In some cases, poor quality of the images may lead to re-imaging, which increases the patient's dose. Today, artificial intelligence has made significant progress in various fields. Deep learning is one of the branches of artificial intelligence, which is widely used in medical imaging. In this article, the generative adversarial network is used as one of the most powerful available neural network models for resolution improvement, noise, and artifact reduction of chest X-rays. The values of RMSE, PSNR, and SSIM are calculated for 150 images with an average of 4.66, 34.92, and 0.923, respectively. These results show that trained networks have a high ability to improve the resolution of chest X-rays and make them more diagnostically valuable. Also, in cases where the image quality is low for any reason, there will be no need for re-imaging, and the patient will not receive the extra dose resulting from the re-imaging.