Diffusion-Weighted Imaging

Diffusion-weighted imaging is a highly sensitive technique for the detection of acute cerebral infarction.[15,29] It confirms hyperacute ischemic damage within the brain before changes are apparent on conventional MR sequences.[29] This is made possible by the ability of diffusion-weighted imaging to detect water movement within the brain parenchyma. Less water movement occurs within a region of severely ischemic brain than in normal brain parenchyma. Diffusion-weighted imaging can detect this difference, allowing the identification of brain tissue that has undergone acute ischemic cellular damage.

 Figure 4. Anisotropic diffusion-weighted images obtained from the same anatomical level. For each image, the diffusion gradient has been applied in a different direction, leading to differences in white matter diffusion (anisotropy). (A) Diffusion gradient applied to the superior-inferior plane; white matter tracts coursing in this direction, such as the posterior limb of the internal capsule, are dark (arrow) indicating fast water diffusion. (B) Diffusion gradient applied to the right-left plane; the anterior corpus callosum is dark (arrow) due to increased diffusion along the tract. (C) Diffusion gradient applied to the anterior-posterior plane; the optic radiations are dark (arrow) due to their anterior-posterior orientation.

 Brownian Motion. To understand diffusion-weighted imaging, it is necessary to appreciate a normal physical process that affects water molecules. Water within the brain parenchyma is continually moving by Brownian motion. This random motion causes molecules to move further away from their original position with time, a process called diffusion. The amount of Brownian motion that a water molecule undergoes depends both on temperature and the presence of physical barriers. Most brain parenchymal water is located in the extracellular space, where few barriers exist. In this location Brownian motion is rapid and leads to significant water diffusion. Much less water is found within the cell, where it is more restrained by the presence of both cell membrane and organelles, leading to less water diffusion.

 

Technique. The amount of diffusion arising from Brownian motion can be detected using a diffusion-weighted MR sequence. Tissues containing a high amount of water diffusion are dark on diffusion-weighted imaging, whereas tissue without diffusion is bright. MR imaging detects the presence of hydrogen protons, making it highly sensitive to the presence of water. Diffusion-weighted imaging is a special type of MR imaging, which not only detects the presence of water but also whether it is stationary or moving. This feature is made possible by the addition of two magnetic diffusion gradients to a MR pulse sequence. These gradients are applied at different times within the pulse sequence and exactly opposite in their actions. If water is stationary, the second gradient exactly cancels the first, leaving no lasting effect. In this case, all the water protons are included in the MR signal, which is therefore bright. Water protons that move during the time between the gradients cannot be included in the MR signal because the second gradient cannot negate the effects of the first, unless the protons are in the same position. In this case, some of the MR signal is lost and therefore appears darker.

These diffusion gradients are commonly applied separately in three different directions: superior-inferior, anterior-posterior, and right-left to allow the separate detection of water movement in these planes (Fig. 4).

 t is important to perform such multidirectional diffusion imaging because white matter exhibits a phenomenon called anisotropy, whereby diffusion varies by direction. Diffusion is faster along the direction of myelinated white matter tracts than it is perpendicular to them. Thus, when a diffusion gradient is applied along the direction of a white matter tract, the signal on diffusion-weighted imaging is darker, indicating more rapid water diffusion in this direction. In contrast, gray matter does not exhibit anisotropy and diffusion is equal in all directions. Anisotropic images can be combined to form an isotropic image, which contains diffusion information from all planes.

 Figure 6. Apparent diffusion coefficient (ADC) maps of a 72-year-old male with right-sided weakness. (A) Diffusion-weighted image showing signal hyperintensity in the territory of the left middle cerebral artery consistent with acute cerebral infarction. (B) The infarcted area appears as an area of low-intensity signal on the ADC map. The apparent diffusion coefficient of the outlined area is decreased 65% from normal values.

 If more detailed directional information is required, tensor diffusion imaging can be performed in which a diffusion gradient is applied along six separate axes. This process generates very detailed information but at the expense of both extra imaging and post-processing time. Although tensor imaging may be useful for detecting white matter pathologies, it is unnecessary for imaging acute strokes.

For routine diagnostic purposes, it is easiest to interpret diffusion information on a diffusion-weighted image where areas of decreased diffusion appear as hyperintense signal. It is also possible to generate absolute values for the amount of diffusion within each MR voxel. This value is termed the apparent diffusion coefficient (ADC), and the value in each voxel can be viewed on an ADC map. An area of decreased diffusion with a low ADC would appear dark on an ADC map but bright on a diffusion-weighted image (Fig. 6).

 Figure 7. Diffusion-weighted imaging can detect hyperacute cerebral infarction before conventional magnetic resonance sequences. In this subject, signal hyperintensity is clearly seen on (A) diffusion-weighted imaging at a point when (B) the changes in T2-weighted signal are just becoming visible.

 ADC maps have the advantage that they only contain diffusion information and remove background MR signal resulting from T2 and proton density.

Detection of Acute Cerebral Infarction. Diffusion-weighted imaging detects a local decrease in water diffusion in areas of acute cerebral infarction. It appears as an area of increased signal and occurs before T2-weighted changes become apparent (Fig. 7).

