Handbook of Neuroimaging for the Ophthalmologist M Tariq Bhatti, Ilona Schmalfuss
INDEX
×
Chapter Notes

Save Clear


Basic principles of computed tomography (CT), magnetic resonance imaging (MRI), and vascular imaging1

  • • Introduction: page 1
  • • Computed tomography: page 4
  • • Magnetic resonance imaging: page 7
  • • Perfusion imaging: page 25
  • • Vascular imaging: page 26
 
Introduction
A number of imaging modalities are available for the evaluation of patients with ophthalmological symptoms, ranging from plain films (X-ray) and ultrasound to CT and MRI. In current practice, plain films are rarely performed and are primarily reserved for detection of radiopaque foreign bodies prior to obtaining an MRI. Ocular ultrasound is excellent for the evaluation of intraocular and orbital abnormalities but has very limited applications for intracranial lesions. In addition, the quality of an ultrasound scan is operator-dependent and affected by strong ultrasound reflectors such as air, calcifications, and bone, that obscure the view of deeper structures. Therefore, the majority of patients with neuro-ophthalmological symptoms requiring radiological evaluation undergo a CT or MRI examination.
The decision between obtaining a CT or MRI is primarily driven by the onset and progression of symptoms, clinical suspicion of disease, differential diagnosis, patient age, scanner availability, and patient factors (e.g. body habitus, weight, allergies, and patient's preference). In general, CT is preferred for the acutely sick patient because it is widely available, easily accessible, and facilitates rapid detection of significantly debilitating or life-threatening disease processes such as an intracranial hemorrhage, tension orbit, orbital fracture, or an intraorbital abscess. In contrast, MRI is typically utilized for patients with subacute to chronic symptoms, to further characterize a CT abnormality, and for specific indications such as optic nerve abnormalities, small vessel ischemic disease, demyelinating lesions, and posterior fossa or spinal cord diseases. In addition, MRI is the preferred imaging modality in children and young adults as well as in patients requiring repetitive scanning (e.g. for evaluation of hydrocephalus) to prevent or minimize 2CT radiation exposure and its biological effects. The advantages and disadvantages of MRI and CT also play an important role in the choice of the appropriate imaging modality. CT is superior in identifying acute blood products, while MRI is much better in the detection of old blood products (hemosiderin). Hemosiderin is invisible on CT. In contrast, calcifications are much easier appreciated with CT. Calcium can assume various signal intensities on standard MRI sequences depending on the compound composition and concentration, which may prevent its detection or result in misinterpretation of the findings (Figure 1.1). Significant differences can also be seen in the evaluation of bony and soft tissue structures between MRI and CT. In general, CT is better in the detection of small bony erosions while MRI is superior in the evaluation of the bone marrow. Both, CT and MRI, perform similarly when gross bony involvement is present. In contrast, MRI is superior in showing soft tissue detail, particularly when different sequences are combined. MRI is also superior to CT in visualizing cranial nerves.
It is important to understand the relative and absolute contraindications of MRI and CT imaging when choosing the appropriate study for an individual patient. Certain implanted devices and metallic foreign bodies are not MRI compatible and may preclude or allow MRI examination only under specific conditions. The MRI Safety website provides annually updated and comprehensive information on MRI compatible medical devices and implants. Claustrophobia is a relative contraindication to MRI that can be usually overcome by administration of sedative medication or general anesthesia.
zoom view
Figure 1.1: Visibility of calcifications on computed tomography (CT) and magnetic resonance imaging. Noncontrast enhanced axial CT image (a) through the basal ganglia demonstrates marked calcifications (arrows) of the globus pallidus bilaterally. These are not visible on the T2-weighted image (b) at the same level.
3Claustrophobia is very rarely an issue with CT as the scanner tube is much shorter and wider compared with the MRI bore. Overall, there are no absolute contraindications for CT; however, caution and a risk–benefit analysis are required in pregnant and pediatric patients to prevent or minimize radiation exposure. For a pregnant patient, the risk to the fetus is negligible when only the head and neck are scanned due to the fact that the radiation beam is remote from the fetus. Additionally, protection of the fetus can be provided by placement of a lead apron over the mother's body during the CT examination. Scanning of the mother's body requires more consideration but is rarely needed for ophthalmological patients. There are no absolute contraindications for MRI during pregnancy using up to 3T MRI scanners. To date, there has been no scientific evidence in humans to suggest increased risk of disability or hearing loss in children exposed to MRI in utero. In fact, MRI is routinely utilized in the imaging of a fetus when fetal disease or deformity is suspected on clinical or ultrasound examination. Gadolinium administration is, however, contraindicated in pregnancy as the Gadolinium chelates cross the placenta into the fetal circulation. The Gadolinium is excreted through the fetal kidneys into the amniotic fluid, and then swallowed by the fetus reentering the fetal circulation. This circular mechanism continues until delivery, and currently it is unknown what long-term effects such extended Gadolinium exposure has upon the fetus. In contrast, iodinated contrast injection is not contraindicated during pregnancy. There is no increased risk of thyroid dysfunction in the newborn.
Intravenous contrast administration is typically required for CT and MRI examinations in the assessment of neuro-ophthalmological disorders. Iodinated contrast agents are used in CT and Gadolinium based in MRI. Both, CT and MRI contrast agents, are contraindicated in patients with acute renal insufficiency because of the risk of permanent renal damage associated with iodinated contrast and the development of nephrogenic systemic fibrosis (NSF) with Gadolinium. The same concerns apply for patients with chronic renal failure undergoing an MRI but this is only partially true for CT. There is no contraindication for the administration of iodinated contrast in patients on chronic dialysis without any urine production because the kidneys are already severely and irreversibly damaged. Caution must be applied to chronic dialysis patients with some urine production as the administration of iodinated contrast might cause additional renal damage and lead to increasing frequency of dialysis. All chronic dialysis patients are also at risk of potential fluid overload following the administration of contrast. Therefore, the CT examination should be scheduled as close as possible to the time of dialysis.
Occasionally, preexisting allergies to contrast require modifications in the patient's imaging evaluation. Allergic reactions are more commonly seen with iodinated contrast than with Gadolinium. These allergies can be classified into the following categories:4
  • Mild (5–8% of patients): mild urticaria and hives, mild angioedema (face, neck, tongue). No treatment is usually required
  • Moderate (1% of patients): severe urticaria or hives, moderate angioedema (tongue, face, neck), short lasting hypotension with tachycardia, bronchospasm. Treatment is often required
  • Severe (very rare): severe, bronchospasmus, severe hypotension, laryngeal edema, mental status changes, cardiac arrest, and other life-threatening condition. Treatment is always required
The recurrence rate of allergic reactions ranges from 25% to 40%. Therefore, premedication with corticosteroids starting about 13 hours prior to contrast injection in cases of mild-to-moderate reaction is usually performed. The corticosteroid injection is often combined with an antihistamine agent that is given 1 hour prior to the contrast injection. In severe reactions, contrast administration is not advised as there is insufficient evidence that the premedication with corticosteroids is effective. Therefore, alternate imaging modalities should be considered. Nausea, vomiting, and sensation of warmth throughout the body represent physiological reactions to contrast and do not require any premedication.
 
