PERFUSION IMAGING USING MR CONTRAST AGENTS

The use of exogenous MR contrast agents for the study of cerebral perfusion has long been recognized (Villringer et al., 1988; Rosen et al., 1990). These contrast agents also provide information about different physiologic parameters related to CBF, cerebral blood volume (CBV) and the mean transit time (MTT) of blood through a volume of tissue. This technique, usually referred to as DSC-MRI, involves the injection of a bolus of contrast agent (typically 0.1 to 0.3 mmol/kg body weight) and the rapid measurement of the MRI signal loss caused by spin dephasing (i.e., decrease in T₂ and T₂*) during its fast passage through the tissue (Villringer et al., 1988). Although the vascular space is a small fraction of the total tissue volume (5% in the human brain), the compartmentalization of contrast agent within the intravascular space leads to a significant transient drop in signal. This susceptibility effect extends beyond the vascular space (Gillis and Koenig, 1987; Villringer et al., 1988) and dominates over the T₁ relaxation enhancement. However, in regions with blood-brain barrier (BBB) disruption, the contrast agent does not remain intravascular, resulting in a decrease in the susceptibility effect and a corresponding increase in the importance of T₁ enhancement.

Since the transit time of the bolus through the tissue is only a few seconds, high temporal resolution imaging is required to obtain sequential images during the wash in and wash out of the contrast material and, therefore, resolve the first pass of the tracer (Fig. 1). The practicability of DSC-MRI was greatly increased with the widespread availability of echo planar imaging (EPI). This fast imaging technique allows an improved characterization of the passage of the bolus and facilitates the acquisition of multislice data, thus increasing the regional coverage of the technique.

Figure 1.

Sequential spin-echo planar images during the first passage of a bolus of contrast agent (repetition time [TR] = 2 seconds) together with the signal intensity time course (in arbitrary units) for three regions of interest (ROI): ROI 1, right basal ganglia; ROI 2, right occipital white matter; ROI 3, peripheral branch of the right middle cerebral artery (MCA). The top left and bottom right images correspond to t = 16 seconds and 34 seconds, respectively. The images show the signal intensity decrease associated with the passage of the bolus. Three different periods can be identified in the time course data: the baseline (before the arrival of the bolus), the first passage of the bolus, and the recirulation period (in this case, a second smaller peak, more clearly seen the arterial region [ROI 3]).

Full figure and legend (292K)

Basic theory

The model used for perfusion quantification is based on the principles of tracer kinetics for nondiffusable tracers (Zierler, 1962; Zierler, 1965; Axel, 1980) and relies on the assumption that in the presence of an intact BBB, the contrast material remains intravascular.

Considering a bolus of contrast material injected into the tissue, the concentration C_VOI(t) of tracer within a given volume of interest (VOI) at a later time t can be defined in terms of three functions (Axel, 1995; Østergaard et al., 1996b):

• Transport function, h(t): Probability density function of transit time t through the VOI following an ideal instantaneous unit bolus injection. This reflects the distribution of transit times through the voxel, which is dependent on the vascular structure and flow.
• Residue function, R(t): Fraction of tracer still present in the VOI at time t following an ideal instantaneous unit bolus injection. By definition, R(t) = [1 - ^t₀h()d], where the integral term represents the fraction that has left the VOI, and R(t = 0) = 1, that is, all of the tracer is present at time t = 0. Also, notice that this residue function represents the fractional tissue concentration, that is, R(t) C_VOI(t).
• Arterial input function (AIF), C_a(t): Concentration of contrast agent in the feeding vessel to the VOI at time t.

With these definitions, the concentration C_VOI(t) can be written in terms of a convolution of the residue function and the arterial input function (Østergaard et al., 1996b): (Eq. 1) where F_VOI is the CBF in the VOI, is the density of brain tissue (needed to provide the correct flow units), and k_H = (1 - H_art)/(1 - H_cap) accounts for the difference in hematocrit (H) between capillaries and large vessels, since only the plasma volume is accessible to the tracer. This expression can be interpreted by considering the AIF as a superposition of consecutive ideal boluses C_a()d injected at time . For each ideal bolus, the concentration still present in the VOI at time t will be proportional to C_a()R(t - )d, and the total concentration C_VOI(t) will be given by the sum (or integral) of all of these contributions.

Therefore, to calculate CBF, Eq. 1 must be deconvolved (see later) to isolate F_VOI R(t) and the flow obtained from its value at time t = 0 (initial point of the deconvolved response curve). However, in practice, the measured AIF may be affected by both a delay and dispersion between the artery where it was estimated and the more peripheral tissue (VOI). As described by Østergaard et al. (1996b), these effects introduce spreading of the deconvolved curve, and instead of the initial point (which would underestimate the flow) the maximum of the curve should be used. To minimize these effects, the AIF should be measured as close as possible to the VOI.

Another important physiologic parameter that can be obtained by using DSC-MRI is CBV. For an intact BBB, CBV is proportional to the normalized total amount of tracer, (Eq. 2)

The normalization to the AIF accounts for the fact that, independent of the CBV, if more tracer is injected, a greater concentration will reach the VOI. Notice that because the arterial input to an organ is ultimately derived from a common source, relative CBV (relCBV) can be measured without any knowledge of the AIF (Rosen et al., 1990). This is one reason for the popular use of regional CBV maps in early DSC studies.

The third physiologic parameter, MTT, is the average time required for any given particle of tracer to pass through the tissue after an ideal bolus injection. By using the definition of the transport function, MTT can be written as the ratio of the first moment of the transport function to its zeroth moment 0: (Eq. 3)

In the classic outflow experiment (where the concentration in the venous output, C_OUT(t), is measured), the MTT can be calculated from the first moment of the tracer concentration (Axel, 1995). However, as pointed out by Weisskoff et al. (1993), this MTT is distinct from the first moment of C_VOI(t). The latter depends on the topologic makeup of the vasculature. Therefore, calculation of MTT cannot be performed without first solving Eq. 1.

Once CBF and CBV are known, MTT also can be calculated directly by using the central volume theorem (Stewart, 1894; Meier and Zierler, 1954): (Eq. 4)

Conversely, some studies have used the central volume theorem to estimate a "perfusion index" ("CBF_i") from the ratio of the CBV to the first moment of concentration-time curve. However, apart from the previously mentioned dependency on the underlying vascular structure, this "perfusion index" is influenced by the shape of the bolus, since the first moment contains contributions from both the MTT and the first moment of the AIF (Axel, 1995). The shorter the MTT, the more important this extra contribution becomes, and a simple linear relation between F_VOI and "CBF_i" is not expected, particularly at high flow values (Wittlich et al., 1995). In addition to the assumptions of an intact BBB and stable flow during the measurement, some extra assumptions are necessary for the model described earlier to be valid: (1) that the contrast agent used is really a tracer, that is, it has no effect on the CBF (Doerfler et al., 1997) and has negligible volume itself; (2) that recirculation of the tracer is negligible (see later); (3) that the change in T₁ relaxation is negligible; and (4) that dispersal and delay of the bolus as it reaches the VOI are not significant. This last assumption is likely to be invalid during ischemia and, as discussed later, can produce an underestimate of the calculated perfusion.

Quantification issues

One of the fundamental points regarding the use of the model described earlier is the need to convert changes in regional MR signal intensity to contrast agent concentration time curves; that is, since MRI is not able to directly measure the tracer concentration, it must be measured indirectly through its effect on signal intensity. It has been shown, both empirically (Villringer et al., 1988; Rosen et al., 1990; Hedehus et al., 1997) and using Monte Carlo simulations (Fisel et al., 1991; Weisskoff et al., 1994; Boxerman et al., 1995; Kennan et al., 1994), that the concentration of the tracer in the VOI is approximately proportional to the change observed in the relaxation rate R₂ (= 1/T₂) or R₂* in normally perfused tissue. By assuming a single exponential relation, the change in relaxation rate (R₂) can be obtained from the change in signal intensity from the baseline signal before contrast administration (S₀). (Eq. 5) where S_VOI(t) is the signal intensity measured in the VOI at time t, and TE is the echo time of the sequence. The proportionality constant _VOI is a constant that depends on the tissue, the contrast agent, the field strength, and the pulse sequence parameters. An equivalent relation is assumed for the concentration of the tracer in the arterial input (with a proportionality constant _art).

Therefore, this technique allows, in principle, the quantification of regional CBF, CBV, and MTT. However, because of practical and theoretical difficulties, quantification using more simplistic approaches has been used commonly. There are three main different approaches to the quantification of data obtained using DSC-MRI: (1) quantification of absolute CBF; (2) quantification of relative CBF (relCBF); and (3) quantification using summary parameters, such as time-to-peak (TTP), bolus arrival time (BAT), maximum peak (MP), full width at half maximum, peak area, first moment of the peak (C⁽¹⁾_VOI), and parameters defined from the coefficients of the Gamma-variate function (see later).

The first two approaches require the deconvolution of the AIF and, therefore, are technically demanding. An accurate characterization of the AIF also is required (see later). The use of summary parameters, on the other hand, does not require the deconvolution of the measured signal and has been widely used, both in animal and human studies because of its much simpler and less time-consuming processing (see Applications of Perfusion Imaging Using MR Contrast Agents). However, it has disadvantages compared with the other approaches, since there is no simple relation between the summary parameters and CBF. They also depend on other factors such as CBV, MTT, bolus volume and shape, injection rate, and cardiac output. This makes the interpretation of summary parameter maps less straightforward, and in general, it depends on assumptions about the underlying vascular structure (Weisskoff et al., 1993; Gobbel et al., 1991). Furthermore, accurate comparison between subjects, or repeated measurements in follow-up studies, is not possible. However, when no information about the AIF is accessible, summary parameters are the only quantitative option and, in many cases, they can be useful in helping to distinguish between various pathologic and physiologic situations.

Deconvolution methods. Several methods to deconvolve Eq. 1 have been proposed. Østergaard et al. (1996b) have compared the performance of some of them, both using Monte Carlo simulations and experimental data. They analyzed two main categories of deconvolution approaches: model-dependent (Jacquez, 1972) and model-independent techniques (Gobbel and Fike, 1994; Rempp et al., 1994). In the first, an empirical analytical expression is chosen to describe the tracer vascular retention, that is, a specific analytical expression for R(t) is assumed. The most common model is to consider the vascular bed as one single, well-mixed compartment, and therefore R(t) = exp(-t/MTT). In the second, the model-independent approach, both CBF and R(t) are determined by nonparametric deconvolution, that is, the residue function is also treated as an unknown variable. This approach can be subdivided in two categories. In the transform approach, the convolution theorem of the Fourier transform is used to deconvolve Eq. 1: (Eq. 6) where ^-1 denotes the inverse of the Fourier transform . In the other category, the algebraic approach, Eq. 1 is rewritten as a matrix equation and solved either by a regularization approach or by using singular value decomposition techniques (Press et al., 1992). For a more detailed description of these methods, refer to Østergaard et al. (1996b) and references therein.

The results obtained by Østergaad et al. (1996b) indicate that model-dependent approaches predict correct absolute CBF values only if the actual residue function is sufficiently described by the assumed model. It also can yield a reasonable estimate of relative flow when the shape of R(t) is relatively uniform across the brain. However, this is not likely to be the case in pathologic tissue or in tissue with altered hemodynamics. In the case of the Fourier transform approach, the estimate of CBF was shown to be biased by the CBV, producing underestimates in cases of high flow and short MTT (relative to the sampling rate). The nonparametric deconvolution using regularization was found to give accurate flow values over a wide range, although it also showed some dependency on the vascular volume. Finally, their results showed that the singular value decomposition technique gave, in general, good accuracy, independent of the underlying vascular structure, R(t), and volume, CBV, even for a signal-to-noise ratio (SNR) typical of EPI.

Other deconvolution approaches have been adopted, such as deconvolution using orthogonal polynomials (Schreiber et al., 1998 and references therein), where the AIF is approximated by a subset of an orthogonal function system, and Eq. 1 is solved analytically, or the use of a maximum likelihood method, where the equation is solved iteratively (Vonken et al., 1998). However, a complete description of deconvolution methods is beyond the scope of this review.

Absolute measurement of CBF. To obtain CBF in absolute units (mL 100 g^-1 min^-1), different approaches have been taken. In one of these approaches, fixed values for the relevant proportionality constants are assumed (Rempp et al., 1994; Schreiber et al., 1998; Gückel et al., 1996). For example, uniform values for hematocrit in the capillaries and large vessels are assumed (typically, H_art = 0.45 and H_cap = 0.25), 1.04 g/mL is assumed for the density of brain tissue (), and a single uniform proportionality constant for tissue and the AIF is assumed in Eq. 5, that is, _VOI = _art for any tissue. With these assumptions, absolute CBF values, which agree with published data using other techniques, have been reported in normal subjects (Rempp et al., 1994; Schreiber et al., 1998). However, the validity of these assumptions in pathologic conditions, or a complete analysis of the error introduced in this way, remains to be investigated. Table 1 summarizes the flow values obtained in quantification studies.

Table 1 - Studies of flow quantification using continuous and pulsed arterial spin labeling and dynamic susceptibility contrast MRI techniques.

Full table

An alternative approach for the calculation of absolute CBF values involves cross-calibration with another technique (Østergaard et al., 1998a, 1998b; Wittlich et al., 1995). In one of these studies, CBF measurements were performed using DSC-MRI and PET in the same animals (normocapnic and hypercapnic pigs) (Østergaard et al., 1998b). By comparing average flow rates through all regions, a common empirical conversion factor (to absolute flow units) was obtained for DSC measurements. Recently, this cross-calibration also was performed on normal human subjects (Østergaard et al., 1998a). However, the validity of the same conversion factor in patients with altered hemodynamics and for different tissue types remains to be shown. A similar approach using DSC-MRI and autoradiography was used to calculate absolute CBF in rats during middle cerebral artery occlusion (MCAO) (Wittlich et al., 1995). However, in this study, the AIF was not measured, and the MTT was approximated by C⁽¹⁾_VOI.

Notice that since the same proportionality constants are assumed for the normalization of CBF and CBV, the ratio of CBV/CBF will be an absolute measure of MTT (Østergaard et al., 1996b).

Recirculation or leakage of the tracer. The model previously described is only for the first pass of the tracer. However, the measured C_VOI(t) can include contributions from recirculation (which can be recognized as a second, smaller concentration peak or an incomplete return to baseline after the first pass; Fig. 1). Therefore, these contributions must be eliminated before any perfusion calculation. This can be achieved, either by considering the curve only until the appearance of the recirculation, or by fitting a portion of the concentration time curve to an assumed bolus shape function, typically a gamma-variate function, (Starmer and Clark, 1970; Berninger et al., 1981).

In a Monte Carlo study of the CBV dependency on the SNR of the base images, Boxerman et al. (1997) showed that numerical integration and -fitting gave similar results for DSC images with SNR greater than 20 to 30. However, for lower SNR, the variability introduced by the covarying parameters in the -fitting yielded CBV maps with decreased SNR. Although similar conclusions are expected to be valid for CBF, further simulations are needed to extend these findings.

The presence of a damaged BBB results in leakage of the tracer to the extracellular space. This leads to a decrease in the susceptibility effects and to an increase in T₁ relaxation rate, which opposes the signal losses from T₂ (usually seen as a positive slope after the return of the peak to baseline). These effects can be minimized with a proper choice of pulse sequence parameters, and, in cases of a moderate BBB disruption, a technique to correct for T₁ enhancement in CBV maps has been proposed (Weisskoff et al., 1994). An alternative in cases of a disrupted BBB is the use of a dysprosium-based contrast agent (Dy-type) instead of a gadolinium-type (Gd-type). Because of its much smaller T₁ enhancement (R₁(Dy) R₁(Gd)/40 (Villringer et al., 1988; Moseley et al., 1991)), this contrast agent should give more accurate perfusion calculations. Also notice that when the BBB is damaged, contrast agents can be used to extract information about BBB permeability by modeling the leakage of the tracer out of the intravascular space. This has been an area of great interest in the study of brain tumors and multiple sclerosis, but its description is outside the scope of this review (see Tofts, 1997 for a recent review).

Measurement of the arterial input function. The AIF depends not only on the shape of the injected bolus but also on the cardiac output, the vascular geometry, and the cerebral vascular resistance (Mottet et al., 1997). The AIF can be estimated directly by measuring the signal loss in a region of interest positioned on (or around) a feeding cerebral artery, such as the carotid artery or the middle cerebral artery (MCA). The signal in such a region would show an early, large increase in R₂ after contrast injection. The AIF generated in this way is commonly used, and in preliminary data, it has been shown to be proportional to measurements obtained invasively by arterial blood sampling in animals (Porkka et al., 1991; Rosen et al., 1991b). However, a full study to confirm these findings remains to be performed. Furthermore, because of partial volume effects with surrounding tissue, the AIF may be underestimated, thereby overestimating the CBF.

For a given contrast agent, higher sensitivity with the DSC technique can be obtained by increasing the injected dose or by using a longer TE. However, because of the greater concentration of the tracer in the arterial supply than in the VOI, a larger arterial signal drop is observed. If the signal falls below the background noise level, the AIF will be underestimated, introducing errors to the CBF quantification. To reduce this source of error, two different echo times can be used, with a shorter TE for the determination of the AIF (Perman et al., 1992; Rempp et al., 1994; Schreiber et al., 1998). Another possibility is to use only those measurements that fall above the background noise (Østergaard et al., 1998b; Ellinger et al., 1998).

Sequence type (gradient-echo versus spin-echo)

Dynamic susceptibility contrast MRI can be used either with spin-echo (SE) or gradient-echo (GE) imaging, since the passage of the bolus affects both T₂ and T₂*. Because of the larger effect observed using GE, most of the original studies used this type of sequence to follow the passage of the tracer through the brain. However, Monte Carlo simulations have shown that the susceptibility contrast in GE images arises from both large and small vessels, whereas in SE images it is dominated by small, capillary-sized vessels (Weisskoff et al., 1994; Boxerman et al., 1995; Kennan et al., 1994). Therefore, more studies are being performed using SE techniques, reflecting more microvascular perfusion. Furthermore, SE images offer improved image quality, particularly in regions such as the temporal lobes and the frontal sinus. However, notice that the CBV determined in this way is not the total CBV (even after accounting for the hematocrit concentration) but just the microvascular CBV (Boxerman et al., 1995; Østergaard et al., 1998b).

APPLICATIONS OF PERFUSION IMAGING USING MR CONTRAST AGENTS

Because of the technical difficulties and the assumptions involved in the quantification of absolute CBF, most of the applications of DCS-MRI have used summary parameters. Table 1 summarizes the studies of flow quantification.

Studies of experimental animal models

Cerebral ischemia. One of the first applications of the DSC technique was in the study of cerebral ischemia in experimental animal models. Unlike conventional MRI (T₁, T₂, and proton density), which is insensitive to acute ischemia, diffusion-weighted imaging (DWI) and DSC-MRI were shown to detect stroke within minutes of the event. Thus, much of the interest in the earlier studies was on the acute identification of ischemic regions (Moseley et al., 1990; Maeda et al., 1993; Wendland et al., 1991; Finelli et al., 1992; Kucharczyk et al., 1991). However, since the area with a perfusion deficit was found to be larger than the area depicted by DWI, the emphasis moved toward the use of DSC-MRI to complement and help in the interpretation of the changes observed using conventional MRI and DWI, with the ultimate aim of identifying and characterizing the ischemic penumbra (Astrup et al., 1981).

