Investigating the feasibility of photon-counting K-edge imaging at high x-ray fluxes using nonlinearity corrections

Pulse pileup occurring at high x-ray fluxes can severely degrade the energy resolution provided by a photon counting detector, which can represent a problem in spectroscopic CT when performing quantitative material discrimination tasks. As the effects of pileup can be most easily seen as a degradation of a detector's count rate linearity at high fluxes, it has been proposed previously to quantify and correct these nonlinearities. While this strategy has been applied successfully to materials without K-edges, it is currently unknown if this still prevails when using medical contrast agents. The purpose of this study is to close this gap.

A Medipix2MXR Hexa detector was employed, featuring a pixel pitch of 165 [micro sign]m and a 1 mm thick CdTe sensor. A phantom containing various concentrations of iodine and gadolinium contrast agents was subject to energy selective CT acquisitions, using a pulsable x-ray source operated at 70 kVp. These acquisitions were obtained at low and high photon fluxes of 1.0 × 10(6) and 1.3 × 10(7) mm(-2) s(-1), respectively. Nonlinearity corrections were applied to the high-flux projections and for each pixel separately. The results were compared to the results at low photon fluxes.

At high fluxes, a general reduction of the reconstructed attenuation coefficients was observed, which could be partially recovered using the correction strategy applied. The spectroscopic separation of iodine from the phantom material, however, degraded with increasing x-ray flux. In contrast to this, gadolinium could still be discriminated almost as well as in the low flux case.

Nonlinearity corrections applied to high flux measurements can help to recover attenuation coefficients normally obtained at low fluxes for low-Z materials, which do not exhibit an absorption edge in the relevant energy range. However, as a result of a significant change of the x-ray spectrum, the spectroscopic contrast normally observed for iodine was found to vanish with increasing x-ray flux. In other words, the authors' results indicate that nonlinearity corrections may be feasible only when the K-edge of interest is sufficiently high compared to the mean photon energy, and that spectroscopic CT at high x-ray fluxes may suffer from less limitations when using high-Z materials as contrast agents. A future study should aim to confirm these findings under clinical conditions.