July 2000, VOLUME 175
NUMBER 1

Recommend & Share

July 2000, Volume 175, Number 1

Review

MR Perfusion Imaging of the Brain
Techniques and Applications

+ Affiliation:
1Both authors: Department of Radiology, Duke University Medical Center, Box 3808, Durham, NC 27710.

Citation: American Journal of Roentgenology. 2000;175: 207-219. 10.2214/ajr.175.1.1750207

Next section

MR imaging has become a powerful clinical tool for evaluation of brain anatomy. Its application has recently expanded into evaluation of brain function via assessment of a number of functional or metabolic parameters. One such parameter is cerebral perfusion, which describes passage of blood through the brain's vascular network. MR perfusion imaging refers to several recently developed techniques used to non-invasively measure cerebral perfusion via assessment of various hemodynamic measurements such as cerebral blood volume, cerebral blood flow, and mean transit time. These techniques have great potential in becoming important clinical tools in the diagnosis and treatment of patients with cerebrovascular disease and other brain disorders. Potential applications include the evaluation of tissue at risk after acute stroke, noninvasive histologic assessment of tumors, evaluation of neurodegenerative conditions such as Alzheimer's disease, as well as assessment of the effects of drugs used to treat these conditions. The purpose of this article is to provide an understanding of basic techniques and applications of MR perfusion imaging and highlight recent developments in this emerging technology.

Techniques
Previous sectionNext section

Measurement of tissue perfusion depends on the ability to serially measure concentration of a tracer agent in a target organ of interest. Exogenous tracers such as iced saline solution, iodinated radiographic contrast material, and radionuclides have been used [1, 2]. More recently, with the advent of MR imaging, exogenous tracer agents, such as paramagnetic contrast material, and endogenous tracer agents, such as magnetically labeled blood, have been used [3].

To obtain hemodynamic parameters from serial tissue tracer concentration measurements, a general model of the method by which that tracer passes through or distributes in the target organ is required [4]. Such a model must be based on an understanding of the manner in which the tracer is infused—that is, bolus injection versus constant infusion, and on assumptions about the pharmacokinetic properties of the tracer in the organ of interest. These assumptions include diffusibility from the intravascular to extravascular space, volume of distribution, and equilibrium half-life of the tracer.

Exogenous Tracer Methods

Exogenous tracer methods in MR perfusion imaging use a model that assumes the tracer is restricted to the intravascular compartment and does not diffuse into the extracellular space. Imaging can be performed either dynamically (rapid imaging over time during a bolus injection) or in the steady state (imaging after a constant infusion has reached an equilibrium concentration in the blood).

Dynamic imaging.—Dynamic imaging takes advantage of transient changes in the local magnetic field of the surrounding tissue induced by a bolus of paramagnetic tracer passing through the organ capillary network (Fig. 1A,1B,1C,1D,1E). These changes in the local magnetic field can be measured as signal changes on MR imaging. Ultrafast imaging techniques, such as echoplanar and spiral MR imaging [5, 6], enable the accurate measurement of rapidly varying signal changes that are due to the first pass of the bolus with adequate temporal resolution (<2 sec for coverage of the entire brain). Signal-time course data can then be converted to relative tracer tissue concentration-time course data [3]. Tracer concentration-time curves can then be analyzed to determine various tissue hemodynamic parameters, such as tissue blood volume, blood flow, transit time, and bolus arrival time (Fig. 2). In this article, the terms “cerebral blood volume,” “cerebral blood flow,” and “mean transit time” are defined as follows: Cerebral blood volume refers to the volume of blood in a given region of brain tissue, commonly measured in milliliters per 100 g of brain tissue. Cerebral blood flow refers to the volume of blood per unit time passing through a given region of brain tissue, commonly measured in milliliter per minute per 100 g of brain tissue. Mean transit time refers to the average time it takes blood to pass through a given region of brain tissue, commonly measured in seconds. Bolus arrival time refers to the time it takes for an IV-injected bolus of contrast material to arrive at a given region of the brain, also commonly measured in seconds.

figure
View larger version (130K)
Fig. 1A. —32-year-old healthy man. Echoplanar images of same slice show bolus of paramagnetic contrast material passing through tissues. Imaging was performed at rate of one image per 1.5 sec. Every fourth image is shown. Echoplanar image obtained at 9 sec.

figure
View larger version (126K)
Fig. 1B. —32-year-old healthy man. Echoplanar images of same slice show bolus of paramagnetic contrast material passing through tissues. Imaging was performed at rate of one image per 1.5 sec. Every fourth image is shown. Echoplanar image obtained at 15 sec.

figure
View larger version (144K)
Fig. 1C. —32-year-old healthy man. Echoplanar images of same slice show bolus of paramagnetic contrast material passing through tissues. Imaging was performed at rate of one image per 1.5 sec. Every fourth image is shown. Echoplanar image obtained at 21 sec. Note transient drop in signal intensity in and around major blood vessels (arrows) that is due to susceptibility effects from bolus.

figure
View larger version (130K)
Fig. 1D. —32-year-old healthy man. Echoplanar images of same slice show bolus of paramagnetic contrast material passing through tissues. Imaging was performed at rate of one image per 1.5 sec. Every fourth image is shown. Echoplanar image obtained at 27 sec.

figure
View larger version (137K)
Fig. 1E. —32-year-old healthy man. Echoplanar images of same slice show bolus of paramagnetic contrast material passing through tissues. Imaging was performed at rate of one image per 1.5 sec. Every fourth image is shown. Echoplanar image obtained at 33 sec.

figure
View larger version (43K)
Fig. 2. —Diagram explaining calculation of relative cerebral blood volume, cerebral blood flow, and mean transit time using dynamic contrast-enhanced T2-weighted technique. Signal-time course data for each voxel is converted to tracer tissue concentration-time course data using well-characterized relationship between T2* signal intensity and tracer tissue concentration [3]. Maps of relative cerebral blood volume are obtained by determining area below tracer concentration-time curve. Maps of relative cerebral blood flow are obtained by determining height of ideal tissue concentration-time curve, or tissue response function. Maps of mean transit time are obtained by dividing area under tissue response function by its height. To obtain tissue response function, arterial concentration-time curve, or arterial input function, must be deconvolved from measured tissue concentration-time curve. This arterial input function may be derived directly from imaging data. EPI = echoplanar imaging.

