Graded Reduction of Cerebral Blood Flow in Rat as Detected by the Nuclear Magnetic Resonance Relaxation Time T 2 : A Theoretical and Experimental Approach

Abstract

The ability of transverse nuclear magnetic resonance relaxation time, T₂, to reveal acutely reduced CBF was assessed using magnetic resonance imaging (MRI). Graded reduction of CBF was produced in rats using a modification of Pulsinelli's four-vessel occlusion model. The CBF in cerebral cortex was quantified using the hydrogen clearance method, and both T₂ and the trace of the diffusion tensor (D_av = 1/3TraceD) in the adjacent cortical tissue were determined as a function of reduced CBF at 4.7 T. A previously published theory, interrelating cerebral hemodynamic parameters, hemoglobin, and oxygen metabolism with T₂, was used to estimate the effects of reduced CBF on cerebral T₂. The MRI data show that T₂ reduces in a U-shape manner as a function of CBF, reaching a level that is 2.5 to 2.8 milliseconds (5% to 6%) below the control value at CBF, between 15% and 60% of normal. This reduction could be estimated by the theory using the literature values of cerebral blood volume, oxygen extraction ratio, and precapillary oxygen extraction during compromised CBF. D_av dropped with two apparent flow thresholds, so that a small 11% to 17% reduction occurred between CBF values of 16% to 45% of normal, followed by a precipitous collapse by more than 20% at CBF below 15% of normal. The current data show that T₂ can be used as an indicator of acute hypoperfusion because of its ability to indicate blood oxygenation level-dependent phenomena on reduced CBF.

Keywords

Brain Acute ischemia Oxygen extraction ratio Oxygen metabolism

Brain function is ultimately dependent on uninterrupted CBF, and a severe drop in CBF below the level of autoregulation inevitably leads to cerebral dysfunction and finally to irreversible ischemic damage. Studies using single photon emission computed tomography or positron emission tomography (PET) show that adaptive responses in regional cerebral hemodynamics and oxygen extraction take place on reduction of cerebral perfusion pressure (CPP) (Grubb et al., 1975; Wise et al., 1983; Powers, 1991). These include an increase in cerebral blood volume (CBV), which results from vasodilation associated with a drop in vascular resistance, and increased oxygen extraction ratio (OER). These adaptations are necessary to secure substrate supply for oxidative metabolism. Thus, increased OER can be considered as an index of viable brain tissue during acutely compromised CBF and indeed, experimental evidence from focal ischemia studies suggest that tissue with high OER and low CBF may exhibit with better prognosis than that with low extraction and flow (Young et al., 1996). Consequently, a noninvasive imaging approach revealing the severity of mismatch between oxygen consumption and delivery would be of great value for the evaluation of tissue status in hypoperfused areas.

Single photon emission computed tomography and PET methods provide regional CBF in quantitative terms, and PET can be used to detect regional CMRO₂ and OER; these methods have been extensively used both in experimental and clinical stroke studies (Frackowiak, 1985; Sette et al., 1989; Heiss and Podreka, 1993). Magnetic resonance imaging (MRI) offers several means to image perfusion including tracking with D₂O, ¹⁹F-labeled probes, Gd-chelates, as well as using magnetic labeling of arterial spins. All of these methods have been successfully applied to reveal reduced perfusion either in qualitative or quantitative terms (for a review, see Hossmann and Hoehn-Berlage, 1995). Recently, T₂* and T₂ MRI were used to reveal the consequences of reduced perfusion in the cat and rat brain (de Crespigny et al., 1992; Roussel et al., 1995; Gröhn et al., 1998). Both of these NMR relaxation times are sensitive to the blood oxygenation level-dependent (BOLD) phenomenon (Ogawa et al., 1990; Thulborn et al., 1982), capable of probing the balance between oxygen consumption and delivery (Davis et al., 1998; Oja et al., 1999). The quantitative interrelationships between physiologic and MRI variables for single spin-echo (SE)-BOLD MRI have been previously described and validated in animal models (van Zijl et al., 1998). Until now, the theory has been successfully applied to derive quantitative data for CBV in hypoxic hypoxia (van Zijl et al., 1998) and in hypercarbia or hypocarbia (Ulatowski et al., 1999) in cats, as well as for OER during visual activation in humans (Oja et al., 1999). Furthermore, we have demonstrated that T₂ reductions from BOLD at 9.4 T can be detected in hypoperfusion and during the first minutes of middle cerebral artery occlusion in rats (Gröhn et al, 1998). In the current study, we assess the applicability of the quantitative SE-BOLD theory under conditions of compromised CBF in rat brain at 4.7 T. Experimental T₂ data are quantitatively compared with calculated ones using the aforementioned theory. We also simulated T₂ changes at 1.5 T to evaluate the relevance of the approach to clinical MRI.

METHODS

Animal preparation

Graded decrease in CBF was induced for male Wistar rats (n = 26) using a modification of the four-vessel global forebrain ischemia model (Pulsinelli et al., 1982). Briefly, both arteria vertebrales were electrocoagulated under halothane anesthesia (1.5% halothane in 70/30 N₂O/O₂). On the following day, rats were reanesthetized with urethane (1.2 to 1.5 g/kg intraperitoneally), which provided stable anesthesia in a single injection, causing minimal depression of the cardiovascular and respiratory system (Buelke-Sam et al., 1978; Flecknell, 1996) and maintaining close-to-normal cerebral autoregulation (Sandor et al., 1986). Both femoral arteries were cannulated for blood pressure measurements (Datex, Helsinki, Finland) and blood gas analysis (ABL-5, Radiometer, Inc., Copenhagen, Denmark). Both common carotid arteries (CCAs) were exposed and surrounded with silicon-coated snares joined to a controllable screw systems, which were similar, as previously described in detail (Allen et al., 1993). These occluding devices made it possible to produce graded, reversible occlusions of the CCAs from outside of the magnet. Core temperatures of the animals were monitored and maintained close to 37°C with an external heating blanket both during surgery and MRI.

Determination of CBF

Cerebral blood flow was quantified using the hydrogen clearance method (Crockard et al., 1980). Briefly, the skull was exposed and two Teflon-coated 0.1-mm platinum/iridium electrodes with 1-mm exposed tips were inserted 1.5 mm into cerebral cortex through burr holes made 2 mm caudal to bregma and 3 mm lateral from midline to both sides of the brain. Ag/AgCl reference electrodes were inserted subcutaneously in the flank. A mixture of H₂/N₂/O₂ 10/65/25 was used for loading, after which the inspired gas was switched to N₂/O₂ 75/25, and a hydrogen clearance curve was recorded for 2 minutes. An initial slope procedure was used in quantitative analysis of CBF (Olesen et al., 1971).

