Diffusion-weighted imaging in the evaluation of intracranial lesions

Seminarsin ULTRASOUND CTandHRI DECEMBER 2000

VOL. 21, NO 6 Advanced MR Imaging Techniques

D i f f u s i o n - W e i g h t e d I m a g i n g in the E v a l u a t i o n of I n t r a c r a n i a l L e s i o n s Mauricio Castillo and Suresh K. Mukherji Conventional magnetic resonance (MR) imaging is able to show pathology early on and to provide the radiologist with some degree of lesion characterization based on the relaxation time of different tissues. Many times, however, conventional MR imaging is not capable of depicting abnormalities at a time when early therapy may be successful, or of differentiating among different types of lesions before surgery. Diffusion-weighted imaging (DWI), a technique that is relatively new, is rapidly gaining popularity. Its increased use stems from the fact that many of the newer MR units are echo-planar capable. Although DWI may be obtained without echo-planar techniques, most DWI is now obtained by using gradients capable of very fast rising times. Echo-planar DWI may be obtained in a matter of seconds and, thus, is much less sensitive to bulk motion than other imaging techniques. Although DWI has been used extensively for the evaluation of acute cerebral infarctions, new uses for it are being explored constantly. In this article we address the nature of DWI and its use in the stroke patient as well as in other clinical situations where w e believe it is useful. "To understand motion is to understand nature."--Aristotle

Copyright © 2000 b y W.B. Saunders Company

MAGE CONTRAST ON diffusion-weighted imaging (DWI) is based on the microscopic motion of water molecules. Normally, this motion is disorganized (or random) and is referred to as Brownian or translational motion. 1 In the brain, contrast on DWI depends mostly on the water molecules located in the extracellular space. Two extra gradient pulses are applied to conventional spin-echo sequences and sensitize them to microscopic random water motion. After the initial gradient application, the water molecules will acquire phase shifts of their transverse magnetization. The second gradient application serves to refocus the stationary spins (similar to the action performed by the 180 ° pulse of the spin-echo sequence). Phase changes caused by motion prevent this refocusing and result in signal loss. Thus, after the application of diffusion-sensitizing gradients, normal water motion results in loss of signal intensity. Water

I

molecules that are moving freely will lose considerable signal intensity, resulting in apparent increased signal intensity from those regions in which water molecules remain relatively stationary. The amount of image signal depends in part on the strength of the diffusion gradient applied. 2 The parameter that reflects the duration and the strength of the diffusion gradient is called the B value. The B value is measured in sec/mm 2. A diffusion gradient may be applied in any or all of the three encoding direc-

From the Department of Radiology, University of North Carolina School of Medicine, Chapel Hill, NC. Address reprint requests to Mauricio Castillo, MD, CB # 7510, UNC-CH, Chapel Hill, NC 27599-7510. E-mail: casti llo @med. uric.edu Copyright © 2000 by W.B. Saunders Company 0887-2171/00/2106-0001 $10. 00/0 dei: l O.l O53/ul.2000.19939

Seminars in Ultrasound, C-g,and MRI, Vol 21, No 6 (December), 2000: pp 405-416

405

406

tions (slice, phase, and frequency). The motion of water will be affected by its surrounding structures, thus, the motion of water molecules in the white matter is in part restricted by the orientation of the neighboring axons. The phenomenon of directional preponderance of water motion is called anisotropy. Anisotropy is greater when measured perpendicular to the direction in which a diffusion gradient was applied. Depending on the direction in which the diffusion gradient was applied, different white matter tracts will have normal increased signal intensity (Fig 1). From a theoretical standpoint it is possible to confuse the normal brightness from these white matter tracts with pathology or for this normal hyperintensity to make it difficult to visualize underlying pathology. 3 To avoid this problem, the diffusion gradient may be applied in the three directions. By mathematically incorporating the information obtained from the application of the diffusion gradient in all three directions (x, y, and z), it is possible to eliminate artifacts caused by anisotropy and produce the so-called isotropic images. In these images, normal brightness from white matter tracts is nearly completely eliminated and hyperintensities nearly always reflect underlying pathology. These combined images map the average diffusion coefficient (the trace) in all three directions and, therefore, are called trace images. Trace images do not map the diffusion coefficient for each pixel, but rather produce images dependent on signal intensity. Most clinical magnetic resonance (MR) units are capable of producing DWI by using unidirectional diffusion gradients or trace imaging. To quantify the degree of water motion, apparent diffusion coefficient (ADC) maps are needed. 2 The term apparent is used as the origin for the motion of the water molecules. Because multiple ADCs exist in a sampled volume element, two or more B values are needed to measure them. At our institution we use three B values (0, 500, and 1,000 sec/mm 2) to generate ADC maps. Contrary to anisotropic and isotropic DWI, zones with restricted water motion will have a lower ADC value and appear as having low-signal intensity on these maps (Fig 2). In routine clinical imaging, ADC maps may not be necessary? To generate the diffusion-sensitizing pulses, strong and rapidly activating gradients are needed. 2 This is the reason why in many institutions DWI is performed with echo-planar-capable MR units.

