Abstract
1657
Objectives: We investigate quantitative performance of an iterative removal of cross-talk in simultaneous Tc-99m/I-123 SPECT-CT imaging.
Methods: Three sets of phantom data were acquired on a SPECT-CT GE Infinia-Hawkeye camera. First, three bottles (33ml, 8-16MBq, I-123) and the heart phantom (120ml, 28MBq, Tc-99m) were scanned in air on the camera bed (Exp.1). Then, the same containers were located inside the thorax phantom filled with cold water (Exp.2) and with Tc-99m activity (Exp.3). Data were acquired using two energy windows LW=[132–148keV]; UW=[151–167keV], separated to minimize cross-talk in UW, and a cardiac protocol with 128x128 matrix and 60 views. Both LW- and UW-images were reconstructed from LW and UW data, respectively, using OSEM (10 subsets, 24 iterations) with corrections for attenuation, resolution loss, and high energy photons. Self-scatter and cross-talk components were updated after every four iterations using our APD algorithm and included in the forward step. The Tc-99m and I-123 activities were measured in LW- and UW-images (showing the magnitude of cross-talk) and in the volumes of interest (showing quantitative accuracy).
Results: In Exp. 1, we obtained 0-2% errors for I-123 sources (located far from Tc-99m insert) and 4% error for the total UW-image, independent of cross-talk correction. For LW, our method decreased the error (due to I-123 primary and high-energy photons) in estimating total activity in the LW-image from 19 to 8%. In Exp. 2, the cross-talk in the LW-image is larger, but decreased by a factor of 2 after application of our method. In Exp. 3, the Tc-99m background degrades UW-image, but our algorithm lessened this error from 17 to 6%.
Conclusions: For our energy windows selection, iterative APD-based corrections for self-scatter and cross-talk allows for improving quantitative accuracy of both images by a factor of 2-3.
- Society of Nuclear Medicine, Inc.