Within this context, combined diffusion and DSC-MRI summary parameters were used to identify a severe ischemic central area surrounded by a more moderately affected peripheral area in transient MCAO studies in rats (Muller et al., 1995a), as well as in permanent MCAO studies in cats (Moseley et al., 1990; Moseley et al., 1993) and rats (Pierce et al., 1997; Quast et al., 1993; Dijkhuizen et al., 1997). Pierce et al. (1997) showed that the apparent diffusion coefficient (ADC) was linearly correlated with MP signal in the periphery and suggested that their combined study could be useful for the quantitative assessment of the variable flow gradients in focal ischemia, including potentially critical areas at risk in the ischemic periphery (Fig. 2).

Figure 2.

(A) Perfusion-sensitive magnetic resonance (MR) images of the first-pass transit of gadodiamide in rat focal ischemia. Gadodiamide transit induces a transient drop in MR signal intensity. Focal ischemia is evident as regions that show no signal change. Imaging speed is one image per second using fast low-angle shot imaging. (B) First-pass transit profile of a bolus of gadodiamide through a contralateral ROI and four ipsilateral ROI. Each ipsilateral ROI was obtained from the corresponding apparent diffusion coefficient (ADC) map based on ADC thresholds: ADC < 4.0 10^-4 mm²/s, 4.0 < ADC < 5.0 10^-4 mm²/s, 5.0 < ADC < 6.0 10^-4 mm²/s, ADC > 6.0 10^-4 mm²/s. ADC < 5.0 10^-4 mm²/s. Values shown are means from seven rats. (Reprinted from Pierce et al. 1997 MRI measurements of water diffusion and cerebral perfusion: their relationship in a rat model of focal cerebral ischemia. J Cereb Blood Flow Metab 17:183–190 with permission.)

Full figure and legend (365K)

A similar approach was used to identify different levels of ischemia in a cat model of partial MCAO (Fig. 3). The combination of high-speed ADC measurements and MP signal allowed the differentiation of mild (no change in ADC and slight reduction in the MP), moderate (slight reduction in ADC and reduced MP), and severe cerebral hypoperfusion (marked drop in ADC and almost complete absence of contrast agent passage) Roberts et al, 1993; Roberts et al, 1996. Similar results were obtained recently using ADC, MP, BAT, and high magnetic field T₂ mapping in models of hypoperfusion and focal cerebral ischemia in the rat (Grohn et al., 1998).

Figure 3.

(A) Time course of R₂* in a cat model of mild ischemic injury with no sign of injury on triphenyltetrazolium chloride (TTC) staining and no apparent hyperintensity on diffusion-weighted imaging (DWI). A drop in the peak R₂* effect of 20% in the ischemic parietal cortex and a delay of 4 to 6 seconds relative to the contralateral normal cortex are typical. Numerical fits to the gamma-variate function, also shown, allow a better estimate of the peak R₂* effect and eliminate effects of recirculation of contrast agent in the asymptotic tail of the R₂* curves. (B) Time course of R₂* for a severe ischemic injury with no uptake of TTC and a pronounced hyperintensity on DWI. A clear absence of transit in the ischemic cortex is confirmed. The R₂* curves show no contrast agent-induced T₂* shortening in the ischemic cortex. Gamma-variate fitting clearly is possible only for the R₂* curve from the normal parietal cortex. (Reprinted from Roberts et al. [1993] High-speed MR imaging of ischemic brain injury following stenosis of the middle cerebral artery. J Cereb Blood Flow Metab 13:940–946 with permission.)

Full figure and legend (83K)

The DSC summary parameters also were used to study the reperfusion state in transient MCAO in the cat (Kucharczyk et al., 1993), distinguishing between successful and failed reperfusion and identifying delayed reperfusion injury. Recently, in a study of delayed damage in a model of temporary hypoxia-ischemia in the rat, it was shown that ADC alone was a bad predictor of long-term outcome (Dijkhuizen et al., 1998). Measurements of ADC, T₂-weighted imaging, DSC summary parameters (MP, TTP, and peak area), laser-Doppler flowmetry, and histologic study were used to show different regional response to reperfusion and an associated tissue-specific sensitivity to delayed damage.

In a study of transient focal ischemia in cats (Caramia et al., 1998), DSC summary parameters were used to identify the hemodynamic spatial heterogeneity in response to ischemia and reperfusion. In particular, a peripheral area of increased vascular transit time (VTT = TTP - BAT) was observed, which was interpreted as a mismatch between CBF and CBV. This area had a different response to reperfusion compared with the core region.

Another application, where DSC summary parameters were used, is to the determination of the relation between apoptosis and perfusion deficits during transient cerebral ischemia in cats. Vexler et al. (1997) showed that a significantly higher number of apoptotic cells were observed in areas that experienced 2.5 hours of severe perfusion deficit (regardless of the state after reperfusion) compared with the number in areas with reduced but maintained perfusion. These results suggest that the apoptotic process is induced in the ischemic core and contributes in the degeneration of neurons associated with transient ischemia.

Cortical spreading depression. The DSC summary parameters (relCBV and TTP) also have been used in the study of cortical spreading depression. By following the ADC changes in ischemia-induced spreading depression, Röther et al. (1996a, 1996b) showed that these transient ADC changes originated in the border of the core region, where there was a moderate perfusion deficit (increased bolus transit time), and that the variation in the ADC recovery time in the ischemic periphery reflected the severity of the tissue perfusion deficit.

De Crespigny et al. (1998) studied the changes in relCBV and ADC during spreading depression in rats, although not using DSC techniques. In this study, steady-state CBV measurements using a blood pool contrast agent were performed simultaneously with high-speed ADC mapping. They found transient regional hypoperfusion after the ADC changes with a variable delay (average value of 16 seconds), thought to be a consequence of the elevated energy requirements during the repolarization after spreading depression.

Therapeutic evaluation. The combined used of DSC-MRI and DWI provides an excellent tool to investigate the effect of new therapeutic interventions in models of embolic stroke in animals. Reith et al. (1996) have assessed the effects of thrombolytic therapy using recombinant tissue plasminogen activator (rt-PA) in a rat model, whereas more recently, Yenari et al. (1997) have examined the combined effect of rt-PA and direct anti-thrombin therapy using hirulog in rabbits. Furthermore, the effect of rt-PA has been studied in a model of cerebral venous thrombosis (Röther et al., 1996d). Bolus tracking techniques also have been used to evaluate the cerebroprotective effects of noncompetitive N-methyl-D-aspartate antagonists (Pan et al., 1995; Minematsu et al., 1993) and treatment with the free radical scavenger U74389G (Muller et al., 1995b) in temporary focal ischemia in rats. Other applications include the evaluation of the neuroprotection using NBQX in permanent focal cerebral ischemia in the rat (Lo et al., 1997) and the assessment of the efficacy of intravenous (as compared with intraarterial) therapy with prourokinase in a model of embolic stroke in rats (Takano et al., 1998). In all of these studies, serial measurements of diffusion and perfusion were performed, both in the treated and control groups of animals.

Clinical applications

Cerebrovascular diseases. Because of the insensitivity of conventional MRI for the detection of acute stroke and the lack of a definite treatment in the early 1990s, few studies of cerebral ischemia were performed during the hyperacute phase (first few hours), and, therefore, the affected area was already visible on T₂-weighted images. In these studies, DSC-MRI was mainly used to evaluate the perfusion state of the affected region and provide extra information, which was not available from conventional MRI (Edelman et al., 1990; Warach et al., 1992). Furthermore, since EPI capabilities were not available in many centres, the initial studies were restricted to single-slice acquisitions using fast GE sequences. In some studies during the hyperacute phase, the selected slice did not include the ischemic region, and false-negative results were obtained (Warach et al., 1992; Röther et al., 1996c). With the advent of thrombolytic therapy and the need for an early assessment of the affected area, the emphasis on DSC studies changed toward hyperacute diagnosis, identification of tissue at risk, and prediction of neurologic outcome.

Many studies address the mismatch between the region of perfusion disturbance, the area with reduced diffusion, and the possible relation with tissue at risk of infarction (Fig. 4) (Sorensen et al., 1996, 1997; Østergaard et al., 1996a; Rordorf et al., 1998; Tong et al., 1998; Schwamm et al., 1998; Baird et al., 1997; Barber et al., 1998). In studies during the hyperacute phase using magnetic resonance angiography, ADC, and relative CBV (Sorensen et al., 1996; Rordorf et al., 1998), it was possible to classify the MCA stroke into different clinical and pathophysiologic subtypes. Group 1 (patients with M1 occlusion) presented a mismatch between the extent of early infarction (as defined by DWI) and the extent of early ischemia (as defined by DSC-MRI), and they usually showed a progression of infarction into the larger region of abnormal perfusion. Group 2 (patients with open M1 on magnetic resonance angiography and stroke from occlusion of a distal branch) presented a matched diffusion/perfusion abnormality and were found to have maximal extent of regional injury on their initial study (Fig. 5). As suggested by the authors, this classification may help to the selection of patients who could benefit from thrombolytic and neuroprotective therapies.

Figure 4.

Images of a 53-year-old man 3 hours after acute onset of left hemiparesis. (A) T₂-weighted fast spin-echo shows minimal sulcal effacement in the right hemisphere. The isotropic diffusion-weighted image (B) correlates well with the size of the infarcted area at the follow-up computed tomography (CT) scan 3 days later (C). The relative cerebral blood volume (CBV) (D) and CBF (E) maps show a small region with low flow and high volume that survive the stroke (arrow). The region of flow-volume mismatch is clearly visible in the mean transit time (MTT) map (F) and might be identifying a region of ischemic tissue with preserved hemodynamic auto regulation. (Reprinted from Østergaard et al. [1996b]; High resolution measurement of cerebral blood flow using intravascular tracer bolus passages. I. Mathematical approach and statistical analysis. Magn Reson Med 36:715–725 with permission.)

Full figure and legend (333K)

Figure 5.

(A) A 62-year-old man had sudden onset of dense right hemiplegia 7 hours before his MRI scan. The T₂ sequence shows only a subtle increase in signal in a gyral pattern, whereas the DWI map shows an ischemic injury (arrow) located mostly in the insular and periinsular cortex. The CBV map demonstrates a perfusion abnormality much larger than the DWI abnormality. Magnetic resonance angiography (MRA) demonstrates a left MCA occlusion (arrow). In the follow-up MRI, the stroke has expanded beyond the initial area of perfusion abnormality. (B) Cortical branch stroke with matching DWI and perfusion abnormality. The initial results of a MRI study of a 75-year-old man who presented to the emergency department 4 hours after the acute onset of mutism are shown. The T₂-weighted image is normal, whereas a matching abnormality consistent with ischemic injury is seen on the DWI and CBV maps (arrows). An MCA stem opened bilaterally (arrow) is shown by MRA. The follow-up head CT scan done 4 days later confirms the presence of a stroke (arrow), which is consistent with the abnormality seen on the initial DWI and CBV maps. (Reprinted from Rordorf G, Koroshetz WJ, Copen WA, Cramer SC, Schaefer PW, Budzik RF, et al [1998] Regional ischemia and ischemic injury in patients with acute middle cerebral artery stroke as defined by early diffusion-weighted and perfusion-weighted MRI. Stroke 29:939–943 with permission.)

Full figure and legend (351K)

In a different approach to the identification of tissue at risk, some studies examine the concentration-time curves in the different areas (Röther et al., 1996c; Tsuchida et al., 1997; Reith et al., 1997). Röther et al. (1996c) analyzed DSC summary parameters during the first 6 hours after symptoms onset. Three different patterns for the passage of the tracer, depending on the perfusion status of the tissue, were identified: in the center of the lesion, a complete loss of the bolus passage (pattern 1) was associated with the infarct core; a delayed bolus, with reduced peak and increased width (pattern 2) was taken to represent the potentially salvageable tissue; and, finally, a minimal bolus delay with a slightly reduced peak (pattern 3) was observed only in follow-up studies in regions that showed a marked recovery.

In a preliminary analysis of patients with severe clinical deficits, Warach et al. (1996) found that the combined use of DWI and DSC-MRI (relCBV) was highly accurate in predicting clinical outcome by identifying the patients who would improve. The advantage of combined DWI/DSC-MRI analysis compared with conventional MRI was found to be more important in the subgroup of patients studied during the first 6 hours. In a study on the correlation between lesion volume (measured by DWI or DSC-MRI) and neurologic deficit (quantified using the National Institutes of Health Stroke Scale), Tong et al. (1998) report a significant correlation between initial lesion size (measured within 6.5 hours of symptom onset) and the score on the National Institutes of Health Stroke Scale for both DWI and DSC-MRI volumes. In a similar study, Barber et al. (1998) found a high correlation between DSC-MRI lesion volumes (using C⁽¹⁾) and acute neurologic state (assessed by the Canadian Neurological Scale), clinical outcome (Canadian Neurological Scale, Barthel Index, and Rankin Scale), and final infarct volume (T₂-weighted imaging). However, the correlation of acute DWI lesion volumes was not as robust. These authors suggest that the difference from the previous study may result from the different clinical scales used or to a different proportion of patients with perfusion deficits larger than the region of reduced diffusion (which are more likely to have lesion enlargement). This highlights the need to study numerous patients before these correlations can be established and emphasizes the danger of extending particular findings to different patient populations.

Although further studies are needed, it is expected that a combined analysis using DWI and DSC-MRI (together with magnetic resonance angiography and conventional MRI) could be used to identify patients who might benefit from various treatment modalities and those who can avoid the potential risks associated with thrombolysis and neuroprotective agents. Furthermore, the duration of a therapeutic window may need to be defined for individual patients, depending on the physiologic state of the tissue, instead of being defined only by the time from symptom onset (Baron et al., 1995; Fisher, 1997). This is supported by recent data that show expansion of ischemic lesions several hours, and even days, after the acute insult (Baird et al., 1997; Schwamm et al., 1998). Therefore, DWI and DSC-MRI could improve patient selection, guide stroke therapy, and, therefore, improve the evaluation of new therapeutic strategies.

In patients with unilateral symptomatic internal carotid artery (ICA) occlusion, DSC-MRI also was used to assess perfusion deficits (Gückel et al., 1994; Nighoghossian et al., 1996; Kluytmans et al., 1998). All of these studies show changes in various summary parameters, suggesting an impairment of perfusion reserve in the ipsilateral side. Furthermore, these hemodynamic changes occurred mainly in white matter regions (Kluytmans et al., 1998), which is consistent with its higher vulnerability in patients with ICA occlusion.

The DSC summary parameters have been used to evaluate the perfusion characteristics in children with homozygous sickle cell anemia (both with and without clinical evidence of cerebrovascular disease) (Tzika et al., 1993), in patients with moyamoya disease (Tzika et al., 1997; Tsuchiya et al., 1998), and, in conjunction with the blood oxygenation level-dependent (BOLD) contrast technique, to study functional brain activation in patients with chronic cortical stroke (Sorensen et al., 1995).

Notice that because of its dynamic characteristic, DSC-MRI can provide perfusion related information not available from PET or single photon emission computed tomography (SPECT) studies. For example, in a study 5.5 hours after stroke onset, Sorensen et al. (1997) showed that despite the apparently normal shape and area of the bolus (probably indicative of unaffected CBV and CBF), there was only a delayed bolus arrival to the ischemic tissue, consistent with slow collateral flow through feeding arteries. Similar results also were obtained in an area adjacent to a chronic stroke (Fig. 6) (Sorensen et al., 1995).

Figure 6.

Evaluation of stroke with perfusion imaging. (A) T₂-weighted image of a 74-year-old man who had two strokes at least 3 weeks before imaging. Abnormal signal consistent with stroke is seen in the right temporal/parietal lobe (open area) and in the left occipital lobe (solid arrow). (B) The regional cerebral blood volume (rCBV) map demonstrates markedly reduced rCBV in the right hemisphere but only slightly reduced rCBV in the left. (C) Map of time to peak. This demonstrates higher peak time (longer from time of injection) in the right hemisphere, particularly in the region of the stroke. (D) Time course for four different ROI. The ROI from the map of time to peak are overlaid onto a T₂-weighted image. The top left image shows a ROI within an area of stroke and demonstrates a delay of about 5 seconds when compared with the contralateral hemisphere at the same location (top right image). In this patient, there also is a delay in peak arrival time of 1.5 seconds in cortex, which appears normal on conventional imaging. This delay in arrival of the bolus of contrast agent implies that there is some residual large vessel disease in this patient. Acquisition parameters for map: echo planar imaging (EPI) SE, TR = 1500 milliseconds, echo time (TE) = 100 milliseconds, 0.2 mmol/kg Gd-DTPA-BMA (gadodiamide). (Reprinted from Sorensen et al. [1995] Functional MR of brain activity and perfusion in patients with chronic cortical stroke. Am J Neuroradiol 16:1753–1762, © by American Society of Neuroradiology, with permission.)

Full figure and legend (684K)

Cerebrovascular reactivity. Acetazolamide (ACZ) is a potent vasodilatory agent that is being increasingly used clinically to test the vascular reserve in cases of large artery cerebrovascular disease. Dynamic susceptibility contrast MRI provides a means to assess cerebrovascular reactivity by performing two bolus injections of contrast agent: one before (to obtain a baseline measurement), and the other after the ACZ administration. Several studies address the ability of DSC-MRI to measure CBV changes in normal volunteers (Levin et al., 1995; Petrella et al., 1997; Berthezene et al., 1998) and the age dependency of this vasodilatory response (Petrella et al., 1998).

Two studies have been performed on the application of the ACZ stimulation test to patients with occlusive cerebrovascular disease (Gückel et al., 1996; Schreiber et al., 1998). These studies not only address the CBV response but also measure the effect on CBF and MTT. They are among the few studies where absolute quantification was performed. Gückel et al. (1996) assessed the cerebrovascular reserve capacity in patients with occlusion of the ICA or MCA. Regions of interest were chosen from the MCA territory but without including infarcted tissue. A statistically significant reduction in CBF response was found in the affected MCA territory, indicative of a severely compromised cerebrovascular reserve capacity. They also found a decoupling between CBV and CBF response to ACZ, suggesting that CBV alone is not a reliable indicator for the cerebrovascular reserve capacity. In a recent study, Schreiber et al. (1998) evaluate the response to ACZ in patients with occlusion or high-grade stenosis of the ICA, and a similar reduction in the cerebrovascular reserve capacity was found in peri-infarcted regions (Fig. 7). An interesting finding of this study was the correlation between the change in CBF and the baseline MTT value, which suggests that MTT baseline measurements could be used to assess cerebrovascular reserve capacity.

Figure 7.

A patient with occlusion of the right internal carotid artery having transient ischemic attacks. (A) Conventional T₂-weighted SE image. The slice was positioned at the ventricles. (B) Parametric regional cerebral blood flow (rCBF) image. Blood flow is normal in both hemispheres, and vasodilation is observed in the right hemisphere on the rCBV (C) and MTT (D) maps. Images (E) and (F) were obtained from a MRI examination at 15 minutes after acetazolamide (ACZ) injection. (G) The percent change of rCBV after ACZ stimulation is given. In the left hemisphere, a rCBF value of only 11% indicates compromised cerebrovascular capacity. Only a small region in the right hemisphere shows a significant rCBF increase (arrow). In most regions, a slight decrease of rCBF is observed. The units are as follows: rCBF, mL 100 g^-1 min^-1; MTT, seconds; rCBF, percentages. (Reprinted from Schreiber et al. [1998] Cerebral blood flow and cerebrovascular reserve capacity: estimation by dynamic magnetic resonance imaging. J Cereb Blood Flow Metab 18:1143–1156 with permission.)

