A Multicenter Validation of Regional Cerebral Blood Flow Quantitation Using [ 123 I]Iodoamphetamine and Single Photon Emission Computed Tomography

Abstract

Recently, two methods have been proposed for regional cerebral blood flow (rCBF) quantitation using [¹²³I]iodoamphetamine (IMP) and single-photon emission computed tomography (SPECT). The table look-up (TLU) method has been shown to provide both rCBF and volume of distribution, V_d, images from two SPECT scans, while a single-scan autoradiographic (ARG) technique provided rCBF using a fixed and assumed V_d. In both methods, a single blood sample was referred to calibrate the previously determined standard input function The present multicenter project was designed to evaluate the accuracy of both methods for use as clinical investigative tools. Ten independent institutions performed [¹²³I]IMP-SPECT studies according to both methods in 76 subjects (10 normal volunteers, 32 patients with cerebrovascular disease, and 34 patients with other diseases). Calculated rCBF values were compared with those obtained by the following reference methods available in the participating institutions; [¹⁵O] H₂O positron emission tomography (PET) (five institutions), [¹³³Xe]SPECT (four institutions), and the [¹²³I]IMP microsphere method (three institutions). Both ARG and TLU methods provided rCBF values that were significantly correlated with those measured by the [¹⁵O] H₂O PET technique (p < 0.001 for all subjects; overall regression equation, y = 15.14 + 0.54×) and those measured by the [¹²³I]IMP-microsphere method (p < 0.001 for all subjects; y = 2.0 + 0.80×). Significant correlation (p < 0.05) was observed in 18 of 24 subjects studied with the [¹³³Xe] SPECT reference technique (overall regression equation, y = 15.0 + 0.55×). Mean cortical gray matter rCBF in a group of normal subject was 43.9 ± 3.3 and 43.4 ± 2.0 ml/min/100 g for the ARG and TLU methods, respectively. Regional V_d of [¹²³I]IMP estimated by the TLU method was 45 ml/ml ± 20% in the normal cortical region. Close agreement between ARG and TLU rCBF values was observed (y = −3.21 + 1.07×, r = 0.97), confirming the validity of assuming a fixed V_d in the ARG method. Results of this study demonstrate that both the ARG and TLU methods accurately and reliably estimate rCBF in a variety of clinical settings.

Keywords

CBF quantitation Single photon emission computed tomography

¹²³Iodine (¹²³I) labeled N-isopropyl-p-iodoamphetamine (IMP) can provide a tomographic map of brain perfusion with single photon emission computed tomography (SPECT) (Winchell et al., 1980; Kuhl et al., 1982; Podreka et al., 1989; Greenberg et al., 1990; Yokoi et al., 1993; Yonekura et al., 1993; Odano et al., 1994; Nishizawa et al., 1989; Takeshita et al., 1992; Hashikawa et al., 1994; Murase et al., 1990; Ito 1995a). However, due to clearance of the tracer from the brain, imaging with conventional SPECT cameras may cause a systematic underestimation of global cerebral blood flow (CBF) and a reduction in contrast between high and low regional CBF (rCBF) regions. Recently, two methods, proposed by Iida et al., (1994a,b), provide absolute, quantitative rCBF using conventional [¹²³I]IMP SPECT in a routine clinical setting. These two methods employ a two-compartment model and take the clearance of [¹²³I]IMP from the brain into account. This allows delayed and prolonged SPECT data acquisition using a conventional rotating gamma camera. The so-called table look-up (TLU) method provides two functional parameters: rCBF and regional distribution volume, V_d, of [¹²³I]IMP from two SPECT studies acquired at different times after the [¹²³I]IMP administration. The second, autoradiographic (ARG), method provides a quantitative rCBF map from a single SPECT scan using an assumed V_d value. In both methods, frequent and/or continuous arterial blood sampling and measurements of the lipophilic fraction in each blood sample are avoided by employing a standard input function previously generated from an independent subject group (Iida et al., 1994a, b ; Munaka et al., 1992; Ito et al., 1992). Use of this standard input function has been validated previously both theoretically and clinically, and it has been demonstrated that individual differences of the input function do not significantly affect rCBF estimates (Iida et al., 1994a, b ; Munaka et al., 1992; Ito et al., 1992).

Previous studies (Iida et al., 1994a, b ) have also shown that quantitative rCBF values from these two methods are in good agreement with those obtained by previously established methods, e.g., nonlinear least square fitting to [¹²³I]IMP kinetic data, the [¹²³I]IMP-microsphere method, and the PET ¹⁵O water ARG technique. While these studies demonstrated the potential of these techniques, the number of subjects was limited and all validation studies were performed at a single institution. To evaluate the general applicability of the techniques in a large, heterogeneous patient population for a variety of SPECT systems, a multicenter project was initiated. Ten independent institutions participated in the project. Each institution compared rCBF values obtained from the IMP-TLU and IMP-ARG methods with rCBF measurements for the same patient using an established reference technique available at the institution.

METHODS

Institutions

Ten independent institutions in Japan (Table 1) participated in this project. All institutions performed [¹²³I]IMP-SPECT on each subject following set protocols for the IMP-TLU and IMP-ARG methods. Each subject was also studied by the reference rCBF method available at the particular institution (Table 2).

TABLE 1.

List of institutions, and SPECT scanners used for the IMP studies in each institution

			Mid-scan time (MST) and duration

Institution (short name)	SPECT scanner	Type of scanner	MST (min)	Duration (min)	Attenuation correction^e	Energy window (keV)	Cross calibration
Research Institute for Brain and Blood	Shimadzu Head tome (SET031)^a	Ring dedicated for brain study	40 180	8 15	Sorenson	159 ± 20%	16 cm^ϕ cylindrical
Vessels-Akita (Akita)
Yamagata University School of Medicine (Yamagata)	Shimadzu Head tome (SET-021)^a	Ring dedicated for brain study	40180	6 6	Sorenson	159 ± 20%	16 cm^ϕ cylindrical
Gunma University School of Medicine (Gunma)	Shimadzu Headtome (SET-031)^a	Ring dedicated for brain study	40180	1616	Sorenson	159 ± 20%	20 cm^ϕ cylindrical
Faculty of Medicine, University of Tokyo (Tokyo)	Toshiba GCA9300A^b	Three-head, fan-beam collimator	40180	3030	None^f	159 ± 10%	20 cm^ϕ cylindrical
Fujita Health University School of Medicine (Fujita)	Shimadzu Headtome (SET-031)^a	Ring dedicated for brain study	4080	1717	Sorenson	159 ± 15%	20 cm^ϕ cylindrical
Kyoto University School of Medicine (Kyoto)	Shimadzu Headtome (SET-030W)^c	Ring for whole body study	40180	2020	Sorenson	159 ± 20%	19 cm^ϕ cylindrical
Ehime University School of Medicine (Ehime)	Hitachi 2000H-40^NP	Four-head	40180	1414	Chang	159 ± 10%	20 cm^ϕ cylindrical
Faculty of Medicine, Kyushu University (Kyushu)	Toshiba GCA9300A^b	Three-head, fan-beam collimator	40180	2424	Chang^g	160 ± 10%	17.5 cm^ϕ cylindrical
Niigata University School of Medicine (Niigata)	Shimadzu Headtome (SET-050)^a	Ring dedicated for brain study	40180	3030	Sorenson	159 ± 25%	16 cm^ϕ cylindrical
Institute of Development, Aging and Cancer, Tohoku University (Tohoku)	Yokokawa neurocam^d	Three-head, parallel-beam collimator	40180	1818	Sorenson	159 ± 10%	18.5 × 13.5 cm eliptic

NP, not published.

