Journal of Neuro-Ophthalmology, June 2002, Volume 22, Issue 2

Diffusion-Weighted Magnetic Resonance Imaging

Diffusion-weighted magnetic resonance imaging is a specialized technique that measures the degree of diffusion of water molecules within extracellular space and between intracellular and extracellular space. Diffusion-weighted imaging signal is high (bright) when diffusion is restricted, as occurs in cytotoxic damage from ischemia, inflammation, trauma, or tumor. This technique, now available on most magnetic resonance imaging units, is especially helpful in detecting early ischemic stroke and multiple sclerosis and in differentiating arachnoid cyst from epidermoid tumor and brain abscess from neoplasm.

STATE OF THE ART Diffusion- Weighted Magnetic Resonance Imaging Suresh K. Mukherji, MD, Thomas L. Chenevert, PhD, and Mauricio Castillo, MD Diffusion- weighted magnetic resonance imaging is a spe­cialized technique that measures the degree of diffusion of water molecules within extracellular space and between in­tracellular and extracellular space. Diffusion- weighted im­aging signal is high ( bright) when diffusion is restricted, as occurs in cytotoxic damage from ischemia, inflammation, trauma, or tumor. This technique, now available on most magnetic resonance imaging units, is especially helpful in detecting early ischemic stroke and multiple sclerosis and in differentiating arachnoid cyst from epidermoid tumor and brain abscess from neoplasm. ( JNeuvo- Ophthalmol 2002; 22: 118- 122) Diffusion- weighted imaging ( DWI) is a specialized magnetic resonance imaging ( MRI) technique that is particularly sensitive to the normally disorganized ( ran­dom, " Brownian") microscopic motion of water molecules ( 1). In the brain, signal intensity on DWI depends primarily on the ability of water molecules to move within and be­tween the intracellular and extracellular spaces via perme­able cell membranes. Thus far, DWI has been used primar­ily for the detection of acute cerebral infarction, which im­pairs water mobility. However, there are other applications in which DWI facilitates diagnostic insight. TECHNICAL CONSIDERATIONS The increased use of DWI arises from advances in newer MR units that allow them to perform echoplanar im­aging. These units contain advanced gradients capable of very fast switching times that permit each image to be ac­quired in approximately one tenth of a second. As a result, a whole brain DWI series can be obtained in seconds rather than minutes. A DWI pulse sequence incorporates two ad­ditional gradient pulses compared with a conventional spin Departments of Radiology ( SKM, TLC) and Otolaryngology Head Neck Surgery ( SKM), University of Michigan Health System, Ann Arbor, Michigan, USA. Department of Radiology, University of North Carolina School of Medicine ( MC), Chapel Hill, North Carolina, USA. Address correspondence to Suresh K. Mukherji, MD, Department of Radiology, 1500 East Medical Center Drive, Ann Arbor, MI 48109, USA; E- mail: mukherji@ umich. edu echo imaging sequence. These additional pulses allow mea­surement of microscopic motion of water molecules. The first gradient pulse alters the magnetization of each water molecule based on the position of the molecule. The second gradient pulse completely removes this alteration as long as the water molecule remains at its original location and does not move. However, any molecular movement over the interval when these pulses are applied will lead to incom­plete restoration of the water magnetization. That is, mo­lecular motion in the presence of the " diffusion gradient pulses" leads to a loss of MRI signal. The sensitivity to dif-fusional signal loss is controlled by choice of an image ac­quisition parameter called the " b- value." For neuroimag-ing, a b- value of about lOOOsec/ mm2 is often used because it offers good diffusion- dependent contrast and image qual­ity ( 2,3). For a given b- value, the amount of signal loss re­flects molecular mobility. Greater mobility, or