CASTILLO AND MUKHERJI

Powerful gradients are capable of applying a strong diffusion gradient in a short period of time. As the echo time (TE) decreases, the signal-to-noise ratio of DWI increases, and this can only be accomplished with echo-planar imaging. In addition, the availability of rapid gradient switching found in echo-planar MR units makes them capable of applying diffusion gradients with multiple B values. With the use of echo-planar DWI artifacts from bulk motion, respiration and lack of cooperation are considerably reduced. At our institution, all DWI is performed by using echo-planar techniques. DWI may be performed in MR units that are not echo-planar capable, but these techniques will not be addressed in this article. Unfortunately, most DWI is hampered by artifacts. 2,3 The most prominent artifacts are caused by magnetic susceptibility effects and are seen at interfaces between air, bone, and brain. The iron contained in blood also results in significant artifacts. Metal surgical clips produce marked artifacts. Metal away from the brain, such as dental braces, may also degrade the images. Thus, many of these artifacts are commonly seen at the base of the skull and close to the convexity. Most artifacts caused by the voluntary or involuntary (including physiologic) motion are not a problem when echo-planar techniques are used to obtain DWI. Instability of the gradient system will result in areas of signal void. Anisotropy effects are not considered true artifacts but may be confused with pathology by less experienced observers. For example, if the diffusion gradient is applied in the slice direction, the splenium of the corpus callosum appears bright. If the diffusion gradient is applied in the phase direction, the cortical spinal tract will appear bright. When using anisotropic DWI, shinethrough effects from T2 relaxation may project on the diffusion image and result in artifacts. These artifacts may be eliminated by using trace DWI or ADC maps. 5 T2 shine-through is occasionally beneficial because it accentuates pathology by adding the T2 effects to the decreased translational water motion. This is particularly true when imaging infarctions involving the brainstem where the earliest finding is caused by restriction of anisotropy. TECHNICAL PARAMETERS With the currently used parameters at 1.5 T, most

diffusion images are of low spatial resolution. At

DWI OF INTRACRANIAL LESIONS

407

Fig 1. Appearance of normal white-matter tracts on DWI when applying the diffusion-sensitizing gradient in different directions. (A) With the diffusion gradient applied in the sagittal direction, the fibers of the splenium (arrows) of the corpus callosum are normally bright. (B) Diffusion gradient applied in the axial direction results in normal brightness in the internal capsules, particularly their posterior limbs (arrows), There is residual brightness in the center of the splenium of the corpus callosum. (C) With trace-diffusion imaging, only some parts of the cortex show hyperintensity. The white-matter tracts are relatively hypointense.

408

CASTILLO AND MUKHERJI

ADC maps is generally performed off-line in an independent console. PHYSIOLOGY OF WATER MOTION

Fig 2. ADC map in a stroke patient. Axial section from ADC map shows a marked hypointensity (reduced diffusion coefficient) in the territory (M) of the right middle cerebral artery. The normal brain is relatively hyperintensive.

