Single photon emission local tomography (SPELT)

Local tomography uses truncated projection data to reconstruct a region of interest, and is important in medical imaging and industrial non-destructive evaluation using micro X-ray CT. The popular filtered backprojection (FBP) algorithm does not reconstruct a reliable image, which varies with the degree and location of truncation due to its global convolution kernel. A typical local tomography method uses a second derivative local operator to replace the global convolution kernel in the filtered backprojection algorithm (LFBP). By using a local filter, the reconstructed region depends only on the local projections. The singularities (edges) are preserved, but the exact image value cannot be recovered. This paper, using the data consistency conditions, developed a pre-processing technique that uses the FBP algorithm, which outperforms direct FBP and LFBP.