MRI Denoising with Residual Connections and Two-Way Scaling Using Unsupervised Swin Convolutional U-Net Transformer (USCUNT)

In medical image processing, denoising is a crucial preprocessing step that lowers noise. This paper employs the combination of U-Net and Swin transformer for MRI diagnosis. We combine a convolutional neural network with these two models. Three modules comprise the USCUNT model: 1. Convolution layers are used for extracting spatial features, and 2. Swin transformer is used for image reconstruction. 3. Using UNet for high-level deep feature extraction. In this research, denoising is performed on the artificial noisy MRI pictures. Up and down sampling are used in the UNet module to retrieve deep features. Modern denoising models are contrasted with the suggested architecture. The suggested approach performs better, with an SSIM of 0.823 and PSNR of 29.45.