MRI methods

The MRI data were acquired in a horizontal 4.7 T magnet (Magnex Scientific, Ltd., Abdington, UK) equipped with actively shielded field gradients (Magnex) interfaced to a s.m.i.s. console (Surrey Medical Imaging, Ltd., Guildford, UK) using a single-loop receiver/transmitter surface coil (diameter 30 mm). A horizontal imaging slice, thickness of 1.5 mm, field of view of 40 mm, matrix size of 128 × 64, zero-filled to 128 × 128 before Fourier transformation, was set according to axial T₁-weighted multislice SE pilot images so that the center of the slice was 2 mm from the ends of the electrodes. The data for T₂ maps were collected using a SE-sequence with a repetition time (TR) of 1500 milliseconds, four echo times (TE) of 20, 50, 80, and 110 milliseconds, and an adiabatic BIR-4 refocusing pulse (Garwood and Ke, 1991). A Hahn SE-sequence was selected instead of a multiecho approach to maximize sensitivity of the sequence to SE-BOLD (van Zijl et al., 1998; Gröhn et al., 1998).

Images of the trace of the diffusion tensor (D_av = 1/3 TraceD) were acquired using the same SE-sequence (TR of 1500 milliseconds, TE of 75 milliseconds) with four bipolar diffusion gradients in each direction (Mori and van Zijl, 1995). Diffusion weighting strengths of 40, 956, and 1376 s/mm² were used to calculate the absolute diffusion image.

Experimental protocol

Two baseline CBF values and two sets of control images were measured, after which CCAs were partially occluded to produce a controlled CBF decrease. After 1 minute of stabilization, T₂ between 2 and 7 minutes, D_av between 7 and 12 minutes, and CBF both between 1 and 5 and 9 and 13 minutes of CCA occlusions were measured, and CCAs were released. This protocol was repeated one to five times for one animal with more severe flow reductions until either the CBF threshold for a diffusion drop by more than 15% was reached or CBF did not normalize after release of the CCAs. The occlusions with greater than a 10% difference between repeated CBF measurements during a single insult were discarded. Data from each hemisphere were used independently, resulting in 152 CBF values measured during occlusions from 26 animals studied. T₂ and D_av maps were computed, and regions of interest from parietal cortex were analyzed.

Statistical analysis

Statistical analyses were performed with analysis of variance with post hoc multiple comparisons using Tukey's HSD test to determine significant differences between individual group means as well as to determine homogeneous subsets for significance level p < 0.05. To compare simulated model with the experimental results, multiple comparisons were performed using simulated T₂ values as contrast coefficients. All values are indicated as mean ± SD.

THEORY AND SIMULATIONS

A quantitative theory taking into account the interrelationships between the SE-BOLD changes and physiologic parameters (van Zijl et al., 1998; Ulatowski et al., 1999) was applied to conditions of reduced CBF from vascular occlusion. In the following, equations and quantitative assumptions different from the earlier studies are summarized. Notice that the validity of the assumptions is evaluated in case of moderate hypoperfusion, (i.e., CBF above 20 mL·100 g⁻¹·min⁻¹, thus being comparable with flow regions studied by others (Busza et al., 1992; Kohno et al., 1995). The parameters for calculations are for rat at 4.7 T and for human at 1.5 T to simulate either our experimental data or the clinically relevant situation, respectively.

The basic physical phenomenon behind the theory is the sensitivity of SE T₂ relaxation to BOLD (Thulborn et al, 1982; Bryant et al., 1990; van Zijl et al., 1998). Although T₂* is more commonly used to reveal in vivo BOLD effect because of its sensitivity to both intravascular and extravascular susceptibility effects, quantification of the T₂* effect is more complex because of its sensitivity to magnetic field inhomogeneity and local susceptibility differences (e.g., at tissue interfaces) (Ogawa et al., 1993). It has been demonstrated that the SE-BOLD effects arise mainly from the intravascular compartment and can be reconciled by protons exchanging rapidly between plasma and erythrocytes, which have different susceptibilities depending on the fraction of deoxygenated hemoglobin (x_deoxy) inside erythrocytes, thus creating a quantitative relationship between blood oxygenation and T₂-relaxation (Bryant et al., 1990; van Zijl et al, 1998).

The value of x_deoxy in blood depends on arterial oxygen saturation fraction, Y_a, and OER and can expressed in individual microvessels as:

x_{d e o x y, i} = 1 - Y_{a} + m_{i} \cdot O E R \cdot Y_{a} i \in a, c, v

(1)

in which

O E R = \frac{C M R O_{2}}{[H b_{t o t}] \cdot C B F \cdot Y_{a}}

(2)

in which m_i is the fraction of the maximum amount of oxygen that can be extracted on average for the particular vessel and it amounts to 0, 0.5, and 1 for arterioles (a), capillaries (c), and venules (v), respectively. CMRO₂ is the cerebral metabolic rate of oxygen consumption and [Hb_tot] is hemoglobin concentration. Under physiologic conditions, diffusive transport of gases between tissue and blood is thought to proceed only in the capillaries, and in the previous reports, we have used morphometric arteriole-capillary-venule volume ratio of 0.21:0.33:0.46 (capillary fraction equals to microvessels of diameter less than 10 μm). This classic concept has been challenged, since significant precapillary oxygen loss evidently takes place in the arterioles, with a diameter ranging from 20 to 50 μm (Popel and Gross, 1979; Ivanov et al., 1982; Intaglietta et al., 1996; Torres Filho et al., 1996). To take these data into consideration, we have modified the vascular fractions to an arteriole-capillary-venule volume ratio of 0.05:0.49:0.46, whereby the capillary fraction includes smaller arterioles and capillaries in which gas exchange takes place (Sharan et al., 1989). These vascular fractions are based on the deoxygenation process in microvessels and should not be confused with the morphometric fractions. Under physiologic conditions, a linear extraction of oxygen is assumed (Torres Filho et al, 1996) yielding m_c = 0.5. However, the deoxygenation process may be different in severe hypoperfusion. Decreased pH and increased CO₂ in the hypoperfused area result in an elevated P₅₀ of hemoglobin, enhancing release of oxygen in precapillary arterioles with high PO₂ (Sharan et al., 1989; Intaglietta et al., 1996). The reduced CBF may further favor precapillary oxygen loss because the prolonged residence time in microvascular network allows more time for gas diffusion (Popel and Gross, 1979). Therefore, we also have simulated the influence of precapillary oxygen loss on T₂ by varying m_c.