our institution we use a single-shot technique, implying that the entire brain is studied during one acquisition. The slice thickness varies from 3 to 6 m m and there is a 20% interslice gap. The matrix is usually 128 × 128 and the field of view is 240 ram. The TE is 103 ms and the repetition time (TR) is 4,900 ms. Only one excitation is possible during image acquisition. Initially, we obtained a set of baseline images by using a low B (30) value. These are basically low-resolution T2-weighted images and serve for comparison with the high B value DWI. As we have gained experience, we do not obtain them anymore but used the fast-spin echo T2-weighted sequences for this purpose. Most of our DWI is generated as trace images and all three axes are displayed as a single set of images. Acquisition of single axis and trace images is approximately 5 and 16 seconds, respectively. We do not routinely obtain ADC maps. If ADC maps are needed (almost always for research purposes), B values of 0, 500, and 1,000 sec/mm2 are applied in the x, y, and z directions. Postprocessing of the

The normal loss of signal intensity in highstrength DWI is caused by loss of phase coherence within the volume element (voxel) studied. Most of the free water is normally found in the extracellular space. Water inside the cells is less free to move because of the complex environment of the cytoplasm and, thus, contributes less to normal signal loss than extracellular water. Two types of edema are commonly recognized. Vasogenic edema implies an increase in the amount of extracellular water and is commonly seen tracking along the white matter. Classically, vasogenic edema is seen as a result of tumors. Cytotoxic edema implies a failure of the sodium-potassium pump in the cell membrane leading to an influx of water into the cell, causing it to swell. 6 A decrease of the energydependent pathways is needed for cytotoxic edema to occur. On conventional and fast-spin echo T2weighted images, both types of edema are bright (Fig 3). On DWI, vasogenic edema is not well seen because the amount of water in the extracellular space is increased and results in theoretically more dephasing and signal loss. In cytotoxic edema the amount of extracellular water is proportionally reduced, swelling of the cells restricts the normal translational motion of extracellular water, and random motion of the relative increased intracellular water is limited because of the complex intracellular environment. Thus, cytotoxic edema results in significant signal gain caused by a relative lack of spin dephasing. This makes DWI an ideal method with which to visualize processes characterized by early onset of cytotoxic edema such as cerebral infarctions. It is in the early evaluation of cerebral infarctions that DWI has made its greatest impact. CEREBRAL INFARCTIONS

Because cytotoxic edema occurs in just a few minutes of a critical decrease in cerebral blood flow, DWI may show most acute infarctions. This is particularly helpful when thrombolytic therapy, which can only be used within a relatively short window of opportunity, is considered. Changes in DWI have shown up as early as 3 minutes and it is generally accepted that most infarctions will show abnormalities by 45 minutes after their onset7 The sensitivity of DWI is said to be close to 100% 2

DWI OF INTRACRANIAL LESIONS

409

Fig 3. Differentiation between vasogenic and cytotoxic edema by using DWI. (A) Axial postcontrast Tl-weighted image shows enhancing mass (arrow) in the left occipital lobe surrounded by hypointense edema. There is also edema (M) and intravascular enhancement in the left temporal lobe secondary to an acute middle cerebral artery occlusion. (B) FLAIR image shows infarction (c) with cytotoxic edema involving both gray and white matter. Vasogenic edema (v) secondary to the tumor has a finger-like configuration and respects the cortex. Both types of edema are equally bright. (C) DWI shows that only cytotoxic edema (c) from infarction is bright. Vasogenic edema is difficult to visualize in this image. (Courtesy of G. Fatterpekar, MD, New York, NY.)

hours after the ictus. Although we generally assume that the presence of cytotoxic edema (and, thus, positive DWI) implies tissue death, this is not always true. Zones of brightness on DWI closely correspond with hypoperfused brain, but if perfusion is reestablished, many of the abnormalities on DWI will resolve. Diffusion-weighted abnormalities are maximal 24 hours after an ictus and start to resolve 7 to 14 days thereafter (Figs 4 and 5). Parenchymal enhancement with gadolinium follows an opposite time course. That is, as diffusion abnormalities begin to resolve, enhancement begins to become more prominent. In humans with acute infarctions, the sensitivity and specificity of

DWI is close to 95%, making it one of the most reliable noninvasive techniques. DWI also shows that the infarcts are heterogenous. Contrary to the popular belief that MR imaging may not be sensitive for identification of the acute hemorrhage, DWI is very sensitive to the presence of blood products of any age (Fig 6). Within the same ischemic territory, ADCs will vary. In many patients the area of DWI abnormality is smaller than that seen on perfusion MR studies. This mismatched area is referred to as the penumbra. The penumbra is tissue at risk and most therapy protocols attempt to save it. In the penumbra, the perfusion is low but the energy-dependent

