Phase Integration in Differential Phase Contrast (DPC) Computed Tomography (CT) via deconvolution

Differential Phase Contrast (DPC) CT has emerged as a focal point of recent research. This technique necessitates an integration step to retrieve phase information from phase gradient data. While the Hilbert filter is a possible parameter-free solution for phase integration within the tomographic reconstruction step, this work models the phase integration as a deconvolution problem and three alternative deconvolution methods are explored: the Wiener filter, sparse deconvolution and Total Variation (TV) deconvolution. Using simulated data, a comparative quantitative analysis was performed, demonstrating how variations in the regularization parameters of the filters impact image quality. The results indicate that iterative algorithms, particularly sparse deconvolution, yield superior outcomes compared to the Hilbert transform.