Quantitation of regional cerebral blood flow corrected for partial volume effect using O-15 water and PET: I. Theory, error analysis, and stereologic comparison

Regional differences in distribution volume of I-123 IMP in the human brain: effect on CBF calculated by ARG method

OBJECTIVE

Two methods of quantitating cerebral blood flow (CBF) with iodine-123-labeled N-isopropyl-p-iodoamphetamine (I-123 IMP) and a two-compartment model had been proposed; one is the table look-up (TLU) method and the other is the autoradiographic (ARG) method. The TLU method provides values of the cerebral blood flow (CBF) values and distribution volume of I-123 IMP (Vd) independently. In the ARG method, a fixed Vd is applied for the entire brain to calculate CBF. Our purpose was to evaluate regional differences in Vd in the human brain, or possible effects of regional differences in Vd on CBF calculated by the ARG method.

METHODS

In the present study, two SPECT scans were acquired from each of eight normal subjects (aged 44.0 +/- 16.7) at 40 min and 180 min of mid-scan-time after intravenous 1 min infusion of 111 MBq IMP. A single arterial blood sampling was performed 10 min after the IMP infusion. All images were anatomically normalized and analyzed with SPM99 and Matlab. We generated CBF and Vd images for each subject by the TLU method and evaluated differences in Vd among brain structures. We subsequently generated another set of CBF images by the ARG method and examined differences between CBF calculated by the TLU method and that by the ARG method.

RESULTS

Significant main effects of subject and brain structure in Vd were observed (two-way ANOVA). Vd values were higher in the deep gray matter than in the cerebral cortical regions. Among the cerebral cortical regions, no significant difference in Vd was observed. In spite of the significant differences in Vd among the brain structures, the voxel-by-voxel analyses as well as the ROI analyses revealed no statistically significant difference between CBF calculated by the TLU method and that by the ARG method.

CONCLUSIONS

Although regional differences in Vd were observed, the present results support the assumption that a fixed Vd does not cause significant error in the calculation of CBF by the ARG method.

Evaluation of a commercial PET tomograph-based system for the quantitative assessment of rCBF, rOEF and rCMRO2 by using sequential administration of 15O-labeled compounds

The purpose of this study was to develop a reliable and practical strategy that generates quantitative CBF and OEF maps accurately from PET data sets obtained with 15O-tracers. Sequential sinogram data sets were acquired after the administration of 15O-tracers, and combined single-frame images were obtained. The delay time between sampled input function and the brain was estimated from the H2(15)O study with the whole brain and the arterial time-activity curves (TACs). The whole-brain TACs were obtained from the reconstructed images (image-base method) and the sinogram data (sinogram-base method). Six methods were also evaluated for the dead-time and decay correction procedures in the process of generating a single-frame image from the dynamic sinogram. The estimated delay values were similar with both the sinogram-based and image-based methods. A lumped correction factor to a previously added single-frame sinogram caused an underestimation of CBF, OEF and CMRO2 by 16% at maximum, as compared with the correction procedure for a short sinogram. This suggested the need for a dynamic acquisition of a sinogram with a short interval. The proposed strategy provided an accurate quantification of CBF and OEF by PET with 15O-tracers.

Quantitative comparison of the bolus and steady-state methods for measurement of cerebral perfusion and oxygen metabolism: positron emission tomography study using 15O-gas and water

