Fig. 1

Flowchart of POCS reconstruction algorithm. Starting with the original sparsely sampled k-space data (K ₀), the central zone extended by zero filling (K ₀′) and reconstructed to make a low-resolution image (I ₀′) from which the low-frequency phase map is extracted (φ ₀′). The K0 data is also used to produce an initial estimate of the full k-space (K ₁), which is used to generate an estimate of the reconstructed image (I₁). The magnitude of the reconstructed image estimate is then multiplied by exp(iφ ₀′) to produce a phase-corrected estimate (I ₁′) from which a new estimated k-space (K ₁′) is obtained. This k-space estimate is used to form a new k-space estimate (K ₂) by substituting missing values for all of the missing data in the original sparsely sampled data. The process repeats until the convergence criteria are met