If water exchange between erythrocytes and plasma is considered to be the major mechanism in T₂ relaxation and fast exchange conditions prevail, the relaxation rate (R₂ = 1/T₂) in blood can be written as (Allerhand and Gutowsky, 1964; van Zijl et al., 1998):

\begin{array}{l} R_{2, b l o o d, i} & = & R_{2, p l a s} + H c t_{i} {(R_{2, e r y, i} - R_{2, p l a s}) \\ + (1 - H c t_{i}) {(Δ ω_{i})}^{2} τ_{i} [1 - (2 τ_{i} / τ_{C P M G}) \\ \cdot tanh (τ_{C P M G} / 2 τ_{i})]} i \in a, c, v \end{array}

(3)

\begin{array}{l} R_{2, e r y, i} - R_{2, p l a s} & = & R_{2, d i a} + R_{2, o x y} + x_{d e o x y, i} (R_{2, d e o x y} \\ - R_{2, o x y}) \end{array}

(4)

Δ ω_{i} = ω_{e r y, i} + ω_{p l a s} = ω_{d i a} + ω_{o x y} + x_{d e o x y, i} (ω_{d e o x y} - ω_{o x y})

(5)

R_2,plasma, R_2,deoxy, R_2,oxy, and R_2,dia denote the relaxation rates for plasma, deoxygenated and oxygenated hemoglobin, and the remaining diamagnetic sites, respectively, and they were recently deduced for 1.5 and 4.7 T (van Zijl et al., 1998): R_2,plas = 2.0 1/s, R_2,dia + R_2,oxy = 7.5 1/s, R_2,deoxy − R_2,oxy = 14 1/s for 1.5 T, and R_2,plas = 4.6 1/s, R_2,dia + R_2,oxy = 9.3 1/s, R_2,deoxy − R_2,oxy = 42 1/s for 4.7 T. Susceptibility difference between plasma and erythrocytes (Δω) is calculated using coefficients as follows (van Zijl et al., 1998): ω_deoxy = −0.2708 ppm, ω_oxy = −0.0538 ppm, and ω_dia = 0.0304 ppm. The time τ_CPMG equals TE in our experiments, thus explaining the greater sensitivity of a single SE to BOLD than a multiecho approach (Gröhn et al., 1998; van Zijl et al., 1998). We used TE times of 20, 50, 80, and 110 milliseconds and 50, 80, 110, and 140 milliseconds in T₂ calculations at 4.7 T and 1.5 T, respectively. Notice that R₂ values and signal intensities have to be calculated for each echo-time independently and that apparent R₂ for selected parameters can be fitted from these signal intensities (Eq. 6). The lifetime τ_i = (1 − Hct_i)τ_cry of exchange between erythrocytes and plasma is calculated using the lifetime inside the erythrocyte τ_ery = 12 milliseconds for humans and τ_ery = 7.7 milliseconds for rat. The latter value is calculated from membrane permeability of 7.1 * 10³ cm/s for humans and 11.0 * 10³ cm/s for rat (Benga and Borza, 1995), assuming similar erythrocyte area-to-volume ratios.

Hematocrit (Hct) influences exchange-based T₂ relaxation of blood by both affecting τ_i and determining the sizes of the exchange compartments (Eq. 3). In cerebral microvessels, Hct is known to be 85% of the one in large vessels. This relation appears not to hold during locally reduced CBF (Yamauchi et al., 1998), and a 50% decrease in cerebral Hct was detected in the rabbit carotid occlusion model when CBF had reduced by 52% (McHedlishvili et al., 1997). A key mechanism for Hct reduction is believed to be irregular distribution of the erythrocytes according to flow velocity in arterial bifurcations, so that gradual separation of erythrocytes and plasma in the arterial branching sequence leads to local decrease of Hct in the hypoperfused region (McHedlishvili and Varazashvili, 1981; McHedlishvili et al., 1997). In the current global animal model, the arteries are occluded at the levels of CCAs, thus lacking the arterial branching sequence proposed for Hct reduction (see McHedlishvili and Varazashvili, 1981). Therefore, we have used values calculated with constant microvascular cerebral Hct = 0.85 × 0.47 for rat (Baker et al., 1979). However, the values calculated with decreased Hct from the baseline value 0.85 × 0.41 for human data were used, since they are relevant to clinical situation of acute arterial occlusion (Yamauchi et al., 1998). Decrease in Hct causes increase in OER (Eq. 2) because [Hb_tot] is directly related to Hct (Ulatowski et al., 1996), and, thus, the combined effect of these two parameters has to be taken into account to estimate the influence of decreased Hct in SE-BOLD.

Quantitative interrelationships for R₂ described earlier are inherent to blood and also hold during reduced CBF. To treat R₂ in a MRI voxel, it is necessary to account for the contribution of tissue R₂. The SE-BOLD theory assumes slow exchange between microvascular bed and tissue (van Zijl et al., 1998), which also is expected to prevail in acute hypoperfusion, and, thus, signal attenuation in a SE experiment can be expressed as follows:

\begin{array}{l} S (T E) / S (0) & = & x_{t i s s u e} e^{- R_{2, t i s s u e} T E} + \sum_{i} x_{b l o o d, i} e^{- R_{2, b l o o d, i} T E} \\ = & e^{- R_{2, a p p} T E} i \in a, c, v \end{array}

(6)

in which the tissue fraction (x_tissue) and blood fractions (x_{blood, i}) are related through Eq. 7.

\begin{array}{l} x_{t i s s u e} & = & 1 - \sum_{i} x_{b l o o d, i} \\ = & 1 - \sum_{i} \frac{C B V_{i}}{V_{d} + C B V - C B V_{n o r m,}} i \in, c, v \end{array}

(7)

The R₂ of the tissue remains unaltered in the early phase of hypoperfusion, as supported by experimental data from ischemia studies (Kato et al., 1986; Knight et al., 1994; Schwab et al., 1997). We use values of R_2,tissue 19.8 1/s for rat at 4.7 T fitted from our measurements assuming OER is 0.36 under normal physiologic state and 12.6 1/s for human gray matter in 1.5 T (Whittall et al., 1997). For urethane-anesthetized Wistar rat cortex CBV under normal physiologic state, CBV_norm is 4.77 ± 0.13% (Sandor et al., 1986), and wet-dry weight ratio of the Wistar rat cortex varies between 80% and 84% (Schwab et al., 1997). Therefore, the corresponding values for human gray matter, CBV 47 μL/g (Leenders et al., 1990) and water content of (V_d) 820 μL/g (Brooks et al., 1980), can be used in calculations of blood and tissue fractions both for the rat and humans.

