A strategy for the quantitative evaluation of mammography image qualit
y phantoms is described and illustrated by the evaluation of a new des
ign of phantom, the TOR(MAM). The performance criteria assessed were s
ensitivity and measurement uncertainty. Sensitivity was assessed as th
e response of the image quality (IQ) score to changes in kV, scatter t
hickness below the phantom, and by use with a range of clinical system
s. The measurement uncertainty was measured as the inter- and intra-ob
server errors from repeated measurements on 12 mammography systems by
5 observers. The IQ score was found to be very sensitive to changes in
image quality but had large observer errors.