Ph. Pretorius et al., ATTENUATION CORRECTION STRATEGIES FOR MULTIENERGY PHOTON EMITTERS USING SPECT, IEEE transactions on nuclear science, 44(3), 1997, pp. 1323-1328
The aim of this study nas to investigate whether the photopeak window
projections from different energy photons can be combined into a singl
e window for reconstruction or if it is better to not combine the proj
ections due to differences in the attenuation maps required for each p
hoton energy. The mathematical cardiac torso (MCAT) phantom was modifi
ed to simulate the uptake of Ga-67 in the human body. Four spherical h
ot tumors were placed in locations which challenged attenuation correc
tion. An analytical 3D projector with attenuation and detector respons
e included was used to generate projection sets. Data were reconstruct
ed using filtered backprojection (FBP) reconstruction with Butterworth
filtering in conjunction with one iteration of Chang attenuation corr
ection, and with 5 and 10 iterations of ordered-subset maximum-likelih
ood expectation-maximization (ML-OS) reconstruction. To serve as a sta
ndard for comparison, the projection sets obtained from the two energi
es were first reconstructed separately using their own attenuation map
s. The emission data obtained from both energies were added and recons
tructed using the following attenuation strategies: 1) the 93 keV atte
nuation map for attenuation correction, 2) the 185 keV attenuation map
for attenuation correction, 3) using a weighted mean obtained from co
mbining the 93 keV and 185 keV maps, and 4) an ordered subset approach
which combines both energies. The central count ratio (CCR) and total
count ratio (TCR) were used to compare the performance of the differe
nt strategies. Compared to the standard method, results indicate an ov
erestimation with strategy 1, an under-estimation with strategy 2 and
comparable results with strategies 3 and 4. In all strategies, the CCR
's of sphere 4 (in proximity to the liver, spleen and backbone) were u
nder-estimated, although TCR's were comparable to that of the other lo
cations. The weighted mean and ordered subset strategies for attenuati
on correction were of comparable accuracy to reconstruction of the win
dows separately. They are recommended for multi-energy photon SPECT im
aging quantitation when there is a need to combine the acquisitions of
multiple windows. This work can also be extended to include other mul
ti-energy photon emitters such as In-111.