When CPP drops below 60 mm Hg, the normal physiologic interrelationship between CBV and CBF given by the equation, CBV/CBV_norm = (CBF/CBF_norm)^0.5(van Zijl et al., 1998), does not hold. The drop in CPP is compensated by decreased microvascular resistance, leading to vasodilation. It is experimentally shown that CBV reaches its maximal value, CBV_max when CPP is reduced to 20 mm Hg (Ferrari et al., 1992; Grubb et al., 1975). The reported data concerning CBV below the CPP levels where CBV_max has been attained appear to be contradictory. Some studies indicate a linear drop of CBV as a function of reduced CPP (Powers et al., 1984; Sette et al., 1989), whereas others show CBV_max at nonzero CPP (Schumann et al., 1988). One potential mechanism reducing CBV even below normal under these conditions might be drop of arterial input pressure below the critical closing pressure in cerebrovascular circulation (Burton, 1951). We assume in our calculations, that CBV_max is reached by CPP of 20 mm Hg, followed by a decline toward normal CBV with lower CPP values. For simplicity, an equal relative volume increase in each microvessel type is assumed to take place, thus maintaining the volume fraction ratio indicated earlier. The OER has been shown to increase on decreasing CPP below 60 mm Hg. A close to linear increase in OER with decreasing CPP below 60 mm Hg has been reported, reaching the level of 0.8 at CPP 20 mm Hg (Sette et al., 1989; Ferrari et al., 1992). Further increase in OER is expected to be impossible because of inhibition of CMRO₂ on a more severe drop in perfusion pressure. To compare simulated results with the experimental ones, we use a linear inter-relationship between CBF and CPP values below 60 mm Hg (Sette et al., 1989; Powers, 1991; Ferrari et al., 1992).

RESULTS

Simulation of the effects of various physiologic parameters

Using the normal physiologic parameters defined in Methods, T₂ changes were computed from the SE-BOLD theory at 4.7 T and 1.5 T as a function of OER and by varying the CBV values between 100% and 200% of normal (Figs. 1A and 1B). Similarly, the effects of varying nonlinear extraction of oxygen in gas exchanging microvasculature (Figs. 1C and 1D) and Hct (Figs. 1E and 1F) were estimated. The key factors affecting T₂ are OER and CBV at both field strengths during hypoperfusion. Notice that the influence of CBV on T₂ differs greatly at the two field strengths because of the field-dependent T₂ relaxation of blood and tissue. At 4.7 T, tissue and average blood T₂, weighted for fractional sizes of given vascular type, are equal at normal OER and, therefore, apparent T₂ is unaffected by CBV change. However, increased OER on reduced CPP decreases blood T₂ below that of tissue, and a simultaneous increase in CBV results in an even steeper drop in apparent T₂. At 1.5 T, blood T₂ approaches T_2,tissue when OER increases to 0.8, and, therefore, high CBV has only a small influence on T₂ during hypoperfusion.

FIG. 1.

Simulated T₂ relaxation time as a function of oxygen extraction ratio (OER). Values are calculated using the theory described in Methods for 4.7 T (A, C, and E) and 1.5 T (B, D, and F). (A and B) Cerebral blood volume (CBV) varies between 100% and 200% of normal value. (C and D) Influence of precapillary oxygen losses is estimated by varying m_c (see Eq. 1) between 0.5 and 0.9. (E and F) Hematocrit (Hct) varies from 50% to 100% of normal cerebral hematocrit. Physiologic values of CBV = 100%, m_c = 0.5, and Hct = 100% of normal are used as starting values in all figures (thick line).

We also simulated the plausible influence of precapillary oxygen losses during compromised CBF by varying the coefficient m_c in Eq. 1 (Figs. 1C and 1D). Because no quantitative information about the magnitude of this factor during hypoperfusion is available, we computed T₂ values corresponding to m_c range of 0.5 to 0.9. At 4.7 T, such an increase in m_c leads to a maximum decrease of 0.5 milliseconds in T₂ when OER is elevated up to 0.5. Further increases in OER would not influence T₂. At 1.5 T, the effect of m_c is synergistic with increasing OER, attaining a maximal effect of 1.2 milliseconds at OER close to unity. Although the effect of increased precapillary oxygen losses on T₂ is relatively small at both fields, its influence is doubled by increased CBV associated with hypoperfusion.

The effect of locally decreased Hct was estimated by varying it between 50% and 100% from normal (Figs. 1E and 1F). At 4.7 T, T₂ increases by less than 0.5 milliseconds on elevated OER (OER more than 0.36) when Hct reduces in this range. At 1.5 T, a decrease in Hct results in an increase in T₂ by 1.5 to 1.8 milliseconds over the entire OER range.

The shapes of the T₂ versus OER curves clearly differ between 1.5 T and 4.7 T (Fig. 1). At 1.5 T, T₂ decreases almost linearly with increasing OER, whereas at 4.7 T an elevation of OER above 0.7 leads only to a minor T₂ decrease. To reconcile this, we calculated T₂ and relative SE signal intensity in arterial, capillary, and venous blood fractions (Fig. 2). It is evident that at OER above 0.55 venous T₂ becomes less than 20 milliseconds at 4.7 T, which is the shortest TE in our experimental design. Consequently, the SE signal intensity arising from the venous compartment drops below 1% of the total signal intensity. This decreases the contribution of venous blood to the total relaxation time, thus moderating the apparent T₂ decrease. At 1.5 T, T₂ is 46 milliseconds for completely deoxygenated blood (Fig. 2B), and thus venous blood will be detectable at all OER values, leading close to linear T₂ dependence on OER (Fig. 1).

FIG. 2.

Simulated T₂ relaxation time (A and B) and the percentage of the total signal intensity (SI) (C and D) in tissue and in each microvessel compartments as a function of OER. The SI is calculated using echo time (TE) of 50 milliseconds for 4.7 T (C) and TE of 80 milliseconds for 1.5 T (D). Values are calculated for 4.7 T (A and C) and 1.5 T (B and D) using normal values for CBV, m_c, and Hct.