Hirose et al.(1982).

Kypros et al.(1993).

Yonekura et al.(1989).

Kouris et al. (1992).

All institutions employed filtered back projection for the image reconstruction. The attenuation correction followed either Sorenson (1974) or Chang (1978).

No attenuation correction was performed. However, no geometrical correction was applied for the fan beam, thus counteracting the effects of attenuation due to increased sensitivity near the center of field of view.

An empirical correction was performed to provide a uniform distribution in a cylindrical phantom.

TABLE 2.

List of reference methods for each institution

			Normal rCBF value

Institution	Method^a	Scanner	CBF	Age (yr)
Akita	PET (H₂¹⁵O autoradiography)	Shimadzu Headtome-^b	37 ± 8^NP(n = 12)	21–71
Yamagata	[¹³³Xe]SPECT (Kanno-Lassen)	Shimadzu Headtome SET-021^c	Age^f dependent	1.4–80(n = 30)
Gunma	PET(C¹⁵O₂ steady state)	Hitachi PCT H1^NP	50^NP(n = 14)
Tokyo	PET (H₂¹⁵O autoradiography)	Shimadzu Headtome-^b	57 ± 10^NP(n = 28)
Fujita	[¹³³Xe]SPECT (Kanno-Lassen)	Shimadzu Headtome SET-031^c	46 ± 5^g(n = 9)	58 ± 8
Kyoto	PET (C¹⁵O₂ steady state)	Hitachi PCT3600W^NP	41 ± 5^h(n = 6)
Ehime	IMP-microsphere [¹³³Xe]SPECT (kanno-Lassen)	Hitachi 2000H-40^NP	Non 48 ± 7^NP(n = 10)	53 ± 14
Kyushu	PET (H₂^I5O autoradiography)	Shimadzu Headtome-111^d	38 ± 5ⁱ(n = 8)	28 ± 7
Niigata	IMP-microsphere	Shimadzu Headtome SET-050^c	53 ± 3^NP(n = 8)	20–70
Tohoku	[¹³³Xe]SPECT (Kanno-Lassen) IMP-microsphere	Yokokawa Neurocam^e	NonNon

NP, not published.

Detailed procedures are different in each institution. See text.

Iida et al. (1989).

Hirose et al. (1982).

Kanno et al. (1985).

Kouris et al. (1992).

Age dependent. Male, CBF = 61.1e^{−0.0046 Agc.}; female, CBF = 70.0e^{−0.0049 Agc} (Komatani et al., 1990).

Toyama et al. (1990).

Yamauchi et al. (1993).

Kuwabara et al. (1992).

Subjects

In total, 76 human subjects were studied [59 men and 17 women, aged 6–80 years (mean ± SD, 52 ± 18 years)]. The study group included 10 healthy volunteers (all men, aged 26–46 years, mean ± SD; 36 ± 7 years), 32 patients with cerebrovascular disease [17 cerebral infarction, five hemorrhage, two transient ischemic attack (TIA), six arteriovenous malformation (AVM), and two Moyamoya disease], 14 patients with degenerative neurological disorder (eight Alzheimer's, three Pick, and three Parkinson disease), three patients with brain tumors, nine patients with brain injuries, and seven patients with other disorders (deafness, headache, carotid-cavernous fistula, hypoxic brain, and motor ataxia). Two patient suffered from cardiac disease (one with heart failure), and there were six patients with lung disease. Nineteen subjects were smokers.

[¹²³I]IMP-SPECT study protocol

The [¹²³I]IMP-SPECT study followed the procedure described previously (Iida et al., 1994a, b ). Briefly, 111–222 MBq (3–6 mCi) of [¹²³I]IMP was infused into the antecubital vein at a constant infusion rate for 1 min. At 10 min after the beginning of the [¹²³I]IMP infusion, one arterial blood sample was taken and its whole blood radioactivity concentration counted using a well counter that was cross calibrated to the SPECT scanner. A single SPECT scan was performed at mid-scan time (MST) of 40 min after [¹²³I]IMP administration. After repositioning the subject, another single SPECT scan was performed at MST of 180 min after the [¹²³I]IMP administration. The duration of the SPECT scan varied between institutions, and ranged from 6 to 30 min (Table 1).

Each institution employed its own image-reconstruction protocol. Nine institutions performed attenuation correction (Table 1). The one institution (Tokyo) that did not perform attenuation correction, used fan-beam collimators, without applying geometrical corrections for the fan-bean collimators during reconstruction. This increased the pixel counts at the center of reconstructed images, thus compensating to some extent for attenuation. Akita was the only institution that performed scatter correction using an empirical procedure (subtraction of the scatter projection simulated for a cylindrical object).

All institutions performed cross calibration scans between the SPECT scanner and the well counter system. A uniform phantom, which contained 20–50 MBq of ¹²³I was scanned by the SPECT scanner, and the solution in the phantom was sampled and its radioactivity concentration (cps/g) counted in the well counter. The shape and the size of the phantoms varied between institutions and are summarized in Table 1.

Reference methods

Within a short interval (minimum 30 min, maximum 1 week) before or after the [¹²³I]IMP-SPECT study, all subjects were studied using an independent and established method of rCBF quantitation.

PET H₂¹⁵O reference technique

Five institutions performed H₂¹⁵O PET studies as the reference rCBF method before the [¹²³I]IMP-SPECT study. Akita, Tokyo, and Kyushu employed intravenous H₂¹⁵O ARG (Raichle et al., 1983; Iida et al., 1986; Iida et al., 1992; Kanno et al., 1987; Iida et al., 1988; Iida et al., 1989). Scan duration was 90 s for Akitau and 75 s for Tokyo and Kyushu. A partition coefficient of p = 0.8 ml/ml was assumed in Akita, and p = 1.0 ml/ml for the other two institutions. Continuous arterial blood sampling was performed in all cases for the duration of the study. Delay and dispersion were corrected according to published techniques for Akita (Iida et al., 1986) and the other two institutions (Kanno et al., 1987). Two institutions (Gunma and Kyoto) performed C¹⁵O₂ steady state rCBF measurements according to Frackowiak et al. (1980). Several arterial blood samples were taken in all studies, and whole blood radioactivity concentrations were measured in a well counter calibrated to the PET scanner. A water partition coefficient of p = 1.0 ml/ml was assumed in these two institutions.

[¹²³1]IMP-microsphere reference technique

The [¹²³I]-IMP-microsphere technique originally proposed by Kuhl et al. (1982) was employed in Kyoto, Niigata, and Tohoku. Data were obtained from the same injection of [¹²³I]IMP as for the IMP-TLU and ARG methods. For the [¹²³I]IMP-microsphere calculation, Kyoto used data from a 5-min SPECT scan performed at a MST of 5 min. In Kyoto, frequent, discrete arterial blood samples were obtained manually, separated into lypophilic fractions, and counted. Arterial blood was continuously withdrawn in Niigata and Tohoku for the first 5 and 8 min of the study, respectively. The SPECT MST were 30 min for Niigata and 40 min for Tohoku. SPECT studies were calibrated to the blood sampling period, with planar images collected at the end of the blood sampling period (Kuhl et al., 1982). Radioactivity concentrations of the octanol extracted fraction were counted in both institutions.

