Noise Characteristics in POCS (Projection Onto Convex Sets)-Reconstructed MR Images
Peng H, Sabati M, Frayne R
Foothills Medical Centre, Calgary Health Region, University of Calgary
Undersampling of the k-space data has been widely used as an effective approach for rapid 2D and 3D MR imaging. Typically, the central zone of k-space is fully sampled but the peripheral zone is partially sampled. The partial Fourier method, for example, reduces the amount of acquired data by asymmetrically truncating the peripheral portions of k-space in either the phase or the frequency encoding directions, or both. Projection-onto-convex-sets (POCS), techniques have been successful in reconstructing such undersampled k-space data. In evaluating the noise properties of the POCS-reconstructed images and comparing them to those obtained in other reconstruction techniques (via SNR and CNR calculations), care is warranted because the iterative non-linear POCS reconstruction method can modify the underlying noise properties of the image. This preliminary study investigates and characterizes the noise in the POCS reconstruction approach.