410

CASTILLO AND MUKHERJI

Fig 4. Acute infarction not clearly seen on T2-weighted image. (A) Fast-spin echo T2-weighted image shows many nonspecific white-matter abnormalities in a patient with left-sided weakness, (B) DWI shows multifocal brightness (arrows) in regions corresponding to acute infarction, Without DWI, one would not be able to identify the presence of the infarction, The ring artifact around the head is caused by chemical shift.

mechanisms of the cellular membrane are relatively preserved, indicating no significant cellular damage. The combination of DWI and perfusion imaging holds great promise in the evaluation of thrombolytic and neuroprotective therapies. Patients with large penumbras are at risk for progression of infarction and in this situation DWI may show an increase in the size of the infarct, particularly during the first week. The lack of abnormalities on DWI in the presence of clinical neurologic findings suggests a transient ischemic attack (TIA) and may be used to direct these patients toward more conservative treatment protocols. Recently, two case reports (a total of four patients) describe the absence of DWI abnormalities in patients with perfusion abnormalities who subsequently developed cerebral infarctions, s,9 This is a rare occurrence and we have not witnessed it in supratentorial abnormalities. As mentioned earlier, we believe

Fig 5. Acute infarction seen on T2-weighted images and DWI. (A) Fast-spin echo T2-weighted image shows abnormal brightness in the left basal ganglia (arrows), (B) Despite that the infarction in the left basal ganglia (B) was seen on A, this DWI makes it easier to visualize it and maps its extent with greater confidence,

DWI OF INTRACRANIAL LESIONS

Fig 6. Acute hemorrhagic infarction. (A) T2-weighted image shows infarction involving the distribution supplied by the right middle cerebral artery and acute hemorrhage (H) in the basal ganglia. (B) DWl shows an area of signal void (H) corresponding to hemorrhage caused by loss of signal as a result of magnetic susceptibility. The zone of cytotoxic edema (c) corresponding to the infarction is typically bright.

411

that DWI may not be as sensitive for posterior fossa infarctions as it is for cerebral hemispheric infarctions. It is important to understand that though diffusion abnormalities will almost always disappear by 2 weeks after the onset of ischemia, this does not imply a reversal to normal. In all of these patients, conventional MR imaging will show the infarct even when DWI has returned to normal. This phenomenon is called pseudonormalization. 2 We recently studied the accuracy of anisotropic versus isotropic DWI in the evaluation of the stroke patient. 1° In all of our patients, both sequences showed the infarcts equally. Because anisotropic images include the signal from T2 shine-through, some infarcts may be slightly more conspicuous and appear larger in them. Anisotropic imaging provides a more accurate delineation of the extent of the infarction. For practical purposes, both sequences may be used with nearly identical results. These results are also similar to those in a different article in which anisotropic images, isotropic images, and ADC maps were compared. In that study, ADC maps offered a significant advantage over the single axis and trace images. In a different article we studied the use of DWI in the evaluation of diffuse cerebral ischemia, n In diffuse cerebral anoxia, the initial computed tomography (CT) and MR imaging findings may be subtle and difficult to appreciate. The confident diagnosis of diffuse cerebral ischemic injury implies a dismal prognosis and may be used to guide therapy and counsel the patient's relatives. We found that acutely, DWI showed signal-intensity abnormalities in the cerebellum, deep gray matter nuclei, and cortex better than conventional MR imaging (Fig 7A). In the subacute period, some abnormalities remained in the gray matter but the white matter was prominently affected. In the chronic period, DWI became normal but conventional MR imaging showed the diffuse nature of the insult (Fig 7B, 7C). Similar results have been found in children. From a clinical standpoint, DWI is very helpful in the evaluation of ischemia in the brain of newborns. In newborns, the relatively large normal content of water and the lack of myelin maturation make cerebral infarctions difficult to identify by CT or conventional MR imaging. In these newborns, DWI clearly shows acute cerebral infarctions. We have also studied the use of DWI in the identification of early Wallerian degeneration3 2

412

CASTILLO AND MUKHERJI

Fig 7. Diffuse cerebral anoxia. (A) DWI obtained 1 hour of cardiopulmonary resuscitation for prolonged arrest shows brightness in basal ganglia, internal capsules, corpus callosum, cortex, and white matter of the forceps minor. (B) Patient in a vegetative state after cardiac arrest image month after the insult. DWI shows only slight brightness in the white matter of the corona radiata. (C) In the same patient, the T1-weighted image shows linear zones of brightness throughout the cortex and gray-matter structures in the midbrain. The findings presumably represent laminar necrosis and explain the patient's clinical condition.