To evaluate a new simplified bolus method for measurement of cerebral perfusion and metabolism, the parametric images with that method were compared with those obtained from the conventional steady-state method with 15O-gas. The new method also provided images of arterial blood volume (V0), which is a different parameter from cerebral blood volume (CBV) obtained using a C15O technique. Seven healthy volunteers and 10 patients with occlusive cerebrovascular diseases underwent positron emission tomography (PET) scans with both methods. Three-weighted integration was applied to calculate regional cerebral blood flow (rCBF) and regional cerebral metabolic rate of oxygen (rCMRO2) in the bolus method. Global and regional CBF and CMRO2 in volunteers were compared between the two methods and used as control data. Regional values in patients also were evaluated to observe differences between the bilateral hemispheres. Both rCBF and rCMRO2 were linearly well correlated between the two methods, although global difference in CMRO2 was significant. The difference in each parametric image except for V0 was significant between the bilateral hemispheres in patients. The bolus method can simplify oxygen metabolism studies and yield parametric images comparable with those with the steady-state method, and can allow for evaluation of V0 simultaneously. Increase in CBV without a change in V0 suggested the increase might mainly be caused by venous dilatation in the ischemic regions.

A computationally efficient algorithm for determining regional cerebral blood flow in heterogeneous tissues by positron emission tomography

Inclusion of brain tissues with different rates of blood flow and metabolism within a voxel or region of interest is an unavoidable problem with positron emission tomography due to its limited spatial resolution. Because regional cerebral blood flow (rCBF) is higher in gray matter than in white matter, the partial volume effect leads to underestimation of rCBF in gray matter when rCBF in the region as a whole is determined. Furthermore, weighted-average rCBF itself is underestimated if the kinetic model used in the analysis fails to account for the tissue heterogeneity. We have derived a computationally efficient method for estimating both gray matter and weighted-average rCBF in heterogeneous tissues and validated the method in simulation studies. The method is based on a model that represents a heterogeneous tissue as a weighted mixture of two homogeneous tissues. A linear least squares algorithm is used to estimate the model parameters.

Quantitation of regional cerebral blood flow corrected for partial volume effect using O-15 water and PET: II. Normal values and gray matter blood flow response to visual activation

One of the most limiting factors for the accurate quantification of physiologic parameters with positron emission tomography (PET) is the partial volume effect (PVE). To assess the magnitude of this contribution to the measurement of regional cerebral blood flow (rCBF), the authors have formulated four kinetic models each including a parameter defining the perfusable tissue fraction (PTF). The four kinetic models used were 2 one-tissue compartment models with (Model A) and without (Model B) a vascular term and 2 two-tissue compartment models with fixed (Model C) or variable (Model D) white matter flow. Furthermore, rCBF based on the autoradiographic method was measured. The goals of the study were to determine the following in normal humans: (1) the optimal model, (2) the optimal length of fit, (3) the model parameters and their reproducibility, and (4) the effects of data acquisition (2D or 3D). Furthermore, the authors wanted to measure the activation response in the occipital gray matter compartment, and in doing so test the stability of the PTF, during perturbations of rCBF induced by visual stimulation. Eight dynamic PET scans were acquired per subject (n = 8), each for a duration of 6 minutes after IV bolus injection of H2(15)O. Four of these scans were performed using 2D and four using 3D acquisition. Visual stimulation was presented in four scans, and four scans were during rest. Model C was found optimal based on Akaike's Information Criteria (AIC) and had the smallest coefficient of variance after a 6-minute length of fit. Using this model the average PVE corrected rCBF during rest in gray matter was 1.07 mL x min(-1) x g(-1) (0.11 SD), with an average coefficient of variance of 6%. Acquisition mode did not affect the estimated parameters, with the exception of a significant increase in the white matter rCBF using the autoradiographic method (2D: 0.17 mL x min(-1) x g(-1) (0.02 SD); 3D: 0.21 mL x min(-1) x g(-1) (0.02 SD)). At a 6-minute fit the average gray matter CBF using Models C and D were increased by 100% to 150% compared with Models A and B and the autoradiographic method. There were no significant changes in the perfusable tissue fraction by the activation induced rCBF increases. The largest activation response was found using Model C (median = 39.1%). The current study clearly demonstrates the importance of PVE correction in the quantitation of rCBF in normal humans. The potential use of this method is to cost-effectively deliver PVE corrected measures of rCBF and tissue volumes without reference to imaging modalities other than PET.