These parameters are dependent on the specific features of the bolus injection, including the amount of contrast material injected, the injection rate, and the paramagnetic properties of the contrast agent. In addition, hemodynamic parameters depend on variables within the subject being imaged, such as total-body vascular volume and cardiac output. As a result, hemodynamic parameters cannot be directly compared between different subjects and may even differ between examinations on the same subject at different times. Nonetheless, semiquantitative or relative values can be obtained using an internal standard of reference such as normal-appearing gray or white matter. This allows intra- and intersubject comparisons. Semiquantitation is valuable for subcategorizing disease processes, following disease course, and monitoring the effects of therapeutic interventions, provided such alterations are local and do not change the hemodynamic parameters of the internal reference. For example, in the case of stroke, relative hemodynamic measurements can be useful in monitoring the periinfarct ischemic tissue zone at risk for infarction as well as in monitoring the effects of thrombolytic therapy [7, 8]. For diffuse disease processes, in which the internal reference may also be affected, absolute quantitation is necessary. Although absolute quantitation of cerebral blood volume and cerebral blood flow has been attempted using dynamic MR methods that measure arterial input to the brain [9,10,11,12,13], the accuracy of these methods remains unproven [14]. With further improvements in signal-to-noise capability and improved techniques to determine true arterial input, quantitation of absolute cerebral blood volume and cerebral blood flow with dynamic MR methods may well be feasible.

Determination of relative cerebral blood volume from tracer concentration-time data is straightforward and robust, accomplished by integrating the area under the tracer concentration-time curve (Fig. 2). This integration may be performed on the curve data points themselves or on an analytic fit of the data points [15]. The latter approach has the advantage of eliminating overestimation from the effects of tracer recirculation, but this approach has the disadvantage of requiring high signal stability and faster imaging over time [16]. Determination of relative cerebral blood flow requires more extensive processing of the imaging data and is more adversely influenced by poor image quality and instability in the MR signal over time. The processing techniques require deconvolution of an arterial input function from tissue concentration-time data to find the true brain clearance, or mean transit time through the cerebral capillary bed (mean transit time). Cerebral blood volume, calculated by integrating the area under the deconvolved tissue concentration-time curve, is then divided by mean transit time to obtain cerebral blood flow [17]. Alternatively, the initial height of the deconvolved tissue concentration-time curve may be taken as the cerebral blood flow, and the mean transit time may then be calculated as the ratio of cerebral blood volume to cerebral blood flow [18] (Fig. 2). Again, image quality and signal stability over time are important requirements for reliably calculating relative cerebral blood flow because the deconvolution technique mentioned previously can amplify noise and artifactually introduce bias [18].

Determination of an accurate arterial input function is also an important requirement for calculating relative cerebral blood flow. An arterial input function may be obtained directly from the imaging data by manually selecting the voxels from which the arterial input function will be obtained [19]. This may be aided by narrowing the selection to a small population of voxels chosen using an automated algorithm that searches the entire imaging volume for voxels with time-concentration curves that satisfy criteria characteristic of arteries, such as a large peak, early arrival time, and a short mean transit time [9, 11]. The use of such an algorithm increases reproducibility because it requires less user interaction. It should be noted, however, that the exclusive reliance on an automated approach may lead to erroneous selection of an arterial input function. For example, in the case in which a cerebral hemisphere is being fed by a diseased middle cerebral artery, deriving the arterial input function from voxels in the diseased hemisphere would lead to a more accurate result than deriving the arterial input function from the voxels in the contralateral hemisphere, even though the latter may better satisfy criteria for a “normal” artery. Voxel-by-voxel determination of cerebral blood flow, in theory, requires determination of the unique arterial input to each voxel. Because this is not possible, most methods assume the arterial input is uniform across the brain and apply a single arterial input function to the entire brain. In the case of a unilaterally diseased middle cerebral artery, this assumption is violated. Applying an arterial input function chosen from voxels in the normal contralateral hemisphere may lead to underestimation of cerebral blood flow and false-positive identification of an ischemic zone.

Dynamic sequences must be ultrafast to monitor the rapid first-pass transit of a bolus of contrast agent through the brain, which is on the order of 18 sec [20]. Either T1- or T2-weighted techniques can be used. The T2-weighted sequences are more commonly used in clinical practice. Using these sequences, injection of a paramagnetic contrast agent causes a transient drop in signal intensity that is due to the susceptibility effects of the paramagnetic contrast agent. A single- or dual-slice dynamic study can be performed on a conventional MR scanner without specialized gradient hardware [9]. Multislice techniques (up to 30 slices per second) are available on systems with specialized gradient hardware for echoplanar imaging or spiral imaging [21]. These techniques can be either T2-weighted (spin echo) or T2*-weighted (gradient echo). The spin-echo technique has the advantage of minimizing artifact at brain-bone and brain-air interfaces and is more sensitive to signal changes from paramagnetic contrast material passing through small vessels, such as capillaries, rather than through large vessels, such as cortical veins [22]. The spin-echo technique has the disadvantage of requiring a larger dose of contrast material, often 1.5-2.0 times that of a standard dose, to produce signal changes equivalent to those of the gradient-echo technique (Aronen H et al., presented at the Society of Magnetic Resonance in Medicine meeting, August 1992). Furthermore, the spin-echo technique may create bias on serial studies, leading to artificially elevated cerebral blood volume measurements, if repeated within 2 hr of the initial study. This bias is caused by a residual contrast material effect that alters the magnitude of signal change from baseline [23]. These effects have been shown to not be significant using a gradient-echo technique [11].

