Impact of attenuation and scatter correction in SPECT for quantification of cerebral blood flow using 99 mTc-ethyl cysteinate dimer

The purpose of this study was to evaluate the effects of the attenuation correction and scatter correction methods, both validated previously, on the quantitative-estimation of rCBF using ^{99 m}Tc-ECD and SPECT. SPECT scans were performed on 7 subjects, and images were reconstructed by OSEM in which uniform and segmented μ maps were used for attenuation correction, with and without scatter correction, which is based on the transmission-dependent convolution subtraction technique. Segmented and uniform μ maps were generated from MR images. The authors also produced uniform μ maps using ECD images obtained at various threshold levels. Scatter correction improved the image contrast dramatically. K₁ image with attenuation and scatter, corrections assuming a homogeneous CL map was consistent as compared with those by segmented μ map in most regions, except for in a deep structure (e.g. 7.3%). This small amount of error was also observed in a phantom study and Monte-Carlo simulation. Absolute K₁ values in the reconstructed images were sensitive to the threshold level when edge of the brain was determined from the ECD images, and varied from 14.2 to 42.3%, corresponding to the threshold level from 10 to 20%, respectively. Using optimal threshold level, absolute K₁ values varied by ±8.8%. This suggests the need for further development of appropriate edge detection technique. This study demonstrated the scatter correction is essential in quantitative SPECT studies with ECD in brain. It was also demonstrated that the use of uniform attenuation μ map could provide reasonable accuracy, despite a small but significant errors in a deep structure regions