CT Image Denoising using Autoencoder and Generative Adversarial Networks
C. Aneesh, Gembali Saumik, K Varun, K Afnaan, Tripty Singh, Adhirath Mandal · 2024
Denoising images plays an important role in analyzing medical images. CT images usually contain a lot of noise which would affect the diagnosis. This project focuses on denoising CT images for better understanding and enhancing the accuracy of medical diagnosis. Proposed an approach for CT image denoising utilizing Convolutional autoencoders and Generative adversarial networks. The architecture integrates an encoder-decoder for Autoencoders and Generative adversarial networks including a Generator for generating denoised images and a discriminator for identifying real and fake images generated by the networks. Experimental evaluations conducted on synthetic and real CT datasets showcase the model's efficiency. When tested, achieved a PSNR value of 74.2 for Autoencoders and a PSNR value of 54.03 for Generative adversarial networks. Simulation results demonstrate the performance of this proposed method for denoising CT images.