To estimate the combined effect of the aforementioned parameters in hypoperfused tissue, we computed T₂ using a schematic dependence of OER, CBV, and CBF on CPP based on literature (see Theory and Simulations). At 4.7 T, T₂ did not change with decreasing CPP until OER started to increase (Fig. 3). A maximal T₂ drop of 1.8 to 2.5 milliseconds, using a CBV increase of 50% to 100%, was computed at CPP of 20 mm Hg, which equals the highest CBV and a CBF of 32% of normal. Further decline in CBF, associated with a decrease in CBV at more severe CPP reductions, returned T₂ toward the baseline value. At 1.5 T, T₂ was slightly elevated as a response to increased CBV in the autoregulation range of CPP. Increased OER caused a maximal T₂ decrease by 3 milliseconds, and this level was maintained despite CBV changes on a severe CPP drop. In hypoperfusion, T₂ was relatively insensitive to the degree of CBV change at 1.5 T, thus making estimation of OER from T₂, in this respect, more accessible at 1.5 T than at 4.7 T.

FIG. 3.

Simulated T₂ relaxation time as a function of reduced cerebral perfusion pressure (CPP). The T₂ is calculated for 4.7 T (A) and 1.5T (B). Dependence of CBV (C), OER (D), and CBF (E) on CPP is based on data presented in literature (see Methods). Normal physiologic values for m_c and Hct are used.

The effects of altered Hct and precapillary oxygen loss were evaluated by assuming a linear change of these parameters as a function of CPP between 0 and 60 mm Hg, as indicated in Fig. 4. At 4.7 T, influence of these parameters is less than 0.5 milliseconds, whereas at 1.5 T they appear to have more pronounced effects on relaxation, thus complicating attempts to quantify OER from cerebral T₂ in hypoperfusion.

FIG. 4.

Simulated T₂ relaxation time as a function of reduced cerebral perfusion pressure (CPP). The T₂ is calculated for 4.7 T (A) and 1.5 T (B), assuming reduced cerebral Hct (D) or enhanced precapillary oxygen loss, m_c(E). Dependence of CBV (CBV_max = 200%), OER, and CBF on CPP are identical to Fig. 3. (C) Influence of decreased Hct to OER is calculated using Eq. 2. Solid line, constant Hct and m_c; dashed line, altered m_c and constant Hct; dotted line, constant m_c and altered Hct; dash-dot line, altered m_c and Hct.

Experimental hypoperfusion

The blood gases, pH, mean arterial blood pressure, and the core temperature remained within normal physiologic range during the experiments (Table 1).

TABLE 1.

Physiologic parameters before and during occlusions

	PaO₂ (mm Hg)	PaCo₂ (mm Hg)	pH	T (°C)	MABP (mm Hg)
Control	110 ± 30	42 ± 7	7.37 ± 0.06	36.9 ± 0.5	102 ± 15
Occlusion	110 ± 30	38 ± 7	7.39 ± 0.07	37.2 ± 0.4	106 ± 23

All values indicated are mean ± SD. There were no statistically significant differences between control and occlusion values (Student's t-test). MABP, mean arterial blood pressure.

We analyzed both D_av and T₂ from the cerebral cortex adjacent to the H₂-electrodes. A statistically significant 11% to 17% decrease in D_av from a control value of 0.69 ± 0.03 × 10⁻³ mm²/s was detected when CBF was 16% to 45% of the preocclusion level of 71 ± 30 mL·100 g⁻¹·min⁻¹. This was followed by a steep 26 ± 7% drop of D_av when CBF was reduced to 15% or below the control level (Fig. 5B). Post hoc analysis differentiated three homogeneous subsets of D_av figures corresponding to the CBF ranges of less than 15%, 16% to 65%, and 46% to 100% of the control level. Unlike D_av, T₂ already was significantly decreased from control figure of 50.5 ± 1.5 milliseconds, when CBF was reduced to 60% to 75% of control (Fig. 5A). The maximal reduction of T₂ by 2.5 to 2.8 milliseconds was determined at CBF values 15% to 60% of normal. Further reduction of CBF was followed by attenuation of T₂ response, however. ΔT₂ as a function of CBF, calculated for 4.7 T using CBV_max of 200%, constant Hct, and elevated m_c as described earlier (Fig. 4E), is overlaid with the experimental ΔT₂ data (Fig. 5A). Experimental data were not significantly different from simulated ΔT₂ curve (p > 0.10), indicating good agreement with the model.

FIG. 5.

Experimentally determined ΔT₂(A) and (B) ΔD_av as a function of CBF from 26 animals. The magnetic resonance imaging data are divided into seven groups according to CBF as follows: 0% to 15% (n = 13), 16% to 30% (n = 7), 31% to 45% (n = 7), 46% to 60% (n = 8), 61% to 75% (n = 14), 76% to 90% (n = 36), 91% to 110% (n = 67) of control CBF. Dashed line indicates the simulated T₂ change for 4.7 T calculated as in Fig. 4. using constant Hct, altered m_c, and CBV_max = 200%. Experimental values indicated are means ± SD. Statistical significance is evaluated using analysis of variance with Tukey's HSD post hoc test between control and occlusion values (*P < 0.05, **P < 0.01).

DISCUSSION

The current data show that mismatch between oxygen consumption and delivery associated with misery perfusion can be imaged by means of single-echo T₂ MRI, and the exchange-based SE-BOLD theory can be applied to explain the effects of reduced T₂ in quantitative terms by adopting literature data for some of the key physiologic parameters. The experimental T₂ data at 4.7 T as a function of CBF show a U-shaped curve with a minimum between 15% and 60% of normal CBF, where T₂ was reduced by approximately 2.5 to 2.8 milliseconds relative to the baseline physiologic conditions, which is of the same order of magnitude as earlier measured in rat cortical hypoperfusion model at 4.7 T (Gröhn et al., 1999). The shape of the T₂ curve matches relatively well with the calculated T₂ graph, taking into account the reported changes in CBV, OER, and m_c.

The current experimental MRI approach was designed to contain diffusion as a “reference,” since reports have been published interrelating CBF and diffusion coefficient using independent methods for their measurements. These studies unambiguously show that severe reduction of diffusion takes place at a level of CBF of approximately 20 mL·100 g⁻¹·min⁻¹ (Busza et al., 1992; Kohno et al., 1995). Even if caution must be exercised in comparing data of orientation-dependent diffusion constants with D_av results (Miyabe et al., 1996), we found a consistent CBF threshold for severe collapse of diffusion. Interestingly, we determined a small drop in D_av below CBF 45% of control, which is of the same order of magnitude as reported for diffusion reduction in periphery of ischemic core (Kohno et al., 1995). Statistical analysis revealed three distinct homogeneous subsets of diffusion values on graded hypoperfusion, suggesting that diffusion may drop in two phases along reduced CBF. A quantitative match between data by Kohno et al. (1995) and our CBF thresholds for diffusion changes support the accuracy of the CBF measurements.