Computed tomography
The current widely available multislice technology has revolutionized CT imaging. Multislice CT allows simultaneous acquisition of up to 320 images with a minimum slice thickness of 0.5–1.25 mm, depending on the scanner type. Multislice CT produces a volumetric data set of the entire head within a few seconds. Such scanning parameters not only reduce motion degradation but also allow reformations of the data in any desirable plane. Therefore, direct coronal imaging is not necessary and has been replaced by high-quality reformations that can correct for suboptimal positioning of the patient (Figure 1.2). Such reformations also allow the clinician to follow the course of a specific anatomical structure such as vessel or nerve, and avoid artifacts related to dental fillings that might compromise the view upon the orbits. A drawback of multislice CT is the production of a large number of images that are often very difficult for the clinician to review in its entirety. Some of the data sets consist of thousands of images, in particular when CT angiography (CTA) or CT perfusion is involved, and require the use of sophisticated Picture Archive Computer System (PACS) units that can manage the large data sets and allow fast reformation of the images into the desired slice thickness and plane(s). Such post-processing capabilities of the CT images are more demanding upon the reading physician, but at the same time allow individualization of the reformations based on patient's unique anatomy or pathology.
Although advancements in technology have improved the technical demands on producing CT images, there still remain some important logistical considerations for the CT technologist and radiologist. The majority of the anatomical structures related to vision are relatively small.5
zoom view
Figure 1.2: Correction of suboptimal patient positioning. Original axial computed tomography (CT) image (a) through the orbit demonstrates asymmetric appearance of the orbital structures, which is most apparent at the orbital apex and lateral orbital wall levels. The right superior orbital fissure is seen on the right (arrow) while no opening is present at the posterior orbit on the left side. Notice also the partial visualization of the anterior zygomatic arch on the right (*), which is not seen on the left. Axial, oblique reformation of the original CT data set (b) was obtained to correct for the suboptimal positioning of the patient. Both orbits are displayed at the optic nerve canal level (black arrowheads). The anterior zygomatic arches (z) are also visible to the same extent on each side. No significant differences in image quality are identified between the two images. This patient had a large medial orbital abscess (white arrowheads) on the left.
Therefore, it is critical to acquire high-resolution images to optimize the detectability of an abnormality. This is accomplished by choosing the smallest field of view possible. In an adult patient, a field of view of 16 cm is typically utilized for the orbits and 22 cm for the brain yielding a spatial resolution of about 0.3 and 0.45 mm, respectively. Magnification of CT images at the PACS station may seem to compensate for a too large scanning field of view, but in reality the resolution of the image remains the same. The magnification only displays each CT pixel element over more monitor kernels, and therefore does not improve the image resolution but rather results in slightly blurred images (Figure 1.3).
Contrast injection technique is another area of CT examination that requires special attention. The current CT scanners are so fast, that often the injected contrast does not reach the capillary and/or venous phase before the scan is completed. Sole enhancement of the arteries might be desirable when a CTA is requested (see ‘Vascular imaging’ section); however, it can lead to significantly compromised image quality of soft tissue or venous abnormalities. In such instances, a nonenhanced vein may be misinterpreted as thrombosed, an abscess mistaken for a noninfected fluid cavity (due to the lack of rim enhancement), or a lesion may not be visible at all (Figure 1.4). Therefore, it is important that the technologists adhere to the protocol prescribed for contrast injection rate and scan delay.
6
zoom view
Figure 1.3: Differences in image resolution. Axial computed tomography (CT) image obtained with a large field of view (a) displays both temporal bones. The large field of view image was zoomed in on the Picture Archive Computer System station (b) to better visualize the right middle ear. Notice the slightly blurred appearance of the bony edges when compared with the original image (a). Image (c) displays the same anatomical structures as image (b) with an optimized field of view to the right temporal bone yielding a higher resolution image. The stapes (between arrowheads) and the incudostapedial joint (arrow) are markedly better visualized in image (c) than on the other images (a or b).
zoom view
Figure 1.4: Enhancing lesion on early and delayed post-contrast enhanced computed tomography (CT) images. Coronal CT reformations through the cerebellum in the arterial phase (a) of contrast injection demonstrate no significant abnormalities, while the delayed contrast enhanced image (b) at the same level reveals an enhancing mass in the superior cerebellum on the right (arrows) caused by metastatic disease from a head and neck cancer.
In addition, the radiologists and ophthalmologists need to be aware of these issues and start their evaluation of the CT images by determining if the images were obtained in the correct contrast enhancement phase.
After the acquisition of the CT data, the images are reconstructed using a soft tissue algorithm, bone algorithm, or both. The bone 7algorithm most optimally demonstrates bone detail by mathematically enhancing the bony edges. This process provides very sharp images but at increased image noise that in turn decreases the contrast resolution of the soft tissues (Figure 1.5a). On the contrary, the soft tissue algorithm focuses on the nonosseous structures and displays them with smoother borders, lower noise, and higher contrast differentiation (Figure 1.5b). This type of post-processing can only be performed by the CT technologist at the CT scanner and is very different from changing the viewing settings from bone to soft tissue windows on the PACS system! The difference between these two concepts becomes most apparent when a ‘bone algorithm’ image displayed in bone window is compared with a ‘soft tissue algorithm’ image displayed in bone window. The sharpness of the bone is lost on the ‘soft tissue algorithm’ images and some smaller bony structures may not be visible (Figure 1.5c). In contrast, when a ‘bone algorithm’ image is displayed in soft tissue window, the image usually looks very ‘grainy,’ because of the increased level of noise, making it impossible to evaluate the soft tissues (Figure 1.5d). Hence, the radiologists and the ophthalmologists need to be aware of the limitations of each of these algorithms and request additional post-processing images as needed.
 
