Offset Masking Improves Deep Learning based Accelerated MRI Reconstructions

Deep learning approaches to accelerated MRI take a matrix of sampled Fourier-space lines as input and produce a spatial image as output. In this work we show that by careful choice of the offset used in the masking procedure, the symmetries in k-space can be better exploited, producing higher quality reconstructions than given by standard equally-spaced masks or randomized masks motivated by compressed sensing.