The simulated T₂ graph using the equations in SE-BOLD theory is solely determined by the physiologic variables chosen. The experimental U-shaped curve of T₂/CBF dependency can be reconciled by reported changes of both CMRO₂ or CBV. Previous experimental data from rat show that CMRO₂ is inhibited below CBF of 35 mL·100 g⁻¹·min⁻¹ (Ferrari et al., 1992). Powers (1991) deduced, using results from human PET studies, that brain oxygen uptake begins to decline below CBF of 30 mL·100 g⁻¹·min⁻¹. The shortest T₂ values that we determine occur at CBF range that is close to 30 mL·100 g⁻¹·min⁻¹, and reversal of T₂ change could reflect inhibition of CMRO₂ at severely compromised substrate delivery. However, the behavior of CBV, as shown in Results, contributes to T₂ at much the same CBF range as the inhibition of CMRO₂, thus making it impossible to distinguish between the quantitative role of these two factors. Nevertheless, the observed BOLD effect appears at the flow and metabolic conditions where tissue exhibits viability.

Whereas many of the parameters mentioned earlier, such as T_2,tissue, can be assumed to be unaltered after a few minutes of moderate hypoperfusion, this is not true after blood-brain barrier breakdown, which is known to occur several hours after ischemia (van Bruggen et al., 1994). Increased permeability of the cerebral vessels may enhance the exchange between microvessels and tissue, collapsing the validity of Eq. 6 and leading to a situation of fast or intermediate exchange. Notice that after 1 to 2 hours of ischemia, V_d and R_2,tissue dramatically change, thus making the relative small reduction in R_2,blood to go undetected unless other contributing parameters are determined independently. Edema has been shown to develop earliest after 30 minutes of CBF dropping below the threshold of the ischemic energy failure and only after extended periods on graded ischemia in baboons (Bell et al., 1985). Thus, we are confident that Eq. 6 holds for the conditions under which our measurements were performed.

The theory treats SE-BOLD as an intravascular phenomenon, and because no alterations in vascular permeability or erythrocyte viability occur in graded hypoperfusion experiments, SE-BOLD is expected to remain an intravascular phenomenon in the experimental data set. The constants for susceptibilities and relaxation parameters were determined from experimental data of other groups. Theory assumes the exchange of water between erythrocyte and plasma as the main mechanism of transferring relaxation effects by deoxyhemoglobin. However, other mechanisms, such as diffusion through field gradients, may operate at high fields (Thulborn et al., 1982; Ogawa et al., 1993).

Implications to imaging of misery perfusion in clinical settings

The current data show that the absolute T₂ and diffusion images data exhibit distinct CBF thresholds, thus indicating different pathophysiologic events of early reduction of blood flow. Because oxygen metabolism is a mandatory requirement for generation of deoxyhemoglobin, a drop in T₂ is regarded as an index of misery perfusion and tissue “viability” whereas a diffusion drop is associated with ischemic energy failure. Such information is expected to be of great clinical impact and would give complementary information to that of perfusion-diffusion mismatch (Sorensen et al., 1996; Baird, et al., 1998). Our simulations suggest that reduction of T₂ also is detectable at 1.5 T during acute reduction of CBF. Misery perfusion has been found up to 2 to 4 days after the onset of stroke (Wise et al., 1983; Sette et al., 1989; Derdeyn et al., 1998), and it fulfills many of the signs associated with definitions of penumbral tissue (Astrup et al., 1981; Hossmann, 1994; Ginsberg and Pulsinelli, 1994). These areas can be salvaged with reperfusion within 6 hours (Young et al., 1996) but often progress into infarction without treatment (Wise et al., 1983; Pappata et al., 1993; Furlan et al., 1996). One of the greatest challenges in imaging of cerebral compromised perfusion and ischemia is to identify tissue at risk of infarction so that the possible intervention can be proceeded. In principle, this can be achieved by detecting compromised hemodynamic status of the tissue and by revealing the metabolic activity in the tissue which more directly indicates the viability of the tissue.

Studies using PET suggest that CBF and CBV alone are not sufficient to predict tissue viability, since they indirectly probe tissue status and lack temporal dimension of misery perfusion (Heiss and Podreka, 1993). These studies conclude that tissue outcome can be different at similar perfusion status (Heiss et al., 1992; Heiss and Podreka, 1993; Hossmann, 1994; Furlan et al., 1996). Studies on acute ischemic stroke (Frackowiak, 1985), carotid stenosis (Grubb et al., 1998), and on experimental acute middle cerebral artery occlusion (Young et al., 1996; Heiss et al., 1997) indicate that elevated OER may be an indicator of tissue outcome. It has been shown in a cat model that CMRO₂ proceeds at control level despite reduction of CBF by more than 60% because of maximized OER (Heiss et al., 1997). In this study, it also was shown that the prognosis of middle cerebral artery occlusion largely followed the behavior of OER during the first hour of the insult. In cats with persistently elevated OER, the death toll was much lower than in those with unaffected or reduced OER, and smaller infarcts were detected in the animals with increased OER and low flow than in those with normal or low OER under compromised CBF (Heiss et al., 1997). An infarction is associated with a fall in CMRO₂ and a subsequent decline in OER indicates poor prognosis (Wise et al., 1983). The current study raises the possibility of using single SE-MRI to image viable oxygen-using tissue and of using T₂ and diffusion data to assess tissue status and viability in the previously suggested manner (Gröhn et al., 1998). We emphasize that absolute single SE-T₂ images are to be used to this end, since the changes are only 5% of the baseline T₂ and would go undetected in T₂-weighted images. Furthermore, our observations are eligible for acute CBF reduction only, since time-dependent phenomena may change the pattern of cerebral adaptations (Hossmann, 1994).

Footnotes

Abbreviations used

Acknowledgements

The authors thank Dr. Kari Mauranen for advice in statistical analysis, and Mrs. Kati Holopainen and Ms. Niina Kuhmonen for expert technical assistance.