[¹³³Xe]SPECT (Kanno-Lassen) technique

Yamagata, Fujita, Ehime, and Niigata employed the [¹³³Xe]SPECT (Kanno-Lassen) technique (Kanno et al., 1979; Celsis et al., 1981) as reference. Following an inhalation of ¹³³Xe for 1 min, dynamic SPECT images were acquired for 4 min, with the exception of Yamagata where data were acquired for 6 min. The end-tidal ¹³³Xe concentration was monitored continuously and the input function obtained from this. In Yamagata and Fugita, the end-tidal partial pressure of CO₂ was monitored during the study, and the calculated rCBF values were corrected for the change of P_aco₂ between the [¹²³I]IMP and the [¹³³Xe]SPECT studies (4% in rCBF per unit mmHg) in Yamagata.

Regions-of-interest (ROIs) definition

ROIs were drawn manually on reconstructed [¹²³I]IMP images for both early and delayed SPECT scan data sets on the reference rCBF images. In each institution, ROIs were drawn on cortical gray matter, white matter, cerebellum, and deep gray matter. Exact definition of ROIs was at the discretion of each institution, but the same shape and size of ROIs were drawn for both the [¹²³I]IMP and the reference rCBF images. Selected ROIs numbered between 15 and 25 in each subject. In total, 1,300 ROIs were obtained from all subjects.

rCBF calculation from [¹²³I]IMP-SPECT

For all ROIs, rCBF values were calculated from the [¹²³I]IMP-SPECT data sets according to both the IMP-TLU and the IMP-ARG algorithms. In both methods, arterial input function was determined by calibrating the previously determined standard input function (Iida et al., 1994a, b ; Munaka et al., 1992; Ito et al., 1992), with the blood sample obtained at 10 min. In the IMP-ARG calculation, the regional distribution volume of [¹²³I]IMP, V_d, was assumed to be 45 ml/ml.

Data analyses

The rCBF values calculated by the IMP-TLU and IMP-ARG methods were compared with those obtained by the reference rCBF study for all ROIs of individual subjects. Calculated rCBF and V_d values were also averaged for four different brain areas in each subject, namely cortical gray matter, white matter, deep gray matter, and cerebellum. As well as comparing IMP-TLU and IMP-ARG against the reference method in individual subjects, data from all subjects were pooled for each institution to provide combined subject comparisons. Pearson's correlation analysis and linear-regression analysis were used to evaluate relationships between two different measurements for each patient and institution. Comparison of multiple data sets were performed using a one-way analysis of variance (ANOVA). Paired t-tests were used to identify specific differences of paired data. Student's t test was used to compared data from different institutions; p < 0.05 was considered statistically significant.

RESULTS

Results of the correlation analysis performed in each institution are summarized in Table 3.

TABLE 3.

Summary of results of the linear regression analysis in each institution

			IMP-TLU versus reference			IMP-ARG versus reference

Institute	Reference method	No. of subjects	r	Slope	Intercept	r	Slope	Intercept
Akita	PET (H₂¹⁵O autoradiography)	5	0.65 ± 0.13^c	0.47 ± 0.19	9.8 ± 4.9	0.67 ± 0.14^c	0.41 ± 0.15	8.8 ± 4.4
Gunma	PET(C¹⁵O₂ steady state)	5	0.60 ± 0.24^c	0.56 ± 0.23	17.5 ± 10.8	0.63 ± 0.16^c	0.68 ± 0.24	14.0 ± 8.2
Tokyo	PET (H₂¹⁵O autoradiography)	4	0.60 ± 0.32^c	0.34 ± 0.19	23.3 ± 5.7	0.70 ± 0.06^c	0.35 ± 0.08	23.9 ± 1.6
Kyoto	PET(C¹⁵O₂ steady state)	4	0.83 ± 0.04^c	0.53 ± 0.23	19.3 ± 7.0	0.80 ± 0.06^c	0.53 ± 0.14	18.6 ± 5.2
Kyushu	PET (H₂¹⁵O autoradiography)	8	0.85 ± 0.09^c	0.51 ± 0.16	17.1 ± 6.1	0.86 ± 0.09^c	0.56 ± 0.18	15.5 ± 5.7
Fujita	Xe Kanno-Lassen	9	0.62 ± 0.23ns for n = 2	0.49 ± 0.21	17.9 ± 8.3	0.61 ± 0.27ns for n = 2	0.53 ± 0.28	17.5 ± 9.5
Ehime	Xe Kanno-Lassen	15	0.50 ± 0.27ns for n = 4	0.62 ± 0.49	8.4 ± 18.5	0.49 ± 0.28ns for n = 4	0.65 ± 0.58	7.5 ± 22.0
Yamagata	Xe Kanno-Lassen	4	0.81 ± 0.17^a	0.57 ± 0.08	18.3 ± 8.6	0.82 ± 0.16^a	0.72 ± 0.15	13.7 ± 8.6
Niigata	Xe Kanno-Lassen	1	0.5^b	0.22	14.5	0.39^a	0.19	13.1
Tohoku	IMP-microsphere	14	0.97 ± 0.04^c	1.02 ± 0.11	–4.8 ± 3.8	0.99 ± 0.02^c	1.11 ± 0.15	–8.8 ± 5.4
Niigata	IMP-microsphere	3	0.98 ± 0.02^c	0.83 ± 0.17	–8.2 ± 4.7	0.98 ± 0.03^c	0.98 ± 0.32	–15.4 ± 10.9
Kyoto	IMP-microsphere	4	0.97 ± 0.01^c	0.82 ± 0.11	2.1 ± 2.5	0.96 ± 0.01^c	0.82 ± 0.13	1.4 ± 3.1

IMP-TLU and IMP-ARG rCBF values are compared against the reference method. Analysis is performed for each subject, and the values of r, slope, and intercept are presented as mean ± SD for each institution. Number of regions-of-interest drawn in each subject ranged between 15 and 25.

NS, not significant.

p < 0.05 for all subjects.

p p < 0.01 for all subjects.

p 0.001 for all subjects.

Five institutions (Akita, Gunma, Tokyo, Kyoto, and Kyushu) showed that the rCBF values obtained by the IMP-TLU and IMP-ARG methods were significantly correlated (p < 0.001) with those from the H₂¹⁵O PET method in all subjects (Table 3). Combined subject correlation was also found to be significant in four institutions for both IMP-TLU and IMP-ARG methods against H₂¹⁵O PET. Correlation for Tokyo combined subjects was not significant (r = 0.42 and 0.41 for IMP-TLU and ARG methods, respectively). IMP-ARG rCBF values for all subjects are plotted against the ¹⁵O PET reference in Fig. 1. Values of rCBF were averaged for each of the four main brain areas in each subject, namely cortical gray matter, white matter, and deep gray matter, and cerebellum.

FIG. 1.

Combined subject correlations of rCBF values calculated by the IMP-ARG method with those by the H₂¹⁵O PET method. Data were obtained from five institutions: Akita (×), Gunma (□), Kyoto (+), Kyushu (▴), and Tokyo (○). Values of rCBF were averaged for each of the four main brain areas in each subject, namely cortical gray matter, white matter, cerebellum, and deep gray matter. Significant correlations were observed in four institutions. Akita: y = 5 + 0.51×, r = 0.74 (p < 0.01); Gunma: y = 20 + 0.49×, r = 0.57 (p < 0.05); Kyoto: y = 19 + 0.49×, r = 0.92 (p < 0.001); and Kyushu: y = 14 + 0.59×, r = 0.81 (p < 0.001). Correlation appeared not to be significant for Tokyo (r = 0.41), probably due to lack of appropriate attenuation correction and, thus, failure of the cross calibration. Significant correlation was also found for all data: y = 15 + 0.54×, r = 0.70 (p < 0.001). Dashed straight line denotes line of identify.

