CT attenuation correction for the quantitative 123I-IMP CBF SPECT image reconstruction

Objectives Conventional attenuation correction (AC) such as Chang’s method for quantitative 123I-IMP cerebral blood flow (CBF) SPECT measurement may not be accurate due to attenuation from cranial bones. We performed CT AC for the quantitative CBF SPECT image reconstruction and compared with conventional Chang’s AC.

Methods Six neuro-psychiatric patients, aged 34-58 yr without any morphological abnormality in MRI were evaluated. Quantitative CBF SPECT study was performed with the injection of 123I-IMP (222 MBq) and one point arterial blood sampling (IMP-ARG method) by using two detectors type SPECT system. Non-contrast head CT study was also done separately by using CT system. Transaxial CT images (slice thickness: 8mm) were then automatically co-registered and re-sliced (1.72mm) identical with trans-axial CBF SPECT images using a fully automatic image registration software (GMS7700R, e.soft (fusion ART), Toshiba, Japan). Then, CT attenuation map was generated. Two types of SPECT images were reconstructed: 1) filtered-back projection (FBP) with CTAC, and 2) FBP with Chang’s AC. CBF values in each region were compared between two AC methods. To verify the SPECT values obtained by CTAC, we measured brain phantom with and without thin or thick plasters surrounding the phantom for imitation of the cranial bones.

Results CBF values (ml/100g brain/min) in all regions with CT-AC (48.5±3.7) were significantly higher than with Chang’s AC (37.7±2.5) by 22-33% (P<0.05). Cerebellum showed the highest underestimation in Chang’s AC (41.9±6.3) as compared with CT-AC (55.7±6.3) by 33% probably due to attenuation by cranial bones in posterior fossa. In brain phantom study, SPECT values with and without thin or thick plasters obtained by CTAC were almost identical with true values. However, all SPECT values obtained by Chang’s AC were lower than true values and further decreased with thicker plasters.

Conclusions CT AC would improve the underestimation of the quantitative 123I-IMP CBF SPECT measurement as compared with conventional Chang’s AC