Abstract
445
Objectives Quantitative imaging of Y90 is gaining broader interest in support of internal radiotherapy applications. The purpose of this study was to compare performance of PET/CT and SPECT/CT for that application using the standard clinical acquisition and reconstruction software.
Methods An elliptical phantom containing four spheres having volumes 1.9, 11, 63, and 100cc was imaged with both PET/CT and SPECT/CT. Initially spheres were filled with 9 mCi of Y90 chloride (Perkin-Elmer) and no background was added. One week later, 11.2 mCi of Y90 was introduced as background and imaging was repeated. SPECT data were acquired using a Siemens Symbia TruePoint SPECT/CT scanner using high energy collimators and a 135-165 keV energy window. Images were reconstructed using Siemens OSEM (20 iterations, 6 subsets, no filter, no scatter correction) with a 150 keV attenuation map. PET images were acquired using a Siemens Biograph 6 LSO-based scanner using a single bed position, a 425-650 keV window, and were reconstructed using 3D OSEM (3 iterations, 21 subsets). The PET countrate was low due to the small positron fraction of Y90. Attenuation maps in both cases were derived from CT. Regions of interest for quantification were defined on the corresponding CT images for each sphere and for background regions.
Results Qualitatively, both PET and SPECT images appeared noisy with PET demonstrating somewhat higher spatial resolution. Without background, errors in SPECT quantification ranged from -9% to -64% (using largest sphere for calibration), while PET errors were -21% to -58%. Addition of background, resulted in errors of +13% to -31% and -0.6% to -67% for PET and SPECT, respectively. In all cases, the sphere-to-background activity ratio was underestimated.
Conclusions While both SPECT/CT and PET/CT may be useful for imaging Y90, both will require considerable corrections to achieve good quantitative accuracy. Without such, it is premature to claim the superiority of one modality over the other