Dear Wang Kei Ma,

You are correct to question the applicability of currently-available computational phantoms for a particular medical imaging exam.  In addition to the absence or presence of organs of interest, you should consider the geometry of the medical imaging exam, including how the human body is represented in relation to the medical imaging equipment.  In mammography, the breast is compressed to achieve the appropriate clinical image.  I am not aware of a computational phantom which allows for the application of forces to achieve compression, but phantoms based on non-uniform rational B-splines, such as the University of Florida/National Cancer Institute hybrid phantom series, do allow for appropriate deformation, which could be derived from knowledge of the compression forces involved.  You can also check out this article, http://pubs.rsna.org/doi/abs/10.1148/radiol.2462070256, in which the Cristy phantom was adjusted for mammography.

Regarding benchmarking, I would suggest comparing the output of your model with measurements from actual mammography machines.  You could also compare the results of your simulations to published data from Sobol and Wu 1997 (http://www.ncbi.nlm.nih.gov/pubmed/9127307).

Best regards,

Amir

Here is a article performing mammography Monte Carlo simulations using GATE:

http://iopscience.iop.org/article/10.1088/0031-9155/58/7/2247/meta;jsessionid=430C351512EA46372DAE31C581D32B83.c4.iopscience.cld.iop.org

I have been using an XCAT phantom to estimate dose in mammography. XCAT is able to produce a phantom with compressed left or right breast. All the necessary organs are there (see attached image).

While I like the XCAT phantom, their license policy is unfriendly.

The license policy shouldn't be unfriendly. We're always glad to help as far as I know.