Enhancing the Resolution of ChestX-ray Images with SRGAN and Sub-Pixel CNN

A crucial diagnostic tool for respiratory and heart disorders is chest X-ray imaging. However, the limited resolution of acquired images can hinder accurate interpretation and subsequent treatment decisions. To address this challenge, we employ deep learning-based super-resolution techniques, specifically SRGAN and ESPCN, to enhance the chest X-ray images. In this study, we proposed a modified ESPCN and present our findings on the application of SRGAN and ESPCN for enhancing chest Xray images. The enhanced image resolution enables radiologists to visualize fine details and subtle abnormalities that were previously difficult to discern. This enhancement significantly improves image interpretation and diagnostic accuracy. Moreover, our findings demonstrate the potential of super-resolution techniques for medical education and training. Through extensive experimentation, we saw significant visual improvements, as evidenced by PSNR(peak signal-to-noise ratio) of SRGAN and proposed ESPCN are 40.46 and 40.17, and achieved SSIM values 0.84 and 0.71 respectively. In conclusion, our study showcases the efficacy of SRGAN and ESPCN in enhancing chest X-ray imaging. By improving image quality and diagnostic accuracy, these techniques have the potential to revolutionize medical diagnostics, leading to better patient outcomes and a more effective healthcare system.