Development of image-based scatter correction for brain perfusion SPECT study: comparison with TEW method

In order to convert scatter uncorrected into corrected SPECT image, an image-based scatter correction (IBSC) method has been developed. The aim of this study was validation of its role as image converter from scatter uncorrected into corrected images equivalent to image corrected by conventional TEW method. IBSC method is executed in the postreconstruction process and only requires an attenuation corrected main photopeak image with broad μ value, I_AC^μb. The scatter component image is estimated by convolving I_AC^μb with a scatter function followed by multiplying with an image-based scatter fraction (SF) function. The IBSC method was evaluated with Monte Carlo simulations and ^99mTc-ECD SPECT human brain perfusion studies obtained from five volunteers. The noise property of the scatter corrected image using IBSC method, I_IBSC, was compared with that by TEW method, I_TEW, with simulated brain phantom images. Image contrast between gray with white matter in the human study was also compared between IBSC and TEW method. The global signal-to-noise (S/N) ratio of I_IBSC was decreased to 14% compared to that of I_AC^μb, whereas that of I_TEW was decreased to 21%. In human brain imaging, significant difference in image contrast between IBSC and TEW method was not observed (p<0.05). In conclusion, the IBSC method could be applied to clinical brain perfusion SPECT as conversion I_AC^μb into a scatter corrected image equivalent to I_TEW. This achieves a better noise property than the TEW method.