Undersampled MR Image Reconstruction with Data-Driven Tight Frame

Figure 2

(a) Original image. (b) Cartesian sampling mask with accelerating factor . (c), (d), and (e) are the PSNR, HFEN, and SSIM values versus the number of iterations. (f), (g), and (h) are reconstruction error magnitudes for CSMRI-TV, DLMRI, and DDTF-MRI, respectively.

(a)

(b)

(c)

(d)

(e)

(f)

(g)

(h)