Magnetic resonance imaging
MRI is technically more challenging than CT because it is hampered by a longer acquisition time, greater prevalence of artifacts, and lower spatial resolution. However, MRI offers a wider range of imaging options with hundreds of available sequences. All these factors and some others as outlined below need to be considered and balanced together when protocolling an MRI examination to maximize its diagnostic image quality.
 
Imaging sequences
The natural magnetization induced in the human body by a magnet forms the physical basis of MRI. The degree of magnetization generated in an anatomical structure is contingent upon its proton density, which in turn depends on the number of hydrogen atoms. Therefore, fluids will have the highest magnetization properties while cortical bone the lowest. When a patient is placed into the magnet, all protons within the patient's body align longitudinally with the main field of the scanner (also known as longitudinal magnetization). The longitudinal magnetization cannot be measured. With the start of the sequence, the protons are ‘knocked out’ from their longitudinal alignment. This process generates the transverse magnetization that can be measured as an ‘echo’ that is expressed as signal intensity. After the ‘knockout,’ the protons immediately attempt to realign with the main magnetic field that takes varying amounts of time depending on tissue composition. As the protons realign with the main magnetic field, the transverse magnetization decreases and the longitudinal magnetization increases.8
The type of MRI sequence depends on:
  • Repetition time (TR): TR is defined as the time between the different ‘knockouts.’ The higher the TR, the longer the time protons have to realign with the main magnetic field and to recover the longitudinal magnetization
  • Echo time (TE): TE is the time between the ‘knockout’ and the reading of the echo. The shorter the TE, the greater the number of ‘knocked out’ protons that is reflected as higher transverse magnetization
    zoom view
    Figure 1.5: Differences in image appearance between bone and soft tissue algorithm. Axial computed tomography image in bone algorithm displayed in bone window (a) shows sharp appearance of the bony edges. Slightly displaced fracture is seen in the left medial orbital wall (arrowhead) and right nasal ridge (white arrow). In addition, a small fracture fragment is noted in a mid ethmoid air cell on the left (black arrow in image a). Only the left orbital fracture (arrowhead) is visible on the soft tissue algorithm image displayed in soft tissue window (b). The image (b), however, better demonstrates the soft tissues [globes (g) and extraocular muscles (….)]. It also reveals small amount of intraorbital fat [black arrow in image (b)] that is located in the same ethmoid air cell as the small fracture fragment [black arrow in image (a)] seen in image (a). The fracture fragment itself is difficult to appreciate in image (b). The image (b) is less grainy in appearance than the image (a).
    The soft tissue algorithm image displayed in bone window (c) is also smoother in appearance than image (a). On this image, only the right nasal ridge fracture (white arrow) and the medial orbital wall fracture on the left (arrowhead) are visible. The small fracture fragment is not seen, as the ‘smoothing’ effect of the soft tissue algorithm results in blurred appearance of such small bony structure. The display in bone algorithm makes distinction of the fat within the ethmoid air cell (black arrow in c) from other soft tissues difficult to appreciate when compared with its appearance on soft tissue window [black arrow in image (b)]. The same image in bone algorithm displayed in soft tissue window (d) is very grainy in appearance. This is caused by the inherently increased noise related to the bone algorithm technique. The image (d) best identifies the small foreign body (black arrowhead) underneath the left lid that has more appearance of a small air bubble on the other images.
    9
  • Flip angle (FA): FA expresses the degree of ‘knocking out’ of the protons. Spin echo sequences apply a 90° FA, while gradient echo sequences utilize <90° FA. The greater the FA, the higher the transverse magnetization and with it the greater the signal intensity. Therefore, spin echo images are of better quality than gradient echo images at the expense of markedly longer acquisition time. Gradient echo images are also prone to susceptibility artifacts that further influence the image quality
 
T1 sequence
T1-weighted images are generated by choosing short TR and short TE times. Such parameters enhance the differences between the various water-heavy tissues [such as cerebrospinal fluid (CSF) or vitreous] assuming a dark signal and fat based tissues appearing as very bright signal (Figure 1.6a). Normal muscle and brain parenchyma are represented by different shades of gray. T1-weighted images provide the best anatomic detail of all sequences and are often referred to as ‘anatomy scan.’
T1-weighted images can be obtained as a spin echo or gradient echo sequence. Utilization of the gradient echo technique shortens the acquisition time but is more prone to artifacts that may be mistaken for lesions. In particular, flow-related artifacts are diagnostically problematic as they may cause spontaneously bright signal on the unenhanced T1 images that can be mistaken for a thrombosed vessel or suggest enhancement that is actually not present. Therefore, spin echo T1-weighted imaging is often the preferred sequence with gradient echo sequences primarily reserved for less cooperative patients or when shorter imaging time is desired.
 
T2 sequence
T2-weighted images are generated by choosing long TR and long TE times. This leads to very bright signal of water-based tissues (Figure 1.6b). T2-weighted sequence best demonstrates pathology since many diseased tissues show edema related to accumulation of fluid. Therefore, the T2 image set is often called the ‘pathology scan.’
T2-weighted images can also be obtained as a spin echo or gradient echo sequence. In contrast to T1 images, gradient echo technology is preferred for T2-weighted imaging because the spin echo sequence has significantly longer acquisition time predisposing the images to motion degradation (Figure 1.6b). The difference between the spin echo and gradient echo T2 images is best seen with adipose tissue, which assumes a dark signal on the spin echo and a bright signal on the gradient echo T2-weighted sequences.
 
Fluid-attenuated inversion recovery (FLAIR) sequence
FLAIR-weighted images are similar to T2-weighted images as they are optimized to detect edema. However, unlike T2-weighted images, the signal from CSF is suppressed on FLAIR images leading to dark appearance of the ventricular system, the sulci over the convexities, and cisterns (Figure 1.7).10
zoom view
Figure 1.6: T1 versus T2-weighted images. Axial T1 (a) and T2 (b) weighted images illustrate the expected signal intensities of the different tissues. The vitreous (v), suprasellar cistern (sc), and the large right temporal lobe resection cavity (T) are dark on the T1-weighted image in contrast to the very bright appearance of the intraorbital (*) and subcutaneous fat (arrows). On the T2-weighted image, all fluid containing structures are the opposite signal intensity to the T1-weighted image assuming a very bright appearance (v, sc, T). The adipose tissues (*, white arrows) are also bright as this sequence was obtained with gradient echo technique. Notice the small blurring bands (arrowheads) in the globes caused by eye motion artifacts. This appearance would be even more accentuated with a spin echo T2-weighted sequence (not shown) because of its longer acquisition time.
The suppression facilitates the conspicuity of small cortical and subcortical infarctions as well as acute to subacute subarachnoid hemorrhage and cortical blood products, which would otherwise blend together with the hyperintensity of CSF on the T2-weighted images (Figure 1.8). Unfortunately, the suppression of the CSF is often incomplete in areas of significant CSF motion, e.g. at the foramen of Monroe and in the 3rd and 4th ventricles as well as in the posterior fossa cisterns. Such suboptimal suppression, referred to as ‘CSF flow artifacts,’ can mimic intracranial abnormalities (Figure 1.7).
 
