Medical CT Image Super-resolution Algorithm based on GAN

Computed Tomography (CT) images are essential for visualizing the internal structures of the human body and are widely applied in the diagnosis of various diseases. However, during the processes of generating, storing, and transmitting medical CT images, the clarity of these images can be degraded by non-human factors, resulting in issues such as reduced resolution and blurred texture details, thereby compromising image fidelity. In this study, we present an approach utilizing Generative Adversarial Networks (GANs). that incorporates a grayscale value loss as an additional constraint while maintaining the original loss function constraints. We conduct experiments for 4x super-resolution reconstruction. To evaluate the reconstruction performance, we compare the Peak Signal-to-Noise Ratio (PSNR) and the Structural Similarity Index (SSIM) of the reconstructed images. The experimental results demonstrate ,which the proposed method exhibits promising performance in both quantitative metrics and visual quality.