MA-RECON: Mask-aware deep-neural-network for robust fast MRI k-space interpolation

Nitzan Avidan, Moti Freiman

Abstract: Highlights•Introduction of a mask-aware DNN methodology for k-space interpolation from under-sampled ‘k-space’ data.•Enhancement of generalization capacity and resilience to variations in acquisition process and anatomical distribution.•Providing empirical evidence for enhanced robustness in clinical regions, utilizing the publicly available fastMRI dataset.