Resolution mesh is what generally used. You should consider that the spatial resolution of the scanner is a few tens of microns (~50) so use a wire mesh of 30 microns or less. Alternatively small pinholes could also be used
You can also measure the "edge response" by having a flat knife-edge high-Z filter (tungsten, gold, etc ...) that is in contact with the I.P. and then expose to x-rays from a distant source. You can translate that into a line-spread-function or point-spread-function since the LSF is the derivative of the edge response function.
We've found that using a BAS-SR image plate, scanned at 25-um with a Fuji 7000 scanner, the FWHM of the LSF (fit to a gaussian) of about 100-um.
Our scanner (Image reader 1800II) has a nominated 50um resolution. The idea is to check how reliable this value is, where it will effect the source size calculation. If the source is magnified with a factor X that gives a calculated source value above the resolution, then it is OK. The problem will be if we get a source size that is less than the scanner's resolution, in which case we have to work out the magnification that will bring the calculated source size to the nominated resolution or less to get the right value (FWHM) when doing deriving LSF out of your ESF.