Full figure and legend (608K)

Notice that because the ACZ test requires multiple bolus injections, special care must be taken to avoid (or take account of) the effect of the residual contrast agent from the first injection on the second measurement (Runge et al., 1994; Levin et al., 1995; Levin et al., 1998).

Other applications. In a recent study, Cutrer et al. (1998) used DSC-MRI and DWI to study brain hemodynamics during spontaneous visual auras in patients with migraine headaches. Although the measurements of MTT, relCBF, and relCBV performed interictally were normal and symmetrical throughout the brain, a moderate focal reduction in all of these parameters was observed in the occipital lobe during visual aura. Interestingly, no significant change in water diffusion was observed, suggesting that the perfusion deficit was not severe or persistent enough to induce ADC changes.

Another clinical application for DSC-MRI is in the perfusion evaluation of chronic neurodegenerative disorders such as Alzheimer's disease. Although no CBF evaluation has been performed, several studies have evaluated changes in relCBV and its correlation with PET (Gonzalez et al., 1995; Caramia et al., 1995), SPECT (Harris et al., 1998), and the scores on the Mini-Mental State Examination (Maas et al., 1997; Harris et al., 1996).

Dynamic susceptibility contrast MRI also has been used in other applications, in particular, to study changes in CBV, such as the increased CBV observed in patients who are positive for human immunodeficiency virus (Tracey et al., 1998), in schizophrenic patients (Cohen et al., 1995), and in patients with epilepsy (Warach et al., 1994), and to detect changes after cocaine administration both in humans (Kaufman et al., 1998) and animals (Li and Suojanen, 1995).

Brain tumors, because of their different blood flow and vascular patterns, also are good candidates for DSC studies. Most of these studies concentrate on the evaluation of CBV, although a few also address the CBF distribution (Fig. 8) or other DSC summary parameters (Østergaard et al., 1996a; Sorensen et al., 1997). A good correlation between CBV and tumor grade (Aronen et al., 1994; Rosen et al., 1991b) has been demonstrated. Cerebral blood volume mapping also has been used to differentiate recurrent tumor from radiation necrosis (Rosen et al., 1991a; Caramia et al., 1995; Sorensen et al., 1997), and to study the changes associated with radiotherapy in a group of patients with astrocytomas (Wenz et al., 1996). Tumor heterogeneity, not observed in conventional MRI (precontrast or postcontrast), has been shown on CBV maps (Rosen et al., 1991b; Caramia et al., 1995), and a recent comparative study using relative CBV, ²⁰¹T1-SPECT and PET (Siegal et al., 1997), has shown the high sensitivity of DSC-MRI to small regional changes. In this study, a 63% and 55% rate of delayed detection compared with CBV was found using SPECT and PET, respectively, which resulted from the higher spatial resolution and sensitivity of CBV maps produced by DSC-MRI. However, notice that BBB breakdown may be present in many cases, resulting in extravascular leakage of the contrast agent. As mentioned in the section on perfusion imaging using MR contrast agents, this effect introduces errors in the measurement, unless one properly accounts for them.

Figure 8.

(A) The relative cerebral blood volume (relCBV) map of a patient with a low-grade tumor in the right temporal lobe shows a moderately elevated CBV over an extensive area. (B) Also shown is a CBF-weighted image acquired using flow-sensitive alternating inversion recovery (FAIR), which displays two distinct areas (arrows) with high flow within the more general region of elevated CBV. These focal regions also can be seen on the dynamic susceptibility contrast (DSC)-MRI relative cerebral blood flow (relCBF) maps obtained using two different deconvolution methods (a model-dependent approach with an exponential residue function [C] and singular value decomposition deconvolution [D]). However, this regional heterogeneity was not well reproduced in the relCBF map using a model-dependent approach with a Gaussian residue function. (E) Symmetrical areas of high flow around the brain stem and corresponding to the course of the MCA are artifacts caused by susceptibility effects around large vessels. (Reprinted from Østergaard et al. [1996] High resolution measurement of cerebral blood flow using intravascular tracer bolus passages. I. Mathematical approach and statistical analysis. Magn Reson Med 36:715–725 with permission.)

Full figure and legend (355K)

PERFUSION MRI USING SPIN LABELING OF ARTERIAL WATER

An MR image can be sensitized to the effect of inflowing blood spins if those spins are in a different magnetic state to that of the static tissue. The family of techniques known as ASL techniques uses this idea by magnetically labeling blood flowing into the slices of interest. Blood flowing into the imaging slice exchanges with tissue water, altering the tissue magnetization. A perfusion-weighted image can be generated by the subtraction of an image in which inflowing spins have been labeled from an image in which spin labeling has not been performed. Quantitative perfusion maps can be calculated if other parameters (such as tissue T₁ and the efficiency of spin labeling, see later) also are measured. Since exogenous contrast agents are not required for these techniques, the perfusion measurement is completely noninvasive. Under the general heading of ASL, two distinct subgroups exist: continuous ASL and pulsed ASL. Despite sharing a similar basis, the two approaches have different advantages and limitations that affect their ability to quantify perfusion.

Continuous arterial spin labeling techniques

Basic theory. The first ASL technique, proposed by Detre et al. in 1992, is known as continuous arterial spin labeling (CASL). They suggest the use of a train of RF pulses to repeatedly saturate blood water spins flowing through the neck. The saturated (or "labeled") spins flow into the brain and, assuming water is a freely diffusible tracer, exchange completely with brain tissue water, thus reducing the overall tissue magnetization. A steady state develops where the regional magnetization in the brain is directly related to CBF. This steady state can be derived from the Bloch equation for longitudinal relaxation when it is modified to include flow (Detre et al., 1992): (Eq. 7) where M_z(t) is the longitudinal magnetization of the brain tissue per unit mass, M⁰_b is the magnetization of fully relaxed tissue per unit mass, T_1b is the longitudinal relaxation time of brain tissue water, f is blood flow (perfusion) (mL 100 g^-1 min^-1), is the blood-brain partition coefficient for water, M_a is the magnetization of water in the inflowing arterial blood per unit volume of blood, and M_v is the equivalent term for the outflowing venous blood. This leads to the following equation for flow quantification (Detre et al., 1992): (Eq. 8) where M^ss_b is the steady-state magnetization per unit mass of brain tissue and T_1app is the constant that describes the exponential decay to the steady state magnetization and is defined by (Eq. 9)

Equation 9 is the fundamental equation of ASL, describing how the apparent longitudinal relaxation time of brain tissue is affected by blood flow. By measuring T_1app and the ratio of magnetization with and without arterial spin saturation, a value for flow can be calculated using Eq. 8.

Soon after CASL was proposed, Williams et al. (1992) improved the technique by using adiabatic fast passage to label the arterial water spins. Making use of the adiabatic theorem (Abragam, 1961), the magnetization of a spin moving through a magnetic field gradient G at velocity is inverted if a RF pulse is used to apply a constant B₁ field perpendicular to the main field, and the following condition is satisfied: (Eq. 10) where is the magnetogyric ratio of the hydrogen nucleus. As the spins flow along the direction of the magnetic field, their frequency sweeps from far below resonance to far above (or vice versa, depending on the polarity of the gradient and the direction of flow). In this way, adiabatic inversion will be induced. If such a RF pulse is applied for several seconds to a plane in the neck of a subject, inflowing arterial water spins are inverted during this period, and a flow-related steady state will be achieved. The advantage of labeling arterial spins by inversion rather than saturation is a twofold increase in the difference between the labeled and unlabeled states, and flow is then given by: (Eq. 11) where M^cont_b is the magnetization of tissue per unit mass without arterial spin inversion (the control image) and M^inv_b is the magnetization per unit mass with arterial spin inversion (the perfusion-weighted image).

Transit time and magnetization transfer effects. Two main problems soon became apparent in the application of continuous spin labeling as a method to measure tissue perfusion accurately. First, the time taken for spins to travel between the labeling plane and the imaging slice (henceforth referred to as the transit time) is nonzero, and therefore T₁ relaxation occurs during this period. Different regions of the brain have different transit times, depending on the distance from the labeling plane and the cerebrovascular anatomy. For accurate quantification, this effect must be taken into account. The transit time problem has been noted in experiments comparing CASL with the radioactive microsphere method of CBF quantification, where an underestimation of flow was observed with the spin labeling technique when the flow rates were low (Walsh et al., 1994). Second, the application of a long (i.e., several seconds) off-resonance RF pulse causes a decrease in the water signal resulting from magnetization transfer (MT) effects. Although no direct saturation of the observed water magnetization in the imaging slice occurs because of the narrow line width of the free water peak, saturation of macromolecular spins does occur, resulting in attenuation of the free water signal through magnetization transfer (Wolff and Balaban, 1989). This reduces the perfusion-dependent signal difference between the control and spin-labeled images. Zhang et al. (1992) made an initial attempt to account for these two effects by incorporating MT effects into the Bloch equation analysis and measuring the time lag between the initiation of spin labeling and the first observable decrease in tissue magnetization. They accounted for transit time effect by modifying the degree of inversion, (equal to 0.5 for saturation, 1 for perfect inversion), by an amount related to T₁ relaxation: (Eq. 15) where (Eq. 12) where T_1a is the longitudinal relaxation time of arterial blood water and is the transit time. It is possible to measure by imaging a slice in the neck just after the spin labeling has been performed and observing the difference in signal intensity in the main arteries in images acquired with and without the inversion pulse (Zhang et al., 1993). To produce quantitative perfusion maps, knowledge of the transit time on a pixel-by-pixel basis is necessary. This has been attempted by Ye et al. (1997b), where the difference between spin-labeled and control images was measured as a function of the duration of the labeling period using single-shot EPI techniques, but the SNR was found to be insufficient to accurately estimate the transit time with high spatial resolution. An alternative approach has been taken by Alsop and Detre (1996) in which a time delay is inserted into the sequence between the end of the labeling period and the image acquisition. Although this reduces the signal difference between the control and labeled images and complicates the quantification procedure, it can be shown (Alsop and Detre, 1996) that if this delay is greater than the arterial transit times across the image, then the resulting CBF maps will be almost completely insensitive to variations in transit time. This technique has been shown to produce fairly good quality perfusion images, even in patients with cerebrovascular disease, and the concomitant large range of transit times throughout the brain (Detre et al., 1998b). However, the length of the postlabeling delay, and thus the tolerance to long transit times is limited by T₁ relaxation and SNR. Nevertheless, this approach appears to be the most robust and promising method available for the elimination of transit time effects.

Magnetization transfer complicates the extension of the methods mentioned earlier to multislice acquisition, since the control image must experience exactly equivalent MT effects as the spin-labeled image. This means that the control "plane" must be symmetrically opposite the inversion plane with respect to the imaging slice, which can be true only for a single imaging slice, and limits the slice orientation to being perpendicular to the arteries in which spin labeling is performed. The control image is acquired by changing the sign of the frequency offset of the labeling pulse (assuming symmetrical MT effects) or by reversing the gradient polarity. It has been suggested (Pekar et al., 1996) that to properly account for MT effects (and possible eddy current effects caused by the labeling gradient) a four-step protocol for image acquisition should be used in which both frequency and gradient polarity are alternated, although this remains a point of debate (see Silva et al., 1997b). Alternatively, a two-coil setup can be used to avoid the saturation of macromolecular spins (Zhang et al., 1995). In this setup, a small surface coil is placed on the neck of the subject immediately adjacent to the artery supplying blood to the brain. The physical range of the B₁ field produced by the surface coil is limited to a localized region. Arterial water spins are inverted using the surface coil, but no MT effects are observed in the brain, since the B₁ field does not extend sufficiently far. A separate (actively decoupled) coil is used to apply the imaging pulses and receive signal. Extension of this approach to multislice perfusion imaging is straightforward, since the control image does not need to contain any MT information (Silva et al., 1995). However, specialized hardware is needed to implement this method, and flow quantification still is complicated by the need to account for a large range of arterial transit times. Recently, a more comprehensive analysis of the MT effects in single-coil CASL has been presented based on a four-compartment model of free and bound solvent and macromolecular protons (McLaughlin et al., 1997). Two alternative approaches for flow quantification were suggested, which differ in the way that the perfusion subtraction image is normalized and the manner in which T₁ is measured. The presence of off-resonance radiation reduces the T₁ of tissue caused by the MT effect, and the resulting relaxation time, T_1sat, is an important parameter in this analysis. Complete saturation of the macromolecular magnetization is not a prerequisite of these approaches, making them directly relevant to the use of CASL in human studies, where RF power deposition is a restricting factor and macromolecular saturation cannot be guaranteed.

Intravascular signal. Another possible source of systematic error in CASL techniques is the presence of intravascular signal in the subtraction (control minus labeled) images. All of the theoretical models describing the relation between flow and signal difference (e.g., Equation. 7 through 12) are based on the assumption that signal comes exclusively from tissue. If a significant amount of signal emanates from the vasculature, perfusion will be overestimated. Most spin labeling studies have been performed with flow-crushing gradients included in the imaging sequence, which are intended to completely eliminate signal from intravascular spins. Studies show that inclusion of a bipolar gradient pair with a "b" value (Le Bihan and Turner, 1992) of approximately 5 s/mm² reduces the perfusion difference signal in human gray matter by almost 50% compared with the signal obtained with no crusher (Ye et al., 1997b). The extraction fraction, E, usually is assumed to be equal to unity (i.e., water is a freely diffusable tracer), and experiments by Silva et al. using CASL with and without macromolecular saturation (Silva et al., 1997a) and diffusion gradients (Silva et al., 1997b) demonstrate that this assumption is reasonable at normal flow rates in the rat brain. However, at higher flow levels, the extraction fraction was shown to be reduced (E 0.6 when f = 400 mL 100 g^-1 min^-1). The transit time insensitive sequence proposed by Alsop and Detre (1996), mentioned previously, allows most of the labeled blood to either exchange with the tissue or "wash through" the vasculature during the postlabeling delay, and therefore suffers less from the associated errors of arterial signal contributions. Notice that since transit time effects cause an underestimation of CBF and intravascular signal causes an over-estimation, the two effects can cancel out one another. However, this will be true only under specific circumstances, and both sources of systematic error need to be addressed. A combination of delayed acquisition and flow-crushing gradients seems to be sufficient to account for both effects (Ye et al., 1997a).

It has been suggested that the problems of intravascular signal contamination and extended transit times in regions of reduced perfusion can, paradoxically, be used to assist in the identification of these areas (Detre et al., 1998b). CASL approaches which include a postlabeling delay should achieve transit time insensitivity except in regions where the transit time is especially increased. These regions, which are likely to correspond to areas of extremely low perfusion, contain significant amounts of intravascular spin labeled water, and so in the ASL subtraction image appear as bright foci of signal (Fig. 9). Surrounding areas of moderately decreased CBF appear dark, reflecting the desired contrast for the perfusion image. Therefore, it is possible to identify the most severely compromised tissue based on the observation of unusually bright spots in the perfusion map. Quantitative interpretation of the signal is difficult, since knowledge of exact transit times is required, but qualitative evaluation of perfusion deficits should be possible, which may be adequate for clinical applications of the technique.

Figure 9.

Intraluminal artifact in continuous arterial spin labeling (CASL) CBF subtraction images of stroke patients. Gray scale shows CBF ranging from 0 to 127 mL 100 g^-1 min^-1. (A) The CBF images from a stroke patient show a focal region of hypoperfusion in the left frontal lobe. Arrows indicate bright intraluminal artifact resulting from markedly delayed arterial transit times to the hypoperfused region. (B) The CBF images from a second stroke patient. Bright intraluminal artifact is observed diffusely and degrades the images severely. Away from this artifact, cortical CBF is reduced. These images demonstrate the problems of using arterial spin labeling (ASL) approaches to obtain quantitative measures of CBF in stroke. (Reprinted from Detre et al. [1998] Noninvasive MRI evaluation of cerebral blood flow in cerebrovascular disease. Neurology 50:633–641 with permission.)

Full figure and legend (553K)

Multislice imaging. Finally, the extension of CASL techniques to multislice perfusion imaging with standard hardware has become possible with another method devised by Alsop and Detre (1998). Rather than acquiring the control image by changing the frequency or gradient polarity of the RF inversion pulse, the amplitude of the pulse is sinusoidally modulated at a certain frequency . Fourier transformation of an RF waveform of base frequency ₀ with sinusoidal amplitude modulation of frequency results in a pair of inversion planes at frequencies ₀ . The result is that as spins flow through, they are inverted and then immediately "uninverted," thereby losing their label. The MT effects of the sinusoidally modulated waveform have been shown to be equivalent to those of the standard labeling RF pulse (except close to plane of inversion), thus enabling multislice perfusion imaging with any image orientation. Some concerns remain regarding the efficacy of the double inversion pulse; however, it is likely that these problems will be overcome in the near future, making this technique extremely promising.

Pulsed arterial spin labeling techniques

Basic theory. As mentioned in the previous section, continuous ASL techniques have two major problems: magnetization transfer effects, caused by the application of a long off-resonance RF pulse before image acquisition, and the loss of the spin label by the blood water as it travels from the labeling plane to the imaging slice. In pulsed arterial spin labeling (PASL) techniques, these sources of error are minimized to facilitate flow quantification. This is achieved by using a short (typically 10 millisecond) RF pulse to label spins and minimizing the distance between the labeling region and the imaging slice. Table 2 summarizes the main features of the different ASL techniques, which are described more fully in the following paragraphs.

Table 2 - Principal classifications of the arterial spin labeling techniques.

Full table

Echo planar imaging and signal targeting with alternating radiofrequency. The first of the PASL group of perfusion imaging approaches was proposed by Edelman et al. in 1994 and is known as echo planar imaging and signal targeting with alternating radiofrequency (EPISTAR) (Edelman et al., 1994). This technique is similar conceptually to CASL. After saturation of the imaging slice (see Transit Time Effects later), a slab proximal to the imaging slice is labeled using a single, short RF inversion pulse. The blood in this slab then is allowed to flow into the imaging slice, and an image is acquired after a time TI. A control image also is acquired for which the label is applied distal to the imaging slab (Edelman et al., 1994) or with no slab selective gradient (a variant known as PICORE [proximal inversion with control for off-resonance effects, Wong et al., 1997]). Recently, in an approach similar in concept to that of Alsop and Detre for CASL described in the previous section, the control image was acquired by applying a "double inversion" pulse pair (Edelman and Chen, 1998). The MT effect on the imaging slices is the same in the control and spin labeled images, and inflowing blood in the control image is fully relaxed, since it experiences an overall 0° nutation. Subtraction of the images results in a flow-dependent image with a signal intensity determined by (Kwong et al., 1995; Calamante et al., 1996): (Eq. 13) where M is the difference in magnetization per unit mass between the labeled and control images, M₀ is the equilibrium tissue magnetization per unit mass of tissue, and TI is the time between spin inversion and image acquisition (inversion time). This theoretical signal difference is less than that of CASL by a factor of exp(-TI/T₁). Equation 13 relies on several simplifying assumptions, such as tissue and blood T₁ being equal and the efficiency of the inversion pulse being equal to 1. Under practical circumstances these assumptions are not valid, and a more general analysis (Calamante et al., 1996) predicts a biexponential relation between M and TI: (Eq. 14) where ₀ is the degree of inversion of the spin labeling pulse and T_1app is defined in Eq. 9. If T_1a T_1b and ₀ is unity, Eq. 14 reduces to Eq. 13. However, for brain tissue such as white matter where the T₁ is approximately 50% of the blood water value, a large error is introduced by making the simplifying assumptions (Calamante et al., 1996). Quantification of blood flow therefore requires the acquisition of a set of image pairs at different inversion times and the differences in signal to be fitted to Eq. 14.