In three institutions (Tohoku, Niigata, and Kyoto), the rCBF values obtained by the IMP-TLU and IMP-ARG methods were shown to be significantly correlated (p < 0.001) with those obtained by the [¹²³I]IMP-microsphere method in all subjects (Table 3). High correlation (p < 0.001) was also found in the combined subjects comparisons in all three institutions for both the IMP-TLU and IMP-ARG methods. IMP-ARG values are plotted against this reference method in Fig. 2. In Kyoto and Tohoku, the rCBF values calculated by the IMP-TLU or IMP-ARG methods were not significantly different from those obtained by the [¹²³I]IMP microsphere method, and the slopes of the regression lines were not significantly smaller than unity. However, in Niigata, the rCBF values obtained by the IMP-TLU and IMP-ARG methods were significantly smaller than those measured by the [¹²³I]IMP-microsphere method (p < 0.001). This appeared to be due to a systematic overestimation of rCBF by the [¹²³I]IMP-microsphere method, rather than underestimation by the IMP-TLU and IMP-ARG methods (see Discussion).

FIG. 2.

Combined subject correlations of rCBF values calculated by the IMP-ARG method with those by the [¹²³I]IMP-microsphere method. Each datum corresponds to the average for each of the four main brain areas in each subject (same as Fig. 1). Data were obtained from Niigata (○), Tohoku (+), and Kyoto (•). Significant correlation was observed for all institutions. Niigata: y = −31 + 1. 3×,r = 0.95 (p < 0.001); Tohoku: y = 1 + 0.86×, r = >0.90 (p < 0.001); Kyoto: y = 0 + 0.85×, r = 0.97 (p < 0.001). Correlation was also significant for all data: y = 2 + 0.80×, r = 0.87 (p < 0.001). The calculated rCBF values by the IMP-ARG methods were significantly (p < 0.001) smaller in Niigata compared with those measured by the [¹²³I]IMP-microsphere method. Dashed straight line denotes line of identify.

The four institutions of Yamagata, Niigata, Fujita, and Ehime demonstrated that the rCBF values obtained by the IMP-TLU and IMP-ARG methods were significantly correlated with those obtained by the ¹³³Xe Kanno-Lassen method, except for six subjects, as shown in Table 3. Averaged p-values for the [¹³³Xe]SPECT method were, however, larger than those from the other two reference methods. Correlations of combined subject data were found to be significant in Fujita, Yamagata, and Niigata, but not in Ehime, for both IMP-TLU and IMP-ARG methods (Fig. 3). It was found in Yamagata that P_aco₂ was greatly different between [¹²³I]IMP and ¹³³Xe studies in two subjects; this was thought to be responsible for the not significant grouped subject correlation of both the IMP-TLU and IMP-ARG methods (r = 0.14 and r = 0.15 for the IMP-TLU and IMP-ARG methods, respectively, before the correction). A correction of 4% in rCBF per unit mmHg in P_aco₂ increased the correlation coefficients (r = 0.51 and r = 0.57 for IMP-TLU and IMP-ARG, respectively), and the correlation became significant (p < 0.05) for both methods. No significant difference was observed in P_aco₂ in Fujita between the [¹²³I]IMP and ¹³³Xe studies, and the grouped correlation was significant without the correction for P_aco₂.

FIG. 3.

Combined subject rCBF values calculated by the IMP-ARG method compared with those by the ¹³³Xe-SPECT (Kanno-Lassen) method. Data correspond to the average for each of the four main brain areas in each subject (same as Fig. 1). Data were obtained from Ehime (crosses), Fujita (open circles), Niigata (filled circles), and Yamagata (filled triangles). Significant correlation was found in Fujita (y = 18 + 0.51×, r = 0.49, p < 0.05), Niigata (y = −3 + 0.62×, r = 0.99, p < 0.001) and Yamagata (y = 10 + 0.80×, r = 0.51, p < 0.05), but not in Ehime (r = 0.08). Correlation was significant for all data (y = 15 + 0.55×, r = 0.42, p < 0.05). The dashed straight line denotes the line of identify. Data for Yamagata were corrected for differences in P_aco₂ between the studies.

Typical functional images obtained from a healthy volunteer (data from Akita) are shown in Fig. 4: images of rCBF and V_d of water obtained by PET with H₂^I5O, images of rCBF and V_d of [¹²³I]IMP obtained by the IMP-TLU method, and an rCBF image obtained by the IMP-ARG method. It can be seen that there is excellent agreement between the IMP-TLU and the IMP-ARG rCBF images. It can also be seen that these two rCBF images compare well with those measured by PET, despite systematic underestimation in gray matter and systematic overestimation in white matter. The V_d of [¹²³1]IMP was relatively homogeneous throughout the whole brain.

FIG. 4.

Functional rCBF images of representative tomographic slice of the brain obtained from a healthy volunteer in Akita; 0–15 water PET, IMP-ARG, and IMP-TLU methods, together with the V_d (distribution volume) image calculated by the IMP-TLU method. The same absolute color scale is used for the three rCBF images. It can be seen that the rCBF images obtained by the IMP-TLU and the IMP-ARG methods show close agreement, but, compared with the PET images, are slightly reduced in high rCBF areas, and somewhat elevated in low rCBF areas.

Figure 5 compares the rCBF values calculated by the IMP-ARG method against those calculated by the IMP-TLU method. There was no significant difference between the two methods.

FIG. 5.

Comparison of calculated rCBF values obtained by the IMP-ARG method with those by the IMP-TLU method. There was no significant difference between the two methods, confirming the validity of fixing the V_d value in the IMP-ARG method.

Values of V_d calculated by the IMP-TLU method (averaged for the four regions in each institution) are summarized in Fig. 6 for all institutions. Calculated V_d values showed no significant difference among the regions. Average V_d values were 47.5 ± 8.6 for cortical gray matter, 45.2 ± 19.2 for white matter, 42.0 ± 11.2 for the cerebellum, and 46.6 ± 11.8 ml/ml for deep grey matter regions. There was no significant difference in the averaged V_d values among the institutions except for Akita (significantly smaller compared with the other nine institutions, p < 0.05).

FIG. 6.

The V_d values obtained by the IMP-TLU method (mean ± 1 SD) in each institution. The V_d value averaged for all subjects was 43 ± 9 ml/ml. The V_d value showed no significant difference among nine institutions, with the exception of Akita with a significantly smaller V_d (p < 0.05).

Normal rCBF values obtained from Gunma are summarized in Table 4. No significant difference was found in rCBF between the IMP-TLU and IMP-ARG methods. Normal rCBF values calculated by the IMP-TLU and IMP-ARG methods were also not significantly different from those measured with the H₂¹⁵O PET method. There was no significant difference in the calculated V_d values among the regions.

TABLE 4.

Summary of rCBF and V_d values obtained from four healthy volunteers at Gunma University (age, 41 ± 6.3 years)

	Cortical gray matter	White matter	Cerebellum	Deep gray matter	Mean ± SD
rCBF by IMP-TLU (ml/min/100 g)	43.4 ± 2.0	27.0 ± 3.1^a	40.7 ± 3.1	38.6 ± 0.5	37.4 ± 6.8
V_d by IMP-TLU (ml/ml)	47.5 ± 8.6	45.2 ± 19.2	42.0 ± 11.2	46.6 ± 11.8	45.4 ± 12.2
rCBF by IMP-ARG (ml/min/100 g)	43.9 ± 3.3	26.8 ± 4.6^a	39.0 ± 2.9	38.5 ± 2.24	37.0 ± 7.2
rCBF by PET (ml/min/100 g)	46.4 ± 4.2	26.1 ± 3.1	44.3 ± 4.0	41.2 ± 2.4	39.5 ± 8.8

p < 0.01 vs. other regions.

