RT Journal Article
SR Electronic
T1 Quantitative Bremsstrahlung SPECT/CT image reconstruction with patient specific Monte Carlo scatter compensation for Y-90 microsphere radioembolization (RE)
JF Journal of Nuclear Medicine
JO J Nucl Med
FD Society of Nuclear Medicine
SP 444
OP 444
VO 53
IS supplement 1
A1 Dewaraja, Yuni
A1 Roberson, Peter
A1 Clinthorne, Neal
A1 Chun, Se Young
A1 Hubers, David
A1 Regan, Denise
A1 Ackermann, Robert
A1 Fessler, Jeffrey
A1 Ljungberg, Michael
YR 2012
UL http://jnm.snmjournals.org/content/53/supplement_1/444.abstract
AB 444 Objectives Accurate Y-90 SPECT is complicated by the continuous energy spectrum of Bremsstrahlung photons. As energy window based-scatter correction (SC) is not possible, we propose Monte Carlo (MC) based scatter estimation using the patient’s SPECT/CT images as input to the simulator. Methods First, 6 acquisition windows were evaluated by SIMIND MC simulation, and based on maximizing the primary/scatter ratio, windows 105-135 keV and 135-165 keV were selected. Y-90 projections were generated both by physical phantom measurements on a SPECT/CT system with a HE collimator and XCAT phantom simulations for this system. Activity distributions were clinically realistic for Y-90 RE and included 2 liver lesions (80 & 40 mL) in XCAT and a 60 mL liver lesion in the physical phantom with total activity of 370 MBq. Initial reconstruction using in-house 3D OS-EM software was without SC, but after 5 iterations the MC scatter estimate was included in the forward projector. All reconstructions included energy window dependent collimator-detector response and CT-based attenuation correction. Results Qualitatively, improved tumor to liver contrast was evident in scatter corrected images compared to OS-EM without SC. Quantitatively, for XCAT, the lesion/liver activity concentration ratio was 3.5 and 3.1 without SC, but improved to 4.9 and 4.4 with MC SC (true ratio = 5). The error in absolute activity quantification of lesions was as high as 29% and 38% without SC, but improved to 1% and 12% with MC SC. For the physical phantom the lesion/liver ratio was 4.1 without SC, but improved to 6.1 with MC SC (true ratio = 6.7). The error in absolute quantification of liver and lesion activities were 44% and 67% without SC, but improved to 8% and 5% with MC SC. Conclusions Despite the continuous spectrum, accurate Bremsstrahlung SPECT is possible with MC SC, making the images well suited for evaluating Y-90 distribution and for patient-specific dosimetry in RE