Flow-sensitive alternating inversion recovery. Measurements of changes in perfusion based on dynamic changes in tissue T₁ were first performed by Kwong et al. (1992) (see Functional Brain Studies). However, measurements of resting values of perfusion were not possible using this method since the intrinsic tissue T₁ was unknown. To identify the component of signal in the slice-selective inversion recovery image, which results purely from the inflow of blood spins during the inversion time, a second image is required, which has relaxed with intrinsic tissue T₁. At approximately the same time, Kwong et al. (1995), Kim (1995), and Schwarzbauer et al. (1996) proposed the sequence that is commonly referred to as flow-sensitive alternating inversion recovery (FAIR). In this technique, two inversion recovery images are acquired. The first follows a slice-selective inversion and thus has a signal intensity determined by T_1app (Eq. 9); the second follows a global inversion and so (assuming blood and tissue relax at the same rate) has no flow enhancement (i.e., the image signal intensity is determined by intrinsic tissue T₁). Subtraction of the two images leaves a signal that is directly related to flow, and the relation is the same that as given earlier for EPISTAR (Eq. 13). In the general case when T_1app T_1a, the biexponential expression of Eq. 14 applies. This is true because EPISTAR and FAIR are equivalent insofar as they involve the acquisition of a pair of images, one of which has inverted blood water flowing into the imaging slice and the other of which has fully relaxed blood water flowing in. In the two techniques, the state of the static tissue magnetization is different at the time of image acquisition (FAIR images derive from an inversion recovery, whereas in EPISTAR the tissue is saturated before the application of the inversion tag). However, since the static component of the signal is removed by subtraction of the spin-labeled image from the control image, the behavior of the difference signal with inversion time is the same in both techniques (Fig. 10).

Figure 10.

Raw spin-labeled images and difference images (control-labeled) of the human brain for EPI and signal targeting with alternating radiofrequency (EPISTAR) and FAIR as a function of inversion time (TI). From top to bottom: EPISTAR raw images; FAIR raw images; EPISTAR difference images; FAIR difference images. The TI values from left to right are as follows: 200, 400, 600, 800, 1000, and 1200 milliseconds. Whereas the static tissue contrast is different for the two techniques, the difference images are dependent only on the inflow of labeled blood. (Reprinted from Wong et al. [1997] Implementation of quantitative perfusion imaging techniques for functional brain mapping using pulsed arterial spin labeling. NMR Biomed 10:237–249. Copyright John Wiley & Sons Limited. Reproduced with permission.)

Full figure and legend (314K)

Transit time effects. A practical consideration for both EPISTAR and FAIR is the interaction of the edges of the inversion slice with the imaging slice. Ideally, the slice profiles of both the imaging and inversion pulses would be sharp "top-hat" functions, and the inversion slice would be placed immediately adjacent to (for EPISTAR) or superimposed over (for FAIR) the imaging slice. However, the actual inversion profiles generated by hyperbolic secant (sech) adiabatic RF pulses are not perfectly rectangular. The edges of the inversion slice have a certain degree of slope caused by the transition between spins meeting and not meeting the adiabatic condition. Spin relaxation during the play out of the pulse also can degrade the shape of the pulse profile (Frank et al., 1997). To minimize the interaction between the inversion and imaging slice profiles, EPISTAR usually is implemented with an in-plane RF presaturation pulse of thickness slightly greater than the imaging slice, applied immediately before spin labeling (Table 2). Also, it is necessary to move the edges of the inversion slice away from the imaging slice, which means that the inversion slice is shifted away from the imaging slice for EPISTAR and widened for FAIR. A typical inversion-imaging slice thickness ration that has been used in FAIR is 3:1 (Kim, 1995), which was found to be the minimum necessary to avoid subtraction artefacts from static tissue signal. As a result of this, a transit time is introduced during which the inflowing blood is not in the state assumed by the standard model (fully relaxed for FAIR, fully inverted for EPISTAR). Attempts have been made to minimize the inversion-imaging slice thickness ratio by using frequency offset corrected inversion (FOCI) adiabatic pulses (Ordidge et al., 1996) for improved slice definition in FAIR (Pell et al., 1998; Yongbi et al., 1998). A second practical point to consider is the width of the inverted slab of spins. If the bolus width is insufficiently wide, fresh spins that have not been inverted will flow into the imaging slice during the inversion time. In EPISTAR, this width is defined in the pulse sequence by the band width of the inversion pulse and the slice-select gradient; in FAIR, the range of spins affected by the nonselective inversion pulse is defined by the physical extent of the RF coil. If inflow of fresh spins occurs, the perfusion signal will be less than predicted by the standard ASL model. Therefore, the presence of either a transit time or inflow affects the accuracy of perfusion quantification (Kim et al., 1997b; Pell et al., 1999a).

An alternative kinetic model has been put forward by Buxton et al. (1998) to describe the behavior of the signal difference in pulsed and continuous ASL experiments, which takes into account a finite transit time and limited bolus width. These factors can be incorporated into the standard T₁ model described earlier (Kim et al., 1997b; Pell et al., 1999a), leading to similar predictions for the behavior of M as a function of inversion time for the two models. Although generally less than in continuous ASL, the transit time in pulsed ASL can be large, particularly in multislice acquisitions. Using standard PASL techniques, it is necessary to acquire image pairs at a range of inversion times to estimate the transit time and account for its effect on perfusion quantification. Alternatively, PASL pulse sequences can be modified to reduce their sensitivity to transit time effects and allow the acquisition of a purely perfusion-dependent subtraction image. The most promising of these modifications have been proposed by Wong et al. under the acronyms QUIPSS (QUantitative Imaging of Perfusion using a Single Subtraction) and QUIPSS II (Wong et al., 1998a). In QUIPSS, a saturation pulse is applied to the imaging slice at a time TI₁ after application of the labeling or control RF pulse, and the image is acquired at a later time TI₂. By doing this, any signal difference that evolves during TI₁ is destroyed, and only spins that enter during the subsequent period TI (= TI₂ - TI₁) before image acquisition contribute to the perfusion difference signal. Under these circumstances, perfusion quantification is transit time independent as long as TI₁ is longer than the longest transit time of the system, and TI is less than the temporal width of the remaining spin-labeled bolus. In QUIPSS II, a saturation pulse is applied to the labeled region at time TI₁, effectively "clipping" the trailing edge of the bolus and thus giving it a well-defined temporal width, and the image is again acquired at TI₂. For this sequence, the conditions for transit time insensitivity are as follows: (1) TI₁ must be less than the natural width of the labeled bolus, and (2) TI must be greater than the longest transit time of the system so that the entire bolus enters the imaging slice before the image acquisition. The conditions for QUIPSS II are more easily met than for QUIPSS, and QUIPSS II also has the advantage of allowing some washout of labeled intravascular signal during the TI period. Contamination of the perfusion signal by the signal from labeled blood in large vessels is minimized, resulting in a more accurate measurement of tissue perfusion. The drawback of the technique is a reduction of the perfusion signal difference compared with standard PASL techniques because of T₁ decay during TI.

The introduction of a transit time-insensitive PASL sequence has allowed the extension of the technique to multislice acquisitions. Initially, FAIR was implemented in a multislice mode merely by increasing the width of the slice-selective inversion so that several slices were contained within the inversion slab (Kim and Tsekos, 1997a). Also, EPISTAR has been used to obtain multislice perfusion data by acquiring several image slices adjacent to the inversion slab (Edelman and Chen, 1998). However, in these approaches, a range of transit times is automatically introduced by the different physical distances between each imaging slice and the closest edge of the inversion slice, thus making accurate quantification difficult. For multislice PASL, QUIPSS II is a natural choice because of its inherent transit time insensitivity. The main problem for QUIPSS II is the need to limit the second inflow time TI to a length that is not much longer than tissue T₁ to prevent the decay of a significant proportion of the perfusion label. Since the condition for transit time insensitivity is that TI is longer than the longest transit time present, a limitation on the length of TI limits the robustness of the sequence and, therefore, limits the number of slices that can obtained in a single acquisition.

Intravascular signal. As with CASL, the contribution of blood water signal needs to be considered in the quantification of the PASL difference signal. The inclusion of a bipolar gradient in the imaging sequence has been shown to reduce the perfusion signal and cause a lengthening of the measured transit time (Wong et al., 1998b). This is interpreted as a reduction in the amount of spin-labeled blood water signal present in the subtraction images, since the rapid motion of the intravascular water should cause its signal to decay rapidly in the presence of a field gradient. However, choice of the magnitude of the gradient strength (or "b" value) is arbitrary and generally is empirically determined as the point at which a further increase in b value does not alter the calculated perfusion value. There remains some debate in the literature as to whether blood signal should contribute in any way to the perfusion signal. In the original T₁ model for ASL, it is assumed that all signal originates from tissue water, since CBV is relatively small and crusher gradients are used. However, it has been suggested (Buxton et al., 1998) that blood water, which is contained within a voxel in the imaging slice and ultimately is destined to perfuse that voxel but which at the time of image acquisition still is in the intravascular space, should be contribute to the PASL signal. This is distinct from blood that is merely passing through a voxel in an artery or arteriole and is traveling to a capillary bed outside of that voxel; this blood should not be included in the perfusion signal. If the signal from intravascular, perfusing blood is not acquired, a quantity slightly different to absolute perfusion is measured. The measurement represents the water extraction fraction-flow product E(f) f (Zaharchuk et al., 1998). In the situation where the extraction fraction is approximately equal to unity, which is true under normal circumstances, this product is equivalent to CBF. However, at higher flow rates where the extraction fraction is known to deviate substantially from unity, a misleading result will be obtained if exclusively tissue water signal is acquired.

In practice, the relative contribution of tissue and blood signal is difficult to assess, since the source of signal is not easily identified. If a bipolar crusher gradient is applied, calculations may be performed to estimate the attenuation of randomly diffusing water signal, although how directly this relates to blood traveling through the arteriole-capillary-venule system is unclear. Complete elimination of signal from blood traveling through a voxel, while simultaneously leaving perfusing intravascular water signal unaffected, is not possible with the use of bipolar gradients (Henkelman, 1994). The extent to which slowly flowing capillary water is attenuated depends on factors such as the exact flow rate and vessel geometry. If blood water signal does contribute to the PASL perfusion signal, a secondary complicating effect is introduced by the different T₂ (or T₂*) values for blood water spins and tissue water spins. In the human brain, T₂ of intravascular water has been estimated to be a factor of approximately 3 greater than that of tissue water (240 versus 80 milliseconds at 1.5 T in Buxton et al., 1998). This could pose a problem for methods that use EPI for image acquisition, since at any given echo time the signal will be weighted toward the intravascular compartment. For a multi-TI experiment, where spin-labeled water progressively shifts from the intravascular to tissue compartments as the inversion time increases, this introduces a systematic bias to the data. This problem has not been addressed in the literature but could potentially affect the values of perfusion obtained using ASL substantially, depending on the intravascular lifetime of spin-labeled water in voxels in the imaging slice. If TurboFLASH (Haase, 1990) image acquisition is used, the problem is minimized because of the extremely short echo times used; however, T₁ relaxation during the imaging period and the saturation effects of repeated RF excitation pulses make the quantification of TurboFLASH ASL images subject to the validity of certain assumptions (e.g., that the image contrast is dominated by the central lines of k-space). Also, TurboFLASH is not amenable to rapid multislice imaging.

Alternative pulsed arterial spin labeling methods. Since the introduction of EPISTAR and FAIR several years ago, a range of methods has been proposed based on the concept of PASL. For example, Un-inverted flow-sensitive alternating inversion recovery (UNFAIR) (Helpern et al., 1997b) is an approach similar to FAIR but modified so that the static tissue component of the signal remains at equilibrium rather than being inverted in both the spin-labeled and control images (Table 2). Other suggested approaches include STAR-HASTE (Chen et al., 1997), PARIS (Helpern et al., 1997a), TILT (Golay et al., 1998), BASE (Schwarzbauer and Heinke, 1998), SMART (Kao et al., 1998), and FAIRER (Zhou et al., 1998). The relative benefits and drawbacks of each of these methods have yet to be fully established, and the reader is referred to the relevant references for details of each of the sequences.

Other arterial spin labeling quantification issues

Accurate quantification of perfusion with the ASL techniques in animals and humans requires consideration of several additional factors. These factors are summarized here:

1. Implementation of the CASL method on clinical scanners: Initial studies using CASL were performed on small animals, where the problems associated with transit times and MT effects are minimal. The clinical implementation is hampered by the increased magnitude of the transit times in humans. Loss of the label also is intensified by the shorter longitudinal relaxation times at the typically lower field strengths. A further issue is related to RF power deposition. The limits on the duty cycle of RF amplifiers and the specific absorption rate of the sequence often restrict the application of a continuous pulse over the required duration (several seconds). Instead, a pulse train is applied consisting of repeated rectangular pulses with a duty cycle of approximately 75% to 90% (Roberts et al., 1994; Ye et al., 1997).
2. Degree of inversion, : An accurate measurement of is necessary for both continuous and pulsed ASL. The method of determination described by Zhang et al. (1993) has been criticized over the measurement location and the bias toward the lower range of laminar flow velocities present in the carotid artery (Maccotta et al., 1997). The degree of inversion is considerably more velocity-dependent for the CASL technique than for the pulsed techniques (Wong et al., 1998b), since the fulfillment of the adiabatic condition in the case of the continuous adiabatic fast passage pulse is intrinsically related to the velocity of spin passage through the labeling gradient (Dixon et al., 1986). Individual calibration of the value of may, therefore, be necessary for the CASL technique.
3. Blood-brain partition coefficient, : A uniform value of often is assumed in previous studies of CBF quantification, but this may be an unrealistic assumption. Different values of the partition coefficient have been reported for gray and white matter and for varying hematocrit levels (Herscovitch et al., 1985). It is also possible that the regional values of will vary as a consequence of evolving pathophysiologic mechanisms. An imaging method for determination on a pixel-by-pixel basis has been described (Roberts et al., 1996).
4. BOLD contrast: Simultaneous flow and T₂*-contrast information can be extracted from a GE-based ASL experiment. Several groups have used FAIR (Kim et al., 1997b, 1997c; Zhu et al., 1998) and QUIPSS (Wong et al., 1997) to study the combined flow and BOLD responses to functional activation paradigms. For flow quantification studies using GE or SE image acquisition, the respective T₂*- or T₂-weighting of the signal must be extracted (see Intravascular Signal), and this procedure usually is based on baseline measurements (Kim, 1995).
5. CSF contamination: The partial volume effect of CSF, which is exacerbated with the larger voxel sizes used in clinical imaging, can result in significant underestimation of flow. Kwong et al. (1995) predicted an underestimation of approximately 30% for a gray matter-CSF volume fraction of 50%:50%.
6. T₁ of blood, T_1a: The measurement of flow with ASL requires a determination of T_1a. Previous studies report a value of this parameter by extrapolation from published in vitro data. The dependence of T_1a on factors such as oxygenation, hematocrit, vessel size, and environment may be significant and requires further study.
7. Venous contamination: The blood in the venous circulation during the experiment may have been labeled by the control or labeling experiments and thereby provide an unwanted contribution to the subtraction signal. This depends on the chosen labeling scheme.
8. Geometrical layout of the tagged vasculature: Early implementations of the CASL technique used a labeling plane in the neck that is perpendicular to the direction of the feeding artery. To reduce the transit time, the labeling plane can be displaced to the level of the blood vessels that lie at the base of the brain (Alsop et al., 1996; Ye et al., 1997b), but the component of the blood velocity along the direction of the labeling gradient must be sufficient to fulfill the adiabatic condition. For both pulsed and continuous techniques, an indirect path of the blood to the imaging slice is reflected as a longer transit time. The techniques are not sensitized to blood that remains in vessels running parallel to and within the imaging slice during the course of the experiment.

ARTERIAL SPIN LABELING APPLICATIONS

Cerebrovascular disease

Since the establishment of the ability of the techniques to reflect tissue perfusion status, ASL has been used increasingly as an experimental tool in several areas of investigation. The quantitative accuracy of the flow measurements of pulsed and continuous techniques has been investigated using animal and human studies (Table 1). Although pulsed ASL has almost exclusively been used in the field of functional magnetic resonance imaging (fMRI) (see Functional Brain Mapping), continuous ASL also has been successively applied to studies of cerebrovascular disease, both in experimental and clinical settings. This section briefly summarizes the main applications of ASL.

Soon after the technique was proposed, CASL was used by Jiang et al. (1993) in a rat model of transient MCAO and by Detre et al. (1993) to demonstrate the redistribution of blood flow through anastomoses after unilateral occlusion of the common carotid artery in the rat. By labeling blood flowing through the unoccluded artery only, the effective territory of that artery and how the territory changed after occlusion of the other artery was shown. In the following year, Jiang et al. (1994) performed a study in which they monitored the effect of whole-body hypothermia on the recovery of perfusion after transient MCAO. Cerebral blood flow was measured quantitatively by CASL with a time resolution of 8.5 minutes using a standard spin echo imaging approach, and it was shown that the degree of postischemic hypoperfusion is reduced if hypothermic conditioning is applied. The same group recently used CASL to study reperfusion levels after anti-intercellular adhesion molecule 1 antibody treatment of MCAO in the rat (Jiang et al., 1998b) and to investigate the thrombolysis of embolic stroke after administration of rt-PA (Jiang et al., 1998a), again in the rat brain. Busch et al. also studied the effect of rt-PA on an embolic stroke insult (Busch et al., 1998), comparing the results from ASL perfusion with ADC measurements. Their results show that an improvement in blood flow as shown by the ASL perfusion maps led to a slowly progressing reversal of ADC changes in the periphery of the ischemic territory, indicative of tissue recovery. In an experiment looking at the acute changes of tissue relaxation parameters after MCAO, Calamante et al. (1999) used CASL perfusion mapping to identify the extent and the severity of ischemia in the affected MCA territory. In a similar study, CASL was used to characterize the changes observed in a model of partial MCAO (Fig. 11). Our group also has been using FAIR perfusion imaging in the study of global ischemia-reperfusion time courses (Pell et al., 1999a). In these experiments, perfusion was measured with a time resolution of 4 minutes, allowing a period of acute postischemic recovery of flow to be defined, followed by a prolonged period of hypoperfusion (Fig. 12). The combination of CBV and CBF measurements with the techniques of contrast agent MRI and CASL, respectively, has been made possible by the development of a contrast agent with modified susceptibility characteristics (Zaharchuk et al., 1998). A study of the flow and volume response to MCAO in the rat was thereby undertaken. Forbes et al. (1997) used ASL to measure CBF and CO₂ reactivity after traumatic brain injury caused by controlled cortical impact. All of these studies demonstrate the ability of ASL methods to provide time courses of perfusion measurements with good temporal resolution in a series of individual subjects.

Figure 11.

An example of the ability of CASL to differentiate between different levels of CBF. The top row shows blood flow in an coronal slice of a rat brain before (a) and 250 minutes after (b) occlusion of the middle cerebral artery (MCAO). (c) Perfusion levels in the MCA territory are severely reduced, as indicated by the concomitant reduction in the trace ADC [Tr(D)] in this area. The bottom row shows the equivalent images for a partial MCAO. The CBF maps show blood flow levels to be perturbed, but to a lesser extent than for the full MCAO, and the reduction is insufficient to cause a change in the ADC.