diffusion, yields less MRI signal. Conversely, a relatively strong MRI signal implies more stationary molecules. Figure 1 illus­trates the stark signal loss in fluid tissues as the b- value is increased from zero ( Fig. \ A) to 1000 sec/ mm2 ( Fig. IB). Molecular motions that have no directional depen­dence are termed " isotropic". For isotropic media, the spe­cific direction of the applied diffusion gradient is of no con­sequence. In some tissues, however, water mobility does vary with direction. These tissues are called " anisotropic". White matter, for example, is highly anisotropic since water mobility parallel to the white matter fiber tracts is substan­tially greater than perpendicular to those tracts. In anisotro­pic media, the direction of the applied diffusion gradient is an important consideration. Thus, by control of the diffu­sion gradient direction, one probes structural directionality of the imaged object ( Fig. 2). In " apparent diffusion coeffi­cient" ( ADC) maps, regions that contain high water mobil­ity, such as the ventricular cavities, show up as bright rela­tive to regions of low water mobility, such as brain tissue. Modern clinical diffusion imaging systems provide several representations of diffusion properties to accentuate or sup­press directional information by appropriate mathematical combination of several directional data sets. As shown in Figure 2, the average of diffusion maps from three orthogo­nal directions ( ADCR/ L, ADCA/ P, and ADCS/ 1) yields a dif­fusion map that, by design, does not exhibit complex fiber tract patterns. This " average" diffusion coefficient map is / i # , pOL10.1097/ 01. WNp, 0. QQ0019698.15867.. 57 . J. Neuro- Ophthalmol. Vol. 22, No. 2,2Q02 , Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. STATE OF THE ART JNeuro- Ophthalmol, Vol. 22, No. 2, 2002 FIG. 1. The effect of changing the acquisition parameter " b value" on diffusion- weighted magnetic resonance image signal intensity. A: Diffusion- weighted magnetic resonance image with a b- value = 0 sec/ mm2 is similar to a T2- weighted image with high signal within the ventricles { arrow) and within a glioma { arrowhead). B: DWI with a b- value = 1000 sec/ mm2 shows signal loss within the ven­tricle { arrow). The loss of signal within the glioma { arrow­head) indicates that the glioma contains fluid. the best single quantitative representation of diffusion in the object. The diffusion- weighted image represents the raw signal strength recorded in the presence of diffusion gradi­ents. As such, the diffusion- weighted image exhibits high mobility areas as dark relative to water confined by tissues. Note that the relative contrast of a diffusion- weighted im­age is nearly reversed relative to an ADC map. The utility of the diffusion- weighted image representation is that tissues with lower restricted diffusion are depicted as bright and with greater conspicuity than on ADC maps ( Fig. 3). DWI clearly distinguishes between two types of " ex­cess water," or edema, in the central nervous system. Va­sogenic edema, caused by incompetence of blood vessels or breakdown of the blood- brain barrier, is an increase in the amount of extracellular water; it is commonly seen along white matter tracts. Cytotoxic edema implies damage to cell membranes, reflecting a failure of the sodium- potassium pump needed to exclude extracellular water from the cell ( 5,6). This type of edema is typically seen in regions of is­chemia or infarcts. On T2- weighted images, both types of edema show up as high ( bright) signal. On diffusion-weighted images, vasogenic edema appears relatively low ( dark) signal intensity, whereas cytotoxic edema results in high ( bright) signal intensity. Occasionally, very high sig­nal on T2- weighted images will result in high signal on dif­fusion- weighted images ( T2 " shine- through"). ADC maps are useful in determining whether the increased DWI signal is due to restricted diffusion or T2 shine- through effects. However, in routine clinical