In this method, the passage through the thoracic aorta to the brain is monitored after bolus injection of a ^99mTc-labeled tracer into the right brachial vein using a large-field gamma camera. Following the injection, a sequence of 100 frames at 1-s intervals is started in a 128 × 128 format, with the patient lying in the supine position, facing the detector (Fig. 4.1).

Regions of interest (ROI) are hand-drawn over the aortic arch (ROI_aorta) and bilateral brain hemispheres (ROI_brain). The size of the ROIs is approximately 55 pixels and 300 pixels for ROI_aorta and ROI_brain, respectively. The activity of the aortic arch is monitored instead of the arterial activity to minimize the invasiveness of the procedure without the need for arterial blood sampling. Time-activity curves for these ROIs are processed with a five-point smoothing technique. Then, the time delay of the brain activity to the aortic arch activity is corrected by shifting the brain activity curve to the left to match the peak times or ascending parts of both curves.

In this graphical approach, a linear phase is discernible during an 8- to 12-s period within the first 30 s after tracer injection, indicating the presence of unidirectional tracer uptake (Fig. 4.2). Thereafter the tracer transfer into the brain is less than expected if transport into the brain tissue continues to be unidirectional. The divergence of expected and measured activities reflects both the loss of radioactivity from the brain to blood and the rapid decrease of the lipophilic content of the tracer in the blood.

Graphically obtained slope (k _u) according to Eq. 4.2 depends on the size ratio of ROI_brain to ROI_aorta. To eliminate this dependence a brain perfusion index (BPI) is developed as follows, where the size ratio of ROI_brain to ROI_aorta is set to 10 for standardization. This standardization makes it possible to compare BPI among subjects.

Direct comparison of BPI for ^99mTc-HMPAO with that for ^99mTc-ECD in the same individuals demonstrated a highly significant correlation between these two tracers but approximately 7 % lower BPI values for ^99mTc-ECD than those for ^99mTc-HMPAO (Matsuda et al. 1995). This lower BPI might be due to low first-pass brain extraction for ^99mTc-ECD as compared with that for ^99mTc-HMPAO. Friberg et al. (1994) reported an average first-pass extraction ratio of 0.60 for ^99mTc-ECD when CBF was 46 ml/100 g/min, while Andersen et al. (1988) reported a higher average first-pass brain extraction ratio of 0.72 for ^99mTc-HMPAO when CBF was 59 ml/100 g/min. BPI represents the unidirectional influx rate from the blood to the brain, which is proportional to a product of CBF and the first-pass extraction ratio. Thus, BPI depends on the first-pass extraction ratio for the used tracer under constant CBF. However, differences between ^99mTc-ECD BPI and ^99mTc-HMPAO BPI were smaller than were expected from the difference in the first-pass extraction ratios, possibly attributable to a difference in radiochemical purity between the two tracers. The radiochemical purity of ^99mTc-HMPAO is somewhat lower than that of ^99mTc-ECD. The lower the radiochemical purity, the smaller BPI becomes.

Reproducibility for calculation of BPI was evaluated. To determine the influence of ROI selection and curve shift for correction of time delay of brain and aortic arch curves, independent observers conducted a repeat analysis of the same radionuclide angiograms. Intra-observer coefficients of variation ranged from 4.0 % (Matsuda et al. 1992) to 5.3, 5.6, and 7.9 % (Van Laere et al. 1999). Interobserver coefficient of variation was reported to be 4.8 % (Zaknun et al. 2008). Automated setting of ROI_aorta can improve both intra- and interobserver reproducibility particularly for naive observers (Groiselle et al. 2000).

Attention must be paid to dead time count loss by gamma camera during radionuclide angiography. The brain activity curve is inevitably delayed in relation to the aortic arch activity curve. The total count rate is higher during estimation of activity in the aortic arch than during estimation of cerebral activity. Most of the injected tracer is present in the field soon after the injection, and thus counting efficiency is lower during the estimation of aortic arch activity. Arterial input appears to be underestimated due to excessive count loss, resulting in overestimation of BPI by over 3 % when injecting 370 MBq of ^99mTc-labeled tracers (Inoue et al. 1997).

Hemispheric BPI values for ^99mTc-HMPAO and CBF values obtained from dynamic SPECT measurements by the ¹³³Xe inhalation method were compared in 38 hemispheres of 19 same individuals (Matsuda et al. 1992). BPI values showed a highly significant positive correlation (correlation coefficient = 0.926) with global CBF (gCBF) obtained from the early picture method in a ¹³³Xe-SPECT study. The regression line equation is global CBF = 2.75*BPI + 17.7.

In comparison with CBF values measured using¹²³I-IMP SPECT with continuous arterial blood sampling (Murase et al. 1999), BPI showed significant correlation coefficients of 0.629 and 0.856 for ^99mTc-HMPAO and ^99mTc-ECD, respectively.

In comparison with CBF values measured using H₂ ¹⁵O PET with continuous arterial blood sampling (Takasawa et al. 2004), BPI for ^99mTc-HMPAO showed a significant correlation coefficient of 0.831.