Full figure and legend (142K)

Figure 12.

Reperfusion time courses in a gerbil model of transient forebrain ischemia (n = 8). Measurements were performed using a modified version of FAIR with improved time efficiency (Pell 1999a). The control blood flow measurements were 158 24 mL 100 g^-1 min^-1 in the cortex and 130 17 mL 100 g^-1 min^-1 in the striatum. On initiation of recirculation, the blood flow recovered and, at approximately 5 minutes of reflow, reached a maximal level; a period of hypoperfusion followed. An expotentially damped polynomial model was fitted to the first 60 minutes of postocclusion data (dark line) to characterize features of the flow response during the acute stage of the reperfusion (Pell et al. 1999b).

Full figure and legend (113K)

In the clinical environment, initial efforts have been made by Detre et al. (1998a) to determine cerebrovascular reserve by assessing the response of the CASL flow measurement to ACZ. The group also has used the technique to evaluate CBF deficits in patients who presented with stenoses and cerebrovascular accidents (Detre et al., 1998b). Evidence of watershed hypoperfusion was found. Similar attempts have been made for EPISTAR (Siewert et al., 1997). However, both techniques have the inherent problems associated with long arterial transit times, and quantitative blood flow measurements have not been reported. Table 1 summarizes the flow values obtained by quantification studies using the ASL techniques.

Functional brain mapping

Initial studies. Since the pioneering work of Roy and Sherrington (1890), it has been known that there is a close spatial and temporal correspondence between local changes in brain neuronal activity and changes in regional cerebral blood flow (rCBF). Whereas the mechanisms of this flow regulation still are poorly understood in detail (Villringer and Dirnagl, 1995), it is well established that blood flow typically changes by 50% or more in gray matter within an area close to that of increased electrical activity (Puce, 1995) and within a few seconds of the onset of increased activity. The well-established technique of PET (Frackowiak et al., 1997) has been used since the early 1980s to monitor rCBF during functional brain activation and is regarded by some as a gold standard for such measurements. The new MRI techniques discussed in this review, measuring ostensibly the same parameter, can be regarded as directly comparable with PET. This is not the case with the more popular fMRI technique of BOLD contrast, which uses the paramagnetic properties of deoxyhemoglobin. Here, the functionally induced image intensity may depend on several parameters, including blood flow, oxygen extraction, hematocrit, local vascular geometry, and baseline blood oxygen saturation. In the special case of SE BOLD contrast (van Zijl et al., 1998), quantification appears to be possible, but this technique has low sensitivity except at high magnetic fields.

The first study using MRI perfusion techniques to observe functional activity in human brain was made by Belliveau et al. (1991). The contrast agent bolus passage method was used, with two venous injections of gadolinium spaced 20 minutes apart. Echo planar imaging was used to monitor the passage of the agent, with a repeat time of 1 second. During the first injection, the subject lay in darkness, whereas during the second he was given visual stimulation with flashing light-emitting diodes. The images from each injection were transformed into relative blood volume maps, and the map made when the subject was in darkness was subtracted from that made during visual stimulation. A significant increase of blood volume in occipital cortex was observed, consistent with PET and other studies of human cerebral visual areas. No attempt was made to quantify perfusion changes in this study. Because of the potential toxicity of gadolinium in repeated doses large enough to give adequate signal changes-only two scans per day can be performed-this technique has not found popularity in functional brain mapping studies.

However, this first publication aroused enormous international interest in MRI as a means of exploring human brain function, and it was not long before the technique of BOLD imaging, using the inherent magnetic properties of deoxyhemoglobin, nature's own contrast agent, was invented by Ogawa (1990) and Turner (1991), working with animal models. Kwong (1992) and Ogawa (1992) soon demonstrated the extreme power of this method in mapping human brain activity, again using visual stimulation. However, in addition to the use of BOLD, Kwong et al. (1992), using an early version of ASL, also obtained results that were entirely consistent with their BOLD data (Fig. 13). The method used RF pulses to invert the spins within the slice to be imaged, and an inversion time of 1100 milliseconds was used to allow noninverted perfusing spins time to move into the imaging slice. This technique is not capable of giving an absolute measure of blood flow, but using Detre's (1992) formalism, Kwong et al. were able to estimate the change of blood flow in occipital cortex produced by visual stimulation. This gave a result (CBF = 55 mL 100 g^-1 min^-1), which is consistent with earlier PET studies. Characteristically, however, the sensitivity of this technique was lower than with BOLD contrast. The next development was made by Edelman et al. (1994), who introduced and applied EPISTAR (see Pulsed Arterial Spin Labeling Techniques) to study local brain activity produced by finger tapping.

Figure 13.

Noninvasive, real-time mapping of V1 activation during visual stimulation. Images are obliquely aligned along the calcarine fissure with the occipital pole at the bottom. Images were acquired at 3.5-second intervals using a slice-selective inversion-recovery sequence. A baseline image acquired during darkness (upper left) was subtracted from subsequent images. Eight of these subtraction images are displayed and were chosen when the image intensities reached a steady-state level, during darkness (OFF), and during 8-Hz photic stimulation (ON). During stimulation, local increases in signal intensity are detected in the medial-posterior regions of the occipital lobes along the calcarine fissure. (Reprinted from Kwong et al. [1992] Dynamic magnetic resonance imaging of human brain activity during primary sensory stimulation. Proc Natl Acad Sci USA 89:5675–5679 with permission.)

Full figure and legend (594K)

Imaging neuroscience studies in humans with arterial spin labeling. A further conspicuous advantage of EPISTAR and other ASL techniques over BOLD and bolus passage contrast agent methods is that the label produced by RF inversion of arterial spins is short lived, lasting only long enough to be observed in capillaries in the imaging slice of interest. Thus, minimal label passes downstream into veins, in contrast with the other methods, and hence ASL techniques have been claimed to provide better spatial specificity for cortical activations. This inspired a more advanced series of experiments on human hand movement using the EPISTAR method (Sanes et al., 1995), which considered the question of somatotopy in M1. Conventional wisdom, dating from Penfield and Boldrey's (1937) work, suggests that motor representations of the digits lay in an orderly sequence along the anterior wall of the central sulcus and the adjacent precentral gyrus. Sanes et al. (1995) were able to show that movements of individual digits each activated overlapping mosaics of areas scattered through the overall hand representation with little or no signs of a spatial sequence, a result consistent with single electrode studies on monkeys by Merzenich and his group (Nudo et al., 1992). In addition, EPISTAR was used to identify the frontal eye fields in a visual saccade task (Darby et al., 1996).

These successes triggered increased efforts to make ASL more quantitative, and thus a more complete alternative to PET. A variety of techniques involving different patterns of RF pulses have been developed. Several of the studies involving specific brain tasks have been undertaken by the group of McLaughlin at the National Institutes of Health using the CASL technique. Whether the local changes in CBF seen using the relatively high spatial resolution of MRI methods is comparable with that obtained using PET is of interest. The first of these studies (Ye et al., 1997c) measured rCBF changes in motor cortex with a finger-tapping paradigm. Without spatial smoothing, the spatial resolution of the functional images was about 4 mm, and the mean change in CBF over five subjects was 91%, considerably larger than that noted using PET (Fig. 14). This value was confirmed in a similar study by Kim et al. (1997a) using the FAIR technique. However, when images were smoothed to a full width at half maximum of 15 mm, more typical of multisubject PET studies, the mean CBF fell to 42%, now in agreement with previously reported ¹³³Xe and PET studies. This strongly suggests that earlier non-MRI studies have been prone to underestimate CBF because of uncorrected partial volume effects. A further study by this group (Ye et al., 1998) investigated a more cognitive, working memory task, involving button press responses to numbers presented on a screen two steps previously-the "two-back" task. This ASL study reproduced results using BOLD well but also provided a significant confirmation of the PET finding that CBF (23% averaged over six subjects in this study) is considerably lower in activated prefrontal cortical areas known to support working memory than blood flow changes in more primary cortex.

Figure 14.

Cerebral blood flow activation images for seven control subjects using the CASL technique. The CBF measurements were convolved with a Gaussian kernel having a full width half maximum of 5.6 mm. Regions marked in black overlay had statistically significant differences in CBF during finger tapping. (Reprinted from Ye et al. [1997] Quantification of regional cerebral blood flow increases during motor activation: a steady-state arterial spin-tagging study. Neuroimage 6:104–112 with permission.)

Full figure and legend (303K)

For functional studies, imaging more than a single slice of the brain in each experimental run clearly is of value (Kim et al., 1997b, Wong et al., 1997). This is discussed more fully elsewhere in this review, but the careful study extending coverage to 10 slices using FAIR with spiral EPI (Yang et al., 1998) is worthy of note. In this study, the areas of increased CBF for a finger movement task were shown to be similar to those found with BOLD contrast, confirming earlier work (Siewert et al, 1996).

Functional arterial spin labeling studies in animals. Meanwhile, functional studies using laboratory animals were being conducted in several laboratories. These investigations have increased understanding of the neurophysiologic basis of BOLD contrast. Kerskens et al. (1996) adapted a rat model showing increases in blood flow using CASL in somatosensory cortex during forepaw stimulation. The CBF effects of various types of anesthesia also were studied. They found that -chloralose anesthesia best ensures persistence of CBF autoregulation to observe effects of hypercapnia on local blood flow changes (Bock et al., 1998). Surprisingly, they found that although the local blood flow increase induced by forepaw stimulation increased by 40% under hypercapnic conditions, there was no significant change in the BOLD contrast signal associated with the stimulation. This is consistent with the hypothesis that when blood flow is globally increased by hypercapnia, during activation local perfusion and oxygen extraction also increase in parallel. These results disagree with those obtained using PET in human brain with a visual stimulus (Ramsay et al., 1993), which showed no variation of local blood flow increase with hypercapnically induced changes in global blood flow. The sources of the discrepancy may be species differences in vascular density and regulation mechanisms.

Another important study (Schmitz et al., 1998) deals with the recovery of rat cerebral vascular autoregulation after cardiac arrest. One hour after successful resuscitation after 10 minutes of cardiac arrest, the ADC did not differ from control, but functional activation associated again with forepaw stimulation was completely suppressed. After 3 hours of reperfusion, functional activity began to reappear, but the recovery of the BOLD signal progressed faster than that of the CASL perfusion-weighted signal. On previous rat studies using electrophysiologic methods and laser Doppler flowmetry, this group found that cortical electrical activity also was slow to recover after cardiac arrest and was mirrored precisely by the recovery in rCBF.

Silva et al. (1995) were the first to perform pharmacologic fMRI using perfusion techniques in a rat model. They examined the CBF response to amphetamine using the two-coil CASL method and found that amphetamine causes significant and spatially specific increases in perfusion in many areas of the brain, including the cortex, cingulate, and caudate putamen (Fig. 15), in agreement with previous results using deoxyglucose uptake to monitor brain activation. This pioneering study indicates the power of perfusion methods to indicate in a simple way the localization of the effect of psychopharmaceutical agents in the brain, although care must be taken to control for any global effects on blood flow caused by the crude vasoactive properties of such drugs. Later work from the same laboratory (Forman et al., 1998) used picrotoxinin (a gamma-aminobutyric acid antagonist) to stimulate rat brain, whereas perfusion was measured by CASL, and brain glutamate was independently assayed using microdialysis. After 30 minutes of picrotoxinin-induced stimulation, glutamate levels decreased but CBF was found to remain elevated, suggesting that additional factors modulate the relation between neuronal neurotransmitters and hemodynamics at these later stages.

Figure 15.

Five-slice perfusion images (two-coil CASL technique) of a rat brain before and after the intraperitoneal administration of 20mg/kg of D-amphetamine in saline: (A) anatomical SE slices; (B) perfusion images before the administration of D-amphetamine; (C) perfusion images after the administration of D-amphetamine; (D) difference images found by subtracting (B) from (C). (Reprinted from Silva AC, Zhang WG, Williams DS, Koretsky AP [1995] Multi-slice MRI of rat brain perfusion during amphetamine stimulation using arterial spin labeling. Magn Reson Med 33:209–214 with permission.)

Full figure and legend (190K)

Although such studies clearly have great potential for understanding regional effects of drugs affecting neurotransmission, major gaps exist in our understanding of the normal energy-providing mechanisms for brain functional activity. Coupled with BOLD contrast, ASL seems likely to shed considerable light on the question of coupling of changes in neuronal activity, CBF, and oxygen extraction. Kim and Ugurbil (1997c) were the first to make quantitative studies of the differences in amplitude between BOLD and FAIR perfusion measures of activation. They report that BOLD signal change is not linearly correlated with the relative CBF increase across the subjects they studied and is negatively correlated with the oxygen consumption change they estimated using the Ogawa (Ogawa et al., 1993) model of the BOLD effect.

Later work by Davis et al. (1998) used the effects of hypercapnia, which is not thought to increase cortical oxygen consumption, on ASL and BOLD measurements to calibrate these measures for subsequent functional activation studies (Fig. 16). This calibration procedure provided a regional measurement of the expected sensitivity of the fMRI BOLD signal to changes in the electrical activity of the brain. Using this more robust experimental design, Davis et al. (1998) were able to show that BOLD signal magnitude was 32% less than expected from the calibration, and this was presumably caused by the action of oxygen metabolism. With a visual stimulus, the oxygen consumption calculated from the BOLD and ASL measurements increased by 16%, whereas blood flow increased by 45%. The finding that oxygen consumption promptly increases with the onset of increased neuronal activity throws doubt on earlier widely cited PET results (Fox and Raichle, 1986), which were consistent with no upregulation of oxygen uptake during functional activity, and contradicts a fMRI study (Kruger et al., 1996) using extended visual stimulation, which suggested that oxygen consumption takes tens of seconds to upregulate. Further studies, perhaps taking advantage of the greater sensitivity to be obtained at higher magnetic field strengths, should shed further light on this disturbingly under-researched area.

Figure 16.

Blood oxygenation level-dependent (BOLD) and FAIR CBF time courses normalized to the hypercapnia signal change, averaged across eight trials. Compared with hypercapnia-induced signal change, the CBF signal outstrips the BOLD signal changes during photic stimulation. Regional cerebral metabolic rate of oxygen (rCMRO₂) time course calculated from the same data shows metabolic response within seconds of photic stimulation onset. No temporal soothing was done: all time points (6-second resolution) were collected and calculated independently. Notice that by analytic design, the average rCMRO₂ during the hypercapnia and baseline periods both are set to 1. (Reprinted from Davis et al. [1998] Calibrated functional MRI: mapping the dynamics of oxidative metabolism. Proc Natl Acad Sci USA 95: 1834–1839 with permission. Copyright 1998 National Academy of Sciences, U.S.A.)

Full figure and legend (153K)

ADVANTAGES AND DISADVANTAGES OF PERFUSION MRI TECHNIQUES

Exogenous versus endogenous contrast agents

Although methods using both exogenous and endogenous contrast agents can be considered as noninvasive (neither requiring ionizing radiation), ASL has the extra benefit of using an endogenous contrast agent. Therefore, multiple measurements can be performed, in principle, without any restriction. Although multiple bolus injections of contrast agent can be performed, the acceptable maximum dose is limited. A lower dose must be administered in multiple measurements, thus reducing the sensitivity of the technique.

Behavior under extreme flow conditions

Although both MRI techniques using exogenous or endogenous contrast agents behave reasonably well in the range of normal CBF values, they have difficulties when dealing with high or low flow. With a low flow, the first obvious limitation is the low SNR of the measurement. This can be improved in the ASL technique by increasing the number of averages (at the expense of increasing the measurement time). On the other hand, in the typical single-bolus DSC studies, because of its dynamic character, averaging is not an option. One possible solution is the use of a larger dose, although the maximum dose is limited (e.g., by Food and Drug Administration regulations). However, SNR is not the only problem when measuring low flows. The second problem is related to the presence of long arrival delays for the arterial blood to the VOI, which may result from collateral circulation in the periphery of stroke regions. Extra delays of up to several seconds can be observed in the ischemic tissue. Although the latest ASL sequences proposed (Alsop and Detre, 1996; Wong et al., 1998a) are less sensitive to differences in transit time (provided an appropriate choice of sequence parameters are chosen), long transit times may make use of the required parameters impractical. In PASL, the transit time insensitivity can be further improved by slice profile optimization (Pell et al., 1998). Another possibility would be to take account of the different transit times, for instance, by measuring them (Zhang et al., 1992; Ye et al., 1997b; Wong et al., 1997; Barbier et al., 1998; Branch et al., 1998), or by using the bolus arrival time information from a DSC experiment. However, this problem is one of the fundamental limitations of the ASL technique for measuring CBF in the presence of long transit time delays. In the case of a long delay, it may be impossible to have tagged signal arriving at the tissue and, therefore, no effect will be observed. There is no simple solution to this problem apart from the use of a higher magnetic field strength with the associated longer T₁.

The presence of long bolus arrival times also is a problem in DSC studies. The transit of the bolus from the site of injection (typically a vein) to the VOI can introduce a delay, as well as dispersion (spread of the bolus shape). This dispersion effectively reduces the labeling (bolus sharpness), and, therefore, the dispersion can be thought of as a "relaxation" process (the more spread the bolus, the smaller the observed effect). If one does not account for this delay and dispersion, it may cause an overestimation of MTT and underestimation of CBF (Østergaard et al., 1996a, 1996b, 1999; Schreiber et al., 1998). However, this apparent increase in MTT is not related to the increase caused by true intravoxel dispersion. The main problem is that the DSC model cannot distinguish between dispersion in the feeding vessels (from the place where the AIF was estimated, e.g., MCA, to the VOI) and intravoxel dispersion by tissue microvasculature within the VOI (Østergaard et al., 1999; Schreiber et al., 1998). Although one can account for the dispersion from the site of injection (vein) to the artery where the AIF is estimated, this can not been done for the extra dispersion from the artery to the input of the VOI (where the true AIF should be obtained). This extra dispersion is particularly significant in cases of collateral flow to ischemic areas. To address this problem, Østergaard et al. (1999) have incorporated a mathematical vascular model into the DSC analysis. They show that the vascular model approach is less sensitive to vascular delay and dispersion than the conventional deconvolution approach. Their results are promising, but further validation on many patients with cerebrovascular diseases is needed.

The case of high flows also is problematic for the ASL technique because the assumption of complete exchange is no longer valid (Silva et al., 1997a, 1997b). If this is not taken into account, the perfusion will be underestimated (since only a fraction of the arterial magnetization is exchanged).

Regional coverage

Both techniques can be used in a multislice protocol, although the maximum number of slices is limited in each case. In DSC-MRI, the coverage is given by a compromise between the number of slices and the time resolution. For a full characterization of the bolus passage, high time resolution is required (the sharper the bolus, the higher the time resolution). For typical repetition time values, a coverage of 10 to 15 slices can be achieved using EPI. Another possibility is the use of three-dimensional acquisitions (Petrella et al., 1997), although with a lower spatial resolution. For the case of ASL, as mentioned in the section on perfusion MRI using spin labeling of arterial water, there are certain limitations to the maximum number of slices. Accurate quantification from about 5 slices typically can be obtained and, in a recent study, up to 10 slices have been acquired using spiral EPI, although with limited spatial resolution (Yang et al., 1998).