DWI OF INTRACRANIAL LESIONS

Wallerian degeneration refers to antegrade degeneration of axons caused by either death of the neuron cell body or proximal axonal injury. The end result of this process is gliosis and demyelination. Most Wallerian degeneration affects the corticospinal white matter tracts as a sequela of large territorial infarctions involving the cerebral hemispheres (Fig 8). Other diseases that may result in Wallerian degeneration are tumors, hemorrhages, and primary white matter disorders such as multiple sclerosis. We were able to identify Wallerian degeneration in some, but not all patients, with large infarctions of the middle cerebral arteries. It is possible that for this specific task, ADC maps are needed. The recognition of Wallerian degeneration is twofold. First, it is important not to confuse it with extension of the infarction or a second infarction. In patients with multiple sclerosis, Wallerian degeneration should not be confused with additional plaques. The presence of Wallerian degeneration also implies further loss of function and persistent clinical disabilities. The quality of life is worse in patients with Wallerian degeneration than in those without it. To end this section, we briefly address chronic infarctions. The sequelae of infarctions is mainly cell and axona] loss, reactive astrocytosis (gliosis), and encephalomalacia. Malacia may be either microcystic or macrocystic. Regardless of the size of these cysts, they contain water whose motion is not restricted. All chronic infarcts are hypointense on DWI. The increased translation movement of water molecules in the malacic cysts results in increased spin dephasing and signal loss. A similar phenomenon is seen in the presence of arachnoid cysts. On

413

Fig 9. DWI in a patient with an acute vasculitis. DWI shows multiple bright areas, mostly subcortical in location, corresponding to infarctions secondary to primary angiitis of the central nervous system (PACNS),

ADC maps, chronic infarctions appear as areas of increased diffusion, that is, high-signal intensity. Chronic infarctions generally appear smaller on DWI and a better delineation is accomplished with fast-spin echo T2-weighted images. Lastly, DWI is extremely sensitive in patients suspected of having acute infarctions caused by vasculitis (Fig 9). In this situation, multiple small focal areas of hyperintensity on DWI are clearly seen. These abnormalities are much brighter than those expected in other multifocal processes such as multiple sclerosis (see later). ARACHNOID CYST VERSUS EPIDERMOID

Fig 8. Early Wallerian degeneration. After an acute left middle cerebral infarction, DWl shows focal bright area (arrow) in the region of the corticospinal tract coursing in the left cerebral peduncle.

We found DWI very useful in differentiating between arachnoid cysts and epidermoid tumors. It is well known that both of these masses may have similar features on T1- and T2-weighted images and that neither enhance after gadolinium is injected. Close attention to the signal intensity of these lesions on the proton density sequence and on high-resolution T2-weighted images (such as those obtained by using constructive interference in the steady state [CISS]) do, however, show features that allow one to differentiate them in the majority

414

CASTILLO AND MUKHERJI

purpose included only small numbers of patients. 14 In our practice, we use DWI for this purpose because it is easy and fast to obtain and, in most cases, will add another degree of certainty to the diagnosis. Ring-enhancing lesions contain a center that generally is assumed to represent necrosis and/or cysts. In abscesses, their necrosis is complex, harboring a complex matrix of proteins, inflammatory cells, cellular debris, and bacteria in high-viscosity pus. In addition, the water molecules in this environment are bound to carboxyl, hydroxyl, and amino acid groups on the surface of macromolecules. 15 All of these characteristics contribute to restrict the Brownian motion of water in the pus of abscesses and results in increased signal intensity on DWI and a low ADC (Fig 11). The central necrosis or cystic components found in tumors contain a less-viscous material composed mostly by cellular debris, serous fluid (often the

