Scatter correction (SC) is important for quantitative SPECT. Among several SC methods proposed so far, transmission dependent convolution subtraction (TDCS) and triple energy window (TEW) methods have been shown to provide an accurate elimination of scatter component. In this work, we investigated the impact of SC and compared both SC methods in brain SPECT with dopaminergic tracer. Phantom data was acquired using I-123 filled striatal brain phantom. In human study, an emission scan was performed after the injection of I-123 β-CIT. SC was applied with both TDCS and TEW. In the reconstruction using OSEM and FBP, five types of data (OSEM with and without SC (by TDCS and TEW), FBP with and without SC (by TEW)) were generated. A post-gaussian filter was applied to the reconstructed image to match the resolution among the reconstructed images. The values of the activities of striatum (STR) and background (BKG), as well as the STR/BKG were used in quantitative evaluation. In the phantom study, both SC methods in three reconstructions showed increased STR/BKG ratio by large increase of STR and small decrease of BKG activities. TDCS with OSEM showed the minimal error in the activity estimation on both regions, compared to TEW with both reconstructions. The values of COV in BKG region for TEW with both reconstructions were higher than those for TDCS with OSEM. TEW with FBP showed the overestimated STR/BKG ratio with large bias, which was mainly due to underestimated BKG activity. In the human study, there was the large increase of STR/BKG ratio, and lower BKG COV by TDCS with OSEM, compared with those by other 4 reconstructions. TDCS with OSEM provided the best accuracy in the quantitative estimation of activity distribution in the striatal phantom. TDCS with OSEM and TEW with FBP provided a comparable accuracy in activity estimation, respectively. However, TDCS with OSEM appears to be better suited for a quantitative study using the region of reference tissue.
|Number of pages
|Published - 2002
|2002 IEEE Nuclear Science Symposium Conference Record - Norfolk, VA, United States
Duration: 2002 Nov 10 → 2002 Nov 16
|2002 IEEE Nuclear Science Symposium Conference Record
|02/11/10 → 02/11/16