Accelerated High-Resolution T1- and T2-weighted Breast MRI with Deep Learning Super-Resolution Reconstruction

Motivation: High spatial and temporal resolutions are required for breast MRI. Goal(s): This study aimed to assess the performance of a deep-learning(DL) algorithm to reconstruct low-resolution Cartesian T1-weighted DCE(T1w) and T2-weighted TSE(T2w) sequences. Approach: In this prospective study, patients underwent 1.5T breast MRI. The study protocol included T1w and T2w, acquired in standard resolution (T1S,T2S) and in low-resolution with following DL reconstructions (T1DL,T2DL). For DL reconstruction, two CNNs were used: (1)Adaptive-CS-Net and (2)Precise-Image-Net. Image quality was analysed qualitatively and quantitatively. BI-RADS agreement between sequences was assessed. Results: Deep-learning for denoising and resolution upscaling reduces acquisition time and improves image quality for breast MRI. Impact: Deep learning reconstruction algorithm for denoising with compressed sensing and resolution upscaling reduces acquisition time and improves image quality for dynamic contrast-enhanced T1w and T2w breast MRI.