A T1-weighted dynamic technique is another method by which to measure cerebral hemodynamics [24] and has the advantage of requiring a smaller contrast material dose and providing better temporal resolution than the T2- or T2*-weighted sequences. The T1-weighted technique measures the relaxivity effects, rather than the susceptibility effects, of an IV-injected dose of paramagnetic contrast material. The relaxivity effect of paramagnetic contrast material refers to the shortening of T1 relaxation time, leading to higher signal on T1-weighted images, whereas the susceptibility effect refers to the shortening of T2 and T2* relaxation times, leading to lower signal on T2- or T2*-weighted images. Because the relaxivity effects of gadopentetate dimeglumine are much stronger than the susceptibility effects, the T1-weighted pulse sequences require a smaller amount of contrast material (approximately 10%) than the T2- or T2*-weighted techniques [25], allowing multiple repeated studies. Moreover, the short injection time allowed by a smaller bolus may result in better quantitation of cerebral blood volume and cerebral blood flow provided that the temporal resolution of the pulse sequence allows tracking the bolus over a sufficient number of time points to extract the corresponding parameters [24]. Subsecond imaging times (300-900 msec) over an anatomic range of one to two slices are currently possible with this technique using fast T1-weighted gradient-echo imaging [25]. The T2- or T2*-weighted technique requires imaging times on the order of 1.5-2 sec, although the anatomic coverage is greater with echoplanar imaging (8-11 slices) or spiral imaging (18-20 slices) [21, 26]. The disadvantage of the T1-weighted technique is that leakage through the blood-brain barrier may lead to errors in measurements of hemodynamic parameters. Although this may be corrected for in the calculations, the effects of blood-brain barrier breakdown are greater with the T1-weighted technique than with the T2- or T2*-weighted technique. Quantitative assessment of permeability through the blood-brain barrier has been examined using both T1- and T2-weighted techniques. Further discussion of such techniques can be found in the literature [27, 28].

Steady-state imaging.—In addition to the more commonly used T1- and T2- or T2*-weighted dynamic perfusion imaging techniques, a T1-weighted steady-state technique may be used to estimate absolute cerebral blood volume with high spatial resolution across the entire brain. This method assumes the tracer is nondiffusible from the intravascular to extravascular space. With this technique, a baseline image is obtained before the injection of the paramagnetic agent, followed by a postinfusion image that is acquired during the “steady state,”—that is, up to 30 min after the contrast material has circulated through the body and reached a point of relative concentration equilibrium. By subtracting the baseline image from the postcontrast steady-state image and normalizing the pixel values to those of a pixel containing only blood, such as in the sagittal sinus, one can obtain a map of absolute cerebral blood volume in units of volume percent [29, 30]. This can be converted to the more conventional units of milliliters per 100 g of brain by normalizing to the density of brain tissue (1.04 g/ml) and multiplying by 100 [9]. Unfortunately, this approach has a number of disadvantages. First, because image subtraction is performed, the resulting images have a low signal-to-noise ratio. Second, patient movement between the pre- and postcontrast scans may affect the accuracy of these measurements. Third, spurious results can be obtained in areas where the blood-brain barrier has been disrupted and the assumption of tracer nondiffusibility has been violated; therefore, the technique is not useful in many cases of infarction and tumor.

Endogenous Tracer Methods

Endogenous tracer methods in MR perfusion imaging use a model that assumes the tracer freely diffuses from the intravascular compartment into the tissue compartment. This model is similar to that used in positron emission tomography and single-photon emission computed tomography (SPECT) in which a tracer is administered and the regional accumulation, influenced by regional blood flow and tracer half-life, is measured [31]. Endogenous tracer MR perfusion methods take advantage of signal loss resulting from magnetically labeled water protons (spins) flowing into the imaging plane and exchanging with tissue protons. Water protons within inflowing arterial blood are magnetically labeled (or “tagged”) by the application of a special radiofrequency pulse designed to invert spins in a thick slab proximal to the slice of interest (Fig. 3). By measuring signal changes between tagged images and baseline untagged images, qualitative or quantitative images of cerebral blood flow can be obtained (Fig. 4). Inflowing blood may be tagged continuously or intermittently [31, 32]. Although continuous-labeling techniques afford twice as much signal contrast compared with pulsed techniques, they produce substantially more radiofrequency pulse-induced power deposition to the subject. This safety consideration can ultimately limit slice coverage and acquisition time.

figure
View larger version (197K)
Fig. 3. —33-year-old healthy man. Unenhanced sagittal T1-weighted MR image of brain shows continuous inversion arterial spin-tagging technique. Solid lines depict imaging slice and dashed line depicts tagging plane where water protons in inflowing arterial blood are magnetically tagged by radiofrequency inversion pulse. Quantitative estimates of cerebral blood flow can be obtained by measuring signal changes between tagged images and baseline untagged images.

figure
View larger version (52K)
Fig. 4. —33-year-old healthy man. Multiple slices of brain obtained using multislice arterial spin-tagging MR perfusion imaging technique. Images show quantitative cerebral blood flow maps. Displayed are five of 10 slice locations extending from level of mid lateral ventricle to level of supraventricular white matter. Artifacts from high flow in superior sagittal sinus are noted anteriorly and, to lesser extent, posteriorly. Total imaging time was approximately 5 min (Courtesy of Yongbi M, Duyn, JH, and Yang Y, Bethesda, MD).

Spin-tagging techniques suffer inaccuracies that are due to two major effects. The first is magnetization transfer effects in the imaging plane as spins are labeled below the imaging plane with a radiofrequency pulse. During this time, the spins in the plane of interest experience an off-resonance radiofrequency pulse that selectively saturates the broad resonance peak of macromolecular-bound protons. This saturation is then transferred to the free-proton pool, resulting in as much as a 60% loss in observed brain signal [33]. To compensate for this effect, one strategy involves applying a radiofrequency pulse during the baseline state at an equal distance above the plane of interest [34, 35].

The second inaccuracy is from the loss of spin labeling during the arterial transit period due to T1 relaxation as blood moves from the tagging plane to the imaging plane [31]. These arterial transit effects may be markedly reduced by introducing a delay after continuous labeling [36] or by tagging spins immediately below the imaging plane using an intermittent pulse to reduce overall transit time [32]. The latter technique, called “EPISTAR” (echoplanar imaging with signal targeting and alternating radiofrequency), provides only a qualitative map of cerebral blood flow because the relationship between cerebral blood flow and EPISTAR signal is complex and depends on differential arterial transit times, the angle of feeding arteries with the imaging plane, and inflow effects [37]. This technique also suffers from low sensitivity and therefore low flow rates (10-25 ml · 100 g-1 · min-1) may be difficult to detect.