Fat suppression techniques
Fat suppression is commonly utilized when imaging anatomical areas containing large amounts of adipose tissue (e.g. orbits). Fat suppression can be applied to the majority of sequences but is most commonly used for T1- and T2-weighted images in the neuro-ophthalmology patients. On T1-weighted images, fat suppression helps to distinguish fat from other hyperintense material such as subacute blood or melanin (Figure 4.18). The application of fat suppression to T2-weighted images facilitates the detection of orbital edema and inflammation. On Gadolinium-enhanced T1-weighted images, fat suppression accentuates abnormal contrast enhancement because the natural ‘brightness’ of adipose structures is suppressed and does not compete with the ‘brightness’ of the Gadolinium (Figure 1.9). As with FLAIR imaging, fat suppression is often incomplete, particularly in areas of strong bone and air interfaces such as between the paranasal sinuses and orbit (e.g. medial orbital wall or orbital floor).11
zoom view
Figure 1.7: Normal appearance of T2 versus fluid-attenuated inversion recovery (FLAIR) images. Axial T2 (a) shows marked hyperintensity of the cerebrospinal fluid (CSF) within the lateral ventricles (*) and sulci (arrowheads). On the FLAIR (b) image, the CSF is suppressed leading to dark appearance of the lateral ventricles (*) and sulci. Notice the small bright structures (arrows) within the lateral ventricles on the FLAIR-weighted images (b). These represent CSF pulsation artifacts that like to occur at the foramina of Monroe and are often asymmetric in distribution.
zoom view
Figure 1.8: Conspicuity of acute to subacute hemorrhage on fluid-attenuated inversion recovery (FLAIR) versus T2. Axial T2 (a) is normal in appearance, while the FLAIR-weighted image at the same level (b) reveals marked hyperintensity in the left paramesencephalic cistern (arrows) consistent with small amount of subarachnoid blood. The blood is more conspicuous on the FLAIR image because of the suppression of the cerebrospinal fluid within the basilar cistern. The subarachnoid hemorrhage (arrows) was confirmed with noncontrast enhanced computed tomography (c).
12Therefore, bright signal (symmetric or asymmetric), related to susceptibility artifacts, is often seen at these interfaces that can be mistaken for Gadolinium enhancement on the fat-suppressed contrast-enhanced T1-weighted images (Figure 1.10). In addition, fat-suppressed images are hampered by markedly lower conspicuity of the anatomical structures because the suppressed fat often assumes similar signal intensity to the adjacent anatomical structures.
zoom view
Figure 1.9: Differences in conspicuity of enhancement on T1-weighted images without and with fat suppression. Gadolinium enhanced, axial T1-weighted image (a) in a patient with eye pain and proptosis shows no significant abnormality. First the application of fat suppression technique (b) reveals an enhancing and infiltrating lesion (arrows) around the medial rectus muscle (….) leading to the diagnosis of orbital inflammatory syndrome.
zoom view
Figure 1.10: Susceptibility artifact mimicking enhancement. Axial fat-suppressed T1-weighted image demonstrates a focal area of hyperintensity in the right posterior medial orbit (arrows) suggesting an enhancing orbital mass. This ‘lesion’ is, however, related to susceptibility artifacts from the maxillary sinus rather than enhancement because no Gadolinium was given. The lack of Gadolinium administration is reflected in the absence of enhancement in the cavernous sinuses (arrowheads). This example illustrates the importance of obtaining fat-suppressed images before and after Gadolinium enhancement to avoid misinterpretation of susceptibility artifacts as abnormalities on the contrast enhanced fat-suppressed T1-weighted image(s).
13
 
