MRI Cross-Modal Synthesis: A Comparative Study of Generative Models for T1-to-T2 Reconstruction

2026-02-05

AI Summary

This study addresses the challenge of efficiently and accurately synthesizing T2-weighted MRI images from T1-weighted scans to reduce clinical acquisition time while preserving diagnostic information. Within a unified experimental framework, it presents the first systematic comparison of Pix2Pix GAN, CycleGAN, and variational autoencoders (VAEs) for cross-modal synthesis on the BraTS 2020 dataset. Quantitative evaluation demonstrates that CycleGAN achieves the highest performance in terms of PSNR (32.28 dB) and SSIM (0.9008), while Pix2Pix GAN yields the lowest mean squared error (MSE = 0.005846). Although the VAE exhibits slightly lower quantitative metrics, it demonstrates strong capabilities in latent space modeling. This work establishes a comprehensive benchmark and offers practical guidance for selecting models in MRI cross-modal synthesis tasks.

Technology Category

• Computer Vision: Generative Adversarial Networks (GANs) for Vision
• Machine Learning: Deep Generative Models & Autoencoders
• Cognitive Modeling & Cognitive Systems: Computational Creativity

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

MRI cross-modal synthesis involves generating images from one acquisition protocol using another, offering considerable clinical value by reducing scan time while maintaining diagnostic information. This paper presents a comprehensive comparison of three state-of-the-art generative models for T1-to-T2 MRI reconstruction: Pix2Pix GAN, CycleGAN, and Variational Autoencoder (VAE). Using the BraTS 2020 dataset (11,439 training and 2,000 testing slices), we evaluate these models based on established metrics including Mean Squared Error (MSE), Peak Signal-to-Noise Ratio (PSNR), and Structural Similarity Index (SSIM). Our experiments demonstrate that all models can successfully synthesize T2 images from T1 inputs, with CycleGAN achieving the highest PSNR (32.28 dB) and SSIM (0.9008), while Pix2Pix GAN provides the lowest MSE (0.005846). The VAE, though showing lower quantitative performance (MSE: 0.006949, PSNR: 24.95 dB, SSIM: 0.6573), offers advantages in latent space representation and sampling capabilities. This comparative study provides valuable insights for researchers and clinicians selecting appropriate generative models for MRI synthesis applications based on their specific requirements and data constraints.

Problem

• MRI cross-modal synthesis
• T1-to-T2 reconstruction
• generative models
• medical image synthesis

Innovation

• MRI cross-modal synthesis
• generative models
• T1-to-T2 reconstruction
• CycleGAN
• variational autoencoder