Medical Image Denoising with Recurrent Residual U-Net (R2U-Net) base Auto-Encoder

Deep learning (DL) approaches have been applied in different sectors of medical imaging applications, i.e. classification, segmentation and detection tasks and shown superior performance. The DL based generative methods are used for image denoising, enhancement and restoration task. In case of image analysis, image denoising is one of the most crucial preprocessing steps. Recently, there are various DL approaches are applied in image denoising problems and achieved state-of-the-art performance. In this work, we apply recurrent residual U-Net (R2U-Net) based autoencoder model for medical image denoising which is applied for digital pathology, dermoscopy, Magnetic Resonance Imaging (MRI) and Computed Tomography (CT) images denoising tasks. The performance of R2U-Net based autoencoder model is also evaluated for Transfer domain (TD) between MRI and CT scan images. The experiments have conducted on different publicly available medical image datasets and shows promising denoising results which can be applied in different medical imaging applications.