imaging, ADC maps are often not necessary, as this distinction can be made by simple visual inspection ( 4). That is, if the T2 signal is not very bright and the diffusion- weighted image is very bright, the DWI high signal likely reflects restricted diffusion. CLINICAL APPLICATIONS Cerebral Infarction DWI has made its greatest contribution to the evalu­ation of acute cerebral infarction. DWI may demonstrate most central nervous system infarctions within minutes of clinical onset. This feature makes it a valuable tool to deter­mine whether thrombolytic therapy should be performed. Cytotoxic edema occurs within a few minutes of a critical decrease in cerebral blood flow. This change is picked up as high signal on DWI as early as 3 minutes after clinical symptoms occur. Most infarctions will show abnor­malities within 45 minutes after onset ( 7). The sensitivity of DWI is said to be close to 100% if 2 hours have elapsed after a brain infarction. In general, sensitivity and specificity of DWI to cytotoxic edema is close to 95%, making it one of the most reliable noninvasive techniques. The presence of cytotoxic edema ( and thus high sig­nal DWI) does not necessarily imply irreversible cell death. Zones of brightness on diffusion- weighted images closely correspond to hypoperfused brain, but if perfusion is rees­tablished, DWI abnormalities may reverse. DWI abnormalities are maximal 24 hours after a stroke and start to resolve within 7 to 14 days thereafter ( Fig. 3). However, disappearance of high DWI signal does FIG. 2. Effect of diffusion gradient direc­tion on signal intensity. A: Gradient di­rection right to left (" ADCRL"). The fibers in the optic radiations run perpendicular to this direction and are seen as " black" { arrowhead). B: Gradient direction an­teroposterior (" ADCA/ P"). Because the optic radiations run perpendicular to this direction, they now have increased sig­nal. C: Gradient direction superoinferior (" ACSS/|"). The fibers of the genu of the corpus callosum, which run from right to left, show signal loss { arrowhead). D: A composite of images in A, B, and C (" ADCo"). It has no specific gradient directionality, so the image is very homogeneous. E: The standard diffusion- weighted magnetic resonance image (' routinely interpreted. DWL -,"). This is the image that is 119 Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited JNeuro- Ophthalmol, Vol. 22, No. 2, 2002 STATE OF THE ART FIG. 3. Diffusion- weighted magnetic resonance image in cerebral infarct. A: Diffusion- weighted magnetic resonance image shows high signal intensity in the left frontal lobe ( arrowhead) in an acute infarct. B: The corresponding ap­parent diffusion coefficient map shows reduced signal in­tensity, which indicates restricted diffusion and confirms the presence of an acute infarct. C, D: Repeat diffusion-weighted magnetic resonance imaging in the same patient 14 days later shows loss of the signal on diffusion- weighted magnetic resonance image ( C) and apparent diffusion co­efficient map ( D). not imply a reversion to normal cell function. In many pa­tients, conventional MRI will show an infarct. This phe­nomenon, known as " pseudonormalization," ( 2) is caused by cellular necrosis, which produces below normal signal intensity on diffusion- weighted images. Infarcts result in cell and axonal loss, reactive astrocytosis ( gliosis), and en-cephalomalacia. The encephalomalacia includes cysts that contain water whose motion is not restricted. Therefore, chronic infarcts are hypointense on diffusion- weighted im­ages. The increased translational movement of water mol­ecules in the encephalomalacic cysts results in increased spin dephasing and signal loss. This is similar to what is seen in arachnoid cysts ( see " Arachnoid Cyst Versus Epi­dermoid Tumor"). On ADC maps, chronic infarctions ac­tually appear as areas of increased diffusion, that is, high signal intensity. Chronic infarctions appear smaller on dif­fusion- weighted images than on fast spin echo