Feasibility study of spectral imaging for differential phase contrast cone beam CT: computer simulations

Abstract

In principle, differential phase contrast (DPC) imaging allows the use of a hospital grade x-ray tube that has a large focal spot size and a wide polychromatic spectrum. It should be noted that due to the integration of interference patterns over the entire spectrum, the fringe contrast in the final intensity image is lower than that from a monochromatic spectrum. Therefore better image quality should be potentially obtained if the energy-dependent interference patterns can be analyzed separately. The key idea of the proposed spectral DPC imaging approach is to acquire DPC images for each photon energy channel, which is named spectral DPC images. The final DPC image can be computed by summing up these spectral DPC images or just computed using certain 'color' representation algorithms to enhance desired features. This research is a feasibility study based on computer simulations to investigate how the spectral DPC method works for a DPC-based cone beam CT (DPC-CBCT) system. The spectral DPC imaging approach is applied to an x-ray spectral centered at 30keV, which is divided into four energy channels in simulation. A simple numerical phantom with low contrast inserts is used and the entire imaging process is simulated using Fresnel diffraction theory. Phase stepping approach is used to manifest and retrieve phase information. The phantom is scanned over a full circular trajectory and the Hilbert filter-based FBP algorithm is used to compute the DPC-CBCT reconstruction. The reconstruction from the proposed spectral DPC-CBCT is compared to that from the conventional DPC-CBCT that only takes detector images for the integrated polychromatic spectrum. The uniformity, noise level and contrast of the inserts in the reconstruction are measured and compared. Simulation results indicate that the spectral DPC imaging approach can improve object contrast and reduce noise for DPC-CBCT.