Background: Attenuation is a well recognized cause of reconstruction artifa
cts in SPECT imaging. Occasionally, we have noted an increase in activity e
xtending from the apical septal portion of the ventricle in women with sign
ificant breast attenuation. Although the idea that attenuation can produce
an increase in activity on the reconstructed images seems paradoxical at fi
rst, it is consistent with the process of filtered back projection. Methods
: We filled a cardiac phantom with 1 mCi of Technetium-99m, placed it in a
water filled anthropomorphic torso phantom and imaged it over a 180 degrees
orbit. Next, a breast phantom designed to simulate a significant degree of
breast attenuation was placed on the torso phantom and imaging was repeate
d. The images were reconstructed first using conventional filtered back pro
jection then with maximum likelihood. Results: When the phantoms with and w
ithout breast attenuation were reconstructed using filtered back projection
and compared, the phantom with breast attenuation had a large 'smear' of a
ctivity extending anteriorly from the apical septal wall which was very sim
ilar to the abnormalities previously noted in clinical images; the phantom
without breast attenuation had no such defect. This artifact was significan
tly less prominent when the images were reconstructed using the maximum lik
elihood technique. Conclusions: Attenuation artifact can also produce a see
mingly paradoxical increase in counts on the reconstructed image but this p
henomenon is consistent with the workings of filtered back projection.