Modified FBP and BPF algorithms for stationary computed tomography with linearly distributed sources

BackgroundStationary computed tomography with linearly distributed sources (SLCT) improves temporal resolution by avoiding gantry rotation. For the dual-SLCT (DSLCT) configuration, it faces challenges associated with redundancy and truncation issues in projection data. Existing FBP-based methods require complicated weighting strategies, whereas BPF-based methods for truncated data require geometry-specific formulations, which may limit their direct applicability to DSLCT.ObjectiveThis study presents two approaches to overcome the above challenges: a weighted Katsevich-type FBP (WK-FBP) algorithm and a locally refined backprojection filtration (LR-BPF) method.MethodsThe WK-FBP incorporates a Parker-like weighting strategy to handle data redundancy, applying weighting only to conjugate ray pairs during backprojection, thus simplifying the implementation process. The LR-BPF refines the classical BPF by locally adjusting the projection derivative, thereby accommodating linear scanning trajectories while effectively avoiding truncation artifacts. The two methods were evaluated under truncated and non-truncated conditions, with and without noise.ResultsBoth algorithms are applicable to DSLCT scanning. For non-truncated data, the WK-FBP achieves accurate image reconstruction. However, under truncation, WK-FBP reconstructions exhibit noticeable artifacts, whereas the LR-BPF consistently provides accurate results.ConclusionsWK-FBP is effective for DSLCT without truncation, whereas LR-BPF provides robust reconstruction performance under truncation.