Diffusion-weighted images (DWI)
DWI is based on the random motion of water through tissues (referred to as diffusion). The diffusion can range from freely mobile to restricted. Acute infarction is the best-known disease process to cause restricted motion of water. During an acute infarction, the interruption of cerebral blood flow leads to rapid breakdown of the cellular energy metabolism and dysfunction of ion exchange pumps, resulting in a massive shift of water from the extracellular to the intracellular space. The water hence gets ‘trapped within the cells’ and results in ‘restricted diffusion.’
Two additional gradients are applied during the scanning process of the DWI sequence to allow the detection of the disturbed diffusivity within tissues. The strength and the width of these gradients as well as the time between each of the gradients determine the ‘diffusion weighting’ of the pulse, also called the ‘b value.’ The lesser the b value, the lower the diffusivity and the higher the T2 effect on an image. Therefore, the low b value images resemble lower resolution T2-weighted images (Figure 1.11a). With increasing b value, the T2 effect decreases and the diffusivity effect increases, reflected on the images as decreasing attenuation of the CSF (Figure 1.11b and c). Therefore, only high b value images are able to display the diffusivity of tissues. On these images, restricted diffusion is reflected as bright signal, while random motion of water shows the same signal intensity (mid gray level) as normal brain parenchyma (Figure 1.12a and b).
Although the T2 effect decreases with increasing b values, none of the images are free of T2 contamination. This leakage of T2 signal is called the ‘T2 shine through phenomenon’ and can lead to misinterpretation of T2 hyperintense lesions for acute infarctions on DWI. To display the ‘apparent diffusion’ without the T2 signal contamination, the apparent diffusion coefficient (ADC) is calculated from images performed with at least two different b values.
zoom view
Figure 1.11: Normal appearance of the diffusion-weighted images using different b values. The b = 0 image appears like a low-quality T2-weighted image with white appearance of the lateral ventricles (*) and subarachnoid spaces. The hyperintensity of the cerebrospinal fluid decreases with increasing b values with similar appearance (*) to brain parenchyma on the b = 500 image and dark appearance (*) on the b = 1000 image.
14
zoom view
Figure 1.12: Restricted diffusion versus T2 shine through phenomenon. Axial diffusion-weighted image (DWI) (a) in a patient with acute onset of right-sided body weakness demonstrates marked hyperintensity in the superior temporal lobe on the left (i) that demonstrates lower signal intensity (i) than the brain parenchyma on the apparent diffusion coefficient (ADC) map (b). This combination of signal intensities is consistent with restricted diffusion as seen in acute infarctions. The DWI (c) of a different patient shows mild-to-moderate hyperintensity in the parietal left lobe (white arrows); however, this lesion is iso- to hyperintense (black arrows) to the adjacent brain on the ADC image (d). This combination of signal intensities is consistent with T2 shine through phenomenon and not restricted diffusion. Patient was ultimately diagnosed with glioma.
Clinically the b 0 and b 800—1000 values are utilized for this calculation. The calculated ADC values are then displayed on ADC map images. Although the ADC images more accurately demonstrate the diffusivity of water in tissue, these are typically not used at the beginning of the image interpretation for two reasons. First, areas of restricted diffusion in acute infarctions are dark on the ADC maps (the opposite of their appearance on DWI) and dark areas are more difficult to perceive than bright areas. Second, the ADC map represents post-processed images and any motion between the b 0 and the high b value image sets will cause false negative results, especially in small infarctions. Therefore, the interpretation of DWI should always start with the review of the high b value images with subsequent confirmation on the ADC images. Areas of acute infarction will show bright signal on the DWI and dark signal on the ADC (Figure 1.12a and b). In contrast, the T2 shine through phenomenon demonstrates intermediate to high signal on the ADC maps (Figure 1.12c and d).
With time, the reparative mechanisms of the body clean up the ‘trapped water’ within the area of the acute infarction converting the region of restricted diffusion into an encephalomalacia defect with adjacent gliosis. This ‘healing’ process results in conversion of the signal on the DWI and ADC images in the opposite direction (dark on DWI and bright on ADC); typically observed by 3–4 weeks with large size infarctions and within a few days with small infarctions. Therefore, DWI not only allows the detection of areas of acute infarctions but also helps to differentiate acute infarcted areas from underlying chronic ischemic changes (Figure 1.13). In addition, DWI can help distinguish acute arterial from venous infarctions with the latter usually lacking any DWI alterations.15
zoom view
Figure 1.13: Distinguishing acute from chronic ischemia with diffusion-weighted image (DWI). Axial fluid-attenuated inversion recovery-weighted images through the high convexities of the hemispheres (a and b) demonstrate bilateral areas of cortical and subcortical hyperintensities (arrowheads) of unknown age. The DWI images at the same levels (c and d) reveal that only the lesions in the left hemisphere show restricted diffusion (arrows) consistent with small areas of acute infarctions. The restricted diffusion was confirmed on the apparent diffusion coefficient images (not shown).
DWI has a very high sensitivity for detecting acute to early subacute infarctions but unfortunately has suboptimal specificity because other disease processes can also display ‘restricted diffusion’ such as infections or neoplastic lesions. Infections (e.g. abscess, encephalitis, and ventriculitis) will show restricted diffusion, while meningitis is often undetectable on DWI or demonstrates only subtle changes (particularly in the early stages of the meningeal process). Bacterial and often fungal abscesses show bright signal on DWI centrally, while the peripheral capsule displays iso- to hypointensity in comparison to the adjacent brain parenchyma (Figure 1.14). The opposite signal pattern is seen on the ADC images. The ADC maps also reveal variable degrees of hyperintensity around the abscess cavity reflecting the surrounding vasogenic edema that is not as apparent on DWI (Figure 1.14c). Many types of encephalitides will show focal areas of restricted diffusion (Figure 1.15) that can be patchy in distribution or localized to specific territories of the brain. Herpes simplex encephalitis typically shows restricted diffusion in the medial temporal lobes in an asymmetric fashion from where it spreads to the inferior frontal lobes. In contrast, Creutzfeldt–Jacob disease classically demonstrates symmetric areas of DWI hyperintensity that preferentially involve the deep gray matter (caudate nucleus, basal ganglia, and thalamus) and the cerebral cortex (sparing the white matter). If restricted diffusion is seen in the brain stem, rhombencephalitis related to Listeria monocytogenes infection needs to be considered. The areas affected by encephalitis may or may not show surrounding vasogenic edema on the ADC images (Figure 1.15). In case of ventriculitis, restricted diffusion is typically observed in the dependent portions of the ventricles, most commonly in the occipital horns of the lateral ventricles (Figure 1.16). In some patients with ventriculitis, all other images will be normal in appearance and only the DWI will lead to the correct diagnosis.16
zoom view
Figure 1.14: Restricted diffusion in a brain abscess. A 25-year-old woman was admitted to the hospital with fever and mental status changes. Computed tomography study without contrast (not shown) demonstrated a lesion in the left frontal lobe. MRI examination was performed for further evaluation. The Gadolinium enhanced T1-weighted image (a) reveals a rim-enhancing lesion (L) in the left frontal lobe with marked surrounding vasogenic edema (arrows) and mild left to right hemispheric shift. On the diffusion-weighted image (b), the lesion is hyperintense centrally (L) and shows a thin hypointense capsule (arrowheads) corresponding to the area of enhancement on the Gadolinium-enhanced image. On the apparent diffusion coefficient map (c), the lesion is hypointense centrally (L) consistent with restricted diffusion and is surrounded by marked hyperintensity reflecting the surrounding vasogenic edema (arrows). The constellation of imaging findings is characteristic for a brain abscess, which was confirmed intraoperatively.
zoom view
Figure 1.15: Restricted diffusion related to encephalitis. A 31-year-old man presented to the emergency room with headache, blurred vision, fever, and photophobia. Computed tomography study (not shown) questioned a possible area of subtle edema in the right peritrigonal region. Magnetic resonance imaging examination was performed for further evaluation. The fluid-attenuated inversion recovery-weighted image (a) shows focal area of nonspecific edema (black arrows). The diffusion-weighted image (b) demonstrates an area of hyperintensity (white arrows) lateral to the trigone of the right lateral ventricle that is markedly hypointense in signal intensity (white arrows) on the apparent diffusion coefficient (ADC) map image (c) consistent with restricted diffusion. The ADC map also reveals focal areas of hyperintensity (arrowheads) around the area of restricted diffusion, in particular anteriorly, indicating associated vasogenic edema. The constellation of imaging findings and clinical presentation are most concerning for localized encephalitis, which was confirmed with cerebrospinal fluid analysis.
17
zoom view
Figure 1.16: Restricted diffusion secondary to ventriculitis. A 45-year-old man with mental status changes following mastoidectomy for coalescent mastoiditis underwent magnetic resonance imaging examination. The Gadolinium-enhanced T1-weighted image (a) shows no abnormalities, while the diffusion-weighted image (b) reveals focal areas of hyperintensity in the dependent portions of the lateral ventricles (arrows). Such an appearance can be caused by blood products or purulent material, the latter being favored in this patient. The subsequently performed lumbar puncture confirmed the diagnosis of bacterial infection.
Any tightly packed blue cell neoplasm as well as embryonal cell type tumors can demonstrate restricted diffusion. In adults, metastatic disease from small cell lung cancer is the most common neoplastic process associated with restricted diffusion. The metastatic lesion may demonstrate areas of central necrosis. Therefore, the appearance on DWI can be mixed with smaller, non-necrotic metastasis being homogeneously bright, while necrotic lesions will show a dark center with a hyperintense rim of viable tumor. In children, Ewing's sarcoma and neuroblastoma represent the most frequent metastatic lesions expressing restricted diffusion (Figure 4.15). Primary central nervous system lymphoma is the most common nonmetastatic malignant brain lesion demonstrating restricted diffusion (Figure 1.17). It classically displays homogenously bright signal on DWI with only mild surrounding vasogenic edema. The vasogenic edema is best visualized on the T2, FLAIR, or ADC-weighted images. Epidermoid tumor is the most well-known benign intracranial lesion associated with restricted diffusion. On CT and the majority of the MRI sequences, it is indistinguishable from an arachnoid cyst. Only DWI is able to reliably differentiate between these two entities (Figure 1.18).
Acute toxic processes (e.g. Wernicke's encephalopathy and carbon monoxide intoxication) may cause restricted diffusion. These diseases often involve specific locations, and the distribution of abnormal DWI signal intensity can lead to the correct diagnosis or narrow down the diagnostic considerations. Symmetric-restricted diffusion in the thalami, mammillary bodies, and periaqueductal gray matter is characteristic for Wernicke's encephalopathy (Figure 7.10). In contrast, bilateral-restricted diffusion in the basal ganglia has a broader differential diagnosis that includes but is not limited to hypoxia, carbon monoxide intoxication, and hyperammonemia. A significant seizure episode, in particular status epilepticus, may also result in variable degrees of restricted diffusion.18
zoom view
Figure 1.17: Restricted diffusion related to corpus callosum lymphoma. Axial Gadolinium-enhanced image (a) shows a homogenously, markedly enhancing mass (M) in the splenium of the corpus callosum with mild surrounding edema (arrows). The mass is hyperintense (M) on the diffusion-weighted image sequence (b) suggesting a small blue cell tumor. The lesion was hypointense in signal intensity on the apparent diffusion coefficient images (not shown). The location of the lesion and the imaging characteristics are classic for a primary central nervous system lymphoma that was confirmed with biopsy.
zoom view
Figure 1.18: Epidermoid tumor versus arachnoid cyst on diffusion-weighted images (DWI). Axial T2-weighted image (a) shows marked mass effect on the left pons (arrowheads) that is caused by a cerebrospinal fluid (CSF) attenuation lesion. On the DWI (b), the lesion assumes a very bright signal (*) consistent with restricted diffusion. The constellation of findings is consistent with epidermoid tumor. In a different patient, similar findings are seen on the T2-weighted image (c) with marked impression on the inferior pons (arrowheads) by a CSF attenuation lesion. In this patient, however, the DWI (d) shows very dark signal in the left prepontine cistern (arrows). These findings are consistent with an arachnoid cyst.
This is, however, typically reversible and not predictive of clinical outcome.
The typical white matter demyelination foci do not demonstrate restricted diffusion. Active demyelination is an exception and is often referred to as ‘myelin edema.’ In myelin edema, tiny cavities are seen on histological specimens. In vivo, these cavities trap fluid that is reflected as restricted diffusion on DWI. Myelin edema is classically seen in leukodystrophies. It can also be observed in multiple sclerosis where the restricted diffusion can aid in distinguishing active from indolent or chronic white matter plaques.
It should be kept in mind that a 'bright signal’ on DWI does not always indicate pathology and might represent an artifact. As with the 19fat suppression technique, any areas of strong bone and air interface can cause susceptibility artifacts on the DWI sequence that are reflected as linear to curvilinear bands of bright signal (Figure 1.19). These artifactual bands of high DWI signal intensity characteristically occur along the posterior margins of the frontal sinuses, superior and posterior borders of the sphenoid sinuses, as well as along the temporal bones. Acute to subacute blood products can also mimic areas of restricted diffusion. This may pose a diagnostic dilemma in patients with a postsurgical fluid cavity because the intrinsic blood products can demonstrate ‘restricted diffusion’ within the cavity that can be mistaken for an abscess. A similar problem can occur with an intraparenchymal hemorrhage. In this instance, it might be difficult to determine if the signal abnormality is related to a hemorrhagic infarction, an underlying hemorrhagic metastasis, or a simple hemorrhage. Continuous follow-up or detection of an additional nonhemorrhagic lesion on MRI can lead to the correct diagnosis in such patients. A summary of common lesions with restricted diffusion and their differentiating imaging features is provided in Table 1.1.
 