CONCLUSION

The demonstration that MRI can be sensitized to detect levels of CBF has generated considerable research activity. As a result, increasingly accurate quantification is possible. Over the last few years, the DSC technique has been used to a greater extent than ASL. This can be explained by the earlier origin of the former technique. In addition, the analysis of DSC-MRI is more straightforward if summary parameters are used. Furthermore, with the advent of EPI in a clinical environment, multislice DSC-MRI has become an immediate option. On the other hand, the original ASL techniques were not possible to extend to multislice (apart from the two-coil case), and it has only recently become a multislice technique. Notice that the only application where the use of ASL has been more commonplace than DSC-MRI is in the field of fMRI. As mentioned before, this is the application where the complete noninvasive character of ASL makes it particularly superior to DSC-MRI.

In conclusion, although certain issues concerning absolute quantification remain to be fully addressed, both perfusion MRI techniques have evolved to be an important tool in research laboratories and clinical units, and they are used in a routine way by an increasing number of centers every day. For clinicians, the high contrast and ease of data acquisition of DSC-MRI make it appealing, whereas for research on brain function, the noninvasive character of ASL with its freely diffusible tracer, blood water, is highly desirable. Both approaches are likely to evolve over the coming years to become faster and more accurate, although which technique will prove to be superior with regard to noninvasive CBF quantification remains to be seen.