An alternative to labeling spins proximal to the imaging plane is to directly label spins in the imaging plane itself using a slice-selective inversion-recovery technique and thereafter measure signal increases from unlabeled inflowing spins. In this case, the unlabeled spins have complete longitudinal magnetization and the T1-relaxation effects that are due to arterial transit are eliminated. Thus, signal changes are indirectly related to cerebral blood flow. By applying an alternating global inversion-recovery pulse along with the slice-selective pulse and comparing the two conditions, one can measure signal changes that are caused solely by inflowing blood. These signal changes are more directly related to absolute cerebral blood flow. The FAIR (flow-sensitive alternating inversion-recovery) technique [38] is one example of this method. A multislice version of FAIR has recently been developed using imaging [39].

Clinical Applications
Previous sectionNext section
Exogenous Tracer Methods

Stroke.—Probably the widest application of exogenous tracer methods in MR perfusion imaging has been in the assessment of cerebral ischemia. A number of investigators have suggested that in the setting of an acute stroke, perfusion imaging in combination with diffusion imaging can help identify surrounding viable ischemic tissue at risk (the so-called ischemic penumbra) [7, 40]. Specifically, it has been hypothesized that the area of decreased cerebral blood volume, decreased cerebral blood flow, or prolonged mean transit time in the ischemic region represents both the infarct core as well as reversible surrounding ischemic tissue at risk, whereas the area of abnormal diffusion represents only the irreversibly ischemic infarct core. The mismatch between the perfusion and diffusion abnormality is thought to represent the potentially salvageable ischemic tissue at risk for infarction (Fig. 5A,5B,5C,5D,5E). Identification of the presence of salvageable tissue surrounding an infarct has taken on critical importance given the availability of recently approved thrombolytic and neuroprotective agents [41]. Because these therapeutic agents are not without risk, it is necessary to select patients with reversibly ischemic tissue who are likely to benefit from this therapy. In the setting of subacute infarction, it is possible to evaluate for the presence of luxury perfusion, characterized by increased cerebral blood volume, surrounding the infarct core. In such cases, it is important to distinguish between cerebral blood volume and cerebral blood flow because cerebral blood volume may be increased adjacent to a recently infarcted area, whereas cerebral blood flow may be decreased because of prolonged transit times [8].

figure
View larger version (162K)
Fig. 5A. —43-year-old man with acute onset of left-sided weakness and visual changes who was found to have left homonmous hemianopsia on examination. Unenhanced CT scan reveals negative finding for cortical infarction.

figure
View larger version (167K)
Fig. 5B. —43-year-old man with acute onset of left-sided weakness and visual changes who was found to have left homonmous hemianopsia on examination. T2-weighted MR image shows increased signal (arrow) in right calcarine cortex.

figure
View larger version (151K)
Fig. 5C. —43-year-old man with acute onset of left-sided weakness and visual changes who was found to have left homonmous hemianopsia on examination. Diffusion-weighted scan demonstrates larger area of signal abnormality (arrow) involving right occipital lobe, consistent with infarction.

figure
View larger version (153K)
Fig. 5D. —43-year-old man with acute onset of left-sided weakness and visual changes who was found to have left homonmous hemianopsia on examination. Color-coded cerebral blood volume map obtained using dynamic T2-weighted technique shows even larger perfusion deficit than that seen in B and C in right occipital lobe, including infarction core, and surrounding tissue at risk. Red denotes high cerebral blood volume; blue, low cerebral blood volume.

figure
View larger version (123K)
Fig. 5E. —43-year-old man with acute onset of left-sided weakness and visual changes who was found to have left homonmous hemianopsia on examination. Color-coded mean transit time map obtained using dynamic T2-weighted technique shows prolonged transit time in right occipital lobe, also corresponding to infarct core, and surrounding tissue at risk. Red denotes prolonged mean transit time; yellow, normal mean transit time.

MR perfusion imaging is useful not only in the assessment of stroke, but also in the assessment of stroke risk. Under normal circumstances, the brain has an autoregulatory mechanism for maintaining adequate cerebral oxygenation in the face of decreasing cerebral perfusion pressure, which allows normal blood flow despite fluctuations in systemic pressure. This mechanism may be impeded in patients with hemodynamically significant carotid artery stenosis who are at high risk for stroke. The ability to maintain an autoregulatory response to hemodynamic stress has been termed “cerebrovascular reserve capacity.” Areas of the brain supplied by a markedly stenotic or occluded artery, in which vasodilatation has already occurred to maintain adequate flow, lack cerebrovascular reserve capacity. As a result, when a pharmacologic vasodilatory challenge is administered, minimal vasodilatory response occurs. Assessment of response to vasodilatory challenge has therefore also been used as an indirect means of measuring cerebrovascular reserve capacity. Perfusion MR imaging may be used to assess cerebral blood volume or cerebral blood flow before and after a vasodilatory challenge using agents such as carbon dioxide or the carbonic anhydrase inhibitor, acetazolamide [10,11,12]. In addition to a poor response to a vasodilatory challenge, perfusion imaging may show other abnormalities in the cerebral hemisphere ipsilateral to a severe carotid stenosis or occlusion, such as delayed bolus arrival time and prolonged mean transit time [42, 43] (Fig. 6A,6B,6C,6D).

figure
View larger version (141K)
Fig. 6A. —66-year-old man with right-sided internal carotid occlusion. Lateral view from conventional arteriogram shows occlusion of right-sided internal carotid artery at origin (arrowhead).

figure
View larger version (111K)
Fig. 6B. —66-year-old man with right-sided internal carotid occlusion. Intracranial MR angiogram reveals no visible flow within intracranial portion of right-sided internal carotid artery until cavernous segment (arrow), where there is reconstitution of right hemispheric circulation via cross-filling (black arrowheads) from Circle of Willis and from external carotid artery (white arrowheads) collaterals.

figure
View larger version (174K)
Fig. 6C. —66-year-old man with right-sided internal carotid occlusion. Baseline mean transit time map reveals mildly prolonged transit times (arrows) in right cerebral hemisphere.

figure
View larger version (164K)
Fig. 6D. —66-year-old man with right-sided internal carotid occlusion. After vasodilatory challenge with acetazolamide, map shows transit time in right hemisphere has normalized, suggesting maintenance of cerebrovascular reserve capacity from adequate collateral circulation.