Fig 10. DWI in the diagnosis of an epidermoid tumor. (A) T2-weighted image of a bright lesion in the perimedullary cistern insinuating into the vallecula. The main differential diagnosis for this extra-axial lesion is epidermoid versus arachnoid cyst. (B) On DWi, the lesion (E) is brighter than the normal CSF (P) within the left side of the perimedullary cistern. The diagnosis is that of epidermoid tumor.

of instances. The use of DWI differentiation between these lesions is straightforward. Because arachnoid cysts tend to contain cerebrospinal fluid (CSF), their signal intensity is low on high-strength DWI. Some arachnoid cysts contain proteinaceous fluid or blood and Signal loss will not be as marked. Potentially, these rare arachnoid cysts may pose difficulties on DWI. We have not yet encountered this problem. Conversely, epidermoids are solid masses and their signal is higher than that of the surrounding CSF (Fig 10). ABSCESSES

DWI has proven to be useful in the preoperative diagnosis of cerebral abscesses. 13 It may help to distinguish nonspecific ring-enhancing lesions into those of an infectious nature from tumor produced by tumor (either secondary or primary). Until now, the articles reporting the use of DWI for this

Fig 11. DWI in the diagnosis of abscess. (A) Contrastenhanced Tl-weighted image shows nonspecific ring-enhancing lesion in the right occipital lobe. The relatively thin rim and extension into the atrium of the right lateral ventricle suggest an abscess, but a necrotic tumor cannot be completely excluded. (B) On this DWl the pus contained in the abscess (A) is very bright. (Courtesy of G. Fatterpekar, MD, New York, NY,)

DWI OF INTRACRANIAL LESIONS

sequela of hemorrhage), and fewer inflammatory cells. This less-complex environment permits the water molecules a greater degree of translational motion freedom when compared with abscesses. Thus, the center of tumors is often of relatively low-signal intensity on D W ! and shows a high ADC. In addition, the presence of blood products that may not be obvious on conventional MR imaging may result in significant susceptibility effects on DWI, contributing to the low-signal intensity present in tumor necrosis. To avoid contamination of T2 shine through when evaluating a potential abscess, trace D W I is recommended. MULTIPLE SCLEROSIS

Although D W I has been used in the evaluation of patients with multiple sclerosis (MS), we have not found a practical use for it. The breakdown of the myelin sheath, which is typically seen in MS plaques, alters the motion of the water molecules in the extracellular space. Because anisotropic is relatively restricted due to the loss of organization of the white-matter fibers, there will be an increase in signal intensity on D W I and lowering of the ADC. 1 Potentially, D W I may be used to show a larger amount of plaques, but from a clinical standpoint this may not be important unless the patient is enrolled in a drug-testing protocol. D W I may also be helpful to date MS plaques. As the lesions become chronic, malacia and gliosis ensue. These histologic features are not lesion dependent and the findings are similar in all chronic lesions regarding their underlying cause. Thus, similar to old infarctions, chronic plaques show lower signal intensity in D W I and an increased ADC. Wholebrain A D C measurements have also shown abnormal values in the regions of the brain that appear morphologically normal. These low A D C values are thought to reflect an increase in the size of the extracellular spaces and/or movement of T cells and macrophages into the white matter.

415 P R I M A R Y BRAIN T U M O R S

We have already addressed the use of D W I in the evaluation of arachnoid cysts, epidermoids, and abscesses versus cystic/necrotic tumors. In this section we address the use of D W I in the evaluation of primary intracranial tumors. ADC values may be used to differentiate the different components of a tumor and establish its margins. 16 Some investigators have shown that there are no significant differences in the ADC values between tumor and surrounding edema, but D W I clearly separates tumor from nonedematous normal tissue. ~7 This has also been our experience (data submitted to the American Journal of Neuroradiology for publication). D W I could potentially also be used to grade tumors38 As the number of cells increase (and, therefore, also the degree of malignancy) in the tumor, the extracellular spaces diminish. Limited size of the extracellular spaces and water motion restricted in the complex intracellular space should give rise to lower ADC values in highly malignant brain tumors. There is also evidence that ADC values tend to become more normal in tumors that respond to treatment. ~9 The long-term significance of that finding is not certain. We have also found that typical meningiomas have a higher A D C than the atypical or malignant ones. This is also a reflection of increased cellularity and decreased extracellular spaces. SUMMARY

D W I is now readily available in the majority of clinical MR imaging units and may be used without any time penalty. Although its use has focused on the early detection of cerebral infarctions, it may be helpful in many other clinical scenarios. It provides an added degree of tissue characterization not available with more conventional M R imaging techniques. There are still many potential applications of D W I that need to be explored.

