Impact of image-based scatter correction for ¹²³IMP-SPECT and SPM analysis

Recently, techniques for converting site-specific normal databases to other databases have been required for SPECT image statistical analysis. To implement scatter correction, we developed an image conversion technique named image-based scatter correction (IBSC). Furthermore, we evaluated the applicability of IBSC for SPM analysis. SPECT studies were performed on 28 normal volunteers and 10 patients with Alzheimer's disease (AD) using ¹²³I-IMP. Two sets of scatter uncorrected images, I_AC^μb, and a corrected image by TEW, I_TEW, were reconstructed with attenuation correction using Chang's method. Scatter corrected images by IBSC, I_IBSC, were generated from I_AC^μb. Normal databases of I_AC^μb, I_TEW and I_IBSC were compared by SPM analysis. The Z-score (=(mean−AD)/S.D.) of I_TEW was compared with that of I_AC^μb. Significant differences (p<0.05) between normal databases of I_AC^μb and I_TEW observed at the global posterior cingulate and precuneus regions almost disappeared when I_TEW and I_IBSC were compared. The Z-score at the posterior cingulate by I_TEW was increased compared with that of I_AC^μb. We conclude that for SPM analysis of IMP-SPECT, IBSC can convert a normal database of scatter uncorrected images into corrected images, which is equivalent to correction by a conventional TEW method. Scatter correction may improve the ability to detect AD.