Brain tumors.—Another area in which MR perfusion imaging may be useful is in the evaluation of brain tumors. Dynamic imaging is performed using either T1-weighted or T2- or T2*-weighted technique [44]. Cerebral blood volume maps can be used to assess neovascularity in tumors, which is thought to correlate with tumor grade and malignant histology. Because of selective sensitivity to small vessels, the T2-weighted technique may be preferred over the T2*-weighted technique [22]. Cerebral blood volume maps may aid in early evaluation of therapeutic agents, especially of a new class of drugs aimed at suppressing growth of tumor blood vessels. These maps can also potentially be used to localize areas of tumor more likely to yield positive results on stereotactic biopsy and to noninvasively differentiate radiation necrosis from recurrent tumor in circumstances in which conventional MR findings are equivocal [45, 46] (Figs. 7A,7B,7C and 8A,8B,8C). Similar techniques have evolved for differentiating radiation necrosis from recurrent tumor on the basis of differences in blood-brain barrier permeability. For detailed discussions of these methods, the reader is referred to the literature [28, 47].

figure
View larger version (143K)
Fig. 7A. —24-year-old woman with previously treated high-grade cerebral neoplasm (anaplastic ependymoma) with an enhancing lesion on follow-up examination. Biopsy revealed radiation necrosis. Contrast-enhanced axial T1-weighted image shows area of abnormal enhancement in right frontoparietal deep white matter.

figure
View larger version (133K)
Fig. 7B. —24-year-old woman with previously treated high-grade cerebral neoplasm (anaplastic ependymoma) with an enhancing lesion on follow-up examination. Biopsy revealed radiation necrosis. Color-coded cerebral blood volume map obtained using dynamic T2-weighted technique illustrates low cerebral blood volume in area of abnormal contrast enhancement seen in A. Red denotes high cerebral blood volume; blue, low cerebral blood volume.

figure
View larger version (151K)
Fig. 7C. —24-year-old woman with previously treated high-grade cerebral neoplasm (anaplastic ependymoma) with an enhancing lesion on follow-up examination. Biopsy revealed radiation necrosis. Overlay of color-coded cerebral blood volume map on T1-weighted image with cerebral blood volume map thresholded so only voxels with cerebral blood volume values equal to or higher than that of gray matter are depicted. Note that area of enhancement in right frontoparietal deep white matter has low cerebral blood volume relative to gray matter, consistent with radiation necrosis.

figure
View larger version (171K)
Fig. 8A. —29-year-old woman with previously treated high-grade astrocytoma with an enhancing lesion on follow-up examination. Biopsy revealed recurrent tumor. Contrast-enhanced axial T1-weighted image depicts area of abnormal enhancement in left frontal lobe periventricular white matter.

figure
View larger version (139K)
Fig. 8B. —29-year-old woman with previously treated high-grade astrocytoma with an enhancing lesion on follow-up examination. Biopsy revealed recurrent tumor. Color-coded cerebral blood volume map obtained using dynamic T2-weighted technique illustrates areas of moderate to high cerebral blood volume in area of abnormal contrast enhancement seen in A. Red denotes high cerebral blood volume; blue, low cerebral blood volume.

figure
View larger version (160K)
Fig. 8C. —29-year-old woman with previously treated high-grade astrocytoma with an enhancing lesion on follow-up examination. Biopsy revealed recurrent tumor. Overlay of color-coded cerebral blood volume map on T1-weighted image with cerebral blood volume map thresholded so only voxels with cerebral blood volume values equal to or higher than that of gray matter are depicted. Note area of enhancement in left frontal lobe periventricular white matter reveals areas of moderate to high cerebral blood volume, consistent with recurrent tumor.

Other disorders.—In addition to evaluation of ischemia and tumors, MR perfusion imaging has been applied to the study of various other neurologic and psychiatric disorders, such as dementia and migraine headaches [48]. The effects of psychoactive drugs, such as cocaine, have been studied as well (Kaufman MJ et al., presented at the International Society of Magnetic Resonance in Medicine, April 1996). In the case of migraine headaches, decreases in cerebral blood volume and cerebral blood flow have been seen during the auras compared with the post-aura state (Sorensen AG et al., presented at the International Society of Magnetic Resonance in Medicine, April 1996). In the case of dementia, decreases in cerebral blood volume in the temporal and parietal lobes of patients with Alzheimer's disease have correlated well with the results of SPECT studies on the same subjects [49, 50].

Endogenous Tracer Methods

Clinical applications of the endogenous tracer methods have been limited compared with those of exogenous methods because of longer acquisition times and sensitivity to patient motion. Furthermore, although FAIR and other spin-labeling techniques are capable of calculating absolute cerebral blood flow in theory, in practice these techniques have not yet shown sufficient signal-to-noise ratio to validate these measurements, especially in low-flow states and in white matter. In such situations, even the theoretical assumptions required to make absolute flow estimates may break down because of, for example, the loss of spin labeling from prolonged transit times. Once these limitations are overcome, however, many clinical applications may be possible.

Stroke.—Quantitative MR cerebral blood flow measurements could potentially be obtained as part of a complete MR evaluation of stroke in the assessment of tissue viability and stroke etiology, for example. Previous work has suggested that tissue viability, in the setting of acute stroke, is related to the degree and duration of ischemia [51, 52]. The degree of ischemia has been assessed in the past through quantitative regional cerebral blood flow measurements using positron emission tomography and 133Xe SPECT, and thresholds of cerebral blood flow measurements have been established below which tissue viability is unlikely. The role of hypoperfusion as the primary cause of stroke can also be examined using quantitative MR measurements of cerebral blood flow in patients with potential border-zone or watershed infarctions [53].