References

Allen

Busza

Proctor

(1993) Controllable graded cerebral ischaemia in the gerbil: studies of cerebral blood flow and energy metabolism by hydrogen clearance and P-31 NMR spectroscopy. NMR Biomed 6:181–186

Allerhand

Gutowsky

(1964) Spin-echo NMR studies of chemical exchange. Part I: Some general aspects. J Chem Phys 41:2115–2126

Astrup

Siesjö

Symon

(1981) Thresholds in cerebral ischemia: the ischemic penumbra. Stroke 12:723–725

Baird

Warach

(1998) Magnetic resonance imaging of acute stroke. J Cereb Blood Flow Metab 18:583–609

Baker

Lindsey

Weisbroth

(1979) The laboratory rat. New York: Academic Press, pp. 109–110

Bell

Symon

Branston

(1985) CBF and time thresholds for the formation of ischemic cerebral edema, and effect of reperfusion in baboons. J Neurosurg 62:31–41

Benga

Borza

(1995) Diffusional water permeability of mammalian red blood cells. Comp Biochem Physiol 112:653–659

Brooks

Di Chiro

Keller

(1980) Explanation of cerebral white-gray contrast in computed tomography. J Comput Assist Tomogr 4:489–491

Bryant

Marill

Blackmore

Francis

(1990) Magnetic relaxation in blood and blood clots. Magn Reson Med 13:133–144

10.

Buelke-Sam

Holson

Bazare

Young

(1978) Comparative stability of physiological during sustained anesthesia in rats. Lab Anim Sci 28:157–162

11.

Burton

(1951) On the physical equilibrium of the small blood vessels. Am J Physiol 164:319–329

12.

Busza

Allen

King

van Bruggen

Williams

Gadian

(1992) Diffusion-weighted imaging studies of cerebral ischemia in gerbils: potential relevance to energy failure. Stroke 23:1601–1612

13.

Crockard

Iannotti

Hunstock

Smith

Harris

Symon

(1980) Cerebral blood flow and edema following cartoid occlusion in the gerbil. Stroke 11:494–498

14.

Davis

Kwong

Weisskoff

Rosen

(1998) Calibrated functional MRI: mapping the dynamics of oxidative metabolism. Proc Natl Acad Sci 95:1834–1839

15.

de Crespigny

Wendland

Deruguin

Koziewska

Moseley

(1992) Real time observation of transient focal ischemia and hyperemia in cat brain. Magn Reson Med 27:391–397

16.

Derdeyn

Powers

Grubb

Jr (1998) Hemodynamic effects of middle cerebral artery stenosis and occlusion. Am J Neuroradiol 19:1463–1469

17.

Ferrari

Wilson

Hanley

Traystmann

(1991) Effects of graded hypotension in cerebral blood flow, volume, and mean transit time in dogs. Am J Physiol 262:1908–1914

18.

Flecknell (1996) Laboratory animal anaesthesia. London: Academic Press, p 116

19.

Frackowiak

RSJ

(1985) Pathophysiology of human cerebral ischemia: studies with positron tomography and ¹⁵oxygen. Assoc Res Nerv Ment Dis 63:139–161

20.

Furlan

Marchal

Viader

Derlon

J-M

Baron

J-C

(1996) Spontaneous neurological recovery after stroke and the fate of ischemic penumbra. Ann Neurol 40:216–226

21.

Garwood

(1991) Symmetric pulses to induce arbitrary flip angles with compensation for RF inhomogeneity and resonance offsets. J Magn Reson 94:511–525

22.

Ginsberg

Pulsinelli

(1994) The ischemic penumbra, injury thresholds, and the therapeutic window for acute stroke. Ann Neurol 36:553–554

23.

Grubb

Jr Derdeyn

Fritsch

(1998) Importance of hemodynamic factors in the prognosis of symptomatic carotid occlusion. J Am Med Assoc 280:1055–1060

24.

Grubb

Jr Raichle

Phelps

Ratcheson

(1975) Effects of increased intracranial pressure on cerebral blood volume, blood flow, and oxygen utilization in monkeys. J Neurosurg 43:385–398

25.

Gröhn

OHJ

Lukkarinen

Oja

(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

26.

Gröhn

OHJ

Lukkarinen

Silvennoinen

Pitkänen

van Zijl

PCM

Kauppinen

(1999) Quantitative MRI assessment of cerebral ischemia in rat using on-resonance T1ρ. Magn Reson Med 42:268–276

27.

Heiss

Graf

Lottgen

(1997) Repeat positron emission tomographic studies in transient middle cerebral artery occlusion in cats: residual perfusion and efficacy of postischemic reperfusion. J Cereb Blood Flow Metab 17:388–400

28.

Heiss

Podreka

(1993) Role of PET and SPECT in the assessment of ischemic cerebrovascular disease. Cerebrovasc Brain Metab Rev 5:235–263

29.

Heiss

Huber

Fink

(1992) Progressive derangement of periinfarct viable tissue in ischemic stroke. J Cereb Blood Flow Metab 12:193–203

30.

Hossmann

(1994) Viability thresholds and the penumbra of focal ischemia. Ann Neurol 36:557–565

31.

Hossmann

Hoehn-Berlage

(1995) Diffusion and perfusion MR imaging of cerebral ischemia. Cerebovasc Brain Metab Rev 7:187–217

32.

Intaglietta

Johnson Winslow

(1996) Microvascular and tissue oxygen distribution. Cardiovasc Res 32:632–643

33.

Ivanov

Derry

Vovenko

Samoilov

Semionov

(1982) Direct measurement of oxygen tension at the surface of arterioles, capillaries and venules of cerebral cortex. Pflugers Arch 393:118–120

34.

Kato

Kogure

Ohtomo

(1986) Characterization of experimental ischemic brain edema utilizing proton nuclear magnetic resonance imaging. J Cereb Blood Flow Metab 6:212–221

35.

Knight

Dereski

Helpern

Ordidge

Chopp

(1994) Magnetic resonance imaging assessment of evolving focal cerebral ischemia: comparison with histopathology in rats. Stroke 25:1252–1261

36.

Kohno

Hoehn-Berlage

Mies

Back

Hossmann

(1995) Relationship between diffusion-weighted MR images, cerebral blood flow, and energy state in experimental brain infarction. Magn Reson Imag 13:73–80

37.

Kohno

Ohta

Kohno

Kumon

Sakaki

Okujima

(1995) Early detection of cerebral ischemic lesion using diffusion-weighted MRI. J Comp Assist Tomogr 19:982–986

38.