 Figure 8. Diffusion-weighted imaging can also detect infarction within the brain stem. Infarction of the left medulla is most clear on a (A) diffusion-weighted image and (B) less well-defined on a T2-weighted image in a patient with lateral medullary syndrome.

 The exact cellular mechanism that decreases water diffusion is controversial. However, it is thought to be partly due to a movement of water from the extracellular space into the cell due to developing cytotoxic edema.[18] After severe or prolonged ischemia, reduced arterial flow leads to a depletion of tissue adenosine triphosphate (ATP) and the function of the Na+-K+ ATPase pump declines. Cellular influx of sodium and calcium follows, attracting water from the extracellular space into the cell. Water diffusion is more restricted within the cell than it is in the extracellular space, overall decreasing tissue diffusion.

 

Diffusion-weighted imaging is both sensitive and specific for acute cerebral infarction. Sensitivities of 88 to 98% [15,29] have been recorded, limited by occasional false-negative examinations caused by artifact or small lacunar infarcts. This technique is also useful for detecting brain stem infarction (Fig. 8).

 Its use in neonates remains more uncertain. Although diffusion-weighted imaging has proven sensitive for detecting global hypoxic episodes affecting the cortical watershed areas, it has been less successful in detecting deep gray matter insults (Fig. 9).[7] This characteristic may reflect differences in the normal water concentration of neonatal brains compared to adults.

 Figure 10. An 87-year-old woman presented with the acute onset of confusion. (A) T2-weighted image demonstrates multiple areas of high-intensity signal in the left thalamus, putamen, and temporal lobe. Whether these changes are due to acute cerebral infarction or whether they represent chronic damage is unclear. (B) In contrast, diffusion-weighted image demonstrates a focal area of high-intensity signal in the left thalamus, indicating that only this lesion is acute.

 The specificity of diffusion-weighted imaging is similarly high, at 95%.[15] Occasional false-positive examinations result from recent seizure activity, Herpes encephalitis, or nonstroke cellular death. To prevent erroneous interpretation of diffusion-weighted images, it is important to examine T2-weighted images before diffusion-weighted images are interpreted because diffusion-weighted imaging is T2-weighted, and T2 signal hyperintensity from nonischemic causes can shine through onto the diffusion-weighted image.

Diffusion-weighted imaging is also a useful technique for distinguishing acute from chronic cerebral infarction (Fig. 10). This distinction is often difficult to make on conventional MR sequences, but it is important for patient management.

 Time Course of Diffusion Changes. Changes in diffusion occur rapidly after the onset of severe ischemia. Animal studies have detected a signal increase on diffusion-weighted images within 2.5 minutes of severe ischemia.[18] In humans, the timing of the first change on diffusion-weighted imaging is less clear due to the inherent delay in imaging stroke patients. Changes, however, have been detected as soon as 39 minutes after symptom onset.[35] The intensity of the signal increases up to 48 hours after insult (Fig. 11).32 The increase indicates a progressive decline in local water diffusion, reflecting increasing water influx into the ischemic cell, and may indicate a target for treatment. After signal intensity peaks, brightness gradually decreases until the abnormality can no longer be detected 4 to 14 days after ictus.[5,22,32] This decline in signal brightness reflects an increase in water diffusion back toward normal values but does not indicate recovery of the ischemic cells. It is caused by the leakage of water, which collects in the extracellular space as vasogenic edema, through the damaged BBB. Consequently, water diffusion increases and ADC is elevated.

 The size of the area where the diffusion-weighted imaging signal is abnormal often increases during the first 24 hours after symptom onset and may continue to do so beyond 48 hours in some subjects (Fig. 12).[2] This finding suggests a progression in tissue damage in the hours after ictus and reflects an important target for both current and future treatments for acute cerebral ischemia.

Does Diffusion-Weighted Imaging Detect Reversible Cerebral Ischemia? Animal studies have indicated that diffusion-weighted imaging signal changes induced by ischemia may be reversible, either partially or fully, if perfusion is restored within an hour of the insult.[17] In areas that show reversibility, ischemic changes are less prominent and a less intense change in signal reflects a smaller decrease in water diffusion.[17] There may be a threshold value of water diffusion at which point ischemic lesions indicate irreversibly damaged tissue.

Whether similar changes in water diffusion induced by ischemia are also reversible in humans is uncertain. Although intravenous rt-PA has clinical value, no studies have examined whether changes occur in diffusion-weighted imaging or perfusion imaging with successful reperfusion. In most clinical studies, the appearance of a diffusion abnormality indicates a clinical deficit lasting more than 24 hours and consistent with cerebral infarction.[29] Changes on diffusion-weighted imaging may also be seen in subjects with a transient clinical deficit that lasts less than 24 hours.[12] Most of these subjects have evidence of cerebral infarction on subsequent conventional MR imaging. In a few, however, the follow-up MR image is normal. It is unclear whether diffusion-weighted imaging changes in these subjects are due to irreversible infarction, which has not been demonstrated by follow-up MR imaging, or to ischemia that has subsequently reversed. In most subjects, however, diffustion changes reflect irreversible tissue damage, with the area of signal change corresponding to the minimum area of infarction.

http://www.thebarrow.org/Education_And_Resources/Barrow_Quarterly/205119