References

1. Abragam A. 1961 Principles of Nuclear Magnetism (Oxford, Clarendon Press).
2. Alsop DC & Detre JA. (1996) Reduced transit-time sensitivity in noninvasive magnetic resonance imaging of human cerebral blood-flow. J Cereb Blood Flow Metab 16: 1236−1249. | Article | ChemPort |
3. Alsop DC & Detre JA. (1998) Multisection cerebral blood flow MR imaging with continuous arterial spin labeling. Radiology 208: 410−416. | ChemPort |
4. Aronen HJ, Gazit IE, Louis DN, Buchbinder BR, Pardo FS & Weisskoff RM et al. (1994) Cerebral blood volume maps of gliomas: comparison with tumor grade and histological findings. Radiology 191: 41−51. | PubMed | ChemPort |
5. Astrup J, Siesjo BK & Symon L. (1981) Thresholds in cerebral ischemia: the ischemic penumbra. Stroke 12: 723−725. | PubMed | ChemPort |
6. Axel L. (1980) Cerebral blood flow determination by rapid-sequence computed tomography. Radiology 137: 676−686.
7. Axel L. 1995 Methods using blood pool tracers: Part II. (In) Diffusion and Perfusion Magnetic Resonance Imaging (Le Bihan D, ed) New York, Raven Press (pp) 205−211.
8. Baird AE, Benfield A, Schlaug G, Siewert B, Lovblad KO & Edelman RR et al. (1997) Enlargement of human cerebral ischemic lesion volumes measured by diffusion-weighted magnetic resonance imaging. Ann Neurol 41: 581−589. | Article | PubMed | ChemPort |
9. Barber PA, Darby DG, Desmond PM, Yang Q, Gerraty RP & Jolley D et al. (1998) Prediction of stroke outcome with echoplanar perfusion- and diffusion-weighted MRI. Neurology 51: 418−426. | PubMed | ISI | ChemPort |
10. Barbier EL, Silva AC, Kim HJ, Williams DS & Koretsky AP. 1998 Dynamic perfusion analysis based on arterial spin labeling. (In: Proceedings of the ISMRM 6th Annual Meeting, Sydney, Australia, p 380).
11. Baron JC, Vonkummer R & Delzoppo GJ. (1995) Treatment of acute ischemic stroke: challenging the concept of a rigid and universal time window. Stroke 26: 2219−2221. | PubMed | ChemPort |
12. Bassingthwaighte JB & Goresky GA. 1984 The cardiovascular system. (In) Handbook of Physiology (Renkin EM, Michel CG, eds) Bethesda, MD, American Physiology Society (pp) 549−626.
13. Bell BA. (1984) A history of the study of the cerebral circulation and the measurement of cerebral blood flow. Neurosurgery 14: 238−246. | ChemPort |
14. Belliveau JW, Kennedy DN, McKinstry RC, Buchbinder BR, Weisskoff RM & Cohen MS et al. (1991) Functional mapping of the human visual cortex by magnetic resonance imaging. Science 254: 716−719. | PubMed | ISI | ChemPort |
15. Berninger WH, Axel L, Norman D, Napel S & Redington RW. (1981) Functional imaging of the brain using computed-tomography. Radiology 138: 711−716. | PubMed | ChemPort |
16. Berthezene Y, Nighoghossian N, Meyer R, Damien J, Cinotti L & Adeleine P et al. (1998) Can cerebrovascular reactivity be assessed by dynamic susceptibility contrast-enhanced MRI? Neuroradiology 40: 1−5. | Article | ChemPort |
17. Bock C, Schmitz B, Kerskens CM, Gyngell ML, Hossmann KA & Hoehn-Berlage M. (1998) Functional MRI of somatosensory activation in rat: effect of hypercapnic up-regulation on perfusion- and BOLD-imaging. Magn Reson Med 39: 457−461. | PubMed | ISI | ChemPort |
18. Boxerman JL, Hamberg LM, Rosen BR & Weisskoff RM. (1995) MR contrast due to intravascular magnetic-susceptibility perturbations. Magn Reson Med 34: 555−566. | PubMed | ISI | ChemPort |
19. Boxerman JL, Rosen BR & Weisskoff RM. (1997) Signal-to-noise analysis of cerebral blood volume maps from dynamic NMR imaging studies. J Magn Reson Imaging 7: 528−537. | ChemPort |
20. Branch CA, Hernandez L, Yongbi M & Helpern JA. 1998 Estimation of arterial arrival time and cerebral response characteristics of perfusion measurements with arterial spin labeling. (In: Proceedings of the ISMRM 6th Annual Meeting, Sydney, Australia, p 375).
21. Buxton RB, Frank LR, Siewert B, Warach S & Edelman RR. 1995 A quantitative model for EPISTAR perfusion imaging. (In: Proceedings of the ISMRM 3rd Annual Meeting, Nice, France, p 132).
22. Buxton RB, Frank LR, Wong EC, Siewert B, Warach S & Edelman RR. (1998) A general kinetic model for quantitative perfusion imaging with arterial spin labeling. Magn Reson Med 40: 383−396. | ChemPort |
23. Busch E, Krueger K, Allegrini PR, Kerskens CM, Gyngell ML & Hoehn-Berlage M et al. (1998) Reperfusion after thrombolytic therapy of embolic stroke in the rat: magnetic resonance and biochemical imaging. J Cereb Blood Flow Metab 18: 407−418. | Article | ChemPort |
24. Calamante F, Williams SR, van Bruggen N, Kwong KK & Turner R. (1996) A model for quantification of perfusion in pulsed labeling techniques. NMR Biomed 9: 79−83. | Article | ChemPort |
25. Calamante F, Lythgoe MF, Pell GS, Thomas DL, King MD & Busza AL et al. (1999) Early changes in water diffusion, perfusion, T₁ and T₂ during focal ischaemia in the rat studied at 8.5 T. Magn Reson Med 41: 479−485. | Article | ChemPort |
26. Caramia F, Aronen HJ, Sorensen AG, Belliveau JW, Gonzalez RG & Rosen BR. 1995 Perfusion MR imaging with exogenous contrast agents. (In) Diffusion and Perfusion Magnetic Resonance Imaging (Le Bihan D, ed) New York, Raven Press (pp) 255−267.
27. Caramia F, Huang Z, Hamberg LM, Weisskoff RM, Zaharchuk G & Moskowitz MA et al. (1998) Mismatch between cerebral blood volume and flow index during transient focal ischemia studied with MRI and Gd-BOPTA. Magn Reson Imaging 16: 97−103. | ChemPort |
28. Chen Q, Siewert B, Bly BM, Warach S & Edelman RR. (1997) STAR-HASTE: perfusion imaging without magnetic susceptibility artifact. Magn Reson Med 38: 404−408. | ChemPort |
29. Cohen BM, Yurgeluntodd D, English CD & Renshaw PF. (1995) Abnormalities of regional distribution of cerebral vasculature in schizophrenia detected by dynamic susceptibility contrast MRI. Am J Psychol 152: 1801−1803. | ChemPort |
30. Cutrer FM, Sorensen AG, Weisskoff RM, Østergaard L, del Rio MS & Lee EJ et al. (1998) Perfusion-weighted imaging defects during spontaneous migrainous aura. Ann Neurol 43: 25−31. | Article | PubMed | ChemPort |
31. Darby DG, Nobre AC, Thangaraj V, Edelman R, Mesulam MM & Warach S. (1996) Cortical activation in the human brain during lateral saccades using EPISTAR functional magnetic resonance imaging. Neuroimage 3: 53−62. | Article | ChemPort |
32. Davis TL, Kwong KK, Weisskoff RM & Rosen BR. (1998) Calibrated functional MRI: mapping the dynamics of oxidative metabolism. Proc Natl Acad Sci USA 95: 1834−1839. | Article | PubMed | ChemPort |
33. De Crespigny A, Röther J, van Bruggen N, Beaulieu C & Moseley ME. (1998) Magnetic resonance imaging assessment of cerebral hemodynamics during spreading depression in rats. J Cereb Blood Flow Metab 18: 1008−1017. | Article | ChemPort |
34. Detre JA, Leigh JS, Williams DS & Koretsky AP. (1992) Perfusion imaging. Magn Reson Med 23: 37−45. | PubMed | ISI | ChemPort |
35. Detre JA, Silva AC, Zhang W, Williams DS & Koretsky AP. 1993 Redistribution of cerebral blood flow following unilateral carotid occlusion: noninvasive determination by perfusion imaging with selective labeling of arterial water. (In: Proceedings of the Society of Magnetic Resonance Medicine, 12th Annual Meeting, New York, p 247).
36. Detre JA, Alsop DC, Samuels OB, Gonzalez-Atavales J & Raps EC. 1998a Cerebrovascular reserve testing using perfusion MRI with arterial spin labeling in normal subjects and patients with cerebrovascular disease. (In: Proceedings of the ISMRM 6th Annual Meeting, Sydney, Australia, p 243).
37. Detre JA, Alsop DC, Vives LR, Maccotta L, Teener JW & Raps EC. (1998b) Noninvasive MRI evaluation of cerebral blood flow in cerebrovascular disease. Neurology 50: 633−641. | ChemPort |
38. Dijkhuizen RM, van der Sprenkel JWB, Tulleken KAF & Nicolay K. (1997) Regional assessment of tissue oxygenation and the temporal evolution of hemodynamic parameters and water diffusion during acute focal ischemia in rat brain. Brain Res 750: 161−170. | Article | ChemPort |
39. Dijkhuizen RM, Knollema S, vanderWorp HB, Ter Horst GJ, De Wildt DJ & van der Sprenkel JWB et al. (1998) Dynamics of cerebral tissue injury and perfusion after temporary hypoxia-ischemia in the rat. Evidence for region-specific sensitivity and delayed damage. Stroke 29: 695−704. | PubMed | ChemPort |
40. Dixon WT, Du LN, Faul DD, Gado M & Rossnick S. (1986) Projection angiograms of blood labeled by adiabatic fast passage. Magn Reson Med 3: 454−462. | PubMed | ISI | ChemPort |
41. Doerfler A, Forsting M, Reith W, Heiland S, Weber J & Hacke W et al. (1997) Bolus injection of MR contrast agents: hemodynamic effects evaluated by intracerebral laser doppler flowmetry in rats. Am J Neuroradiol 18: 427−434. | ChemPort |
42. Edelman RR, Mattle HP, Atkinson DJ, Hill T, Finn JP & Mayman C et al. (1990) Cerebral blood flow: assessment with dynamic contrast-enhanced T₂* weighted MR imaging at 1.5 T. Radiology 176: 211−220. | PubMed | ChemPort |
43. Edelman RR & Chen Q. (1998) EPISTAR MRI: Multi-slice mapping of cerebral blood flow. Magn Reson Med 40: 800−805. | ChemPort |
44. Edelman RR, Siewert B, Darby DG, Thangaraj V, Nobre AC & Mesulam MM et al. (1994) Qualitative mapping of cerebral blood-flow and functional localization with echo-planar MR-imaging and signal targeting with alternating radio-frequency. Radiology 192: 513−520. | PubMed | ChemPort |
45. Ellinger R, Kremser C, Schocke M, Auer A, Felber S & Aichner F. 1998 The impact of the arterial input function on quantitative evaluation of susceptibility contrast enhanced MRI studies. (In: Proceedings of the ISMRM 6th Annual Meeting, Sydney, Australia, p 1207).
46. Finelli DA, Hopkins AL, Selman WR, Crumrine RC, Bhatti SU & Lust WD. (1992) Evaluation of experimental early acute cerebral ischemia before the development of edema: use of dynamic, contrast-enhanced and diffusion-weighted MR scanning. Magn Reson Med 27: 189−197. | PubMed | ChemPort |
47. Fisel CR, Ackerman JL, Buxton RB, Garrido L, Belliveau JW & Rosen BR et al. (1991) MR contrast due to microscopically heterogeneous magnetic-susceptibility: numerical simulations and applications to cerebral physiology. Magn Reson Med 17: 336−347. | PubMed | ChemPort |
48. Fisher M. (1997) Characterizing the target of acute stroke therapy. Stroke 28: 866−872. | PubMed | ChemPort |
49. Forbes ML, Hendrich KS, Kochanek PM, Williams DS, Schiding JK & Wisniewski SR et al. (1997) Assessment of cerebral blood flow and CO₂ reactivity after controlled cortical impact by perfusion magnetic resonance imaging using arterial spin-labeling in rats. J Cereb Blood Flow Metab 17: 865−874. | ChemPort |
50. Forman SD, Silva AC, Dedousis N, Barbier EL, Fernstrom JD & Koretsky AP. (1998) Simultaneous glutamate and perfusion fMRI responses to regional brain stimulation. J Cereb Blood Flow Metab 18: 1064−1070. | Article | ChemPort |
51. Fox PT & Raichle ME. (1986) Focal physiological uncoupling of cerebral blood flow and oxidative metabolism during somatosensory stimulation in human subjects. Proc Natl Acad Sci USA 83: 1140−1144. | PubMed | ChemPort |
52. Frackowiak RSJ. 1997 (In) Human Brain Function (Frackowiak RSJ, Friston KJ, Frith CD, Dolan RJ, Mazziotta JC, eds) San Diego, Academic Press.
53. Frank LR, Wong EC & Buxton RB. (1997) Slice profile effects in adiabatic inversion: application to multi-slice perfusion imaging. Magn Reson Med 38: 558−564. | ChemPort |
54. Gillis P & Koenig SH. (1987) Transverse relaxation of solvent protons induced by magnetized spheres: application to ferritin, erythrocytes, and magnetite. Magn Reson Med 5: 323−345. | ChemPort |
55. Ginsburg MD, Smith DW, Wachtel MS, Gonzalez-Carvajal M & Busto R. (1986) Simultaneous determination of local cerebral glucose utilization and blood flow by carbon-14 double-label autoradiography: method of procedure and validation studies in the rat. J Cereb Blood Flow Metab 6: 273−285. | PubMed | ChemPort |
56. Gobbel GT, Cann CE & Fike JR. (1991) Measurement of regional cerebral blood flow using ultrafast computed tomography: theoretical aspects. Stroke 22: 768−771. | PubMed | ChemPort |
57. Gobbel GT & Fike JR. (1994) A deconvolution method for evaluating indicator-dilution curves. Phys Med Biol 39: 1833−1854. | Article | ChemPort |
58. Golay X, Stuber M, Pruessmann K, Luechinger R, Weber OM & Scheideger MB et al. 1998 Multi-slice functional perfusion imaging using a magnetisation transfer insensitive labeling technique (TILT). (In: Proceedings of the ISMRM 6th Annual Meeting, Sydney, Australia, p 1206).
59. Gonzalez RG, Fischman AJ, Guimaraes AR, Carr CA, Stern CE & Halpern EF et al. (1995) Functional MR in the evaluation of dementia: correlation of abnormal dynamic cerebral blood volume measurements with changes in cerebral metabolism on positron emission tomography with fludeoxyglucose F-18. Am J Neuroradiol 16: 1763−1770. | ChemPort |
60. Gröhn OHJ, Lukkarinen JA, Oja JME, vanZijl PCM, Ulatowski JA & Traystman RJ et al. (1998) Noninvasive detection of cerebral hypoperfusion and reversible ischemia from reductions in the magnetic resonance imaging relaxation time, T₂. J Cereb Blood Flow Metab 18: 911−920. | Article |
61. Gückel F, Brix G, Rempp K, Deimling M, Röther J & Georgi M. (1994) Assessment of cerebral blood volume with dynamic susceptibility contrast-enhanced gradient-echo imaging. J Comput Assist Tomogr 18: 344−351. | PubMed | ChemPort |
62. Gückel FJ, Brix G, Schmiedek P, Piepgras A, Becker G & Kopke J et al. (1996) Cerebrovascular reserve capacity in patients with occlusive cerebrovascular disease: assessment with dynamic susceptibility contrast-enhanced MR imaging and the acetazolamide stimulation test. Radiology 201: 405−412. | PubMed |
63. Haase A. (1990) Snapshot FLASH MRI: applications to T₁, T₂ and chemical-shift imaging. Magn Reson Med 13: 77−89. | PubMed | ChemPort |
64. Harris GJ, Lewis RF, Satlin A, English CD, Scott TM & Yurgelun Todd DA et al. (1996) Dynamic susceptibility contrast MRI of regional cerebral blood volume in Alzheimer's disease. Am J Psychol 153: 721−724. | ChemPort |
65. Harris GJ, Lewis RF, Satlin A, English CD, Scott TM & Yurgelun-Todd DA et al. (1998) Dynamic susceptibility contrast MR imaging of regional cerebral blood volume in Alzheimer disease: a promising alternative to nuclear medicine. Am J Neuroradiol 19: 1727−1732. | ChemPort |
66. Hatakeyama T, Sakaki S, Nakamura K, Furuta S & Matsuoka K. (1992) Improvement in local cerebral blood flow measurement in gerbil brains by prevention of postmortem diffusion of [¹⁴C] iodoantipyrine. J Cereb Blood Flow Metab 12: 296−300. | PubMed | ChemPort |
67. Hedehus M, Steensgaard A, Rostrup E & Larsson HBW. 1997 Investigation of the linear relation between R₂* and gadolinium concentration in vivo. (In: Proceedings of the ISMRM 5th Annual Meeting, Vancouver, Canada, p 1792).
68. Helpern JA, Huang N, Branch CA & Yongbi M. 1997a Perfusion assessment by repetitive inversion of spins (PARIS). (In: Proceedings of the ISMRM 5th Annual Meeting, Vancouver, Canada, p 1759).
69. Helpern JA, Branch CA, Yongbi M & Huang N. (1997b) Perfusion imaging by un-inverted flow-sensitive inversion recovery (UNFAIR). Magn Reson Imaging 15: 135−139. | ChemPort |
70. Henkelman RM, Neil JJ & Xiang QS. (1994) A quantitative interpretation of IVIM measurements of vascular perfusion in the rat brain. Magn Reson Med 32: 464−469. | PubMed | ChemPort |
71. Hossmann KA, Mies G, Paschen W, Csiba L, Bodsch W & Rapin JR et al. (1985) Multiparametric imaging of blood flow and metabolism after middle cerebral artery occlusion in cats. J Cereb Blood Flow Metab 5: 97−107. | PubMed | ChemPort |
72. Jacquez JA. 1972 Compartmental Analysis in Biology and Medicine: Kinetics of distribution of tracer-labeled materials Amsterdam, Elsevier (pp) 84−101.
73. Jiang Q, Zhang ZG, Chopp M, Helpern JA, Ordidge RJ & Garcia JH et al. (1993) Temporal evolution and spatial-distribution of the diffusion constant of water in rat-brain after transient middle cerebral-artery occlusion. J Neurol Sci 120: 123−130. | Article | PubMed | ISI | ChemPort |
74. Jiang Q, Chopp M, Zhang ZG, Helpern JA, Ordidge RJ & Ewing JR et al. (1994) The effect of hypothermia on transient focal ischemia in rat brain evaluated by diffusion- and perfusion-weighted NMR imaging. J Cereb Blood Flow Metab 14: 732−741. | PubMed | ChemPort |
75. Jiang Q, Zhang RL, Zhang ZG, Ewing JR, Divine GW & Chopp M. (1998a) Diffusion-, T₂, and perfusion-weighted nuclear magnetic resonance imaging of middle cerebral artery embolic stroke and recombinant tissue plasminogen activator intervention in the rat. J Cereb Blood Flow Metab 18: 758−767. | ChemPort |
76. Jiang Q, Zhang ZG, Zhang RL, Ewing JR, Divine GW & Jiang P et al. (1998b) Diffusion, perfusion and T₂ magnetic resonance imaging of anti-intercellular adhesion molecule 1 antibody treatment of transient middle cerebral artery occlusion in rat. Brain Res 788: 191−201. | Article | ChemPort |
77. Kao Y, Wan X & MacFall JR. (1998) Simultaneous multi-slice acquisition with arterial-flow tagging (SMART) using echo planar imaging (EPI). Magn Reson Med 39: 662−665. | ChemPort |
78. Kaufman MJ, Levin JM, Maas LC, Rose SL, Lukas SE & Mendelson JH et al. (1998) Cocaine decreases relative cerebral blood volume in humans: a dynamic susceptibility contrast magnetic resonance imaging study. Psychopharmacology 138: 76−81. | Article | ChemPort |
79. Kennan RP, Zhong JH & Gore JC. (1994) Intravascular susceptibility contrast mechanisms in tissues. Magn Reson Med 31: 9−21. | PubMed | ChemPort |
80. Kerskens CM, Hoehn-Berlage M, Schmitz B, Busch E, Bock C & Gyngell ML et al. (1996) Ultrafast perfusion-weighted MRI of functional brain activation in rats during forepaw stimulation: comparison with T₂-weighted MRI. NMR Biomed 9: 20−23. | Article | PubMed | ChemPort |
81. Kim SG. (1995) Quantification of relative cerebral blood flow change by flow-sensitive alternating inversion recovery (FAIR) technique: application to functional mapping. Magn Reson Med 34: 293−301. | PubMed | ISI | ChemPort |
82. Kim SG & Tsekos NV. (1997a) Perfusion imaging by a flow-sensitive alternating inversion recovery (FAIR) technique: application to functional brain imaging. Magn Reson Med 37: 425−435. | ChemPort |
83. Kim SG, Tsekos NV & Ashe J. (1997b) Multi-slice perfusion-based functional MRI using the FAIR technique: comparison of CBF and BOLD effects. NMR Biomed 10: 191−196. | Article | ChemPort |
84. Kim SG & Ugurbil K. (1997c) Comparison of blood oxygenation and cerebral blood flow effects in fMRI: estimation of relative oxygen consumption change. Magn Reson Med 38: 59−65. | PubMed | ISI | ChemPort |
85. Kluytmans M, van der Grond J, Folkers PJM, Mali WP & Viergever MA. (1998) Differentiation of gray matter and white matter perfusion in patients with unilateral internal carotid artery occlusion. J Magn Reson Imaging 8: 767−774. | ChemPort |
86. Kruger G, Kleinschmidt A & Frahm J. (1996) Dynamic MRI sensitized to cerebral blood oxygenation and flow during sustained activation of human visual cortex. Magn Reson Med 35: 797−800. | PubMed | ISI | ChemPort |
87. Kucharczyk J, Asgari H, Mintorovitch J, Vexler Z, Rocklage S & Watson A et al. (1993) Cerebrovascular transit characteristics of DyDTPA-BMA and GdDTPA-BMA in normal and ischemic cat brain. Am J Neuroradiol 14: 289−296. | ChemPort |
88. Kucharczyk J, Mintorovitch J, Asgari HS & Moseley M. (1991) Diffusion/perfusion MR imaging of acute cerebral ischemia. Magn Reson Med 19: 311−315. | PubMed | ChemPort |
89. Kucharczyk J, Vexler ZS, Roberts TP, Asgari HS, Mintorovitch J & Derugin N et al. (1993) Echo-planar perfusion-sensitive MR imaging of acute cerebral ischemia. Radiology 188: 711−717. | PubMed | ChemPort |
90. Kwong KK, Belliveau JW, Chesler DA, Goldberg IE, Weisskoff RM & Poncelet BP et al. (1992) Dynamic magnetic resonance imaging of human brain activity during primary sensory stimulation. Proc Natl Acad Sci USA 89: 5675−5679. | PubMed | ChemPort |
91. Kwong KK, Chesler DA, Weisskoff RM, Donahue KM, Davis TL & Østergaard L et al. (1995) MR perfusion studies with T₁-weighted echo-planar imaging. Magn Reson Med 34: 878−887. | PubMed | ChemPort |
92. Lassen NA, Henriksen O & Sejrsen P. 1984 The cardiovascular system. (In) Handbook of Physiology (Shepherd JT, Abboud FM, eds) Bethesda, MD, American Physiology Society (pp) 21−64.
93. Le Bihan D, Breton E, Lallemand D, Grenier P, Cabanis E & Laval-Jeantet M. (1986) MR imaging of intravoxel incoherent motions: application to diffusion and perfusion in neurologic disorders. Radiology 161: 401−407. | PubMed | ChemPort |
94. Le Bihan D & Turner R. 1992 Magnetic Resonance Imaging (Stark DD, Bradley WG, eds) St Louis, Mosby (pp) 355−371.
95. Leenders KL, Perani D, Lammertsma AA, Heather JD, Buckingham P & Healy JR et al. (1990) Cerebral blood flow, blood volume and oxygen utilization: normal values and effects of age. Brain 113: 27−47. | PubMed | ISI |
96. Lev MH, Kulke SF, Sorensen AG, Boxerman JL, Brady TJ & Rosen BR et al. (1997) Contrast-to-noise ratio in functional MRI of relative cerebral blood volume with sprodiamide injection. J Magn Reson Imaging 7: 523−527. | ChemPort |
97. Levin JM, Kaufman MJ, Ross MH, Mendelson JH, Maas LC & Cohen BM et al. (1995) Sequential dynamic susceptibility contrast MR experiments in human brain: residual contrast agent effect, steady-state, and hemodynamic perturbation. Magn Reson Med 34: 655−663. | PubMed | ChemPort |
98. Levin JM, Wald LL, Kaufman MJ, Ross MH, Maas LC & Renshaw PF. (1998) T₁ effects in sequential dynamic susceptibility contrast experiments. J Magn Reson 130: 292−295. | Article | ChemPort |
99. Li KL & Suojanen JN. (1995) Cocaine-induced changes in time course of regional cerebral blood volume and transit time as determined by dynamic MR imaging. J Magn Reson Imaging 5: 715−718. | ChemPort |
100. Lo EH, Pierce AR, Mandeville JB & Rosen BR. (1997) Neuroprotection with NBQX in rat focal cerebral ischemia: effects on ADC probability distribution functions and diffusion-perfusion relationships. Stroke 28: 439−446. | PubMed | ChemPort |
101. Maas LC, Harris GJ, Satlin A, English CD, Lewis RF & Renshaw PF. (1997) Regional cerebral blood volume measured by dynamic susceptibility contrast MR imaging in Alzheimer's disease: a principal components analysis. J Magn Reson Imaging 7: 215−219. | ChemPort |
102. Maeda M, Itoh S, Ide H, Matsuda T, Kobayashi H & Kubota T et al. (1993) Acute stroke in cats: comparison of dynamic susceptibility-contrast MR imaging with T₂- and diffusion-weighted MR imaging. Radiology 189: 227−232. | PubMed | ChemPort |
103. Marcy VR & Welsh F. (1984) Correlation between cerebral blood flow and ATP content following tourniquet-induced ischemia in cat brain. J Cereb Blood Flow Metab 4: 362−367 (1984). | ChemPort |
104. McLaughlin AC. (1997) Quantification of regional cerebral blood flow increases during motor activation: a steady-state arterial spin tagging study. Neuroimage 6: 104−120.
105. McLaughlin AC, Ye FQ, Pekar JJ, Santha AKS & Frank JA. (1997) Effect of magnetization transfer on the measurement of cerebral blood flow using steady-state arterial spin tagging approaches: a theoretical investigation. Magn Reson Med 37: 501−510. | ChemPort |
106. Meier P & Zierler KL. (1954) On the theory of the indicator-dilution method for measurement of blood flow and volume. Appl Physiol 6: 731−744. | ChemPort |
107. Minematsu K, Fisher M, Li LM & Sotak CH. (1993) Diffusion and perfusion magnetic resonance imaging studies to evaluate a noncompetitive N-methyl-D-aspartate antagonist and reperfusion in experimental stroke in rats. Stroke 24: 2074−2081. | PubMed | ChemPort |
108. Moseley ME, De Crespigny AJS, Roberts TPL, Kozniewska E & Kucharczyk J. (1993) Early detection of regional cerebral ischemia using high-speed MRI. Stroke 24: I60−I65. | ChemPort |
109. Moseley ME, Kucharczyk J, Mintorovitch J, Cohen Y, Kurhanewicz J & Derugin N et al. (1990) Diffusion-weighted MR imaging of acute stroke: correlation with T₂-weighted and magnetic susceptibility-enhanced MR imaging in cats. Am J Neuroradiol 11: 423−429. | PubMed | ISI | ChemPort |
110. Moseley ME, Vexler Z, Asgari HS, Mintorovitch J, Derugin N & Rocklage S et al. (1991) Comparison of Gd- and Dy-chelates for T₂* contrast-enhanced imaging. Magn Reson Med 22: 259−264. | PubMed | ChemPort |
111. Mottet I, Quast MJ, Dewitt DS, Hillman GR, Wei JN & Uhrbrock DH et al. (1997) N-G-Nitro-L-arginine methyl ester modifies the input function measured by dynamic susceptibility contrast magnetic resonance imaging. J Cereb Blood Flow Metab 17: 791−800. | Article | ChemPort |
