SummaryThe aim of this study was to evaluate the effect of scatter and attenuation correction in region of interest (ROI) analysis of brain perfusion single-photon emission tomography (SPECT), and to assess the influence of selecting the reference area on the calculation of lesion-to-reference count ratios. Patients, methods: Data were collected from a brain phantom and ten patients with unilateral internal carotid artery stenosis. A simultaneous emission and transmission scan was performed after injecting 123I-iodoamphetamine. We reconstructed three SPECT images from common projection data: with scatter correction and nonuniform attenuation correction, with scatter correction and uniform attenuation correction, and with uniform attenuation correction applied to data without scatter correction. Regional count ratios were calculated by using four different reference areas (contralateral intact side, ipsilateral cerebellum, whole brain and hemisphere). Results: Scatter correction improved the accuracy of measuring the count ratios in the phantom experiment. It also yielded marked difference in the count ratio in the clinical study when using the cerebellum, whole brain or hemisphere as the reference. Difference between nonuniform and uniform attenuation correction was not significant in the phantom and clinical studies except when the cerebellar reference was used. Calculation of the lesion-to-normal count ratios referring the same site in the contralateral hemisphere was not dependent on the use of scatter correction or transmission scan-based attenuation correction. Conclusion: Scatter correction was indispensable for accurate measurement in most of the ROI analyses. Nonuniform attenuation correction is not necessary when using the reference area other than the cerebellum.