Automatic L1-SPIRiT Regularization Parameter Selection Using Monte-Carlo SURE

Keywords

acknowledgments adjustment affect alley allowing almost although amplification amplified apply appropriate approximate arbor auto automatic available axial block blue brain calibrating calibration channel choice choose closely coarse coil complex compute consented consistent constant contain convolution covariance criteria criterion decreases depends depicted depicts determines deviation disc domain ease eight enforced error estimator expect experiment extract fine form full generalized greater green ground impede implementation improvement included incorporates intervention iterations iterative joint kernel knowledge like locations match matches matching matrix measured middle minimize minimizes noise nonlinear norm optimal original outperforms output paper parallel portray practical practice promise proposes quality random real realization reconstruct reconstructed reconstruction reduce reduced regularization regularized resolution retaining risk sample selecting selection shape significantly simulated since slice soft space sparsity spatial spirit squared stat stein structure subject substantial suggest sure sweeps term thresholding traditionally trans transform true truth tune unbiased unit unknown useful user validated variance varying vector wavelet windowed