Motivation: Multi-contrast and quantitative MRI provide valuable diagnostic information for neurological diseases like multiple sclerosis (MS). However, long scan times make it less practical than CT in time- and motion-sensitive settings. Goal(s): To develop a deep learning framework to synthesize high-resolution multi-contrast MR images and T1&T2 maps of the brain from a rapid acquisition protocol. Approach: We used a U-Net-based supervised learning model, taking rapidly acquired low-resolution images and one high resolution image as input and using high-resolution images as the reference. Results: The framework achieves high-resolution images and quantitative maps with high accuracy, verified by statistical analysis on volunteer and MS patient data. Impact: The proposed approach enables high-resolution multi-contrast MRI and quantitative mapping of the entire brain in 3 minutes. It can improve and facilitate diagnosis and monitoring of neurological diseases like MS by making detailed brain imaging feasible in time-sensitive clinical settings.
Alyuz et al. (Tue,) studied this question.