Dear Neil,

to calculate MTF from a slanted edge object, you can use e.g. this method:

https://www.spiedigitallibrary.org/conference-proceedings-of-spie/5030/1/Simple-method-for-modulation-transfer-function-determination-of-digital-imaging/10.1117/12.479990.short

It works as follows:

1) identify the number of pixel lines in which the edge is shifted by one pixel (e.g. by Sobel-Filter)

2) resort pixel values of these lines and create an oversampled edge spread function (oESF)

3) differentiate the oESF to create a line spread function (LSF)

4) Fourier-transform the LSF to create a MTF

You can also check my thesis page 134 and find a scheme for this method

https://tud.qucosa.de/api/qucosa%3A70604/attachment/ATT-0/

There are also other articles with methods described, e.g.

https://www.isprs.org/proceedings/XXXV/congress/comm1/papers/2.pdf

https://doi.org/10.1093/rpd/nch506

It woud be easier to directly start with an LSF, e.g. using a thin wire for imaging instead of a slanted edge object.

And, it is also possible to create an MTF from bar patterns as you mentioned. Bar patterns, in my opinion, are very simple for QA purpose for the user as you only need to find the pattern with the thinnest lines you can visually distinguish (e.g. 2 lp/mm is vissible but 2,2 lp/mm not). Of course, this is not a complete MTF, but sufficient for a constancy check without sophisticated calculations required from the acquired image.

Best, Falk

Dear Neil

To calcultad the MTF from edges or slits method you can use some available free softwares. I suggest the following:

(PDF) Free software for performing physical analysis of systems for digital radiography and mammography (DOI: 10.1118/1.4870955)

Best, Alessandra