DISCUSSION

Validity of IMP-TLU and IMP-ARG methods

This multicenter project demonstrated the validity of recently proposed IMP-TLU and IMP-ARG methods for noninvasive quantitation of rCBF in humans. It was shown in five institutions that these two methods provided rCBF values that were well correlated and quantitatively consistent with those obtained by PET and H₂¹⁵O. Normal rCBF values were also found to be in good agreement with those measured by the H₂¹⁵O PET and normal values published in the literature (Frackowiak et al., 1980; Yamaguchi et al., 1986; Pantano et al., 1985). The high correlation with the [¹²³I]IMP-microsphere technique further confirms the validity of the two methods. The correlation for combined subject data (p < 0.001 for both methods) shown in Fig. 2 clearly demonstrates that a standard input function calibrated with a single blood sample provides reliable results in the heterogeneous patient population encountered in the routine clinical practice.

This study clearly demonstrates that both the IMP-TLU and IMP-ARG methods provide reliable estimates of rCBF in heterogeneous patient populations using a diverse range of SPECT equipment. The techniques do not require fast dynamic SPECT acquisitions, but can utilize data from prolonged and delayed conventional SPECT scans with good counting statistics. Thus, the techniques are generally applicable and not restricted to using specific instrumentation or specialized techniques not generally available. This study also validates the use of a single standardized input function across institutions, eliminating the need for frequent arterial blood sampling or the requirement for each institution to determine its own standardized input function.

The technical procedures of the IMP-TLU and IMP-ARG methods are simple enough for routine clinical use. In this study, the single arterial blood sample was not found too onerous, even in a clinical setting. A recent study by the Tohoku group (Ito et al., 1994) confirmed that arterial blood sampling may be replaced by venous blood sampling, further simplifying the protocol. However, arterial blood sampling would still be needed for measuring the P_aco₂, which greatly influences absolute rCBF.

Validity of fixing V_d in the IMP-ARG method

Previous simulation studies (Iida et al., 1994b) showed that the error sensitivity of rCBF to assumed V_d values was related to the rCBF value. High sensitivity to errors in the assumed V_d value were observed only in regions of relatively high rCBF, with little error being introduced in areas of low CBF. Since V_d appears to be primarily decreased only in areas of low rCBF and is generally normal in high flow areas, the assumption of a fixed V_d was expected to introduce little error in the rCBF value. The good agreement of rCBF values between the IMP-TLU and IMP-ARG methods in our heterogeneous patient population (Fig. 5) confirms the simulation results and the validity of assuming a fixed V_d value in the IMP-ARG method. Since the IMP-ARG method requires only a single SPECT scan, whereas the IMP-TLU method requires two SPECT scans, the IMP-ARG method is the method of choice for routine clinical use, unless the V_d map is also required.

The V_d image, however, provides additional information, which might be clinically and prognostically useful. Odano et al. (1993) demonstrated that rCBF was reversible in areas that exhibited [¹²³I]IMP redistribution in patients with cerebrovascular diseases. V_d, which is independent of flow, relates to the affinity of [¹²³I]IMP in brain tissue (physiological partition coefficient) and the number of cells (amount of brain tissue in units of gram of tissue within a volume element), in a quantitative manner. Quantitative evaluation of V_d might provide identification of the ischemic penumbra in patients with cerebrovascular disease. Further studies are warranted to investigate the clinical significance of the quantitative V_d map.

Comparison with H₂¹⁵O PET method

Despite the high correlation of the two methods with the H₂¹⁵O PET method for all subjects in all five institutions, the slopes of the linear regression lines were significantly smaller than that of the identity line. This may be due to errors in the SPECT reconstruction, i.e., imperfect attenuation correction, scatter, and limited spatial resolution of the scanner [partial volume effect, (PVE)]. Another possible explanation for the systematic underestimation at high flow may be due to the limited first pass extraction fraction of [¹²³I]IMP in the brain. It was shown by the Ehime group (Murase 1991) that rCBF values can be significantly underestimated at high rCBF, e.g., >50 ml/min/100 g. Although [¹²³I]IMP has a relatively high first-pass extraction among the commercially available rCBF SPECT tracers, a correction would be required in quantitation at a high rCBF range (see Fig. 7). Several other factors are also likely to have contributed to the observed difference with the PET technique: misalignment of ROIs between the SPECT and PET images, poorer (different) spatial resolution of SPECT compared with PET, and physiological change of rCBF between the studies.

FIG. 7.

Underestimation of observed rCBF due to the limited first-pass extraction fraction. The underestimation was estimated for four permeability-surface area product (PS) values, corresponding to four rCBF tracers of H₂¹⁵O, [¹²³I]IMP, ^99mTc-hexamethyl-propylene-amineoxime (^99mTc-HMPAO), and ^99mTc-bicisate (^99mTc-ECD), respectively, according to the Renkin-Crone's equation (Renkin, 1959; Crone, 1963). PS values were obtained by Eichung et al. (1974) for H₂¹⁵O, and by the Ehime group for other tracers (Murase et al., 1991).

The combined subject correlations were not significant in Tokyo for both IMP-TLU and ARG methods. This is probably due to the attenuation correction not being performed in the SPECT reconstruction and, thus, failure in the cross calibration. It is likely that the geometrical factor of the fan-beam collimators do not correct for attenuation sufficiently to provide reliable cross calibration. Proper attenuation correction might have produced significant correlation in this institution as well.

Comparison with the [¹²³I]IMP microsphere method

The correlation with the [¹²³I]IMP microsphere method was better than that with the H₂¹⁵O PET method. This is because of the fact that errors associated with the physical factors (attenuation correction error, scatter and PVE) are similarly included in the IMP-TLU, ARG and [¹²³I]IMP microsphere methods. In addition, ROIs were applied to the same SPECT data set for the comparison analyses.

The institution dependence of the slopes of the regression analysis shown in Fig. 2 (Niigata showed significant difference from others) is probably due to [¹²³I] IMP-microsphere protocol differences between institutions. The integration period of the input function corresponded to the SPECT MST in Kyoto (MST and the integration period, 5 min). On the other hand, Tohoku and Niigata performed a delayed SPECT scan at MST = 40 and 30 min, respectively, well after the end of the input function integration periods of 8 min in Tohoku and 5 min in Niigata. The disparate (incomplete) integration period would cause systematic underestimation in the integration of the input function, and, therefore, systematic overestimation is expected in rCBF, with proportionately larger errors for shorter integration periods (i.e., more overestimation in Niigata than Tohoku). Planar images collected at the end of the blood sampling period were used in the IMP-microsphere method to calibrate the delayed SPECT images back to the input function integration period (Kuhl et al., 1982). Despite this correction, a systematic error (underestimation) can be expected in the rCBF estimates when using delayed SPECT scans, as demonstrated by Takahashi (1994). Again, this effect is expected to be more pronounced in Niigata, with the shorter input function integration period of 5 min, compared to 8 min in Tohoku. Another error source might be due to intravascular radioactivity when calibrating SPECT data with the planar image. Blood radioactivity, particularly in extracranial vessels, would be another error source in the microsphere calculation (Ito et al., 1995b), which would overestimate the global CBF in the calibration procedure of using the planar image. These factors can cause systematic errors in Niigata and Tohoku with different magnitudes, and, thus, might be the reason for the institution dependency.

