RT - Step wedge - Analysis

The "digital radiography" and "image plate" models make it possible to obtain better matches when taking into account the scattered radiation into materials. This is particularly the case when the depth of the scintillator (1000 μm of CsI) is adapted. However, such a scintillator thickness is quite far from the expected value. The strong diffusion observed experimentally on the ferritic steel wedge associated with the image plate detector (profile of rounded steps) is not reproduced by the simulation. We attribute this discrepancy to the fact that the backscattering phenomena, on the support of the image plate in particular, were not taken into account in our simulations.

Measured noises levels obtained on the experimental and simulated images are sum up in the table below.

The signal noise levels are systematically higher in our simulations compared to the experiment (up to a factor of 4.2 in the last case).

In CIVA, the DQE is not a function of the spatial frequency. Even setting a value of 1 for the DQE (the theoretical limit and the maximum value accepted by CIVA), it was not possible to obtain an agreement between experiment and simulation. Furthermore, it is observed that a modification of the Modulation Transfer Function (FTM) does not affect the measured noise level for a given detector response (gray level). In other words, there is no link between FTM and DQE in CIVA modeling when these two quantities are physically linked. This seems to be the source of the difficulties of these detector models in getting the right noise levels.

Experimentally, IQI visibility is low. Most of the time, the first 3 to 4 wires of the first IQI (NF EN ISO 19232-1 W10), and rarely those of the second IQI (NF EN ISO 19232-1 W13), are seen whatever the configuration. The visibility is generally better for the ferritic steel wedge, compared to the Dural wedge.

This last observation remains valid for the simulations, which make it possible to detect qualitatively the wires perceived experimentally. The simulations based on the photostimulable screen model have a slight tendency to reinforce the contrast of the IQI, compared to the experiment.