Benchmarking GANs, Diffusion Models, and Flow Matching for T1w-to-T2w MRI Translation

Daniele Ravi

Abstract: Medical image synthesis for cross-modality translation has become an important task in medical imaging. In this work, we benchmark different generative models for T1-weighted to T2-weighted MRI synthesis. We compare several state-of-the-art approaches including generative adversarial networks (GANs), diffusion models, and recent flow matching techniques. We evaluate these methods on standard MRI datasets using quantitative metrics including SSIM, PSNR, and FID scores, as well as qualitative assessments from radiologists. Our comprehensive evaluation provides insights into the strengths and weaknesses of different generative modeling approaches for medical image synthesis, which can guide future model selection in clinical applications.