Clinical research.—Another major benefit of absolute quantitation of cerebral blood flow is in clinical research, which depends on accurate inter- and intrasubject comparisons. Intrasubject comparisons are useful in following the natural history of a disease process or in assessing the effect of therapeutic interventions in conditions such as stroke, neoplasms, and neurocognitive disorders such as dementia, especially in cases when no normal areas of internal reference are available. Intersubject comparisons of cerebral blood flow between different patient populations may allow assessment of drug efficacy or may be a useful tool in investigating disease mechanisms.

Functional brain mapping.—A third major benefit of absolute quantification of cerebral blood flow is in the area of functional MR brain mapping. Cerebral activation is usually determined through qualitative assessment of local changes in deoxyhemoglobin concentration. This phenomenon, in which local differences in cerebral blood oxygenation levels are related to the degree of neuronal activity, is known as the blood oxygenation level-dependent (BOLD) effect. The BOLD effect is based on close coupling of neuronal activity to hemodynamic response in the brain [54]. Although BOLD changes are more sensitive to cerebral activation than absolute cerebral blood flow changes, BOLD changes are dependent on a number of other physiologic variables and MR parameters [38, 55]. Direct measurement of cerebral blood flow changes during cerebral activation may enable better localization of neuronal activity than measurement of BOLD changes and may allow more physiologically meaningful comparison of activation between different brain regions [38] (Fig. 9A,9B). Quantitative cerebral blood flow changes during motor and working memory tasks have been measured with the arterial spintagging technique and yield results similar to those obtained with other techniques such as 133Xe SPECT and positron emission tomography [56, 57].

figure
View larger version (42K)
Fig. 9A. —33-year-old healthy man. Comparison of finger-tapping activation task using arterial spin-tagging and blood oxygenation level-dependent (BOLD) techniques. (Reprinted with permission from [38]) t test activation maps (red) superimposed on T2*-weighted images using multislice FAIR (flow-sensitive alternating inversion-recovery) technique; FAIR is sensitive to increases in local cerebral blood flow during task.

figure
View larger version (42K)
Fig. 9B. —33-year-old healthy man. Comparison of finger-tapping activation task using arterial spin-tagging and blood oxygenation level-dependent (BOLD) techniques. (Reprinted with permission from [38]) t test activation maps (red) superimposed on T2*-weighted images using BOLD technique, which is sensitive to changes in blood oxygenation. Note patterns of activation similar to those in A.

Conclusion
Previous sectionNext section

In summary, MR perfusion imaging is an emerging clinical tool that enables assessment of regional cerebral hemodynamics using a variety of techniques (Table 1). The most common clinically applicable technique uses rapid T2- or T2*-weighted imaging to monitor the first pass of a bolus injection of exogenous paramagnetic contrast material. Using tracer analysis techniques, one may obtain semiquantitative or relative cerebral blood flow, cerebral blood volume, and mean transit time maps. Spin-tagging techniques use magnetically labeled blood as an endogenous contrast agent and may enable absolute quantitation of cerebral blood flow. These techniques are gaining increasing use and have the potential to become an important clinical tool in the diagnosis and treatment of patients with cerebrovascular disease, neoplasms, and other disorders.

TABLE 1 Advantages and Disadvantages of Current MR Perfusion Imaging Techniques

Address correspondence to J. R. Petrella

We thank Joseph Frank and Alan McLaughlin at the National Institutes of Health for their helpful comments and Jimmie Wong and Luiz Celso H. Cruz for assistance in preparing the figures.

