Fig. 4

Image roughness as a function of the number of net trues. Data reconstructed from the mCT (set as the reference) using 3 iterations, 21 subsets, and no post-filtering (matrix size 200 × 200). Vision data were reconstructed using 5 subsets, no post-filtering, and between 3 to 5 iterations. a 440 × 440 and b 220 × 220 matrix size. Data were fitted using a bi-exponential to ease the visual interpretation