T2- weighted images ( 8,9,10,11,12). Arachnoid Cyst Versus Epidermoid Tumor DWI is helpful in distinguishing between arachnoid cysts and epidermoid tumors ( Fig. 4). These lesions may have similar features on Tl and T2 MRI weighted images and do not enhance. Differentiation between them is straightforward with DWI. Because arachnoid cysts tend to contain pulsating cerebrospinal fluid ( CSF), their signal in­tensity is low on high b- value DWI ( and high on ADC maps). By contrast, epidermoid tumors are solid masses whose water mobility is relatively low. Therefore, signal is higher on DWI and lower on ADC maps than that of the surrounding CSF. Abscess DWI has proved useful in the preoperative diagnosis of cerebral abscesses ( 13). In patients with no antecedent history, it is helpful to divide enhancing lesions into infec­tions and tumors ( 14). Ring- enhancing lesions contain a FIG. 4. Diffusion- weighted magnetic resonance imaging differentiation of arachnoid from epidermoid cyst. A: Axial T2- weighted magnetic resonance image of arachnoid cyst shows high signal intensity mass ( arrow). B: Axial T2- weighted magnetic resonance image of epidermoid cyst shows high signal intensity mass ( arrow) similar to that of the arachnoid cyst. C: Apparent diffusion coefficient map of the arachnoid cyst has high signal ( arrow). D: Apparent diffusion coefficient map of the epidermoid cyst has low signal ( arrow). n 120 . , „.. , „ . . © 2002 Lippincott Williams & WUkins , Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. STATE OF THE ART JNeuro- Ophthalmol, Vol. 22, No. 2, 2002 FIG. 5. Diffusion- weighted magnetic reso­nance imaging of moderate grade glioma. A: Axial T2- weighted image shows a high signal mass in the left basal ganglia { arrow). B: Corresponding diffusion- weighted mag­netic resonance image shows increased signal within the mass { arrow). From this image alone, it is unclear whether the in­creased signal is due to restricted diffusion or " T2 shine- through" effect. C: The ap­parent diffusion coefficient map shows the mass to be heterogeneous. The central area { short arrow) has reduced signal inten­sity, which indicates restricted diffusion, whereas the periphery { long arrow) has high signal due to T2 " shine- through" effect. center that represents necrosis and/ or cysts. In abscesses, the necrotic area harbors a complex matrix of proteins, in­flammatory cells, cellular debris, and bacteria in high vis­cosity pus. The water molecules are bound to carboxyl, hydroxyl, and amino acid groups on the surface of macro-molecules ( 15). These characteristics contribute to re­stricted Brownian motion and result in increased signal intensity on DWI and low signal intensity on ADC. By con­trast, the necrotic center of tumors contains a less viscous material composed mostly of cellular debris, serous fluid, blood products, and relatively fewer inflammatory cells. This less complex environment allows water molecules a greater degree of translational motion than occurs in ab­scesses. Thus, the center of tumors is often of relatively low signal intensity on diffusion- weighted images and shows a high signal intensity ADC. In addition, the blood products, which may not be obvious on conventional magnetic reso­nance images, result in significant effects on diffusion-weighted images that contribute to low signal intensity. To avoid contamination of T2 shine- through when evaluating a potential abscess, review of ADC maps is helpful. Multiple Sclerosis The breakdown of the myelin sheath in acute multiple sclerosis ( MS) plaques reduces the motion of the water mol­ecules in the extracellular space. Because anisotropy is rela­tively reduced due to the loss of organization of the white matter fibers, there will be an increase in signal intensity on DWI and a lowering on ADC ( 1). DWI may measure im­provement in molecular motion as a response to therapy before signal changes reverse on conventional T2W or FLAIR images. DWI may also be helpful in dating MS plaques. As with