Murase et al. (1999, 2001) proposed spectral analysis for the estimation of BPI using radionuclide angiography for ^99mTc-labeled tracers. This analysis has been mostly applied to dynamic positron emission tomography measurements. This analysis has been introduced as a non-compartmental tracer kinetic modeling technique, which can be used to characterize the reversible and irreversible components of acyclic, connected systems and to estimate the minimum number of compartments. This technique is applicable to systems in which the measured data can be expressed as a positively weighted sum of convolution integrals of the input function with an exponential function that has real-valued nonpositive decay constants. BPI measured using graphical analysis and spectral analysis showed highly significant correlations (Murase et al. 1999; Van Laere et al. 2001) with equal interobserver reproducibility (Takasawa et al. 2003). Since BPI measured using spectral analysis appears to be proportional to CBF because of independence from the first-pass brain extraction ratio, it is larger by over 20 % and more highly correlated to CBF values obtained from ¹²³I-IMP SPECT with continuous arterial blood sampling than BPI measured using graphical analysis (Murase et al. 1999). These results suggest that spectral analysis provides a more reliable BPI than graphical analysis particularly in acetazolamide challenge (Takasawa et al. 2002). However, this spectral analysis has not been prevailed to users because of unavailability of an automated program.

To calculate regional CBF (rCBF) and to correct for incomplete retention of ^99mTc-labeled tracers in the brain, Lassen et al. (1988) proposed the following linearization algorithm of a curve-linear relationship between the brain activity and CBF. This algorithm is based on kinetic models for ^99mTc-HMPAO (Fig. 4.3) and ^99mTc-ECD (Fig. 4.4):


where F _i and F _r represent rCBF values for a region i and a reference region, respectively, and C _i and C _r are the SPECT counts for the region i and the reference region, respectively. Acquisition of SPECT data usually starts at 10 min after tracer injection. Global brain is used as the reference region in this method. α is a correction factor for the linearization. The optimum value of the correction factor α is estimated from the following equations:

where k ₃ is the conversion rate constant from lipophilic to hydrophilic tracer in the brain, k _2r is the rate of the back diffusion from a reference region in the brain to the blood, and K _1r is the influx rate constant from the blood to a reference region in the brain. λ is the brain-blood partition coefficient of the lipophilic tracer, and E describes the first-pass brain extraction of the lipophilic tracer across the blood-brain barrier. Average values for these rate constants and λ have been reported (Table 4.1). Lassen’s correction algorithm assumes that k ₃ and λ are the same in all regions. k ₃ and λ were set to the reported mean values of 0.80 and 0.67 for ^99mTc-HMPAO and 0.57 and 1.33 for ^99mTc-ECD, respectively. An almost threefold smaller k_2r value for ^99mTc-ECD than that for ^99mTc-HMPAO results in the higher default α value of 2.6 in the linearization algorithm for ^99mTc-ECD than the default α value of 1.5 in that for ^99mTc-HMPAO. When CBF in the reference region ranged from 20 to 60 ml/100 g/min, optimum α values estimated using Eqs. 4.5 and 4.6 ranged from 3.2 to 1.2 for ^99mTc-HMPAO and from 6.3 to 2.1 for ^99mTc-ECD. The use of the fixed α value in the linearization algorithm gives rise to deviation of the obtained rCBF values from true rCBF values when using optimized α values that vary with CBF in a reference region. However, less deviation of obtained rCBF values from true values may be expected in the presence of a higher α value.

A flow chart of the rCBF measurement procedure comprising radionuclide angiography to estimate gCBF and SPECT to estimate rCBF using ^99mTc-labeled tracers is demonstrated in Fig. 4.5. Global mean SPECT counts are calculated from threshold at 40–50 % of maximum SPECT counts. Thus, original SPECT images (Fig. 4.6a) are converted to rCBF SPECT images without any blood sampling (Fig. 4.6b).

Takeuchi et al. (1997) applied these noninvasive rCBF measurements using ^99mTc-ECD to a split dose protocol (Fig. 4.7) for the consecutive evaluation of baseline rCBF and cerebral perfusion reserve with acetazolamide challenge. Radionuclide angiography after bolus injection of ^99mTc-ECD is performed for 100 s to obtain gCBF at baseline using graphical analysis. Ten minutes after the first injection of ^99mTc-ECD, first SPECT at baseline is started. Then 10 min before the second injection of ^99mTc-ECD, 1,000 mg acetazolamide is intravenously injected during acquisition of projection data for the first SPECT. Acetazolamide does not modify the tracer distribution of the brain during this acquisition of the first SPECT since tracer distribution has been already determined during radionuclide angiography. Immediately after the completion of the first SPECT, an additional dose of ^99mTc-ECD is injected, and 10 min later the second SPECT is started. Total time span required for this procedure is less than 50 min.

To extract SPECT data for the acetazolamide challenge, the first SPECT data were subtracted from the second SPECT data after decay correction of ^99mTc. Post-acetazolamide gCBF is estimated from the baseline gCBF, baseline global mean SPECT count, and post-acetazolamide global mean SPECT count as follows: