A New Magnetic Resonance Imaging Method for Mapping the Cerebral Blood Volume Fraction: The Rapid Steady-State T 1 Method

Abstract

This paper describes a new rapid steady-state T₁ (RSST₁) method for mapping the cerebral blood volume fraction (CBVf) by magnetic resonance imaging (MRI). The principle is based on a two-compartment model of the brain (intra- and extravascular), and the effects of paramagnetic contrast agents on the intravascular longitudinal relaxation time T₁. Using appropriate parameters, an Inversion-Recovery-Fast-Low-Angle-Shot sequence acts like a low pass T₁ filter, suppressing signals from tissues with T₁≫TR (TR=repetition time). It was shown in vivo that, exceeding a particular contrast agent dose, the signal reaches its maximum (corresponding to the intravascular equilibrium magnetization), and is maintained for a duration related to the dose. Acquisitions during this steady state divided by an additional measure of the overall (intra- and extravascular) magnetization at thermal equilibrium provides the CBVf. Experiments were performed on healthy rats at 2.35 T using P760 (Gd³⁺—compound from Guerbet Laboratories) and Gd-DOTA. Because of its high longitudinal relaxivity, P760 is more convenient, and was used to show the feasibility of the method. The CBVf in different structures of the rat brain was compared. The average CBVf for the whole brain slice is 3.29%±0.69% (n=15). The influence of transendothelial water exchange was quantified and transversal relaxation effects were found negligible in microvasculature. Finally, the sensitivity of the method to CBVf increases under hypercapnia was evaluated (1%/mm Hg PaCO₂), demonstrating its potential for longitudinal studies and functional MRI. Clinical applications are feasible since equivalent results were obtained with Gd-DOTA.

Keywords

cerebral blood volume fraction hypercapnia intravascular paramagnetic contrast agent longitudinal relaxation time magnetic resonance imaging rat brain

Introduction

The quantification of the cerebral blood volume (CBV) is useful for studying cerebral pathophysiology. For example, it is a valuable parameter for tumor grading (Aronen et al, 1994) and for the prediction of the final cerebral ischemic infarct volume, and highly specific for the detection of infarcted tissue (Schaefer et al, 2002). In general, absolute CBV values are necessary for repeated measures in the control of therapeutic interventions, in particular when studying an antiangiogenic drug. The CBV also allows the detection of activated areas in functional imaging, and its quantification is helpful to analyze the complex BOLD signal (Ogawa et al, 1993).

The standard magnetic resonance imaging (MRI) methods for CBV evaluation are based on the properties of intravascular paramagnetic or superparamagnetic contrast agents (CAs), which affect longitudinal (T₁) and transversal (T₂ or T₂∗) relaxation times. Two common techniques are available: dynamic and steady-state approaches.

The dynamic methods, including the Dynamic Susceptibility Contrast method (Moonen et al, 1992; Rosen et al, 1991; Villringer et al, 1988) and the T₁ bolus method (Dean et al, 1992; Hacklander et al, 1996, 1997), are based on the intravascular indicator dilution theory (Meier and Zierler, 1954) and require high temporal resolution to acquire series of images during the first pass of the bolus of the injected tracer in the brain tissue. The area under the curve describing the temporal change of the relaxation rate ΔR₁=(ΔT₁)⁻¹ or ΔR₂∗=(ΔT₂∗)⁻¹ during bolus passsage has been demonstrated to be proportional to the CBV (Belliveau et al, 1990; Hacklander et al, 1996).

Steady-state methods (An and Lin, 2002; Hamberg et al, 1996; Kuppusamy et al, 1996; Lin et al, 1997; Payen et al, 1998; Schwarzbauer et al, 1997) consist in acquiring two signals, one before and one after injection of a CA when it is uniformly distributed in the vascular system. The differences ΔR₁ or ΔR₂∗ between the relaxation rates measured before and after CA administration have been shown to be proportional to the CBV of macro- and microvasculature, while the change ΔR₂ in transverse relaxation rate R₂ has been shown to be mainly proportional to the CBV of microvasculature (Boxerman et al, 1995).

Currently, most of these methods provide only relative values (the CBV ratio between a region of interest (ROI) and a reference region) (Hamberg et al, 1996; Payen et al, 1998). Absolute quantification defined as mL blood/100 mg brain tissue or as the CBV fraction (CBVf) in mL blood/mL brain tissue is generally difficult and additional measurements are needed.

Quantifying CBV with dynamic methods necessitates accurate determination of the arterial input function. The arterial input function measured in a supplying artery may differ from that in the tissue of interest resulting in miscalculation of the CBV (Calamante et al, 2004).

With steady-state methods, quantification requires the knowledge of the blood signal before and after CA administration. This signal difference is either directly used for normalization (Kuppusamy et al, 1996; Lin et al, 1997) or to calculate the relaxation rate difference (ΔR₁) (Schwarzbauer et al, 1997) or the susceptibility difference (Payen et al, 2000). This requires either blood sampling (Schwarzbauer et al, 1993; Tropres et al, 2001) or the determination of the signal from voxels located in larger blood vessels (Kuppusamy et al, 1996; Lin et al, 1997; Schwarzbauer et al, 1997).

Moreover, in case of the ΔR₂∗ techniques, the signal interpretation is further complicated by the signal dependence on the vascular morphology (Boxerman et al, 1995), and by the hematocrit difference between capillaries and large vessels (Cremer and Seville, 1983).

The vascular space occupying imaging technique (Lu et al, 2003) is based on the suppression of the blood water signal. In its first version, it does not use CAs, and the signal variation associated with brain activation was shown to be proportional to the CBV variation. More recently a second version (Lu et al, 2005) for absolute CBV mapping was proposed, based on the same principle but using an intravascular CA to obtain a fully relaxed blood signal.

In this paper, the rapid steady-state T₁ method (RSST₁), an alternative MRI method for mapping CBVf in a simple and direct way, is presented. In contrast to the vascular space occupying technique, which suppresses no more than the blood signal before CA injection, the RSST₁ method takes advantage of the T₁ shortening of intravascular CAs and of rapid MRI to suppress all signals from extravascular tissues characterized by long T₁. Consequently, the blood signal, completely relaxed by the CA, being acquired selectively and directly, no image subtraction is necessary. For CBVf mapping, an additional image is acquired at thermal equilibrium to obtain the overall (intra- and extravascular) equilibrium magnetization.

The method was developed and validated at 2.35 T on the normocapnic and hypercapnic healthy rat brain using either bolus injections or continuous CA infusion of P760 and Gd-DOTA.

Materials and methods

Theory

The principle is based on a two-compartment model of the brain parenchyma, assuming the water exchange across the blood—brain barrier (BBB) to be negligible (Kuppusamy et al, 1996) in a first approach. The idea of this method is to distinguish these compartments by two different T₁ clusters: high T₁ for the extravascular compartment and low T₁ for the intravascular compartment, in the presence of an intravascular paramagnetic CA.

An Inversion-Recovery-Fast-Low-Angle-SHot (IR-FLASH) rapid sequence was used (Haase, 1990). In the steady state, the longitudinal magnetization M_Z normalized to its thermal equilibrium M₀ is described by

M_{Z} / M_{0} = 1 - \frac{2 exp (- T_{i n v} / T_{1})}{1 + exp (- T R / T_{1})}

(1)

In Figure 1, the ratio M_Z/M₀ from equation (1) is plotted as a function of T₁ for different inversion times (T_inv) and for a particular repetition time (TR) of 750 ms.

Figure 1

Fractional longitudinal magnetization M_Z/M₀ as a function of the longitudinal relaxation time constant T₁ plotted for TR=750 ms and for different T_inv according to equation (1). For tissues characterized by T₁>1000 ms, M_z/M₀ approaches zero. For T₁<T_inv/5, M_Z/M₀=1.

Suppression of the longitudinal tissue magnetization can be achieved for the inversion time

T_{i n v} (T R, T_{1}) = T_{1} ln 2 - T_{1} ln [1 + exp (- T R / T_{1})]

(2)

From equation (2) it can be deduced that for TR≪T₁, the nulling point T_inv tends towards TR/2. This is illustrated in Figure 1 for the particular example of TR=750 ms. The sequence used appears as a low-pass T₁ filter. If T_inv=325 ms, which approaches TR/2, the signal is practically suppressed for tissues characterized by a T₁ beyond 1000 ms, such as extravascular brain tissue and blood without CA. In contrast, for tissues with short T₁ with respect to T_inv (T₁<T_inv/5), such as for blood in the presence of the CA, the longitudinal magnetization is equal to the equilibrium magnetization M₀ and the signal becomes independent of the T₁ value. This condition is referred to as a ‘rapid steady-state regime’, during which the intravascular signal is constant.

Under these conditions, if transverse relaxation effects are negligible, the CBVf is simply given by

C B V f = \frac{S_{p o s t}}{S_{0}}

(3A)

where S_post is the signal acquired after CA administration, and S₀ is the signal corresponding to the thermal equilibrium magnetization from both compartments. S₀ is acquired for normalization to obtain the CBVf.

Since the tissue T₁ values strongly depend on the static field strength, particular tissues may have low T₁ values, which cannot be completely suppressed by the rapid IR-FLASH sequence for a given TR. In this case, the TR can be shortened but this requires further reduction of the intravascular T₁. It is also possible to subtract an image acquired before CA injection from the postcontrast image to eliminate any residual tissue signal.

Magnetic Resonance Imaging

Experiments were performed on a 2.35 T small animal imager (horizontal bore Bruker magnet, 40 cm inner diameter equipped with an SMIS spectrometer). Two coil systems were used. Either a 1H radio frequency surface coil (34 mm diameter) was used for emission and reception, or an actively decoupled system composed of a homogenous coil for emission and a surface coil for reception.

The inversion pulse of the IR-FLASH sequence, a nonslice-selective adiabatic sech/tanh pulse, is followed by a rapid gradient echo imaging sequence consisting of a series of radio frequency readout pulses with a low flip angle α=10°, and a short repetition time TR_FLASH=10.5 ms. The echo time TE was set to its minimum value 3.2 ms to minimize the transversal relaxation effect. The TR between successive images was 750 ms.

The image acquisition started at T_inv=325 ms at the center of k-space. This T_inv value is approximately TR/2 and was experimentally found to most efficiently suppress the tissue signal. The slice thickness was set to 2 mm. A 32 × 32 acquisition matrix was used for a field of view of either 32 × 32 mm² or 24 × 24 mm², leading to in-plane spatial resolutions of either 1 × 1 mm² or 0.75 × 0.75 mm², respectively. A series of images was acquired with these parameters before CA injection to obtain a baseline of the signal intensity (S_pre), and after CA injection (S_post). The S₀ signal was obtained with the same sequence using a long TR of 10 secs and a T_inv of 8 secs. Signals were acquired for at least 1 min for the bolus injection mode, and for up to 20 mins under continous CA infusion. Inversion-Recovery-Fast-Low-Angle-SHot images of the same slice acquired with TR=3.5 secs, T_inv=325 ms and with a 128 × 64 matrix provided good tissue contrast. Spin-Echo T₂-weighted anatomic (TR=2 secs, TE=80 ms) and angiographic MR images (TR=100 ms, TE=25 ms) were also acquired (matrix size 128 × 64). The brain slices were chosen in the transversal or coronal orientation and contained part of the basal nuclei and frontal and/or parietal cortex, respectively.

Contrast Agents

P760 and Gd-DOTA, both from Guerbet Laboratories, were used as CAs.

P760 is an experimental Gadolinium (Gd³⁺)-based compound characterized by a high relaxivity: r₁=19.7±0.7 (mmol/L)⁻¹ sec⁻¹ in rat plasma at 2.35 T (versus r₁=3.4±0.1 (mmol/L)⁻¹ sec⁻¹ for Gd-DOTA) (Fonchy et al, 2001). The r₁-relaxivity is defined as ΔR₁/[CA], where ΔR₁ represents the change in the longitudinal relaxation rate of the blood induced by the CA, and [CA] stands for its concentration in the blood. Injected at an equal Gd³⁺ dose, the Gd³⁺ concentration in the blood remains higher with P760 than with Gd-DOTA during the postbolus phase, because of its intravascular restriction (Corot et al, 2002). However, its renal elimination rate is equal to that of Gd-DOTA (Port et al, 1999).

Assuming that the total blood volume of a rat weighting 300 g is approximately 15 to 21 mL (Lee and Blaufox, 1985), the P760 blood concentration after a bolus injection of 0.1 mmol/kg P760 ranges between 1.43 and 2 mmol/L. According to the r₁ relaxivity of P760, the T₁ value of blood is then reduced from 1450 ms to 36 or 25 ms, which allows full relaxation of the intravascular magnetization at T_inv=325 ms.

The infusion technique consisted of an initial bolus injection of 0.1 mmol/kg P760 followed by a continuous infusion at a rate of 7.8 μmol kg⁻¹ min⁻¹. The infusion rate was chosen so as to maintain a plasmatic P760 concentration of 1.3 mmol/L, leading to a blood T₁ of approximately 40 ms. In this case, the blood signal keeps its equilibrium value throughout the infusion duration (up 60 mins). For a rat weighting 300 g, the P760 volume injected as a bolus is 0.43 ml (2.3% of the total blood volume), and the volume administered during an infusion of 60 mins is 2 ml (0.2% of the total blood volume per minute).

Gd-DOTA bolus injections were performed at a dose of 0.2 mmol/kg, except for experiments under infusion and hypercapnia. For the hypercapnia experiments, bolus injections of 0.3 mmol/kg Gd-DOTA were administered, resulting in a steady state of about 5 mins, as could be obtained with 0.1 mmol/kg P760. The use of Gd-DOTA for the infusion experiment necessitated a bolus injection of the same dose to avoid a signal drop before the arrival of the continuous infusion at a rate of 63 μmol kg⁻¹ min⁻¹ because of the length of the intravenous-line used in these experiments.

Animals

In vivo studies were performed on healthy male rats (320 to 450 g) from Sprague—Dawley or Wistar stains. They were oxygenized and anesthetized with 1.3% to 1.8% isoflurane. Procedures and animal care conformed to the ‘Guidelines of the French Government’ (decree 87 to 848, 19 October 1987, licenses 006683 and A38071). Catheters were inserted into the femoral vein for CA administration and into the femoral artery for blood sampling. Arterial hemoglobin, pH, and blood gas levels were measured and corrected for rectal temperature before and after each CA injection with the use of a pH/blood gas analyzer (Radiometer Copenhagen ABL™ 510) to confirm full oxygenation and a normocapnic arterial carbon dioxide tension (PaCO₂) (40 mm Hg±5 mm Hg) as well as normal hematocrit (>35%). As CBV is known to increase with PaCO₂ (Payen et al, 1998; Shockley and LaManna, 1988), rats not fulfilling normal physiologic conditions were excluded from the study.

The rats were positioned prone. A warm-water circuit maintained rectal body temperature at 37°C±1°C during surgery and throughout the experiment. To minimize the possibility of motion artifacts, the rat's head was fixed by means of earbars and a bitebar.

Rats undergoing the hypercapnia experiment were additionally equipped with a tracheal cannula for mechanical ventilation. Pulmonary and arterial blood pressures were monitored throughout the experiment. An intravenous injection of 2 mL pancuronium bromide (Pavulon^®, Organon, 0.1 mg/mL) was performed before signal acquisition to avoid spontaneous breathing under hypercapnia.

Hypercapnia was induced with an addition of 5% to 7% CO₂ during 30 to 45 mins, while the oxygen flow (33%) was maintained. Repeated CBVf measurements were performed on the same rat. For each hypercapnia level, a delay of 15 mins was necessary to reach a stable PaCO₂, measured before signal acquisition. Recovery from hypercapnia (30 to 40 mins) under air and 33% oxygen ventilation was allowed before performing a final acquisition.

Preliminary Experiments

To verify whether postinjection the rapid steady-state regime is attained, in vivo dose studies were performed to establish the duration of the steady state for the injected CA dose. The minimal dose required was determined from a study on six rats using different doses of Gd-DOTA (0.01 to 0.15 mmol/kg; n=3; Figure 2A) and P760 (2 to 35 μmol/kg; n=3; Figure 2B). In another experiment, the steady state resulting from 0.1 or 0.2 mmol/kg P760 was compared (n=2; Figure 2C). The CA doses were delivered in increasing order. The delay between successive injections was 30 mins for Gd-DOTA and 60 mins for P760, except for the dose of 0.1 mmol/kg P760 after which an elimination interval of 90 mins was allowed.

Figure 2

CA dose studies on healthy rats. Each symbol corresponds to one signal acquisition averaged over the whole brain. The injection of the CA is administered 15 secs after the start of the acquisition. (A) Gd-DOTA: 0.15 mmol/kg of Gd-DOTA results in a constant signal of about 30 secs duration between the two time points marked with arrows. (B) P760: a constant signal of at least 30 secs duration is achieved for P760 at a dose of 35 μmol/kg. (C) Comparison of the steady-state duration and of the signal amplitude resulting from 0.1 and 0.2 mmol/kg P760 injected to the same rat. The maximum intensity of the signal is the same for the two CA doses, indicating that the intravascular magnetization has attained thermal equilibrium. For the dose of 0.2 mmol/kg, the signal is constant for a longer time.

Moreover, the T₁ of the cerebral parenchyma was measured in vivo, as well as in vitro in blood sampled before and during the first minute after 0.1 mmol/kg P760 injection (at 20°C). The IR-FLASH imaging sequence was used in vivo and also in vitro in addition to a spectroscopic IR-sequence.

Finally, to ensure the validity of equation (3A), the blood volume was measured in NMR-tubes with blood sampled as described above, anticipating a value of 100%. In this case, the S_post and the S_pre signals are obtained from ROIs in the tubes containing blood with and without CA, respectively. The S₀ acquisition provides the corresponding S_0post and S_0pre signals, respectively.

Image Processing and Analysis

Image processing and analysis were carried out on Sun Sparc workstations (Sun Microsystems Inc., Mountain View, CA, USA) using home-made software written in IDL (Interactive Data Language). Images were Fourier interpolated on a 128 × 128 matrix. Regions of interests were manually delineated guided by the angiographic image, so as to avoid large vessels. Signal intensities within ROIs were averaged and plotted as a function of time. S_post was obtained as an average signal from at least 90 acquisitions (about 70 secs) performed during the steady-state regime after bolus injection. The S_pre and S₀ signals were obtained by averaging over an equal time interval. With the infusion-technique, all signals were averaged over 20 mins.

The program displays a CBVf histogram for the chosen ROI, computes its statistic parameters, and displays a CBVf map in a gray scale.

The CBVf maps were computed pixelwise using the following equation, where the signal S_pre is subtracted to exclude residual signal contribution:

C B V f (%) = \frac{100}{N} \sum_{i = 1}^{N} (\frac{S_{p o s t i} - S_{p r e i}}{S_{0 i}})

(3B)

where N is the number of averaged acquisitions, and the index i refers to the image acquisition. The regional CBVf is computed similarly, on the basis of averaged CBVf values within an ROI. Global CBVf measures were obtained from an ROI covering most of the brain structures visible on the slice. When the CBVf is given for a particular rat, the standard deviation (s.d.) refers to the signal fluctuations over time. All data averaged over n individual experiments (in vivo or in vitro) are given as mean±s.d. Cerebral blood volume fraction ratios are calculated for each individual rat, before averaging over n rats.

Statistical significance of CBVf differences between cerebral structures were tested using one-way analysis of variance for repeated measures, followed by a Bonferroni multiple comparison test.

The SNR was measured for an ROI covering the whole brain slice. S_post is calculated by averaging over 90 acquisitions obtained during the rapid steady-state regime:

S N R = \frac{(1 / N) \sum_{i = 1}^{N} S_{p o s t i}}{σ_{S_{p o s t}}}

(4)

where $σ_{S}$ is the square root of the temporal variance of the S_post signal in the same ROI.

Results

Results of the Preliminary Experiments

Figure 2A shows that a steady state of about half a minute (between the two arrows) is obtained with a dose of 0.15 mmol/kg Gd-DOTA, while a dose of 35 μmol/kg P760 already results in a constant signal intensity for more than half a minute. Other experiments show that a dose of 0.2 mmol/kg Gd-DOTA leads to a steady-state duration of more than one minute, so that a series of 90 acquisitions can be averaged to lead to a sufficient SNR. In experiments performed with 0.1 and 0.2 mmol/kg of P760 (Figure 2C), the same maximum intensity is reached, demonstrating that the intravascualar signal indeed reaches its equilibrium value and that it remains constant over a longer period with 0.2 mmol/kg, as expected. Subsequent CBVf experiments were performed with 0.1 mmol/kg of P760 because, with this dose, the intravascular signal kept its equilibrium value for almost 5 mins with convenient SNR for CBVf measurements, and because it is equivalent to the Gd³⁺ dose used with Gd-DOTA in clinics.

All T₁ measurements, in vivo for the cerebral parenchyma (1206±214 ms, n=7) and in vitro on blood sampled before (1450±34 ms, n=5) and during the first minute after administration of 0.1 mmol/kg of P760 (47±6 ms, n=5), confirm that at TR=750 ms the conditions of the RSST₁ method are fulfilled (Figure 1). Therefore, under these experimental conditions, the blood signal completely recovers after inversion, while the tissue signal is suppressed.

Normocapnic Cerebral Blood Volume Fraction Measurements

Figure 3A shows the anatomic details of the rat brain on an IR-FLASH image taken in the coronal orientation. Figure 3B shows the corresponding IR-FLASH image obtained 75 secs after a P760 injection of 0.1 mmol/kg with the sequence parameters for suppression of the signal from extravascular brain tissue. Only larger vessels are clearly visible, such as the superior sagittal sinus and straight sinus, and part of the pterygoid plexus, as well as the contrast in extracerebral tissue. The diagram in Figure 3C illustrates the averaged signal intensity in the depicted ROI before, during, and for about 5 mins after CA injection.

Figure 3

(A) Inversion-Recovery-Fast-Low-Angle-SHot image (TR=3.5 secs, T_inv=325 ms, 0.25 × 0.5 × 2 mm³ resolution) showing anatomic details in a coronal plane. The ROI comprises the parietal cortex, the corpus callosum, thalamus, and striatum. (B) Inversion-Recovery-Fast-Low-Angle-SHot image (TR=750 ms, T_inv=325 ms, 0.75 × 0.75 × 2 mm³ resolution) acquired 75 secs after bolus injection of 0.1 mmol/kg P760. (C) Plot illustrating the stable signal amplitude after P760 injection from the ROI delineated in (A). (D) Coronal CBVf map (0.75 × 0.75 × 2 mm³ resolution) of the slice in (A). The global CBVf for this rat is 3.28%±0.24% including the central vein. The s.d. results from averaging over almost 70 secs starting 30 secs after bolus injection of 0.1 mmol/kg P760. (E) Histogram showing the regional CBVf heterogeneity in such an ROI.

For the normocapnic rat brain under isoflurane anesthesia, the average global CBVf values were 3.29%±0.69% (n=15) using 0.1 mmol/kg P760, and 3.37%±0.41% (n=4) using 0.2 mmol/kg Gd-DOTA. The s.d. reflects intersubject variability, variations in slice position and noise. These results are similar and in good agreement with values reported in the literature for rats (Table 1).

Table 1

Literature review: regional CBV (healthy regions) in normocapnic, normothermic, anesthetized rats

ROI	CBV value	Unit	Technique	Reference
Whole brain parenchyma^a	2.51	ml/100 g	Autoradiography^b	Todd et al (1992)
Whole brain parenchyma^a	2.96±0.57	ml/100 g	Autoradiography^b	Todd et al (1993)
Whole brain parenchyma^c	2.77±0.24	ml/100 g	Autoradiography^b	Todd and Weeks (1996)
Whole brain parenchyma^d	2.40±0.34	%	3D SS T₁ MRI	Lin et al (1997)
Whole brain parenchyma^d	2.96±0.82	%	3D SS T₁ MRI	Lin et al (1999)
Whole brain parenchyma^c	3.14±0.32	%	SS susceptibility contrast MRI	Dunn et al (2004)
Cortex^e	4.3±0.7	%	SS susceptibility contrast MRI	Tropres et al (2004)
Striatum^f	2.2±0.6	ml/100 g	SS susceptibility contrast MRI	Julien et al (2004)
Cortex^a	3.01±0.43	%	SS susceptibility contrast MRI	Payen et al (2000)
Striatum^a	2.94±0 .49
Cortex^a	4.07	%	SS susceptibility contrast MRI	Tropres et al (2001)
Striatum^a	2.87
Whole brain parenchyma	1.90±0.39	%	Histological^g	Pathak et al (2001)
Whole brain parenchyma	1.92±0.32	ml/100 g	Synchrotron radiation quantitative computed tomography	Adam et al (2003)
Without blood structures^h
Whole brain parenchyma	4.18±1.06
With blood structures^h
Cortex^h	2.27
Caudate+putamen^h	2.01

Whole brain parenchyma	3.29±0.69	%	RSST₁-method	This study
Cortex	3.13±0.91
Striatum	2.92±0.68

Rat anesthetized with halothane.

¹⁴C-dextran labeled plasma and ^99mTc-labeled red blood cells.

Rat anesthetized with isoflurane.

Rat anesthetized with intraperitoneal pentobarbital.

Contralateral to C6 glioma, under moderate hypoxia, rats anesthetized with halothane.

Contralateral striatum to C6 glioma, rats anesthetized with intraperitoneal thiopental.

With stereo correction for slice thickness; SS, steady state method.

Anesthetized by intraperitoneal infusion of chloral hydrate.

Figure 3D displays a representative CBVf map (same slice as in Figures 3A and 3B). Extracerebral vessels and tissue are masked. For this rat, the global CBVf on this slice yields 3.28%±0.24% including a macroscopic intracerebral vessel. Figure 3E illustrates the spatial distribution of the CBVf values within the slice. A step size of 0.5% was chosen allowing histograms of a given ROI from different rats to be comparable.

Global CBVf values were obtained with a mean SNR of 40±18 calculated according to equation (4) and averaged over 15 rats. As the SNR is defined, it also reflects the temporal stability of the signal S_post. This SNR allows consistent regional CBVf measurements for different structures of the rat brain, as shown in Figure 4A (n=6). For instance, the blood volume fraction of 2.92%±0.68% in the striatum, which is a mixture of gray and white matter as it is traversed by fibers from the cortex, is lower than the blood volume fraction of pure gray matter such as the cortex (3.13%±0.91%) and higher than the one of white matter (2.00%±0.50%) in accordance with (Adam et al, 2003). The ratios of regional CBVf between cortex and striatum CBVf_cortex/CBVf_striatum of 1.11±0.18 as well as between gray matter and white matter CBVf_GM/CBVf_WM of 1.59±0.51 obtained in healthy normocapnic rats (n=6) are also in the range reported in the literature (Table 2). The difference between the CBVf in the white matter and in the striatum is significant (P<0.05). The difference between the CBVf in the white matter and the one in the cortex, in the subcortical gray matter as well as of the whole slice is highly significant (P<0.01). Although these measures experience partial volume effect, regional CBVf values appear consistent.

Figure 4

(A) Average regional CBVf (n=6) of different cerebral tissues of the rat brain. subc. GM, subcortical gray matter (comprising striatum and thalamus). (B) Representative CBVf histogram for the left parietal cortex, average cortical CBVf for this rat: 3.12%±0.88%. (C) Representative CBVf histogram for the left striatum, average striatal CBVf for this rat: 2.77%±1.51%. (D) Inversion-Recovery-Fast-Low-Angle-SHot image showing the positions of ROIs in the left parietal cortex and in the left striatum.

Table 2

Literature review: regional CBV ratios in normocapnic, normothermic rats

ROI	CBV ratio	Technique	Reference
Cortex/striatum^a	1.43	SS susceptibility contrast MRI	Payen et al (1998)
Cortex/striatum^a	1.02	SS susceptibility contrast MRI	Payen et al (2000)
Cortex/striatum^a	1.42	SS susceptibility contrast MRI	Tropres et al (2001)
Cortex/corpus callosum	1.52±0.42	Histological^b	Pathak et al (2001)
Cortex/corpus callosum^c	1.47	Histological^b	Pathak et al (2003)
Cortex/striatum^d	1.13	Synchrotron radiation quantitative computed tomography	Adam et al (2003)
Cortex/striatum	1.11±0.18	RSST₁ method	This study
GM/WM	1.59±0.51

Rats anesthetized with halothane.

With stereo correction for slice thickness; SS, steady state method; GM, gray matter; WM, white matter.

Contralateral to 9L gliosarcoma.

Anesthetized by intraperitoneal infusion of chloral hydrate.

Figures 4B and 4C illustrate the CBVf histograms of two ROIs centered on the left parietal cortex and the left striatum. While the cortical CBVf is typically approximately 3% to 4%, a large portion of the striatum has a CBVf less than 2%. This may reflect the heterogeneous structure of the striatal tissue, but the partial volume effect probably also contributes to this heterogeneity. Figure 4D depicts the shape and the position of the ROIs within the brain slice. For this rat, the average cortical and striatal CBVf were 3.12%±0.88% and 2.77%±1.51%, respectively. An ROI placed in a large vessel never yielded 100% blood volume, but values around 70%.

A steady-state regime was also obtained under continuous CA infusion and maintained up to 1 hour. Under normocapnic conditions, the CBVf fluctuations did not exceed 9% of the mean CBVf throughout the infusion duration, which could be interpreted as physiologic variabilities.

Cerebral Blood Volume Fraction-Measurement Under Hypercapnia

Figure 5A shows global CBVf changes as a function of the PaCO₂. The CBVf changes are given relative to the normocapnic CBVf, calculated for each rat from the experimental data for a PaCO₂ value of 40 mm Hg, assuming a linear relationship between the global CBVf and the PaCO₂. The slopes obtained with P760 (n=5) and Gd-DOTA bolus injections (n=3) are identical, and equal to about 1% mm Hg⁻¹. Regional analysis reveals that CBVf increases under hypercapnia are steepest for the cortex and surprisingly more pronounced on the left side (1.3% per mm Hg, n=8, Gd-DOTA and P760 experiments combined). The CBVf increase under hypercapnia is in the range of those reported in the literature (Payen et al, 1998; Shockley and LaManna, 1988). A steeper increase in the left neocortex in rats has also been observed (Payen et al, 1998). This study confirms the sensitivity of the method to physiologic CBVf variations.

Figure 5

(A) Global CBVf in % of the global normocapnic CBVf as a function of PaCO₂. The linear regression for experiments with Gd-DOTA (n=3, dashed line) and P760 (n=5, solid line) yield practically identical slopes (Gd-DOTA: 0.99%/mm Hg, R²=0.67; P760: 0.98%/mm Hg, R²=0.85). (B) CBVf measures (example for one rat performed with Gd-DOTA bolus injections) showing the reversibility of the CBVf increase after recovery from hypercapnia to an almost normal PaCO₂.

Figure 5B shows the result of an experiment in which 30 mins recovery has been allowed after three increasing levels of hypercapnia, showing the reversibility of the CBVf change. While the PaCO₂ measured during recovery is still higher than the normocapnic one, the CBVf decreases to a lower value than under normocapnia. A similar undershoot of CBVf has been observed in rats by Payen et al (1998) predominantly for the striatum during recovery after inhalation of CO₂, and in cats by Hamberg et al (1996) after transient ischemia.

Discussion

In this study it was shown how the CBVf can be obtained in a direct way, using rapid MRI in combination with the T₁-shortening effect of intravascular paramagnetic CAs. The experimental conditions (sequence parameters and CA dose) for obtaining a rapid steady-state regime with the proposed method were studied, resulting in the selective measurement of the intravascular signal. The method leads directly to the blood volume, and not to the plasma volume, because of the rapid water exchange between erythrocytes and plasma (Herbst and Goldstein, 1989). This is confirmed by the measurements on blood samples (n=7), in which the S_post/S_0post ratio equals 0.997±0.090 (i.e., 100% blood volume) anywhere in the tube, independent of the local hematocrit value.

In spite of the short TE in this study, the acquired signal might be attenuated by transverse relaxation effects, which have been evaluated.

The way it has been measured, the ratio S_0post/S_0pre corresponds to the factor exp(−(ΔR₂+ΔR₂∗)·TE). ΔR₂ is due to the transverse relaxivity of the CA and the ΔR₂∗ effect is due to the dephasing of the extravascular spins in the gradient field created by the susceptibility difference between the extra- and intravascular compartments. This latter effect does not need to be taken into account in this method since the extravascular magnetization is suppressed.

In vivo, a CBVf of 100% was never attained in a large vessel, which might be because of partial volume contribution from adjacent tissue and the R₂ attenuation of the S_post signal induced by the CA (transverse relaxivity of P760 r₂=33.9±2.3 (mmol/L)⁻¹ sec⁻¹ (Fonchy et al, 2001)). The ratio S_0post/S_0pre measured on blood samples was equal to 0.90±0.17 (n=4). For a P760 concentration of 1.5 mmol/L in blood, the factor exp(−ΔR₂·TE) is 0.86, almost equal to the ratio found experimentally. In brain parenchyma, the transverse relaxation effect appears negligible, since in this case the ratio S_0post/S_0pre equals 1.02±0.03 (n=2). Moreover, the signal amplitude during the early part of the steady state (when the R₂∗ effect is most pronounced) was equal for 0.1 and 0.2 mmol/kg P760, as can be seen in Figure 2C, also demonstrating a negligible transverse relaxation effect in the microvasculature.

The different results obtained from the microvasculature and in the blood samples, simulating large vessels, might be explained by various mechanisms. In the microvasculature, the R₂ effect is probably masked by the small volume fraction or by transverse relaxation mechanisms in blood (Gillis et al, 1995). In blood, transverse relaxation is predominantly due to the diffusion of water through the field gradient arising from the susceptibility difference Δχ_ery/plasma between erythrocytes and plasma. The Δχ_ery/plasma depends on the oxygenation, on the hematocrit and on the susceptibility of the plasma modified by the CA injection. In the blood samples or in large veins, the transverse relaxation effect is probably more pronounced, because the blood oxygenation is lower and the hematocrit is higher (Cremer and Seville, 1983) than in microvasculature. To avoid CBVf underestimation because of the transverse relaxivity of CA, S₀ has to be measured after injection, or the signal can be acquired in a spiral or projection—reconstruction mode with an echo time of 10 to 100 μs.

Another aspect affecting the accuracy of the CBVf measurement is the water exchange across the BBB. Whether it can be considered negligible was evaluated here. The exchange regime (slow or fast) depends on the difference between the longitudinal relaxation rates of the intravascular T_1(iv) ⁻¹ and extravascular T_1(ev) ⁻¹compartment, and on the water exchange rate τ_exch ⁻¹.

Assuming an approximate microvascular blood volume of CBVf=3%, and an intracapillary residence time τ_iv of 500 to 650 ms (Labadie et al, 1994), the extravascular residence time τ_ev=(V_ev/V_iv)τ_iv is in the order of 16 to 21 secs, respectively. The water exchange rate (τ_exch⁻¹=τ_iv⁻¹+τ_ev⁻¹) is thus in the range of 1.6 to 2.1 secs⁻¹. The difference of the longitudinal relaxation rates between the two compartments is in the order of 20 secs⁻¹ (calculated from the experimental data), fulfilling the condition for the slow exchange regime τ_exch⁻¹≪T_1(iv)⁻¹−T_1(ev)⁻¹ and therefore supporting the two-compartment model without exchange. However, the exchange effect on the CBVf measure can be evaluated using the model described in Moran and Prato (2004), defining an exchange regime as K=(1+τ_exch/m)⁻¹, where m is the characteristic time of the two relaxing compartments m=((1/T_1(iv))−(1/T_1(ev)))⁻¹. The impact of the exchange regime (K=0 no exchange, K=1 fast exchange) modifies the intrinsic T₁ of either compartment resulting in apparent relaxation times (T₁^app). These are calculated according to

\frac{1}{T_{1 (a)}^{a p p}} = \frac{1}{T_{1 (a)}} + K P_{(b)} (\frac{1}{T_{1 (b)}} - \frac{1}{T_{1 (a)}})

where a and b denote the compartments and P_(a) or P_(b) are the corresponding compartment volume fractions.

Using the magnetization values supplied by equation (1) for T₁^app and T_inv=325 ms, equation (3B) yields an apparent CBVf of 3.27%, which corresponds to an overestimation of less than 10%. For an exchange time of τ_exch=500 ms the overestimation is less than 12%.

For the vascular space occupying method (Lu et al, 2005), a rough estimation of the water exchange impact using the reported intrinsic and extrinsic MRI parameters for 1.5 T indicates that for τ_exch=630 ms (K=0.2841 after CA administration) the CBVf is overestimated by almost 75% (5.2% instead of 3% for T_1(ev)=1000 secs).

Since the permeability of the BBB to water increases for several brain pathologies, it is important to minimize the sensitivity of the RSST₁ method to transepithelial water exchange. The water exchange effect can be reduced by shortening T_inv, allowing the system less time to exchange water across the vascular boundary (Larsson et al, 2001). For example, with the couple of parameters TR=500 ms and T_inv=225 ms, the overestimation would be reduced to less than 4% for T_1(iv)=T_inv/5 and τ_exch=630 ms.

The RSST₁ method, as it is conceived, is nonsubtractive. Experimentally, for an ROI comprising the whole brain, a residual signal S_pre of 2% to 3% of the total S₀ was detected before injection, higher than the theoretical value for an average tissue T₁ of 1200 ms (0.6% of S₀). This residual signal was eliminated by subtraction of the precontrast acquisition from the postcontrast acquisition. The level of the residual signal was systematically higher by a factor of 1.5±0.2 (n=6) for the white matter of the corpus callosum than for the cortical gray matter, suggesting that S_pre is predominantly a result of low white matter T₁. Another contribution comes from the fact that the lines of k space are not acquired at exactly the same inversion time. Studies performed under CA infusion (data not shown) demonstrated that with TR=500 ms and T_inv=225 ms, the suppression of the extravascular signal is more efficient (2.4%±0.4% instead of 2.9%±0.4% of S₀, n=3), with a theoretical value of 0.07% of S₀ for T₁=1200 ms.

Comparison of regional CBVf obtained with the RSST₁ method and a ‘Gold Standart’ technique on the same animal is the best way to validate the method. Unfortunately, at the laboratory, the only alternative technique available for CBV quantification requires the injection of USPIO (ultrasmall superparamagnetic iron oxide). The absolute CBV could only be comparable between the two measures if the first injected CA were completely eliminated. During the washout time (at least 2 h after injection of P760), it was observed that it is difficult to maintain constant the physiologic parameters of the rat, which is a sine qua non-condition for CBVf comparison.

The experiments reported in Tropres et al (2004), Julien et al (2004) and Payen et al (2000) were performed on Wistar and Sprague—Dawley rats under the same conditions using the same equipment and the regional CBVf values obtained were in good agreement (Table 1).

The experimental CBVf values reported in Table 1 are obtained with different techniques and depend on the origin of the measured signals and on the available SNR. Although comparison is not easy, all CBVf values reported from literature are in the same range, and even regional measurements can be compared with the ones obtained with the RSST₁ method, which has the advantage of leading directly to absolute CBV values.

This method is insensitive to inflow—outflow effects since a spatially nonselective inversion pulse is used. The maps were obtained with a satisfactory SNR required for measuring the small blood signal with a spatial resolution of 0.75 × 0.75 × 2 mm³. In the human brain, this spatial resolution is largely sufficient to distinguish brain structures without partial volume effect, but in the rat brain, this effect is certainly not negligible.

Since the technique is independent of the type and speed of the CA injection, the feasibility of the CBVf measurement under continuous intravenous infusion was demonstrated. The conservation of a steady-state regime over the duration of the infusion offers the possibility to acquire images with increased in-plane resolution or in 3D mode, as well as to study CBVf variations induced by drugs, cerebral activation, or modified physiologic parameters (e.g., PaCO₂). This technique could also be interesting for the evaluation of BBB disruption. In this case, CA extravasation from the brain capillaries should be observable as an irreversible increase of the steady-state signal amplitude, similar to the findings of Seo et al (2002). However, it requires large amounts of CA. Accumulation of the CA could take place in the body after prolonged infusion, associated with an enhanced R₂∗ effect. However, CBVf measures up to 60 mins after the initial bolus injection did not show large variations.

To apply the RSST₁ method in the presence of brain pathologies, and in particular brain tumors, Gd-DOTA cannot be used, since this method requires CA that do not leak across an impaired BBB during the measuring time.

Finally, the CBVf response to changes in PaCO₂ demonstrates the sensitivity of the RSST₁ method. The CBVf reversibility under recovery from hypercapnia shows that the CBVf changes are neither due to prolonged isoflurane anesthesia nor to repetitive CA injection.

Brain activation results in CBV variations of similar amplitude as under hypercapnia (Ito et al, 2003). The RSST₁ method could therefore be a promising tool for functional MRI particularly because absolute CBV values are directly obtained. Furthermore, its knowledge is useful to interpret the complex BOLD signal.

Conclusion

In this article, we show the feasibility of the RSST₁ method, a new MRI method for CBV quantification. This method is rapid, simple, and the measure leads directly to the blood volume. The RSST₁ method has been validated on the rat brain and its sensitivity to CBV increases induced by hypercapnia has been shown. Compared with other quantitative MRI techniques, neither additional relaxation time measurements nor knowledge regarding the arterial input function are required. It is also insensitive to inflow—outflow effects of circulating blood protons. The transverse relaxation effects were evaluated and were found negligible in the microvasculature. The water exchange effect was quantified, indicating that the mean CBVf of 3.3% obtained experimentally corresponds to a CBVf of 3.0% after correction.

P760, an experimental CA, was used for its high longitudinal r₁ relaxivity to obtain full relaxation of the intravascular magnetization when using a short TR needed for tissue signal suppression. However, with 0.2 mmol/kg Gd-DOTA (twice the usual clinical dose), which is well tolerated, similar CBVf measures were obtained. Therefore, the method could be extended to human studies. In this case, because of the better spatial resolution, the regional CBVf measures would be less affected by the partial volume effect.

To apply this method to cerebral pathologies, high relaxivity CAs, which remain confined to the intravascular space even in presence of pathologies of the BBB are needed. A new CA composed of modified cyclodextrins and complexed to Gd³⁺ is under evaluation (Lahrech et al, 2005, 2006).

Footnotes

Acknowledgements

The authors gratefully acknowledge cooperation with Guerbet Laboratories (Aulnay-sous-Bois, France). They also acknowledge their debt to Christoph Segebarth, Boudewijn van der Sanden, and Jean-François Payen for constructive discussions to Regine Farion for the technical support and Association pour la Recherche sur le cancer (ARC) for financial PhD support.

Appendix A

In Figure 6A, the ratios M_Z/M₀ of the intra- and extravascular compartments are compared in the absence (K=0) and in the presence of exchange (K=0.0765, reflecting experimental conditions). The plots are derived from equation (1) as a function of T_inv with TR=750 ms, and using T₁^app instead of T₁ in the presence of exchange. The different T₁^app are calculated according to the exchange model above with T₁ comparable to those obtained experimentally for each compartment before and after CA injection. The K-value used results from τ_exch=630 ms. The simulations were performed for an intravascular volume fraction of 3%. Figure 6A demonstrates that the intravascular signal is not affected by the water exchange, while the extravascular signal is less suppressed when exchange takes place.

To show directly to which extent the CBVf is overestimated for K=0.0765, the sum of the intra- and extravascular magnetization is plotted in Figure 6B. Before injection, the exchange regime is characterized by K=0.9170, but the residual longitudinal magnetization is practically identical with or without exchange.

References

Adam

Elleaume

Le Duc

Corde

Charvet

Tropres

Le Bas

Esteve

(2003) Absolute cerebral blood volume and blood flow measurements based on synchrotron radiation quantitative computed tomography. J Cereb Blood Flow Metab 23:499–512

Lin

(2002) Cerebral venous and arterial blood volumes can be estimated separately in humans using magnetic resonance imaging. Magn Reson Med 48: 583–588

Aronen

Gazit

Louis

Buchbinder

Pardo

Weisskoff

Harsh

Cosgrove

Halpern

Hochberg

(1994) Cerebral blood volume maps of gliomas: comparison with tumor grade and histologic findings. Radiology 191:41–51

Belliveau

Rosen

Kantor

Rzedzian

Kennedy

McKinstry

Vevea

Cohen

Pykett

Brady

(1990) Functional cerebral imaging by susceptibility-contrast NMR. Magn Reson Med 14:538–46

Boxerman

Hamberg

Rosen

Weisskoff

(1995) MR contrast due tointravascular magnetic susceptibility perturbations. Magn Reson Med 34: 555–566

Calamante

Morup

Hansen

(2004) Defining a local arterial input function for perfusion MRI using independent component analysis. Magn Reson Med 52: 789–797

Corot

Violas

Robert

Port

(2002) Arterial concentration profiles of two blood pool agents and Gd-DOTA after intravenous injection in rabbits. Acad Radiol 9(Suppl 1):S137–9

Cremer

Seville

(1983) Regional brain blood flow, blood volume, and haematocrit values in the adult rat. J Cereb Blood Flow Metab 3:254–6

Dean

Lee

Kirsch

Runge

Dempsey

Pettigrew

(1992) Cerebral hemodynamics and cerebral blood volume: MR assessment using gadolinium contrast agents and T1-weighted Turbo-FLASH imaging. AJNR Am J Neuroradiol 13:39–48

10.

Dunn

Roche

Springett

Abajian

Merlis

Daghlian

Makki

(2004) Monitoring angiogenesis in brain using steady-state quantification of DeltaR2 with MION infusion. Magn Reson Med 51:55–61

11.

Fonchy

Lahrech

Francois-Joubert

Dupeyre

Benderbous

Corot

Farion

Rubin

Decorps

Remy

(2001) A new gadolinium-based contrast agent for magnetic resonance imaging of brain tumors: kinetic study on a C6 rat glioma model. J Magn Reson Imaging 14:97–105

12.

Gillis

Peto

Moiny

Mispelter

Cuenod

(1995) Proton transverse nuclear magnetic relaxation in oxidized blood: a numerical approach. Magn Reson Med 33:93–100

13.

Haase

(1990) Snapshot FLASH MRI. Applications to T1, T2, and chemical-shift imaging. Magn Reson Med 13:77–89

14.

Hacklander

Hofer

Reichenbach

Rascher

Furst

Modder

(1996) Cerebral blood volume maps with dynamic contrast-enhanced T1-weighted FLASH imaging: normalvalues and preliminary clinical results. J Comput Assist Tomogr 20:532–9

15.

Hacklander

Reichenbach

Modder

(1997) Comparison of cerebral blood volume measurements using the T1 and T2* methods in normal human brains and brain tumors. J Comput Assist Tomogr 21:857–66

16.

Hamberg

Boccalini

Stranjalis

Hunter

Huang

Halpern

Weisskoff

Moskowitz

Rosen

(1996) Continuous assessment of relative cerebral blood volume in transient ischemia using steady state susceptibility-contrast MRI. Magn Reson Med 35: 168–173

17.

Herbst

Goldstein

(1989) A review of water diffusion measurement by NMR in human red blood cells. Am J Physiol 256:C1097–104

18.

Ito

Kanno

Ibaraki

Hatazawa

Miura

(2003) Changes in human cerebral blood flow and cerebral blood volume during hypercapnia and hypocapnia measured by positron emission tomography. J Cereb Blood Flow Metab 23:665–70

19.

