Algorithm for propagation-based phase-contrast CT reconstruction

Pedro Roberto

Master Semester Project

Section of Life Sciences and Technologies , EPFL

July 2016

Abstract

Differential phase contrast imaging (DPCI) is a recent imaging technique analogous to the conventional x-ray or CT but whose signograms are the first or second derivative of the conventional x-ray sinogram. The term propagation-based phase-contrast CT (PBI) refers to the technique that outputs the second derivative. This technique might be able to produce CTs that require less radiation. As of 2016, this technique has to be done in specialized laboratories like the Paul Scherrer Institute (Villingen, Switzerland). This work's goal is to experimentally determine which is the following algorithms provides better reconstructions for PBI sinograms: inverting the PBI sinograms and then applying a standard iterative reconstruction algorithm or modifying the mentioned algorithm to reconstruct PBI sinograms. The algorithm in question was recently developed by Masih Nilchian and Michael McCann of the Biomedical Imaging Group and is implemented in Matlab. On a smoothed analytical ellipsoid phantom developed by me, I show that the modified algorithm provides better results than inverting the PBI sinograms. This contrasts with the non-PCI modality in which both approaches provide satisfactory results. In the figure below, differences in SNR smaller than 5 are not significant.