Full text loading...
PET scanning geometry showing the annihilation and detection processes.
neighborhood of a central pixel showing the directional derivatives along .
[(a) and (e)] Original phantom. Reconstructed map after 50 iterations of [(b) and (f)] ML algorithm, [(c) and (g)] MRP algorithm, and [(d) and (h)] proposed algorithm.
(a) Gaussian noise corrupted image. Image reconstruction from noisy projection data using (b) ML algorithm, (c) MRP algorithm, and (d) proposed algorithm. The corresponding line plots are shown below the reconstructed images.
(a) Original phantom and (b) Poisson noise corrupted image. (c), (d), and (e) are, respectively, the reconstructed images using ML, MRP, and proposed algorithms.
NMSE vs iteration plot for ML, MAP, and proposed algorithms.
(a)–(c) and (d)–(f) are the reconstructed images using ML, MRP, and proposed approaches for 50 000 and 100 000 emission counts.
Article metrics loading...