深層学習を利用した圧縮センシング再構成の高速化［大会長賞記録］

抄録

　Compressive sensing (CS) is an effective approach for fast magnetic resonance imaging (MRI). To improve the reconstruction accuracy and computational speed, we propose a novel deep architecture using deep learning. Experiments on MR image reconstruction demonstrate that proposed method significantly accelerates the reconstruction time, with image quality comparable to that of traditional iterative reconstruction.