112. Muller TB, Haraldseth O, Jones RA, Sebastiani G, Godtliebsen F & Lindboe CF et al. (1995a) Combined perfusion and diffusion-weighted magnetic resonance imaging in a rat model of reversible middle cerebral artery occlusion. Stroke 26: 451−457. | PubMed | ChemPort |
113. Muller TB, Haraldseth O, Jones RA, Sebastiani G, Lindboe CF & Unsgard G et al. (1995b) Perfusion and diffusion-weighted MR imaging for in-vivo evaluation of treatment with U74389G in a rat stroke model. Stroke 26: 1453−1458. | PubMed | ChemPort |
114. Nighoghossian N, Berthezene Y, Philippon B, Adeleine P, Froment JC & Trouillas P. (1996) Hemodynamic parameter assessment with dynamic susceptibility contrast magnetic resonance imaging in unilateral symptomatic internal carotid artery occlusion. Stroke 27: 474−479. | PubMed | ChemPort |
115. Nudo RJ, Jenkins WM, Merzenich MM, Prejean T & Grenda R. (1992) Neurophysiological correlates of hand preferences in primary motor cortex of adult squirrel monkeys. J Neurosci 12: 2918−47. | PubMed | ISI | ChemPort |
116. Ogawa S, Lee T-M, Nayak AS & Glynn P. (1990) Oxygenation-sensitive contrast in magnetic resonance image of rodent brain at high magnetic fields. Magn Reson Med 14: 68−78. | PubMed | ISI | ChemPort |
117. Ogawa S, Menon RS, Tank DW, Kim SG, Merkle H & Ellermann JM et al. (1993) Functional brain mapping by blood oxygenation level-dependent contrast magnetic resonance imaging: a comparison of signal characteristics with a biophysical model. Biophys J 64: 803−12. | PubMed | ISI | ChemPort |
118. Ogawa S, Tank DW, Menon R, Ellerman JM, Kim S-G & Merkle H et al. (1992) Intrinsic signal changes accompanying sensory stimulation: functional brain mapping with magnetic resonance imaging. Proc Natl Acad Sci USA 89: 5951−5955. | PubMed | ChemPort |
119. Ordidge RJ, Wylezinska M, Hugg JW, Butterworth E & Franconi F. (1996) Frequency offset corrected inversion pulses for use in localized spectroscopy. Magn Res Med 36: 562−566. | ChemPort |
120. Østergaard L, Sorensen AG, Kwong KK, Weisskoff RM, Gyldensted C & Rosen BR. (1996a) High resolution measurement of cerebral blood flow using intravascular tracer bolus passages. II. Experimental comparison and preliminary results. Magn Reson Med 36: 726−736. | PubMed | ChemPort |
121. Østergaard L, Weisskoff RM, Chesler DA, Gyldensted C & Rosen BR. (1996b) High resolution measurement of cerebral blood flow using intravascular tracer bolus passages. I. Mathematical approach and statistical analysis. Magn Reson Med 36: 715−725. | ChemPort |
122. Østergaard L, Johannsen P, Host Poulsen P, Vestergaard Poulsen P, Asboe H & Gee AD et al. (1998a) Cerebral blood flow measurements by magnetic resonance imaging bolus tracking: comparison with [O-15] H₂O positron emission tomography in humans. J Cereb Blood Flow Metab 18: 935−940.
123. Østergaard L, Smith DF, Vestergaard Poulsen P, Hansen SB, Gee AD & Gjedde A et al. (1998b) Absolute cerebral blood flow and blood volume measured by magnetic resonance imaging bolus tracking: comparison with positron emission tomography values. J Cereb Blood Flow Metab 18: 425−432.
124. Østergaard L, Chesler DA, Weisskoff RM, Sorensen AG & Rosen BR. 1999 Modeling cerebral blood flow and flow heterogeneity from magnetic resonance residue data. J Cereb Blood Flow Metab (in press).
125. Pan Y, Lo EH, Matsumoto K, Hamberg L & Jiang H. (1995) Quantitative and dynamic MRI of neuroprotection in experimental stroke. J Neurol Sci 131: 128−134. | Article | ChemPort |
126. Pekar J, Jezzard P, Roberts DA, Leigh JS, Frank JA & Mclaughlin AC. (1996) Perfusion imaging with compensation for asymmetric magnetization-transfer effects. Magn Reson Med 35: 70−79. | PubMed | ISI | ChemPort |
127. Pell GS, Thomas DL, Lythgoe MF, Calamante F, Howseman AM & Williams SR et al. 1998 Measurement of perfusion using arterial spin tagging with the FOCI Pulse. (In: Proceedings of the ISMRM 6th Annual Meeting, Sydney, Australia, p 1190).
128. Pell GS, Thomas DL, Lythgoe MF, Calamante F, Howseman AM & Gadian DG et al. (1999a) The implementation of quantitative FAIR perfusion imaging with a short repetition time in time course studies. Magn Reson Med 41: 829−840. | ChemPort |
129. Pell GS, Lythgoe MF, Thomas DL, Calamante F, King MD & Gadian DG et al. 1999b A study of reperfusion in a gerbil model of forebrain ischemia using serial magnetic resonance FAIR perfusion imaging. Stroke (in press).
130. Penfield W & Boldrey E. (1937) Somatic and sensory representation in the cerebral cortex of man as studied by electrical stimulation. Brain 60: 389−443.
131. Perman WH, Gado MH, Larson KB & Perlmutter JS. (1992) Simultaneous MR acquisition of arterial and brain signal time curves. Magn Reson Med 28: 74−83. | PubMed | ChemPort |
132. Petrella JR, DeCarli C, Dagli M, Duyn JH, Grandin CB & Frank JA et al. (1997) Assessment of whole brain vasodilatory capacity with acetazolamide challenge at 1.5 T using dynamic contrast imaging with frequency-shifted burst. Am J Neuroradiol 18: 1153−1161. | ChemPort |
133. Petrella JR, DeCarli C, Dagli M, Grandin CB, Duyn JH & Frank JA et al. (1998) Age-related vasodilatory response to acetazolamide challenge in healthy adults: a dynamic contrast-enhanced MR study. Am J Neuroradiol 19: 39−44. | ChemPort |
134. Pierce AR, Lo EH, Mandeville JB, Gonzalez RG, Rosen BR & Wolf GL. (1997) MRI measurements of water diffusion and cerebral perfusion: their relationship in a rat model of focal cerebral ischemia. J Cereb Blood Flow Metab 17: 183−190. | Article | PubMed | ChemPort |
135. Porkka L, Neuder M, Hunter G, Weisskoff RM, Belliveau J & Rosen BR. 1991 Arterial input function measurement with MRI. (In: Proceedings of the Society of Magnetic Resonance Medicine, 10th Annual Meeting, San Francisco, USA, p 120).
136. Potchen EJ, Haacke EM, Siebert JE & Gottschalk A eds. 1993 Magnetic Resonance Angiography (St Louis, Mosby).
137. Press WH, Teukolsky SA, Vetterling WT & Flannery BT. 1992 Numerical Recipes in C: The Art of Scientific Computing Cambridge, Cambridge University Press.
138. Puce A. (1995) Comparative assessment of sensorimotor function using functional magnetic resonance imaging and electrophysiological methods. Clin Neurophysiol 12: 450−459. | ChemPort |
139. Quast MJ, Huang NC, Hillman GR & Kent TA. (1993) The evolution of acute stroke recorded by multimodal magnetic resonance imaging. Magn Reson Imaging 11: 465−471. | Article | PubMed | ChemPort |
140. Ramsay SC, Murphy K, Shea SA, Friston KJ, Lammertsma AA & Clark JC et al. (1993) Changes in global cerebral blood flow in humans: effect on regional cerebral blood flow during a neural activation task. J Physiol 471: 521−534. | PubMed | ChemPort |
141. Reith W, Dardzinski BJ, Sotak CH & Fisher M. (1996) Monitoring of tissue plasminogen activator treatment with multi-slice diffusion mapping and perfusion imaging in a thromboembolic stroke model. Int J Neuroradiol 2: 397−407.
142. Reith W, Heiland S, Erb G, Benner T, Forsting M & Sartor K. (1997) Dynamic contrast-enhanced T₂*-weighted MRI in patients with cerebrovascular disease. Neuroradiology 39: 250−257. | Article | ChemPort |
143. Rempp KA, Brix G, Wenz F, Becker CR, Gückel F & Lorenz WJ. (1994) Quantification of regional cerebral blood flow and volume with dynamic susceptibility contrast-enhanced MR imaging. Radiology 193: 637−641. | PubMed | ChemPort |
144. Roberts TPL, Vexler Z, Derugin N, Moseley ME & Kucharczyk J. (1993) High-speed MR imaging of ischemic brain injury following stenosis of the middle cerebral artery. J Cereb Blood Flow Metab 13: 940−946. | PubMed | ChemPort |
145. Roberts TPL, Vexler ZS, Vexler V, Derugin N & Kucharczyk J. (1996) Sensitivity of high-speed "perfusion-sensitive" magnetic resonance imaging to mild cerebral ischemia. Eur Radiol 6: 645−649. | ChemPort |
146. Rordorf G, Koroshetz WJ, Copen WA, Cramer SC, Schaefer PW & Budzik RF et al. (1998) Regional ischemia and ischemic injury in patients with acute middle cerebral artery stroke as defined by early diffusion-weighted and perfusion-weighted MRI. Stroke 29: 939−943. | ChemPort |
147. Rosen BR, Belliveau JW, Aronen HJ, Kennedy D, Buchbinder BR & Fischman A et al. (1991a) Susceptibility contrast imaging of cerebral blood volume: human experience. Magn Reson Med 22: 293−299. | PubMed | ChemPort |
148. Rosen BR, Belliveau JW, Buchbinder BR, Mckinstry RC, Porkka LM & Kennedy DN et al. (1991b) Contrast agents and cerebral hemodynamics. Magn Reson Med 19: 285−292. | PubMed | ChemPort |
149. Rosen BR, Belliveau JW, Vevea JM & Brady TJ. (1990) Perfusion imaging with NMR contrast agents. Magn Reson Med 14: 249−265. | PubMed | ISI | ChemPort |
150. Röther J, de Crespigny AJ, Darceuil H, Iwai K & Moseley ME. (1996a) Recovery of apparent diffusion coefficient after ischemia-induced spreading depression relates to cerebral perfusion gradient. Stroke 27: 980−986.
151. Röther J, de Crespigny AJ, Darceuil H & Moseley ME. (1996b) MR detection of cortical spreading depression immediately after focal ischemia in the rat. J Cereb Blood Flow Metab 16: 214−220.
152. Röther J, Gückel F, Neff W, Schwartz A & Hennerici M. (1996c) Assessment of regional cerebral blood volume in acute human stroke by use of single-slice dynamic susceptibility contrast-enhanced magnetic resonance imaging. Stroke 27: 1088−1093. | PubMed | ChemPort |
153. Röther J, Waggie K, van Bruggen N, de Crespigny AJ & Moseley ME. (1996d) Experimental cerebral venous thrombosis: evaluation using magnetic resonance imaging. J Cereb Blood Flow Metab 16: 1353−1361. | Article | PubMed |
154. Roy CS & Sherrington CS. (1890) On the regulation of the blood-supply of the brain. J Physiol 11: 85−108.
155. Runge VM, Kirsch JE, Wells JW, Dunworth JN, Hilaire L & Woolfolk CE. (1994) Repeat cerebral blood volume assessment with first-pass MR imaging. J Magn Reson Imaging 4: 457−461. | ChemPort |
156. Sanes JN, Donoghue JP, Thangaraj V, Edelman RR & Warach S. (1995) Shared neural substrates controlling hand movements in human motor cortex. Science 268: 1775−1777. | ChemPort |
157. Schmitz B, Hoehn-Berlage M, Kerskens CM, Bottiger BW & Hossmann KA. (1998) Recovery of the rodent brain after cardiac arrest: a functional MRI study. Magn Reson Med 39: 783−788. | PubMed | ChemPort |
158. Schreiber WG, Gückel F, Stritzke P, Schmiedek P, Schwartz A & Brix G. (1998) Cerebral blood flow and cerebrovascular reserve capacity: estimation by dynamic magnetic resonance imaging. J Cereb Blood Flow Metab 18: 1143−1156. | Article | ChemPort |
159. Schwamm LH, Koroshetz WJ, Sorensen AG, Wang B, Copen WA & Budzik R et al. (1998) Time course of lesion development in patients with acute stroke-, serial diffusion- and hemodynamic-weighted magnetic resonance imaging. Stroke 29: 2268−2276. | ChemPort |
160. Schwarzbauer C, Morrisey SP & Haase A. (1996) Quantitative magnetic resonance imaging of perfusion using magnetic labeling of water protons with the detection slice. Magn Reson Med 35: 540−546. | ChemPort |
161. Schwarzbauer C & Heinke W. (1998) BASE imaging: a new spin labeling technique for measuring absolute perfusion changes. Magn Reson Med 39: 717−722. | ChemPort |
162. Siegal T, Rubinstein R, TzukShina T & Gomori JM. (1997) Utility of relative cerebral blood volume mapping derived from perfusion magnetic resonance imaging in the routine follow up of brain tumors. J Neurosurg 86: 22−27. | ChemPort |
163. Siewert B, Bly BM, Schlaug G, Darby DG, Thangaraj V & Warach S et al. (1996) Comparison of the BOLD- and EPISTAR-technique for functional brain imaging by using signal detection theory. Magn Reson Med 36: 249−255. | ChemPort |
164. Siewert B, Schlaug G, Edelman RR & Warach S. (1997) Comparison of EPISTAR and T₂*-weighted gadolinium-enhanced perfusion imaging in patients with acute cerebral ischemia. Neurology 48: 673−679. | PubMed | ChemPort |
165. Silva AC, Zhang WG, Williams DS & Koretsky AP. (1995) Multi-slice MRI of rat brain perfusion during amphetamine stimulation using arterial spin labeling. Magn Reson Med 33: 209−214. | PubMed | ChemPort |
166. Silva AC, Zhang WG, Williams DS & Koretsky AP. (1997a) Estimation of water extraction fractions in rat brain using magnetic resonance measurement of perfusion with arterial spin labeling. Magn Reson Med 37: 58−68. | ChemPort |
167. Silva AC, Williams DS & Koretsky AP. (1997b) Evidence for the exchange of arterial spin labeled water with tissue water in rat brain from diffusion-sensitized measurements of perfusion. Magn Reson Med 38: 232−237. | ChemPort |
168. Singer JR. (1959) Blood flow rates by nuclear magnetic resonance measurements. Science 130: 1652−1653.
169. Sorensen AG, Buonanno FS, Gonzalez RG, Schwamm LH, Lev MH & Huang-Hellinger FR et al. (1996) Hyperacute stroke: evaluation with combined multisection diffusion- weighted and hemodynamically weighted echo-planar MR imaging. Radiology 199: 391−401. | PubMed | ISI | ChemPort |
170. Sorensen AG, Tievsky AL, Østergaard L, Weisskoff RM & Rosen BR. (1997) Contrast agents in functional MR imaging. J Magn Reson Imaging 7: 47−55. | ChemPort |
171. Sorensen AG, Wray SH, Weisskoff RM, Boxerman JL, Davis TL & Caramia F et al. (1995) Functional MR of brain activity and perfusion in patients with chronic cortical stroke. Am J Neuroradiol 16: 1753−1762. | ChemPort |
172. Starmer CF & Clark DO. (1970) Computer computations of cardiac output using the gamma function. J Appl Physiol 28: 219−220. | ChemPort |
173. Stewart GN. (1894) Researches on the circulation time in organs and on the influences which affect it: parts I-III. J Physiol 15: 1.
174. Takano K, Carano RAD, Tatlisumak T, Meiler M, Sotak CH & Kleinert HD et al. (1998) Efficacy of intra-arterial and intravenous prourokinase in an embolic stroke model evaluated by diffusion-perfusion magnetic resonance imaging. Neurology 50: 870−875. | ChemPort |
175. Todd NV, Picozzi P, Crockard HA & Russell RR. (1986) Reperfusion after cerebral-ischemia: influence of duration of ischemia. Stroke 17: 460−466. | PubMed | ChemPort |
176. Tofts PS. (1997) Modeling tracer kinetics in dynamic Gd-DTPA MR imaging. J Magn Reson Imaging 7: 91−101. | PubMed | ISI | ChemPort |
177. Tong DC, Yenari MA, Albers GW, Obrien M, Marks MP & Moseley ME. (1998) Correlation of perfusion- and diffusion-weighted MRI with NIHSS score in acute (<6.5 hour) ischemic stroke. Neurology 50: 864−870. | ChemPort |
178. Tracey I, Hamberg LM, Guimaraes AR, Hunter G, Chang I & Navia BA et al. (1998) Increased cerebral blood volume in HIV-positive patients detected by functional MRI. Neurology 50: 1821−1826. | ChemPort |
179. Tsekos NV, Zhang FY, Merkle H, Nagayama M, Iadecola C & Kim SG. (1998) Quantitative measurements of cerebral blood flow in rats using the FAIR technique: correlation with previous iodoantipyrine autoradiographic studies. Magn Reson Med 39: 564−573. | ChemPort |
180. Tsuchida C, Yamada H, Maeda M, Sadato N, Matsuda T & Kawamura Y et al. (1997) Evaluation of peri-infarcted hypoperfusion with T₂*-weighted dynamic MRI. J Magn Reson Imaging 7: 518−522. | ChemPort |
181. Tsuchiya K, Inaoka S, Mizutani Y & Hachiya J. (1998) Echo-planar perfusion MR of Moyamoya disease. Am J Neuroradiol 19: 211−216. | ChemPort |
182. Turner R. 1988 Perfusion studies and fast imaging. (In) Cerebral Blood Flow (Rescigno A, Boicelli A, eds) New York, Plenum Publishing (pp) 245−258.
183. Turner R, Le Bihan D, Moonen CTW, Des Pres D & Frank J. (1991) Echo-planar time course MRI of cat brain oxygenation changes. Magn Reson Med 22: 159−166. | PubMed | ISI | ChemPort |
184. Tzika AA, Massoth RJ, Ball WAS, Majumdar S, Dunn RS & Kirks DR. (1993) Cerebral perfusion in children: detection with dynamic contrast-enhanced T₂*-weighted MR images. Radiology 187: 449−458. | ChemPort |
185. Tzika AA, Robertson RL, Barnes PD, Vajapeyam S, Burrows PE & Treves ST et al. (1997) Childhood moyamoya disease: hemodynamic MRI. Pediatr Radiol 27: 727−735. | Article | ChemPort |
186. van Zijl PC, Eleff SM, Ulatowski JA, Oja JM, Ulug AM & Traystman RJ et al. (1998) Quantitative assessment of blood flow, blood volume and blood oxygenation effects in functional magnetic resonance imaging. Nature Med 4: 159−167. | Article | PubMed | ISI | ChemPort |
187. Verhaegen MJ, Todd MM, Warner DS, James B & Weeks JB. (1992) The role of electrode size on the incidence of spreading depression and on cortical cerebral blood-flow as measured by H₂ clearance. J Cereb Blood Flow Metab 12: 230−237. | PubMed | ChemPort |
188. Vexler ZS, Roberts TPL, Bollen AW, Derugin N & Arieff AI. (1997) Transient cerebral ischemia: association of apoptosis induction with hypoperfusion. J Clin Invest 99: 1453−1459. | ChemPort |
189. Villringer A & Dirnagl U. (1995) Coupling of brain activity and cerebral blood flow: basis of functional neuroimaging. Cerebrovasc Brain Metab Rev 7: 240−276. | PubMed | ISI | ChemPort |
190. Villringer A, Rosen BR, Belliveau JW, Ackerman JL, Lauffer RB & Buxton RB et al. (1988) Dynamic imaging with lanthanide chelates in normal brain: contrast due to magnetic-susceptibility effects. Magn Reson Med 6: 164−174. | PubMed | ISI | ChemPort |
191. Vonken EPA, Bakker CJG & Viergever MA. 1998 Maximum likelihood estimate of the response function in dynamic susceptibility contrast MRI. (In: Proceedings of the ISMRM, 6th Annual Meeting, Sydney, Australia, p 1214).
192. Walsh EG, Minematsu K, Leppo J & Moore SC. (1994) Radioactive microsphere validation of a volume localized continuous saturation perfusion measurement. Magn Reson Med 31: 147−153. | PubMed | ChemPort |
193. Warach S, Dashe JF & Edelman RR. (1996) Clinical outcome in ischemic stroke predicted by early diffusion-weighted and perfusion magnetic resonance imaging: a preliminary analysis. J Cereb Blood Flow Metab 16: 53−59. | Article | ChemPort |
194. Warach S, Levin JM, Schomer DL, Holman BL & Edelman RR. (1994) Hyperperfusion of ictal seizure focus demonstrated by MR perfusion imaging. Am J Neuroradiol 15: 965−968. | ChemPort |
195. Warach S, Wei L, Ronthal M & Edelman RR. (1992) Acute cerebral ischemia: evaluation with dynamic contrast-enhanced MR imaging and MR angiography. Radiology 182: 41−47. | PubMed | ChemPort |
196. Weisskoff RM, Boxerman JL, Sorensen AG, Kulke SM, Campbell TA & Rosen BR. 1994 Simultaneous blood volume and permeability mapping using a single Gd-based contrast injection. (In: Proceedings of the Society of Magnetic Resonance Medicine 2nd Annual Meeting, San Francisco, USA, p 279).
197. Weisskoff RM, Chesler D, Boxerman JL & Rosen BR. (1993) Pitfalls in MR measurement of tissue blood flow with intravascular tracers: which mean transit-time? Magn Reson Med 29: 553−559. | PubMed | ChemPort |
198. Weisskoff RM, Zuo CS, Boxerman JL & Rosen BR. (1994) Microscopic susceptibility variation and transverse relaxation: theory and experiment. Magn Reson Med 31: 601−610. | PubMed | ChemPort |
199. Wendland MF, White DL, Aicher KP, Tzika AA & Moseley ME. (1991) Detection with echo-planar MR imaging of transit of susceptibility contrast-medium in a rat model of regional brain ischemia. J Magn Reson Imaging 1: 285−292. | ChemPort |
200. Wenz F, Rempp K, Hess T, Debus J, Brix G & Engenhart R et al. (1996) Effect of radiation on blood volume in low-grade astrocytomas and normal brain tissue: quantification with dynamic susceptibility contrast MR imaging. Am J Roentgenol 166: 187−193. | ChemPort |
201. Williams DS, Detre JA, Leigh JS & Koretsky AP. (1992) Magnetic resonance imaging of perfusion using spin inversion of arterial water. Proc Natl Acad Sci USA 89: 212−216. | PubMed | ChemPort |
202. Wittlich F, Kohno K, Mies G, Norris DG & Hoehn-Berlag M. (1995) Quantitative measurement of regional blood flow with gadolinium diethylenetriaminepentaacetate bolus track NMR imaging in cerebral infarcts in rats: validation with the iodo[C-14]antipyrine technique. Proc Natl Acad Sci USA 92: 1846−1850. | PubMed | ChemPort |
203. Wolff S & Balaban R. (1989) Magnetisation transfer contrast (MTC) and tissue water proton relaxation in vivo. Magn Reson Med 10: 135−144. | PubMed | ISI | ChemPort |
204. Wong EC, Buxton RB & Frank LR. (1997) Implementation of quantitative perfusion imaging techniques for functional brain mapping using pulsed arterial spin labeling. NMR Biomed 10: 237−249. | Article | PubMed | ChemPort |
205. Wong EC, Buxton RB & Frank LR. (1998a) Quantitative imaging of perfusion using a single subtraction (QUIPSS and QUIPSS II). Magn Reson Med 39: 702−708. | ChemPort |
206. Wong EC, Buxton RB & Frank LR. (1998b) A theoretical and experimental comparison of continuous and pulsed arterial spin labeling techniques for quantitative perfusion imaging. Magn Reson Med 39: 702−708. | ChemPort |
207. Yang Y, Frank JA, Hou L, Ye FQ, McLaughlin AC & Duyn JH. (1998) Multi-slice imaging of quantitative cerebral perfusion with pulsed arterial spin labeling. Magn Reson Med 39: 825−32. | ChemPort |
208. Ye FQ, Mattay VS, Frank JA, Weinberger DR & McLaughlin AC. 1997a Calculation of cerebral blood flow using "delayed acquisition" arterial spin tagging approaches. (In: Proceedings of the SMR 5th Annual Meeting, Vancouver, Canada, p 1756).
209. Ye FQ, Mattay VS, Jezzard P, Frank JA, Weinberger DR & McLaughlin AC. (1997b) Correction for vascular artifacts in cerebral blood flow values measured by using arterial spin tagging techniques. Magn Reson Med 37: 226−235. | ChemPort |
210. Ye FQ, Smith AM, Yang Y, Duyn J, Mattay VS & Ruttimann UE et al. (1997c) Quantification of regional cerebral blood flow increases during motor activation: a steady-state arterial spin-tagging study. Neuroimage 6: 104−112. | Article | ChemPort |
211. Ye FQ, Smith AM, Mattay VS, Ruttimann UE, Frank JA & Weinberger DR et al. (1998) Quantification of regional cerebral blood flow increases in prefrontal cortex during a working memory task: a steady-state arterial spin-tagging study. Neuroimage 8: 44−49. | Article | ChemPort |
212. Yenari MA, de Crespigny A, Palmer JT, Roberts S, Schrier SL & Albers GW et al. (1997) Improved perfusion with rt-PA and hirulog in a rabbit model of embolic stroke. J Cereb Blood Flow Metab 17: 401−411. | Article | ChemPort |
213. Yonas H, Darby JM, Marks EC, Durham SR & Maxwell C. (1991) CBF measured by Xe-CT: approach to analysis and normal values. J Cereb Blood Flow Metab 11: 716−725. | PubMed | ChemPort |
214. Yongbi MN, Branch CA & Helpern JA. (1998) Perfusion Imaging Using FOCI RF Pulses. Magn Reson Med 40: 938−943. | ChemPort |
215. Zaharchuk G, Bogdanov AA, Marota JJA, Shimizu-Sasamata M, Weiskoff RM & Kwong KK et al. (1998) Continuous assessment of perfusion by tagging including volume and water extraction (CAPTIVE): a steady-state contrast agent technique for measuring blood flow, relative blood volume fraction, and the water extraction fraction. Magn Reson Med 40: 666−678. | PubMed | ChemPort |
216. Zhang WG, Silva AC, Williams DS & Koretsky AP. (1995) NMR measurement of perfusion using arterial spin labeling without saturation of macromolecular spins. Magn Reson Med 33: 370−376. | PubMed | ISI | ChemPort |
217. Zhang WG, Williams DS & Koretsky AP. (1993) Measurement of rat brain perfusion by NMR using spin labeling of arterial water: in vivo determination of the degree of spin labeling. Magn Reson Med 29: 416−421. | PubMed | ChemPort |
218. Zhang WG, Williams DS, Detre JA & Koretsky AP. (1992) Measurement of brain perfusion by volume-localized NMR spectroscopy using inversion of arterial water spins: accounting for transit-time and cross-relaxation. Magn Reson Med 25: 362−371. | PubMed | ChemPort |
219. Zhou J, Mori S & van Zijl PCM. (1998) FAIR excluding radiation damping (FAIRER). Magn Reson Med 40: 712−719. | PubMed | ISI | ChemPort |
220. Zierler KL. (1962) Theoretical basis of indicator-dilation methods for measuring flow and volume. Circ Res 10: 393−407.
221. Zierler KL. (1965) Equations for measuring blood flow by external monitoring of radioisotopes. Circ Res 16: 309−321. | ChemPort |

Acknowledgements

The authors thank their colleagues, especially J.A. Wiersma, at the Department of Medical Physics, RCS Unit of Biophysics, and the Wellcome Department of Cognitive Neurology, University College London, U.K., for assistance and advice during the writing of this review. The authors also thank the following people for their contribution of figures for this article: E.H. Lo, T.P.L. Roberts, L. Østergaard, R.G. Gonzalez, A.G. Sorensen, W.G. Schreiber, J.A. Detre, E.C. Wong, M.F. Lythgoe, K.K. Kwong, F.Q. Ye, A.C. Silva, and T.L. Davis.