References
Previous sectionNext section
1. Frackowiak RSJ, Lenzi G, Jones T, Heather JD. Quantitative measurement of regional cerebral blood flow and oxygen metabolism in man using O15 and positron emission tomography: theory, procedure, and normal values. J Comput Assist Tomogr 1980; 4:727-736 [Google Scholar]
2. Gobbel GT, Cann CE, Fike JR. Measurement of regional cerebral blood flow using ultrafast computed tomography: theoretical aspects. Stroke 1991; 22:768-771 [Google Scholar]
3. Rosen BR, Belliveau JW, Chien D. Perfusion imaging by nuclear magnetic resonance. Magn Reson Q 1989; 5:263-281 [Google Scholar]
4. Lassen NA, Perl WA. Tracer kinetic methods in medical physiology. New York: Raven, 1979:1-155 [Google Scholar]
5. Edelman RR, Wielopolski P, Schmitt F. Echo-planar MR imaging. Radiology 1994; 193:600-612 [Google Scholar]
6. Hou L, Yang Y, Mattay VS, Frank JA, Duyn JH. Optimization of fast acquisition methods for whole-brain relative cerebral blood volume (rCBV) mapping with susceptibility contrast agents. J Magn Reson Imaging 1999; 9:233-239 [Google Scholar]
7. Sorensen AG, Buonanno FS, Gonzalez RG, et al. Hyperacute stroke: evaluation with combined multisection diffusion-weighted and hemodynamically weighted echo-planar MR imaging. Radiology 1996; 199:391-401 [Google Scholar]
8. Tsuchida C, Hiroki Y, Maeda M, et al. Evaluation of peri-infarct hypoperfusion with T2*-weighted dynamic MRI. J Magn Reson Imaging 1997; 7:518-522 [Google Scholar]
9. Rempp M, Gunnar B, Wenz F, Becker C, Guckel F, Lorenz W. Quantification of regional cerebral blood flow and volume with dynamic susceptibility contrast-enhanced MR. imaging. Radiology 1994; 193:637-641 [Google Scholar]
10. Guckel F, Brix G, Schmiedek P, et al. Cerebrovascular reserve capacity in patients with occlusive cerebrovascular disease: assessment with dynamic susceptibility contrast-enhanced MR imaging and the acetazolamide stimulation test. Radiology 1996; 201:405-412 [Google Scholar]
11. Petrella JR, DeCarli C, Dagli M, et al. Assessment of whole-brain vasodilatory capacity with acetazolamide challenge at 1.5 T using dynamic contrast imaging with frequency-shifted burst. AJNR 1997; 18:1153-1161 [Google Scholar]
12. Schreiber WG, Guckel F, Stritzke P, Schmiedek P, Schwartz A, Brix G. Cerebral blood flow and cerebrovascular reserve capacity: estimation by dynamic magnetic resonance imaging. J Cereb Blood Flow Metab 1998; 18:1143-1156 [Google Scholar]
13. Simonsen CZ, Ostergaard L, Vestergaard-Poulsen P, Rohl L, Bjornerud A, Gyldensted C. CBF and CBV measurements by USPIO bolus tracking: reproducibility and comparison with Gd-based values. J Magn Reson Imaging 1999; 9:342-347 [Google Scholar]
14. Jezzard P. Advances in perfusion MR imaging. Radiology 1998; 208:296-299 [Google Scholar]
15. Thompson HK, Starmer CF, Whalen RE, McIntosh HD. Indicator transit time considered as a gamma variate. Circ Res 1964; 14:502-515 [Google Scholar]
16. Lu D, Monahan WG. Effect of sample numbers on the kinetic data analysis of MR contrast agents. Magn Reson Med 1993; 30:131-134 [Google Scholar]
17. Weisskoff RW, Chesler D, Boxerman JL, Rosen BR. Pitfalls in MR measurement of tissue blood flow with intravascular tracers: which mean transit time? Magn Reson Med 1992; 29:553-559 [Google Scholar]
18. Ostergaard L, Weisskoff RM, Chesler DA, Gyldensted C, Rosen BR. High resolution measurement of cerebral blood flow using intravascular tracer bolus passages. I. Mathematical approach and statistical analysis. Magn Reson Med 1996; 36:715-725 [Google Scholar]
19. Ostergaard L, Sorenson AG, Kwong KK, Weisskoff RM, Gyldensted C, Rosen BR. High resolution measurement of cerebral blood flow using intravascular tracer bolus passages. II. Experimental comparison and preliminary results. Magn Reson Med 1996; 36:726-736 [Google Scholar]
20. Belliveau JW, Rosen BR, Kantor HL, et al. Functional cerebral imaging by susceptibility-contrast NMR. Magn Reson Med 1990; 14:538-546 [Google Scholar]
21. Frank JA, Ostuni JL, Yang Y, et al. Technical solution for an interactive functional MR imaging examination: application to a physiologic interview and the study of cerebral physiology. Radiology 1999; 210:260-268 [Google Scholar]
22. Fisel CR, Ackerman JL, Buxton RB, et al. MR contrast due to microscopically heterogeneous magnetic susceptibility: numerical simulations and applications to cerebral physiology. Magn Reson Med 1991; 17:336-347 [Google Scholar]
23. Levin JM, Kaufman MJ, Ross MH, et al. Sequential dynamic susceptibility contrast MR experiments in human brain: residual contrast agent effects, steady state, and hemodynamic perturbation. Magn Reson Med 1995; 34:655-663 [Google Scholar]
24. Dean BL, Lee C, Kirsch JE, Runge VM, Dempsey RM, Pettigrew LC. Cerebral hemodynamics and cerebral blood volume: MR assessment using gadolinium contrast agents and T1-weighted turbo-FLASH imaging. AJNR 1992; 13:39-48 [Google Scholar]
25. Hacklander T, Reichenbach JR, Hofer M, Modder U. Measurement of cerebral blood volume via the relaxing effect of low-dose gadopentetate dimeglumine during bolus transit. AJNR 1996; 17:821-830 [Google Scholar]
26. Duyn JH, Yang Y. Fast spiral magnetic resonance imaging with trapezoidal gradients. J Magn Reson 1997; 128:130-134 [Google Scholar]
27. Tofts PS. Modeling tracer kinetics in dynamic Gd-DTPA MR imaging. J Magn Reson Imaging 1997; 7:91-101 [Google Scholar]
28. Weisskoff RM, Boxerman JL, Sorensen AG, Kulke SM, Campbell TA, Rosen BR. Simultaneous blood volume and permeability mapping using a single Gd-based contrast injection (abstr). Proceedings of the Society of Magnetic Resonance second meeting. San Francisco: Society of Magnetic Resonance 1994; 1:279 [Google Scholar]
29. Kuppusamy K, Lin W, Cizek G, Haacke EM. In vivo regional cerebral blood volume: quantitative assessment with 3D T1-weighted pre- and postcontrast MR imaging. Radiology 1996; 201:106-112 [Google Scholar]
30. Lin W, Celik A, Paczynski RP. Regional cerebral blood volume: a comparison of the dynamic imaging and the steady state methods. J Magn Reson Imaging 1999; 9:44-52 [Google Scholar]
31. Detre JA, Williams DS, Zhang W, Roberts DA, Leigh JS, Koretsky AP. Noninvasive perfusion MR imaging using spin labeling of arterial water. In: Le Bihan D, ed. Diffusion and perfusion magnetic resonance imaging. New York: Raven, 1995:296-304 [Google Scholar]
32. Edelman RR, Siewert B, Darby D, et al. Qualitative mapping of cerebral blood flow and functional localization with echo-planar MR imaging and signal targeting with alternating radio frequency. Radiology 1994; 192:513-520 [Google Scholar]
33. Wolff SD, Balaban RS. Magnetization transfer contrast (MTC) and tissue water proton relaxation in vivo. Magn Reson Med 1989; 10:135-144 [Google Scholar]
34. Pekar J, Jezzard P, Roberts DA, Leigh JS, Frank JA, McLaughlin AC. Perfusion imaging with magnetization transfer frequency offset compensation. Magn Reson Med 1996; 35:70-79 [Google Scholar]
35. McLaughlin AC, Ye F, Pekar JJ, Santha AKS, Frank JA. Effect of magnetization transfer on the measurements of cerebral blood flow using arterial spin tagging approaches. Magn Reson Med 1997; 37:501-510 [Google Scholar]
36. Alsop DC, Detre JA. Reduced transit-time sensitivity in noninvasive magnetic resonance imaging of human cerebral blood flow. J Cereb Blood Flow Metab 1996; 16:1236-1249 [Google Scholar]
37. Ye FQ, Mattay VS, Jezzard P, Frank JA, Weinberg DR, McLaughlin AC. Correction for vascular artifacts in cerebral blood flow values measured using arterial spin tagging techniques. Magn Reson Med 1997; 37:226-235 [Google Scholar]
38. Kim S. Quantification of relative cerebral blood flow change by flow-sensitive alternating inversion recovery (FAIR) technique application to functional mapping. Magn Reson Med 1995; 34:293-301 [Google Scholar]
39. Yang Y, Frank J, Hou L, Ye FQ, McLaughlin AC, Duyn JH. Multislice imaging of quantitative cerebral perfusion with pulsed arterial spin labeling. Magn Reson Med 1998; 39:825-832 [Google Scholar]
40. Sorensen AG, Copen WA, Ostergaard L, et al. Hyperacute stroke: simultaneous measurement of relative cerebral blood volume, relative cerebral blood flow, and mean tissue transit time. Radiology 1999; 210:519-527 [Google Scholar]
41. The National Institute of Neurological Disorders and Stroke rt-PA Stroke Study Group. Tissue plasminogen activator for acute ischemic stroke. N Engl J Med 1995; 333:1581-1587 [Google Scholar]
42. Duyn JH, van Gelderen P, Barker P, Frank JA, Mattay VS, Moonen CT. 3D bolus tracking with frequency-shifted BURST MRI. J Comput Assist Tomogr 1994; 18:680-687 [Google Scholar]
43. Masayuki M, Yuh WTC, Ueda T, et al. Severe occlusive carotid artery disease: hemodynamic assessment by MR perfusion imaging in symptomatic patients. AJNR 1999; 20:43-51 [Google Scholar]
44. Bruening R, Kwong KK, Vevea MJ, et al. Echoplanar MR determination of relative cerebral blood volume in human brain tumors: T1 versus T2 weighting. AJNR 1996; 17:831-840 [Google Scholar]
45. Aronen H, Gazit IE, Louis DN, et al. Cerebral blood volume maps of gliomas: comparison with tumor grade and histologic findings. Radiology 1994; 191:41-51 [Google Scholar]
46. Knopp EA, Cha S, Johnson G, et al. Glial neoplasms: dynamic contrast-enhanced T2*-weighted MR imaging. Radiology 1999; 211:791-798 [Google Scholar]
47. Hazle JD, Jackson EF, Schomer DF, Leeds NE. Dynamic imaging of intracranial lesions using fast spin-echo imaging: differentiation of brain tumors and treatment effects. J Magn Reson Imaging 1997; 7:1084-1093 [Google Scholar]
48. Lev MH, Rosen BR. Clinical applications of intracranial perfusion MR imaging. Neuroimaging Clin N Am 1999; 9:309-331 [Google Scholar]
49. Mattay VS, Frank JA, Duyn JH, et al. Three-dimensional “BURST” functional magnetic resonance imaging: initial clinical applications. Acad Radiol 1996; 3[suppl 2]:S379-S383 [Google Scholar]
50. Harris GJ, Lewis RF, Satlin A, et al. Dynamic susceptibility contrast MR imaging regional cerebral blood volume in Alzheimer disease: a promising alternative to nuclear medicine. AJNR 1998; 19:1727-1732 [Google Scholar]
51. Powers W, Grubb RL, Raichle ME. Physiological responses to focal cerebral ischemia in humans. Ann Neurol 1984; 16:546-552 [Google Scholar]
52. Raichle ME. The pathophysiology of brain ischemia. Ann Neurol 1983; 13:2-10 [Google Scholar]
53. Detre JA, Alsop DC, Vives LR, Maccotta L, Teener JW, Raps EC. Noninvasive MRI evaluation of cerebral blood flow in cerebrovascular disease. Neurology 1998; 50:633-641 [Google Scholar]
54. Ogawa S, Lee TM, Kay AR, Tank DW. Brain magnetic resonance imaging with contrast dependent on blood oxygenation. Proc Natl Acad Sci U S A 1990; 87:9868-9872 [Google Scholar]
55. Hathout GM, Gambhir SS, Gopi RK, et al. A quantitative physiologic model of blood oxygenation for functional magnetic resonance imaging. Invest Radiol 1995; 30:669-682 [Google Scholar]
56. Ye FQ, Smith AM, Yang Y, et al. Quantitation of regional cerebral blood flow increases during motor activation: a steady-state arterial spin tagging study. Neuroimage 1997; 6:104-112 [Google Scholar]
57. Ye FQ, Smith AM, Mattay VS, et al. Quantitation of regional cerebral blood flow increases in prefrontal cortex during a working memory task: a steady-state arterial spin-tagging study. Neuroimage 1998; 8:44-49 [Google Scholar]

