Key points are not available for this paper at this time.
Medical imaging is essential for precise diagnosis and treatment planning in contemporary healthcare, yet getting high-resolution pictures can be difficult because of things like low-dose CT scans' reduced radiation exposure. In order to tackle this problem, a unique strategy called Super-Resolution Generative Adversarial Networks (SRGAN) is presented in this research. To improve image resolution, SRGAN uses a generator and discriminator architecture with residual and upscale blocks. By combining adversarial and content losses, the generator's loss function enables it to produce high-quality images while keeping key elements of their low-resolution counterparts. A thorough literature study demonstrates how effective GAN-based models are for the super-resolution of medical images. After 10 training epochs, experimental findings utilizing a Lung CT dataset show a significant improvement, with PSNR and SSIM scores reaching 32.95 and 0.996, respectively. According to these findings, SRGANs are a viable enhancement technique for low-dose CT medical pictures, with significant advantages for precise diagnosis and patient care in healthcare settings.
Madhav et al. (Thu,) studied this question.