Cancer therapies based on administered radionuclides require accurate
information on tumor dose. One of the major factors influencing the di
stribution of absorbed-dose characteristics is the uniformity of the r
adiolabel distribution in tissue. To study the effect of nonuniformiti
es, we used image analysis techniques to measure automatically the coo
rdinates of autoradiographic grains (sources) and cell nuclei in cut s
ections from three different tumors, following treatment with radiolab
eled antibodies. The spatial distribution data of sources and cell nuc
lei from these tumor sections were assessed and the pattern of energy
deposition in the cell nuclei calculated, assuming that each autoradio
graph grain corresponded to a source of the alpha emitter astatine-211
(At-211) or the beta emitter yttrium-90 (Y-90). The distribution of d
eposited energy obtained for the real grain distributions was compared
to the distribution assuming a locally uniform source distribution, i
.e., simulating grain count averaging as produced by a microdensitomet
ric method within a 100 x 100 mum2 frame size (frame averaging), and a
uniform distribution across the entire section (section averaging). T
he results show first that when the grain distribution is uniform, the
average dose within the section is an adequate estimate of the dose t
o the cell nuclei. Second, when the grain distribution is nonuniform,
the distribution of doses to the cell nuclei is significantly less whe
n calculations use the measured grain coordinates, or frame averaging,
than when section averaging is used. Third, when the sources are loca
ted on or in the cells, both frame and section averaging produce under
estimates of the dose to the cell nuclei.