Recommended Articles

MR Perfusion Imaging of the Brain

Full Access, ,
American Journal of Roentgenology. 2015;205:840-847. 10.2214/AJR.14.13415
Abstract | Full Text | PDF (847 KB) | PDF Plus (860 KB) 
Full Access, , ,
American Journal of Roentgenology. 2012;198:52-62. 10.2214/AJR.10.7255
Abstract | Full Text | PDF (3913 KB) | PDF Plus (1708 KB) 
Full Access
American Journal of Roentgenology. 2005;185:S196-S204. 10.2214/AJR.05.1014
Abstract | Full Text | PDF (2218 KB) | PDF Plus (612 KB) 
Full Access, , , , , ,
American Journal of Roentgenology. 2011;196:1206-1213. 10.2214/AJR.10.4877
Abstract | Full Text | PDF (1065 KB) | PDF Plus (1128 KB) 
Full Access, , , ,
American Journal of Roentgenology. 2013;200:1089-1095. 10.2214/AJR.12.9738
Abstract | Full Text | PDF (887 KB) | PDF Plus (831 KB) 
Full Access, , , ,
American Journal of Roentgenology. 2005;185:406-417. 10.2214/ajr.185.2.01850406
Abstract | Full Text | PDF (3458 KB) | PDF Plus (3627 KB)