Julien

Payen

Tropres

Farion

Grillon

Montigon

Remy

(2004) Assessment of vascular reactivity in rat brain glioma by measuring regional blood volume during graded hypoxic hypoxia. Br J Cancer 91: 374–380

20.

Kuppusamy

Lin

Cizek

Haacke

(1996) In vivo regional cerebral blood volume: quantitative assessment with 3D T1-weighted pre- and postcontrast MR imaging. Radiology 201:106–12

21.

Labadie

Lee

Vetek

Springer

Jr (1994) Relaxographic imaging. J Magn Reson B 105:99–112

22.

Lahrech

Perles-Barbacaru

Gadelle

Aous

Farion

Le Bas

Debouzy

Fries

(2005) Potential of a new contrast agent: gadolinium per (3,6) anhydro alpha cyclodextrin for resonance magnetic neuroimaging. Basle (Switzerland): European Society for Magnetic Resonance in Medicine and Biology

23.

Lahrech

Perles-Barbacaru

Gadelle

Aous

Farion

Le Bas

Debouzy

Fries

(2006) Characterization and potential of Gd-ACX, a new contrast agent for Magnetic Resonance Neuroimaging. Seattle (USA): International Society for Magnetic Resonance in Medicine

24.

Larsson

Rosenbaum

Fritz-Hansen

(2001) Quantification of the effect of water exchange in dynamic contrast MRI perfusion measurements in the brain and heart. Magn Reson Med 46:272–81

25.

Lee

Blaufox

(1985) Blood volume in the rat. J Nucl Med 26:72–6

26.

Lin

Celik

Paczynski

Hsu

Powers

(1999) Quantitative magnetic resonance imaging in experimental hypercapnia: improvement in the relation between changes in brain R2 and the oxygen saturation of venous blood after correction for changes in cerebral blood volume. J Cereb Blood Flow Metab 19:853–62

27.

Lin

Paczynski

Kuppusamy

Hsu

Haacke

(1997) Quantitative measurements of regional cerebral blood volume using MRI in rats: effects of arterial carbon dioxide tension and mannitol. Magn Reson Med 38:420–8

28.

Golay

Pekar

Van Zijl

(2003) Functional magnetic resonance imaging based on changes in vascular space occupancy. Magn Reson Med 50:263–74

29.

Law

Johnson

van Zijl

Helpern

(2005) Novel approach to the measurement of absolute cerebral blood volume using vascular-space-occupancy magnetic resonance imaging. Magn Reson Med 54:1403–11

30.

Meier

Zierler

(1954) On the theoryof the indicatordilution method for measurement of blood flow and volume. J Appl Physiol 6:731–44

31.

Moonen

Liu

van Gelderen

Sobering

(1992) A fast gradient-recalled MRI technique with increased sensitivity to dynamic susceptibility effects. Magn Reson Med 26:184–9

32.

Moran

Prato

(2004) Modeling (1H) exchange: an estimate of the error introduced in MRI by assuming the fast exchange limit in bolus tracking. Magn Reson Med 51:816–27

33.

Ogawa

Menon

Tank

Kim

Merkle

Ellermann

Ugurbil

(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

34.

Pathak

Rand

Schmainda

(2003) The effect of brain tumor angiogenesis on the in vivo relationship between the gradient-echo relaxation rate change (DeltaR2*) and contrast agent (MION) dose. J Magn Reson Imaging 18:397–403

35.

Pathak

Schmainda

Ward

Linderman

Rebro

Greene (2001) MR-derived cerebral blood volume maps: issues regarding histological validation and assessment of tumor angiogenesis. Magn Reson Med 46:735–47

36.

Payen

Briot

Tropres

Julien-Dolbec

Montigon

Decorps

(2000) Regional cerebral blood volume response to hypocapnia using susceptibility contrast MRI. NMR Biomed 13:384–91

37.

Payen

Vath

Koenigsberg

Bourlier

Decorps

(1998) Regional cerebral plasma volume response to carbon dioxide using magnetic resonance imaging. Anesthesiology 88:984–92

38.

Port

Meyer

Bonnemain

Corot

Schaefer

Rousseaux

Simonot

Bourrinet

Benderbous

Dencausse

Devoldere

(1999) P760 and P775: MRI contrast agents characterized by new pharmacokinetic properties. Magma 8:172–6

39.

Rosen

Belliveau

Aronen

Kennedy

Buchbinder

Fischman

Gruber

Glas

Weisskoff

Cohen

, (1991) Susceptibility contrast imaging of cerebral blood volume: human experience. Magn Reson Med 22:293–9; discussion 300–293

40.

Schaefer

Hunter

Hamberg

Sorensen

Schwamm

Koroshetz

Gonzalez

(2002) Predicting cerebral ischemic infarct volume with diffusion and perfusion MR imaging. AJNR Am J Neuroradiol 23:1785–94

41.

Schwarzbauer

Morrissey

Deichmann

Hillenbrand

Syha

Adolf

Noth

Haase

(1997) Quantitative magnetic resonance imaging of capillary water permeability and regional blood volume with an intravascular MR contrast agent. Magn Reson Med 37:769–77

42.

Schwarzbauer

Syha

Haase

(1993) Quantification of regional blood volumes by rapid T1 mapping. Magn Reson Med 29:709–12

43.

Seo

Takamata

Ogino

Morita

Nakamura

Murakami

(2002) Water permeability of capillaries in the subfornical organ of rats determined by Gd-DTPA(2-) enhanced 1H magnetic resonance imaging. J Physiol 545:217–28

44.

Shockley

LaManna

(1988) Determination of rat cerebral cortical blood volume changes by capillary mean transit time analysis during hypoxia, hypercapnia and hyperventilation. Brain Res 454:170–8

45.

Todd

Weeks

(1996) Comparative effects of propofol, pentobarbital, and isoflurane on cerebral blood flow and blood volume. J Neurosurg Anesthesiol 8:296–303

46.

Todd

Weeks

Warner

(1992) Cerebral blood flow, blood volume, and brain tissue hematocrit during isovolemic hemodilution with hetastarch in rats. Am J Physiol 263:H75–82

47.

Todd

Weeks

Warner

(1993) The influence of intravascular volume expansion on cerebral blood flow and blood volume in normalrats. Anesthesiology 78:945–53

48.

Tropres

Grimault

Vaeth

Grillon

Julien

Payen

Lamalle

Decorps

(2001) Vessel size imaging. Magn Reson Med 45:397–408

49.

Tropres

Lamalle

Peoc'h

Farion

Usson

Decorps

Remy

(2004) In vivo assessment of tumoral angiogenesis. Magn Reson Med 51:533–41

50.

Villringer

Rosen

Belliveau

Ackerman

Lauffer

Buxton

Chao

Wedeen

Brady

(1988) Dynamic imaging with lanthanide chelates in normalbrain: contrast due tomagnetic susceptibility effects. Magn Reson Med 6:164–74