A Dictionary Learning Method with Total Generalized Variation for MRI Reconstruction

Figure 3

(a) Reference image. (b) Sampling mask in -space with 6.09-fold undersampling. (c) PSNR versus noise level for DLMRI, SHTGV, and DLTGV. ((d), (e), (f)) Reconstructed images using DLMRI, SHTGV, and DLTGV. ((g), (h), (i)) Reconstruction error magnitudes for DLMRI, SHTGV, and DLTGV with noise .

(a)

(b)

(c)

(d)

(e)

(f)

(g)

(h)

(i)