While the procedures for the IMP-ARG and IMP microspheres methods may appear technically very similar and provide similar results, the IMP-ARG method has several advantages over the [¹²³I]IMP microsphere method in a routine clinical setting. Firstly, arterial input function can be estimated from single blood sampling, whereas the [¹²³I]IMP microsphere method requires continuous blood withdrawal from an arterial line, which is more laborious and invasive. Secondly, correction for the clearance of [¹²³I]IMP from the brain is included in the mathematical formulation in the IMP-ARG method, and, thus, acquisition of planar images can be avoided. Thirdly, the use of the standard input function obviates the need for separating the lypophilic fraction.

Comparison with ¹³³Xe-Kanno-Lassen method

Four institutions demonstrated that rCBF values calculated by the IMP-TLU and IMP-ARG methods were significantly correlated with those measured by the ¹³³Xe-Kanno-Lassen technique in most studies (23 of 29 subjects). The correlations were, however, not as good as comparisons against the H₂¹⁵O PET or the [¹²³I]IMP-microsphere techniques. As demonstrated in previous works, this can be attributed to limitations of the ¹³³Xe Kanno-Lassen technique: (a) large error sensitivity to the gray and white matter mixture, i.e., PVE (Iida et al., 1995); (b) systematic overestimation in the area of small structure (Rezai et al., 1988; Iida et al., 1995); (c) large error sensitivity to ambiguous V_d of ¹³³Xe (Gong et al., 1994); and (d) errors in determination of the input function from the end-tidal radioactivity (Hansen et al. 1990; Hazelrig et al., 1985). In this project, large rCBF values (e.g., >60 ml/min/100 g) were unexpectedly obtained by the [¹³³Xe]-SPECT method in several white matter and infarcted regions, possibly caused by the above error sources.

Of the four institutions, Ehime selected the largest size ROIs, and this might be the reason for the nonsignificant correlation in the combined subject comparisons (error sources described above would be maximized with large ROIs).

Correction for the P_aco₂ appeared to be important, particularly in the ¹³³Xe Kanno-Lassen method, as demonstrated by the Yamagata group (Komatani et al., 1990). P_aco₂ was not monitored in two institutions. The mask system for breathing the ¹³³Xe is likely to cause a change in respiration conditions in some studies (Komatani et al., 1990), and, hence, change in P_aco₂, resulting in a global change of rCBF.

The cortical rCBF values obtained by the IMP-TLU and IMP-ARG methods were significantly smaller than those determined by the [¹³³Xe] technique. This is mainly due to the poor spatial resolution of the SPECT camera. The limited resolution causes gray matter and white matter mixing, and, therefore, the [¹²³I]IMP technique provides rCBF values that were nearly equal to an average over the given ROI, and is often considerably smaller than the microscopic rCBF of the cortical gray matter, which may be measured only by an in vitro microsphere experiment. On the other hand, the ¹³³Xe technique is rather insensitive to this underestimation of the radioactivity counts and provides rCBF values that were highly weighted to the high flow areas (e.g., cortical gray matter) (Rezai et al., 1988; Iida et al., 1995). It is worth noting that the rCBF values calculated by the [¹²³I]IMP techniques were underestimated only by 6% compared with the H₂¹⁵O PET method (see Table 4), and the ¹³³Xe provide rCBF values that are highly overestimated compared with those by the H₂¹⁵O PET technique. Another reason is probably due to the smaller first-pass extraction fraction of [¹²³I]IMP compared with ¹³³Xe. This may cause underestimation of rCBF by ∼30% at rCBF = 80 ml/min/100 g with [¹²³I]IMP, whilst ¹³³Xe has almost the maximum first-pass extraction fraction (approximately unity) for a wide rCBF range and, therefore, provides significantly larger rCBF values than does the [¹²³I]IMP technique.

Cross calibration

The cross calibration between the SPECT scanner and the well counter is an important procedure. Despite relatively simple and standard reconstruction methods employed in each institution, systematic errors in image reconstruction and individual differences in sensitivity between SPECT scanners can be eliminated. The nonsignificant correlation in the combined subject comparison in Tokyo is probably due to lack of the attenuation correction. Significant correlation would probably have been achieved with proper attenuation correction. The reason for significantly smaller V_d values in one institution (Akita) compared with the other nine is probably due to the scatter correction performed in Akita. Note that the two institutions of Gunma and Fujita used the same SPECT scanner as Akita, but employed a conventional reconstruction program without scatter correction, and yielded V_d values significantly greater than those obtained in Akita. These results demonstrate that while the methods are not overly sensitive to the exact reconstruction and attenuation correction methods used, large variations from standard reconstruction techniques, such as no attenuation correction or scatter correction, can influence the accuracy of the rCBF and V_d estimates.

Advantages and disadvantages

Similar to reference methods, the IMP-ARG and IMP-TLU methods provide absolute quantitation of rCBF over a range of blood flows in both normal subjects and patients with cerebrovascular disease. However, the studies can be performed with conventional SPECT scanners and do not require frequent arterial blood sampling. This allows the technique to be much more readily used in assessing patients with cerebrovascular disease in routine clinical practice. The ability of this technique in differentiating normal from ischemic region is thought to be similar to that of other SPECT techniques, particularly the IMP microsphere methods. However, further studies are required to fully investigate the clinical utility of rCBF quantitation, and particularly the usefulness of the added information provided by the V_d map from the IMP-TLU method. Similar to other SPECT techniques and, to a lesser extent, PET techniques, there are a number of limitations of the two methods presented in this paper. Quantitative rCBF values obtained by the [¹²³I]IMP SPECT methods were shown to have poorer contrast between high rCBF and low CBF areas compared with those obtained by the H₂¹⁵O PET method. The lower contrast may reduce accurate delineation between normal and ischemic areas. In addition, the normal rCBF value in the cortical gray matter (Table 4) appeared to be ∼6% smaller than those obtained by the H₂¹⁵O method. These are probably due to several limitations: (a) poorer spatial resolution of SPECT (12–18 mm FWHM) compared with PET [8–10 mm full width at half maximum (FWHM)], (b) lower first-pass extraction fraction of [¹²³I]IMP compared with H₂¹⁵O and ¹³³Xe (see Fig. 7), (c) Compton scatter, and (d) errors in the attenuation correction. Scatter and limited spatial resolution reduce the image contrast, hence gray matter counts are likely to be underestimated and white matter counts overestimated (if a uniform cylinder is used for cross calibration, as in this study). It may be possible to reduce these effects with more accurate reconstruction algorithms incorporating accurate scatter corrections and instruments with improved spatial resolution.

Conclusion

The IMP-TLU and ARG methods are found to be reliable and robust in providing quantitative rCBF values for a heterogeneous patient population. The validity of these two methods was also demonstrated with a variety of SPECT systems and reconstruction programs. Of the two methods, the simpler IMP-ARG method would be the choice for routine use in a clinical environment, unless the V_d map is required. Further improvement of the image reconstruction technique would be highly desired in SPECT for more accurate quantitation of rCBF in clinical study.