old infarctions, chronic MS plaques show low signal intensity in DWI and high signal on ADC. Primary Brain Tumor DWI and ADC values have the potential to differen­tiate the components of a tumor and assess overall cellular-ity ( Fig. 5). Studies ( 16- 18) have shown that ADC values are generally low in solid/ cellular tumors relative to necrotic/ cystic tumors. These observations have also lead investigators to apply ADC maps to quantify therapy-induced necrosis. The use of DWI as an indicator of therapy response has been validated with animal models and in pre­liminary human studies ( 19- 21). REFERENCES 1. Gray L, MacFall J. Overview of diffusion imaging. MRI Clinics of North America 1998; 6: 125- 38. 2. Provenzale JM, Sorensen AG. Diffusion- weighted MR imaging in acute stroke: theoretical considerations and clinical applications. AJR 1999; 173: 1459- 67. 3. Castillo M, Mukherji SK. Practical applications of diffusion mag­netic resonance imaging in acute cerebral infarction. Emergency Radiology 1997; 4: 249- 54. 4. Chong J, Lu D, Aragao F, et al. Diffusion- weighted MR of acute cerebral infarction: comparison of data processing methods. AJNR Am JNeuroradiol 1998; 19: 1733- 9. 5. Provenzale JM, Engelter ST, Petrella JR, et al. Use of MR exponen­tial diffusion- weighted images to eradicate T2 " shine- through" ef­fect. AJR 1999; 172: 537- 9. 6. Sevick RJ, Kanda F, Mintorovitch J, et al. Cytotoxic brain edema: assessment with diffusion- weighted MR imaging. Radiology 1992; 185: 687- 90. 7. Sorensen AG, Wu O, Copen WA, et al. Human acute cerebral is­chemia: detection of changes in water diffusion anisotropy by using MR imaging. Radiology 1999; 212: 785- 92. 8. Lefkowitz D, LaBenz M, Nudo SR, et al. Hyperacute ischemic stroke missed with diffusion- weighted imaging. AJNR Am J Neu-roradiol 1999; 20: 1871- 5. 9. Wang PY, Barker PB, Wityk RJ, et al. Diffusion- negative stroke: a report of two cases. AJNR Am JNeuroradiol 1999; 20: 1876- 80. 10. Castillo M, Mukherji SK, Isaacs D, et al. Cerebral infarctions: evaluation with single axis versus trace diffusion- weighted MR im­aging. AJR 2000; 174: 853- 7. 11. Arbelaez A, Castillo M, Mukherji SK. Diffusion- weighted MR im­aging of global cerebral anoxia. AJNR Am JNeuroradiol 1999; 20: 999- 1007. 12. Castillo M, Mukherji SK. Early abnormalities related to postinfarc­tion Wallerian degeneration: evaluation with MR diffusion-weighted imaging. JCAT 1999; 23: 1004- 7. 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 1998; 171: 1487- 90. 121 Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited JNeuro- Ophthalmol, Vol. 22, No. 2, 2002 STATE OF THE ART 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 Neurora-diol 1999; 20: 1252- 7. 15. Castillo M. Imaging brain abscesses with diffusion- weighted and other sequences. AJNR Am JNeuroradiol 1999; 20: 1193- 4. 16. Tien RD, Felsberg GJ, Friedman H, et al. MR imaging of high- grade cerebral gliomas: value of diffusion- weighted echoplanar pulse se­quences. AJR 1994; 162: 671- 7. 17. Eis M, Els T, Hoehn- Berlage M, et al. Quantitative diffusion MR imaging of cerebral tumor and edema. Acta Neurochir Supplemen-tum 1999; 60: 344- 6. 18. Sugahara T, Korogi Y, Kochi M, et al. Usefulness of diffusion-weighted MRI with echo- planar technique in the evaluation of cel-lularity in gliomas. J Mag Res Imaging 1999; 9: 53- 60. 19. Chenevert TL, McKeever PE, Ross BD. Monitoring early response of experimental brain tumors to therapy using diffusion magnetic resonance imaging. Clin Cancer Res 1997; 3: 1457- 66. 20. Zhao M, Pipe JG, Bonnett J, et al. Early detection of treatment re­sponse by diffusion- weighted 1H- NMR spectroscopy in a murine tumor in vivo. Br J Cancer 1996; 73: 61- 4. 21. Chenevert TL, Stegman LD, Taylor JM, et al. Diffusion magnetic resonance imaging: an early surrogate marker of therapeutic effi­cacy in brain tumors. J Nat Cancer Inst 2000; 92: 2029- 36. _ 122 _ „ , . © 2002 Lippincott Williams & WUkins , Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.