Blood-sensitive sequences
There are two types of blood-sensitive sequences that are currently utilized in clinical practice: gradient-recalled echo (T2* GRE) and susceptibility weighted imaging (SWI). T2* GRE is the most commonly applied sequence. It is based on the chemical properties of iron within hemoglobin that transforms from a diamagnetic to a paramagnetic molecule in the early stages of hemorrhage. The paramagnetic state of the iron causes localized magnetic field inhomogeneities that result in regional susceptibility artifacts. The T2* GRE sequence is very sensitive in the detection of such field inhomogeneities and accentuates their size. Therefore, blood products in the early stage of development and when small in size are easier to detect on the T2* GRE images when compared with other sequences. As the blood products undergo further degradation with time, it becomes easier to detect them on other sequences, in particular the T1-weighted images (Table 1.2).
zoom view
Figure 1.19: Susceptibility artifacts on diffusion-weighted image (DWI). Axial DWI images obtained just superior to the temporal bones (a) and sphenoid sinuses (b) demonstrate linear to curvilinear areas of marked hyperintensity (arrows) that represent susceptibility artifacts caused by the strong air–bone interfaces along the temporal bones and sphenoid sinuses bilaterally.
20
Table 1.1   Distinguishing imaging features of common lesions causing restricted diffusion
Lesion type
Appearance on DWI
Appearance on ADC
Other helpful imaging findings
Acute to early subacute infarction
Hyperintense
Hypointense
Usually gray and white matter involved, progressive mass effect
Late subacute to chronic infarction
Iso- to hypointense
Hyperintense
No mass effect
Bacterial brain abscess
Hyperintense centrally with hypointense rim
Hypointense centrally surrounded by marked hyperintensity
Rim enhancing lesion on Gd T1 with marked surrounding vasogenic edema on FLAIR and T2
Encephalitis
Hyperintense
Hypointense with variable degrees of adjacent hyperintensity
Herpes encephalitis: Bilateral medial temporal lobe involvement
Rhombencephalitis: Brain stem involvement
Ventriculitis
Hyperintensity in the dependent portions of the ventricles
Iso- to hypointensity in the dependent portions of the ventricles
May see variable enhancement of the ependymal surface of the affected ventricle(s)
Lymphoma
Hyperintense
Hypointense with mild amount of adjacent hyperintensity
Homogeneously and markedly enhancing on the GdT1 with relative mild surrounding edema in relation to size of enhancing lesion
Small cell lung cancer metastasis
Homogenously hyperintense and/or peripherally hyperintense with hypointense center
Homogenously hypointense and/or peripherally hypointense with variable degrees of surrounding hyperintensity
Homogenously enhancing and/or peripherally enhancing when centrally necrotic; often both type of lesions coexist; variable degrees of vasogenic edema around each lesion
Neuroblastoma
Homogenously hyperintense
Homogenously hypointense
Often involves the orbital walls
Epidermoid tumor
Homogenously hyperintense
Homogenously iso- to hypointense
Follows CSF signal intensity on all other sequences
Arachnoid cyst
Homogenously hypointense
Homogenously hyperintense
Follows CSF signal intensity on all other sequences
Blood products
Variable
Variable
Variable depending on age
Susceptibility artifacts
Curvilinear hyperintensity
Curvilinear hypointensity
Most commonly along the temporal bones, posterior to the frontal sinuses, superior posterior to the sphenoid sinuses
CSF, cerebrospinal fluid; DWI, diffusion-weighted images; ADC, apparent diffusion coefficient; FLAIR, fluid-attenuated inversion recovery.
At the end of the degradation process, a so-called ‘hemosiderin stain’ remains. The hemosiderin stain is rarely visible on the T1-weighted images and may also not be visible on the T2 or FLAIR-weighted images, particularly when small in size. In this late stage, T2* GRE imaging is again superior to the other MRI sequences. It accentuates the size of the hemosiderin stain, thereby making it more apparent.
On the T2* GRE images, paramagnetic blood products are displayed as an area of very low signal intensity (referred to as ‘black dot’).21
Table 1.2   Time evolution of blood products and their appearance on magnetic resonance imaging
Stage of hemorrhage
Age of hemorrhage
Composition of hemorrhage
Signal intensity on T1 images compared with brain parenchyma
Signal intensity on T2 images compared with brain parenchyma
Hyperacute
<24 hours
Oxyhemoglobin
Iso- to hypointense
Hyperintense
Acute
1–3 days
Deoxyhemoglobin
Hypointense
Hypointense
Early subacute
4–7 days
Intracellular methemoglobin
Hyperintense
Hypointense
Late subacute
8–14 days
Extracellular methemoglobin
Hyperintense
Hyperintense
Chronic
>14 days
Ferritin and hemosiderin
Hypointense
Hypointense
Unfortunately, other paramagnetic material can lead to a ‘black dot’ appearance such as calcium. SWI is a new blood-sensitive technique that is able to differentiate between calcium deposition and blood products. Similar to GRE, SWI is also based on the paramagnetic properties of tissue, but is more complex in nature as it captures not only the magnitude but also the phase information of an area of magnetic field inhomogeneity. This information is displayed on two different image sets called magnitude and phase data sets (Figure 1.20). The phase images are able to differentiate blood products from calcium because calcium remains dark in appearance, while blood assumes a bright signal (Figures 1.20c and d). In contrast to T2* GRE, SWI also displays venous structures as black signal intensity since the veins carry deoxygenated blood that has paramagnetic properties relative to arteries and the surrounding brain parenchyma. These veins are easily identified on the third image set that is provided by the scanner. This third image set displays the SWI data of a thicker brain parenchyma volume as minimum intensity projection (mIP) images (Figure 1.20b). On the mIP images, the various sizes of veins can be visualized and followed similar to the MR angiographic images but as negative contrast (referred to as minimum intensity). Although the interpretation of these images might seem more cumbersome at first, the display of veins can become advantageous in patients with venous thrombosis, in particular when small veins are affected. The thrombosed vein takes on a bright signal on SWI that stands out against the black appearing normal veins. In addition, the thrombosed vein can also be easily distinguished from adjacent hemorrhage because the hemorrhage will remain dark in signal intensity on the magnitude and mIP images, nicely contrasting it from the bright appearance of the involved vein.
 