Leenders

Perani

Lammertsma

(1990) Cerebral blood flow, blood volume and oxygen utilization: normal values and effect of age. Brain 113:27–47

39.

McHedlishvili

Varazashvili

(1981) Red cell/plasma ratio in blood flowing in microvascular beds under control and ischemic conditions. Microvasc Res 21:302–307

40.

McHedlishvili

Varazashvili

(1987) Hematocrit in cerebral capillaries and veins under control and ischemic conditions. J Cereb Blood Flow Metab 7:739–744

41.

McHedlishvili

Varazashvili

Mamaladz

Momtselidze

(1997) Blood flow structuring and its alterations in capillaries of the cerebral cortex. Microvasc Res 53:201–210

42.

Miyabe

Mori

van Zijl

PCM

(1996) Correlation of the average water diffusion constant with cerebral blood flow and ischemic damage after transient middle cerebral artery occlusion in cats. J Cereb Blood Flow Metab 16:881–891

43.

Mori

van Zijl

PCM

(1995) Diffusion weighting by the trace of the diffusion tensor within a single scan. Magn Reson Med 33:41–52

44.

Ogawa

Lee

Barrere

(1993) The sensitivity of magnetic resonance image signals of rat brain to changes in the cerebral venous blood oxygenation. Magn Reson Med 29:205–210

45.

Ogawa

Lee

Nayak

Glynn

(1990) Oxygenation-sensitive contrast in magnetic resonance image of rodent brain at high magnetic fields. Magn Reson Med 14:68–78

46.

Ogawa Menon

Tank

(1993) Functional brain mapping by blood oxygenation level-dependent contrast magnetic resonance imaging: a comparison of signal characteristics with a biophysical model. Biophys J 64:803–812

47.

Oja

JME

Gillen

Kauppinen

Kraut

van Zijl

PCM

(1999) Imaging the balance between oxygen consumption and delivery J Cereb Blood Flow Metab 19:1289–1295

48.

Olesen

Paulson

Lassen

(1971) Regional cerebral blood flow in man determined by the initial slope of the clearance of intra-arterially injected ¹³³Xe. Stroke 2:519–540

49.

Pappata

Fiorelli

Rommel

(1993) PET study of changes in local brain hemodynamics and oxygen metabolism after unilateral middle cerebral artery occlusion in baboons. J Cereb Blood Flow Metab 13:416–424

50.

Popel

Gross

(1979) Analysis of oxygen diffusion from arteriolar networks. Am J Physiol 237:681–689

51.

Powers

(1991) Cerebral hemodynamics in ischemic cerebrovascular disease. Ann Neurol 29:231–240

52.

Powers

Grubb

Jr Raichle

(1984) Physiological responses to focal cerebral ischemia in humans. Ann Neurol 16:546–552

53.

Pulsinelli

Brierley

Plum

(1982) Temporal profile of neuronal damage in a model of transient forebrain ischemia. Ann Neurol 11:491–498

54.

Roussel

van Bruggen

King

Gadian

(1995) Identification of collaterally perfused areas following focal ischemia in the rat by comparison of gradient echo and diffusion-weighted MRI. J Cereb Blood Flow Metab 15:578–586

55.

Sandor

Cox van Put

de Jong

de Weid

(1986) Continuous measurement of cerebral blood volume in rats with the photoelectric technique: effect of morphine and noloxone. Life Sci 39:1657–1665

56.

Sorensen

Buonanno

Gonzalez

(1996) Hyperacute stroke: evaluation with combined multisection diffusion-weighted and hemodynamic weighted echo-planar imaging. Radiology 199:391–401

57.

Schumann

Touzani

Young

Morello

Baron

MacKenzie

(1998) Evaluation of the ratio of cerebral blood flow to cerebral blood volume as an index of local cerebral perfusion pressure. Brain 121:1369–1379

58.

Schwab

Bauer

Zwiener

(1997) The distribution of normal brain water content in Wistar rats and its increase due to ischemia. Brain Res 749:82–87

59.

Sette

Baron

Mazoyer

Levasseur

Pappata

Crouzel

(1989) Local brain haemodynamics and oxygen metabolism in cerebrovascular disease: positron emission tomography. Brain 112:931–951

60.

Sharan

Jones

MDJ

Koehler

Traystman

Popel

(1989) A compartmental model for oxygen transport in brain microcirculation. Ann Biomed Eng 17:13–38

61.

Thulborn

Waterton

Matthews

Radda

(1982) Oxygenation dependence of the transverse relaxation time of water protons in whole blood at high field. Biochem Biophys Acta 714:265–270

62.

Torres Filho

Kerger

Intaglietta

(1996) Po₂ measurements in arteriolar networks. Microvasc Res 51:202–212

63.

Ulatowski

Bucci

Nishikawa

(1996) Cerebral O₂ transport with hematocrit reduced by cross-linked hemoglobin transfusion. Am J Physiol 270:466–475

64.

Ulatowski

Oja

JME

Suarez

Kauppinen

Traystman

van Zijl

PCM

(1999) In vivo determination of absolute cerebral blood volume using hemoglobin as a natural contrast agent: an MRI study employing altered arterial carbon dioxide tension. J Cereb Blood Flow Metab 19:809–817

65.

van Bruggen

Roberts

TPL

Cremer

(1994) The application of magnetic resonance imaging to the study of experimental cerebral ischemia. Cerebrovasc Brain Metab Rev 6:180–210

66.

van Zijl

Eleff

Ulatowski

(1998) Quantitative assessment of blood flow, blood volume and blood oxygenation effects in functional magnetic resonance imaging. Nat Med 4:159–167

67.

Whittall

MacKay

Graeb

Nugent

Paty

(1997) In vivo measurement of T₂ distributions and water contents in normal human brain. Magn Reson Med 37:34–43

68.

Wise

Bernardi

Frackowiak

Legg

Jones

(1983) Serial observations on the pathophysiology of acute stroke: the transition from ischaemia to infarction as reflected in regional oxygen extraction. Brain 106:197–222

69.

Yamauchi

Fukuyama

Nagahama

Katsumi

Okawaza

(1998) Cerebral hematocrit decreases with hemodynamic compromise in carotid artery occlusion: a PET study. Stroke 29:98–103

70.

Young

Sette

Touzani

(1996) Relationships between high oxygen extraction fraction in the acute stage and final infarction in reversible middle cerebral artery occlusion: an investigation in anesthetized baboons with positron emission tomography. J Cereb Blood Flow Metab 16:1176–1188