High-dose Computed Tomography (CT) scans, while invaluable in medical diagnostics, pose significant risks due to increased patient radiation exposure. To mitigate these risks, Low-Dose CT (LDCT) imaging techniques have been developed, aiming to reduce radiation dose while maintaining diagnostic image quality. However, LDCT images often suffer from increased noise levels, which can compromise diagnostic accuracy. In this research paper, we investigate the efficacy of transfer learning by implementing them in autoencoder (CNN) architectures for LDCT denoising. Transfer learning utilizes information gained from training on complex datasets to increase performance on a target task with limited data. Inspired by the harmful effects of high-dose CT scans and the need for LDCT, we explore the potential of transfer learning with autoencoders in enhancing LDCT image quality. Our primary objective is to maintain a strong structural similarity between LDCT and denoised CT images. Among the four models utilized for image denoising, VGG 16 exhibited the highest performance, achieving an SSIM of 0.9967.