Steady-state free procession sequence
The steady-state free procession sequence is a volumetric heavy T2-weighted sequence that accentuates the contrast between CSF and solid structures. Since it is performed with slice thickness of about 1 mm or less, it is capable of depicting all of the cranial nerves. In addition, the very small slice thickness allows reformations in any desirable plane that facilitates detection of lesions compressing the cranial nerves (Figures 10.4 and 10.12).22
zoom view
Figure 1.20: Multiple cavernomas on susceptibility weighted imaging (SWI). Axial noncontrast-enhanced computed tomography (CT) (a) shows a right frontal encephalomalacia defect (*) and a calcified lesion in the left basal ganglia (arrow). The minimum intensity projection image of the SWI sequence (b) confirms the left basal ganglia lesion (arrow) and reveals numerous additional ‘black dots.’ The large area of very low attenuation projecting into the medial frontal lobes (es) is caused by susceptibility artifacts from the ethmoid sinuses. The comparison between magnitude (c) and phase (d) images demonstrates different behaviors of the various lesions. The largest lesion (arrow) remains partially dark in appearance consistent with calcifications. This is concordant with the findings on the CT image. In contrast, all the lesions in the right hemisphere (arrowheads) are bright in appearance on the phase image consistent with blood products. These were not appreciable on the CT examination but were visible on some of the other sequences (not shown). This patient had numerous additional partially calcified cavernomas of different sizes. The encephalomalacia defect in the right frontal lobe is a sequela from prior resection of a cavernoma.
 
