Correlation between ISCV of radiation attenuation and ISCV-NDBs

The intersubject variation in radiation attenuation because of anatomical variation is expected to have a different impact on the ISCV-NDBs depending on the AC method that is used. To evaluate this effect, an attenuation coefficient (mu) measured using CT and the difference in ISCV-NDBs between the two AC methods were compared. The CT and SPECT images of the NDBs were spatially normalized using a transaxial template oriented along the orbitomeatal line. In the normalizing process using SPM8, the “Preserve Amount” option was specified to preserve the total amount of attenuation coefficient in the head. Brain mask ROIs were created by masking the CT image at a threshold of 0 Hounsfield Units (HU) in each transaxial plane. The rationale for choosing this threshold was that the range of head attenuation is more than 0 HU and the relationships between the measured attenuation coefficient using CT (HU) and the mu of low energy gamma rays using I-123 was defined through bilinear fitting [16]. The summation of the measured attenuation coefficient and the mean SPECT count in each ROI were evaluated in each slice from the top of the parietal lobe to the bottom of the posterior cranial fossa (Fig 5A). The difference in the ISCV of the SPECT count between the two AC methods was calculated and compared with the ISCV of the sum of the measured attenuation coefficients in each slice.

A) (a) A CT attenuation coefficient (HU) map of each subject was normalized to the transaxial template oriented along the orbitomeatal line and smoothed using an isotropic Gaussian kernel (FWHM, 12 mm). (c) The positive attenuation coefficients were summed in each slice using the brain mask ROI (b) created by masking the CT mu map (see text). (d) The ISCV of (c) for the NDB (combination of the Y-NDB and the O-NDB) was calculated. (e) The difference between the ISCV map of NDB-Chang’s-AC (e) and the ISCV map of NDB-CT-AC (f) was used to create a map of the difference in ISCV (g). The mean of (g) in each transaxial brain mask ROI (h) was then evaluated. B) The ISCV of the slice-based summation of positive attenuation coefficients (HU) using CT (d) and the mean difference in the ISCV-NDB between the two AC methods (i) for each ROI is plotted. The mean ISCV-NDB-Chang’s-AC was larger than the ISCV-NDB-CT-AC for every ROI. A significant and strong correlation was found between the two parameters.