Fig. 3.
Matched slices showing cone-beam reconstructions from the full (2θmax = 24.95°) projection dataset (bottom row) and the truncated (θmax + Δθ) projection datasets (top 3 rows) emulating the laterally-shifted detector geometry. For the top 3 rows, Δθ of 0.25°, 1.49° and 2.72° correspond to half-overlap region [0,u0] of 10, 60 and 110 pixels of 0.388 mm pitch in the detector plane, respectively. For the top 3 rows, each column represents the three weighting schemes investigated. For Δθ = 0.25°, artifacts are observed centered at the AOR (arrow in top left panel). For Δθ ≥ 1.49°, these artifacts are no longer apparent. For a given Δθ, visually the reconstructions using the three investigated weighting schemes appear similar. Image display range for the reconstructed linear attenuation coefficients is maintained the same, μ ∈ [0.15,0.35] cm−1 for all panels.