REFERENCES 1. GrayL, MacFall JR: Overview of diffusion imaging. MRI 4. Chong J, Lu D, Arago F, et al: Diffusion-weightedMR of Clin N Am 6:125-138, 1998 acute cerebral infarction: Comparisonof data processing methods. AJNR Am J NeuroradioI 19:1733-1739, 1998 2. Provenzale JM, Sorensen GA: Diffusion-weighted MR 5. ProvenzaleJM, Engelter ST, Petrella JR, et al: Use of MR imaging in acute stroke: Theoretic considerations and clinical exponential diffusion-weightedimages to eradicate T2 "shineapplications. AJR Am J Roentgenol 173:1469-1467, 1999 through" effect.AJR Am J Roentgenol 172:537-539, 1999 3. Castillo M, Mukherji SK: Practical applications of diffu6. SevickRJ, Kanda F, MintorovirtchJ, et al: Cytotoxicbrain sion magnetic resonance imaging in acute cerebral infarction. edema:Assessmentwith diffusion-weightedMR imaging. RadiEmerg Radiol 4:249-254, 1997 ology 185:687-690, I992

416

7. Sorensen GA, Copen WA, Davis TL: Human acute cerebral ischemia: Detection of changes in water diffusion anisotropy by using MR imaging. Radiology 212:785-792, 1999 8. Lefkowitz D, LaBenz M, Nudo SR, et al: Hyperacute ischemic stroke missed with diffusion-weighted imaging. AJNR Am J Neuroradio120:1872-1875, 1999 9. Wang PYK, Barker PB, Wityk RJ, et al: Diffusionnegative stroke: A report of two cases. AJNR Am J Neuroradiol 20:1876-1880, 1999 10. Castillo M, Mukherji SK, Isaacs D, et al: Cerebral infarctions: Evaluation with single axis versus trace diffusionweighted MR imaging. AJR 174:853-857, 2000 11. Arbelaez A, Castillo M, Mukherji SK: Diffusionweighted MR imaging of global cerebral anoxia. AJNR Am J Neuroradio120:999-1007, 1999 12. Castillo M, Mukherji SK: Early abnormalities related to postinfarction Wallerian degeneration: Evaluation with MR diffusion-weighted imaging. J Assist Comput Tomography 23: 1004-1007, 1999 13. Kim YJ, Chang KH, Song IC, et al: Brain abscess and necrotic or cystic brain tumor: Discrimination with signal intensity on diffusion-weighted MR imaging. AJR Am J Roentgenol 171:1487-1490, 1998

CASTILLO AND MUKHERJI

14. Desprechins B, Stadnik T, Koerts G, et al: Use of diffusion-weighted MR imaging in differential diagnosis between intracerebral necrotic tumors and cerebral abscesses. AJNR Am J Neuroradiol 10:1252-1257, 1999 15. Castillo M: Imaging brain abscesses with diffusionweighted and other sequences. AJNR Am J Neuroradio120:11931194, 1999 16. Tien RD, Flesberg GJ, Friedman H, et al: MR imaging of high-grade cerebral gliomas: Value of diffusion-weighted echoplanar pulse sequences. AJR Am J Roentgenol 162:671-677, 1994 17. Eis M, Els T, Hoehn-Berlage M, et al: Quantitative diffusion MR imaging of cerebral tumor and edema. Acta Neurochir 60:344-346, 1999 18. Sugahara T, Korogi Y, Kochi M, et al: Usefulness of diffusion-weighted MRI with echo-planar technique in the evaluation of cellularity in gliomas. J Magn Res Imaging 9:53-60, 1999 19. Chenevert TL, McKeever PE, Ross BD: Monitoring early response of experimental brain tumors to therapy using diffusion magnetic resonance imaging. Clin Cancer Res 3:14571466, 1997