MRI synthesis using Deep Learning models

Abstract

In this report, I introduce the project “MRI Synthesis Using Deep Learning Models” - a GANs-based generative network synthesizing multiple MRI images from a given single one. In fact, MRI data acquisition can be time-consuming, prone to motion artifacts, and limited by hardware constraints. The proposed method leverages the power of generative adversarial networks (GANs) to generate multi-domain MRI images from a single input MRI image. Experimental results on IXI and BraTS2020 datasets demonstrate the effectiveness of the proposed method compared to an existing method in metrics SSIM, PSNR and NMAE. The synthesized images can serve as valuable resources for medical professionals in research, education, and clinical applications.