Footnotes

Acknowledgment:

We wish to particularly thank the staff of the 10 participating institutions for their tremendous help and commitment. We also wish to thank Mr. Stefan Eberl, M.Sc, from Royal Prince Alfred Hospital, Sydney, Australia, for invaluable advice on design of the dataanalysis. Thanks are also given to Mrs. Yuko Miura, M.Sc, from the Department of Radiology and Nuclear Medicine, Research Institute for Brain and Blood Vessels, Akita, Japan, for advice on the statistical analysis, and Mr. Hiromi Nishimura, B.Sc, from Department of Neurosurgery, Research Institute for Brain and Blood Vessels, Akita, Japan, for help with programming. We wish to thank also Miss Mayumi Nakazawa and Mr. Yoshihiro Onishi from Nihon Medi-Physics, Hyogo, Japan, for helping with the secretarial work.

Abbreviations used

References

Celsis

Goldman

Henriksen

Lassen

(1981) A method for calculating regional cerebral blood flow from emission computed tomography of inert gas concentrations. J Comput Assist Tomogr 5:641–645

Chan

(1978) A method for attenuation correction in radionuclide computed tomography. IEEE Trans Nucl Sci NS-25:638–643

Crone

(1963) The permeability of capillaries in various organs determined by use of the indicator diffusion method. Acta Physiol Scand 58:292–305

Eichung

Raichle

Grubb

Ter-Pogossian

(1974) Evidence of the limitations of water as a freely diffusible tracer in brain in the rhesus monkey. Circ Res 35:358–364

Frackowiak

RSJ

Lenzi

Jones

Heather

(1980) Quantitative measurement of regional cerebral blood flow and oxygen metabolism in man using 15O and positron emission tomography; theory, procedure, and normal values. J Comput Assist Tomogr 4:727–736

Gong

Devous

(1994) Dynamic Xe-133 SPECT rCBF calculation using a new convolution algorithm. J Nucl Med 35(suppl):189

Greenberg

Kushner

Rango

Alavi

Reivich

(1990) Validation studies of iodine-123-iodoamphetamine as a cerebral blood flow tracer using emission tomography. J Nucl Med 31:1364–1369

Hansen

Jakobsen

Enevoldsen

Flemming

(1990) Problems in cerebral blood flow calculation using xenon-133 in patients with pulmonary diseases. Stroke 21:745–750

Hashikawa

Matsumoto

Moriwaki

Oku

Okazaki

Uehara

Handa

Kusuoka

Kamada

Nishimura

(1994) Split dose iodine-123-IMP SPECT: Sequential quantitative regional cerebral blood flow change with pharmacological intervention. J Nucl Med 35:1226–1233

10.

Hazelrig

Halsey

Wilson

Wills

(1985) Comparison of external lung monitoring with end-tidal air detection using the 133Xenon inhalation method. Stroke 16:964–968

11.

Hirose

Ikeda

Higashi

Koga

Hattori

Kanno

Miura

Uemura

(1982) A hybrid emission CT-HEADTOME 11. IEEE Trans Nucl NS-29:520–523

12.

Iida

Kanno

Miura

Murakami

Takahashi

Uemura

(1986) Error analysis of a quantitative cerebral blood flow measurement using H215O autoradiography and positron emission tomography: With respect to the dispersion of the input function. J Cereb Blood Flow Metab 6:536–545

13.

Kida

Higano

Tomura

Shishido

Kanno

Miura

Murakami

Takahashi

Sasaki

Uemura

(1988) Evaluation of regional differences of tracer appearance time in cerebral tissues using [15O]water and dynamic positron emission tomography. J Cereb Blood Flow Metab 8:285–288

14.

Iida

Miura

Kanno

Murakami

Takahashi

Uemura

Hirose

Amano

Yamamoto

Tanaka

(1988) Design and evaluation of HEADTOME-IV, A whole-body positron emission tomograph. IEEE Trans Nucl Sci 36:1006–1010

15.

Iida

Kanno

Miura

Murakami

Takahashi

Uemura

(1989) A determination of the regional brain/blood partition coefficient of water using dynamic positron emission tomography. J Cereb Blood Flow Metab 9:874–885

16.

Iida

Kanno

Miura

(1992) Rapid measurement of cerebral blood flow with positron emission tomography. Wiley Chichester (CIBA Foundation Symposium 163, Exploring brain functional anatomy with positron emission tomography) pp 23–42

17.

Iida

Ito

Munaka

Bloomfield

Higano

Murakami

Inugami

Eberl

Aizawa

Kanno

Uemura

(1994) A clinical method to quantitative CBF using a rotating gamma camera and 1–123 amphetamine (IMP) with one blood sampling. Eur J Nucl Med 21:1072–1084

18.

Iida

Ito

Nakazawa

Hatazawa

Nishimura

Onishi

Uemura

(1994) Quantitative mapping of regional cerebral blood flow using [123I]N-isopropyl-p-iodoamphetamine (IMP) and single photon emission tomography. J Nucl Med 35:2019–2030

19.

Iida

Narita

Ardekani

Hatazawa

Nakazawa

Kanno

Uemura

(1995) Evaluation of partial volume effect in quantitative measurement of regional cerebral blood flow in single photon emission computed tomography-effects of limited spatial resolution and first-pass extraction fraction. Jpn J Nucl Med 32:155–162

20.

Ito

Iida

Murakami

Bloomfield

Miura

Okudera

Lnugami

Ogawa

Hatazawa

Fujita

Shimoseg-Awa

Kanno

Fukuda

Uemura

(1992) A method for measurement of regional cerebral blood flow using N-isopropyl-p-[123I]iodoamphetamine(123I]-IMP)SPECT; two scans with one point blood sampling technique. Jpn J Nucl Med 29:1193–1200

21.

Ito

Akaizawa

Goto

Koyama

Kawashima

Ono

Yoshioka

Ishii

Knoshita

Inukai

Abe

Atsumi

Sato

Fukuda

(1994) A simple method for measurement of cerebral blood flow using [1231]IMP SPECT with calibrated standard input function by one point blood sampling; validation of calibration by one point venous blood sampling as a substitute for arterial blood sampling. Jpn J Nucl Med 31:1071–1076

22.

Ito

Iida

Bloomfield

Murakami

Inugami

Kanno

Fukuda

Uemura

(1995a) Rapid calculation of regional cerebral blood flow and distribution volume using iodine-123-iodoamphetamine and dynamic SPECT. J Nucl Med 36:531–536

23.

Ito

Ishii

Atsumi

Kinoshita

Kawashima

Ono

Yoshioka

Iida

Uemura

Fukuda

(1995b) Error analysis of Table-Look-Up method for cerebral blood flow measurement by 123I-IMP brain SPECT. Comparison with conventional microsphere model method. Ann Nucl Med 9:75–80

24.

Kanno

Lassen

(1979) Two methods for calculating regional cerebral blood flow from emission computed tomography of inert gas concentrations. J Comput Assist Tomogr 3:71–76

25.

Kanno

Miura

Yamamoto

Iida

Murakami

Takahashi

Uemura

(1985) Design and evaluation of a positron emission tomograph: HEADTOME III. J Comput Assist Tomogr 9:931–939

26.

Kanno

Iida

Miura

Murakami

Takahashi

Sasaki

Inugami

Shishido

Uemura

(1987) A system for cerebral blood flow measurement using an H215O autoradiographic method and positron emission tomography. J Cereb Blood Flow Metab 7:143–153

27.

Komatani

Yamaguchi

Kera

(1990) Cerebral blood flow and brain size in healthy subjects-sensitivity to age and gender. Brain Nerve 42:889–893

28.

Komatani

(1990) Fluctuation of end-tidal CO2 concentration at rCBF measurement by Xe-133 inhalation and the necessity of its correction for CO2. J Nucl Med 27:65–68

29.

