Experimental validation of a computer simulation of radiographic film

In radiographic films, the behavior of characteristic curve is very important for the image quality. Digitization/visualization are always performed by light transmission and the characteristic curve is known as a behavior of optical density in function of exposure. In a first approach, in a Monte-Carlo computer simulation trying to build a Hurter-Driffield curve by a stochastic model, the results showed the same known shape, but some behaviors, like the influence of silver grain size, are not expected. A real H and D curve was build exposing films, developing and measuring the optical density. When comparing model results with a real curve, trying to fit them and estimating some parameters, a difference in high exposure region shows a divergence between the models and the experimental data. Since the optical density is a function of metallic silver generated by chemical development, direct proportion was considered, but the results suggests a limitation in this proportion. In fact, when the optical density was changed by another way to measure silver concentration, like x-ray fluorescence, the new results agree with the models. Therefore, overexposed films can contain areas with different silver concentrations but it can't be seen due to the fact that optical density measurement is limited. Mapping the silver concentration in the film area can be a solution to reveal these dark images, and x-ray fluorescence has shown to be the best way to perform this new way to digitize films. (author)