Kouris

Jarritt

Costa

Ell

(1992) Physical assessment of the GE/CGR neurocam and comparison with a single rotating gamma-camera. Eur J Nucl Med 19:236–242

30.

Kuhl

Barrio

Huang

Selin

Ackermann

Lear

Lin

Phelps

(1982) Quantifying local cerebral blood flow by N-isopropyl-p[I-123]iodo amphetamine (IMP) tomography. J Nucl Med 23:196–203

31.

Kuwabara

Khiya

Otsuka

Masuda

Ichimiya

Fujishima

(1992) Cerebrovascular responsiveness to hypercapnia in Alzheimer's dementia and vascular dementia of Binswanger. Stroke 23:594–598

32.

Kypros

Clarke

Jarritt

Townsend

Thomas

(1993) Physical performance evaluation of the Toshiba GCA-9300A triple-headed system. J Nucl Med 34:1778–1789

33.

Munaka

Iida

Murakami

(1992) Quantification of regional cerebral blood flow (rCBF) measurement with one point sampling by [123I]IMP SPECT. Jpn J Nucl Med 29: 263–267

34.

Murase

Tanada

Kawamura

Mogami M

Miyagawa

Yamada

Higashino

Iio

Hamamoto

(1990) Validity of microsphere model in cerebral blood flow measurement using JV-isopropyl-p-(l-123) iodoamphetamine. Med Phys 17: 79–83

35.

Murase

Tanada

Inoue

Ochi

Fujita

Sakaki

Kimura

Hamamoto

(1991) Measurement of the blood-brain-barrier permeability of I-123 IMP, Tc-99m HMPAO and Tc-99m ECD in the human brain using compartment model analysis and dynamic SPECT [abstract], J Nucl Med 32:911

36.

Nishizawa

Tanada

Yonekura

Fujita

Mukai

Saji

Fukuyama

Miyoshi

Harada

Ishikawa

Kikuchi

Konishi

(1989) Regional dynamics of N-isopropyl-(123I)p-iodo-amphetamine in human brain. J Nucl Med 30: 150–156

37.

Odano

Tsuchiya

Nishihara

Sakai

Abe

Tanaka

(1993) Regional cerebral blood flow measured with N-isopropyl-p-[123I]iodoamphetamine and its redistribution in ischemic cerebrovascular disease. Stroke 24:1167–1172

38.

Odano

Ohkubo

Takahashi

Higuchi

(1994) A new method of regional cerebral blood flow measurement using one-point arterial sampling based on the microsphere model with N-isopropyl-p-[123I]-iodoamphetamine SPECT. Nucl Med Commun 15:560–564

39.

Pantano

Baron

Crouzel

Collard

Sirou

Samson

(1985) The 15O continuous-inhalation method: Correction for intravascular signal using C15O. Eur J Nucl Med 10:387–391

40.

Podreka

Baumgartner

Suess

Muller

Brucke

Lang

Holzner

Steiner

Deecke

(1989) Quantification of regional cerebral blood flow with IMP-SPECT: Reproducibility and clinical relevance of flow values. Stroke 20:183–191

41.

Raichle

Martin

Herscovitch

Markham

(1983) Brain blood flow measured with intravenous H215O. II. Implementation and validation. J Nucl Med 24:790–798

42.

Renkin

(1959) Transport of potassium-42 from blood to tissue in isolated mammalian skeletal muscles. Am J Physiol 197: 1205–1210

43.

Rezai

Kirchner

Armstrong

Ehrhardt

Heistad

(1988) Validation studies for brain blood flow assessment by radioxenon tomography. J Nucl Med 29:348–355

44.

Sorenson

(1974) Quantitative Measurement of Radioactivity In Vivo by Whole Body Counting ( Hine

Sorenson

, eds., p. 311, Academic Press, New York

45.

Takahashi

(1994) Quantifying regional cerebral blood flow with N-isopropyl-p-[123I]iodoamphetamine by ring-type single photon emission computed tomography: Validity of method to estimate early reference value by means of regional brain time-activity curve. Ann Nucl Med 8:253–258

46.

Takeshita

Maeda

Nakane

Toyama

Sakakibara

Komai

Takeuchi

Koga

Ono

Nakagawa

(1992) Quantitative measurement of regional cerebral blood flow using N-isopropyl-(iodine-123)p-iodoamphetamine and single-photon emission computed tomography. J Nucl Med 33: 1741–1749

47.

Toyama

Takeshita

Takeuchi

Anno

Ejiri

Maeda

Katada

Koga

Ishiyama

Kanno

Yamaoka

(1990) Cerebral hemodynamics in patients with chronic obstructive carotid disease by rCBF, rCBV, and rCBV/rCBF ratio using SPECT. J Nucl Med 31:55–60

48.

Winchell

Horst

Braun

Oldendorf

Hattner

Parker

(1980) N-isopropyl[123I]p-iodoamphetamine: Single-pass brain uptake and washout; binding to brain synaptosomes; and localization in dog and monkey brain. J Nucl Med 21:947–952

49.

Yamaguchi

Kanno

Uemura

Shishido

Inugami

Ogawa

Murakami

Suzuki

(1986) Reduction in regional cerebral metabolic rate of oxygen during human aging. Stroke 17:1220–1228

50.

Yamauchi

Fukuyama

Ogawa

Ouchi

Kimura

(1993) Hemodilution improves cerebral hemodynamics in internal carotid artery occlusion. Stroke 24:1885–1890

51.

Yokoi

Iida

Ito

Kanno

(1993) A new graphic plot analysis for cerebral blood flow and partition coefficient with iodine-123-iodoamphetamine and dynamic SPECT validation studies using oxygen-15-water and PET. J Nucl Med 34:498–505

52.

Yonekura

Fujita

Nishizawa

Mukai

Koide

Yamamoto

Tamaki

Konishi

Hirose

(1989) Multidetector SPECT scanner for brain and body: System performance and applications. J Comput Assist Tomogr 13:732–740

53.

Yonekura

Nishizawa

Mukai

Iwasaki

Fukuyama

Ishikawa

Tamaki

Konishi

(1993) Functional mapping of flow and back-diffusion rate of N-isopropyl-p-iodoamphetamine in human brain. J Nucl Med 34:839–844

A Multicenter Validation of Regional Cerebral Blood Flow Quantitation Using [ 123 I]Iodoamphetamine and Single Photon Emission Computed Tomography

Abstract

Keywords

METHODS

Institutions

Subjects

[123I]IMP-SPECT study protocol

Reference methods

PET H215O reference technique

[1231]IMP-microsphere reference technique

[133Xe]SPECT (Kanno-Lassen) technique

Regions-of-interest (ROIs) definition

rCBF calculation from [123I]IMP-SPECT

Data analyses

RESULTS

DISCUSSION

Validity of IMP-TLU and IMP-ARG methods

Validity of fixing Vd in the IMP-ARG method

Comparison with H215O PET method

Comparison with the [123I]IMP microsphere method

Comparison with 133Xe-Kanno-Lassen method

Cross calibration

Advantages and disadvantages

Conclusion

Footnotes

Acknowledgment:

Abbreviations used

References

[¹²³I]IMP-SPECT study protocol

PET H₂¹⁵O reference technique

[¹²³1]IMP-microsphere reference technique

[¹³³Xe]SPECT (Kanno-Lassen) technique

rCBF calculation from [¹²³I]IMP-SPECT

Validity of fixing V_d in the IMP-ARG method

Comparison with H₂¹⁵O PET method

Comparison with the [¹²³I]IMP microsphere method

Comparison with ¹³³Xe-Kanno-Lassen method