Imaging logistics
 
Acquisition time
Each standard MRI sequence requires 3–6 minutes of scanning time for every plane. A standard brain MRI with and without Gadolinium is usually composed of at least seven sequences (sagittal T1, axial FLAIR, T1, T2, DWI, and postcontrast T1-weighted images in axial and coronal planes) resulting in an acquisition time of about 30 minutes. Inclusion of additional sequences, e.g. for the optic nerve sheath complex, which is best imaged in the coronal plane, will add about 15 more minutes to the scanning time. This might seem a reasonable time period, but rapidly it can become a very long procedure to tolerate, especially if the patient is experiencing significant pain. One common scenario is the evaluation of a multiple sclerosis patient that requires additional imaging of the optic nerves and the entire spinal cord. In such cases, 23scanning in two different sittings might be easier to bear by the patient and produce less motion-degraded images.
In general, MRI examinations lasting longer than 45 minutes are not well tolerated by the patient. Therefore, careful protocolling of each MRI examination by the radiologist is required with preferential performance of critical sequences and imaging planes at the beginning of the study to minimize the risk of suboptimal image quality secondary to motion degradation or deliverance of an incomplete study. For example, in an acute stroke patient, the DWI is most critical to detect an acute or early subacute infarction. The DWI scan needs to be supplemented by a blood-sensitive sequence to identify hemorrhagic transformation of an infarction that might preclude anticoagulation therapy; and by MR angiography (MRA) of the circle of Willis and the cervical vasculature to detect areas of significant vascular stenosis and/or occlusion. Additional sequences such as FLAIR, T1- and T2-weighted images are desirable to determine the degree of prior ischemic brain damage but are not as critical for the initiation of the appropriate treatment of the patient's current symptoms and can be performed at a separate time point if needed.
 
Spatial resolution
Spatial resolution, signal intensity, and image noise are closely related concepts to each other. As mentioned above, the degree of magnetization (expressed as signal intensity) is contingent upon the amount of protons within the scanned volume. Therefore, any decrease in the scanned volume will lead to loss of signal intensity and to an increase in image noise.
Each image is composed of many volume elements called voxels. The number of voxels determines the spatial resolution of an image with smaller voxels leading to better resolution. The size of a voxel can be changed by modifying the slice thickness, field of view, or image matrix. The field of view and the image matrix are directly dependent on each other since the matrix represents the number of elements that the field of view of an image is spread over. Therefore, an increase in the image matrix from 256 to 512 elements will reduce the voxel size by a factor of 4 when the field of view and the slice thickness remain constant. In contrast, the reduction in the voxel size can be partially compensated by increasing the field of view and/or the slice thickness. Knowledge of these interconnections is critical for the development of MRI protocols, since any decrease in the voxel size increases the spatial resolution at the expense of signal intensity that in turn directly influences the image quality. There are multiple ways to compensate for the signal loss, but it comes at the expense of increasing the acquisition time. Therefore, there is a fine balance between the most optimal spatial resolution, efficient scanning time, and sufficient signal intensity to produce images of adequate diagnostic quality. Which specific scanning parameters are chosen depends not only on the capabilities of the MRI scanner but also on the anatomical structure of interest. For 24example, imaging with 5 mm slice thickness is sufficient for evaluation of hydrocephalus; however, it is inadequate for the demonstration of an ocular motor cranial nerve abnormality in a patient with diplopia. In addition, any technical upgrade (hardware or software) of the MRI equipment requires review of MRI protocols as this balance is expected to change with advancements in MRI technology.
 
Imaging plane
The majority of MRI sequences cannot be acquired as a volumetric acquisition within a reasonable time frame. Therefore, there are only limited options for multiplanar reformations. In addition, even if a volumetric data set is available, the performed reformations are usually less sharp than the direct coronal or sagittal images because of the slice thickness with which the images were acquired (Figures 9.5c and 10.4). As a consequence, the most optimal imaging plane for each structure of interest needs to be considered when protocolling MRI studies. The optic nerve is best demonstrated in the coronal plane and less optimally in the axial and sagittal planes. Therefore, imaging of disease processes affecting the optic nerve sheath complex and its surrounding tissues should include direct coronal images. On the contrary, suspected cases of carotid artery dissections are best visualized on axial images as this plane best delineates the patent vessel lumen from the intramural hematoma or the decreased, turbulent flow within the false lumen. Often two different imaging planes are needed to optimally demonstrate the structure of interest for two main reasons. First, many structures change their course and assume a different direction. Second, different imaging planes better delineate the relationship of the pathology to adjacent anatomical structures. Volumetric acquisitions are usually reserved for surgical planning purposes and for structures that require oblique or curved reformations for optimal assessment (Figures 9.5c and 10.4).
 
Communication
The ophthalmologist can contribute significantly to the success of MRI studies by providing detailed information to the radiologist in regard to patient signs and symptoms and the clinical working diagnosis. Such communication will facilitate the appropriate choice of study protocol and prioritization of MRI sequences. Abnormal vision for example is a common ophthalmological problem. Without additional clinical information, it is very difficult for the radiologist to provide high-quality images of all the potential disease processes that may cause abnormal vision. Here are some examples of how additional information might alter the imaging protocol of a patient with abnormal vision:
  • In a case of papilledema, a space occupying intracranial lesion, sinus thrombosis, or idiopathic intracranial hypertension will be the most concerning diagnosis requiring pre- and postcontrast-enhanced images through the brain as well as MR venography (MRV)25
  • A patient with unilateral vision loss should undergo imaging of the affected orbit and optic nerve sheath complex. MRA of the cervical and intracranial vasculature needs to be added in case of retinal ischemia, while pre- and postcontrast enhancement images through the brain ought to be included in suspected cases of multiple sclerosis
  • In a case of diplopia, images of the orbit as well as of the cavernous sinuses and posterior fossa are required to evaluate CN III, IV, and VI from their origin in the brain stem to the extraocular muscles in the orbit
 
Perfusion imaging
Perfusion imaging represents dynamic tracking of a contrast bolus through the brain parenchyma over time and can be obtained with either a CT or MRI scan. Perfusion imaging was primarily developed to detect areas of major acute infarctions, but it has gained broader utilization recently in the evaluation of primary brain tumors and in distinguishing radiation necrosis from recurrent tumor.
CT and MRI perfusion scans share the same basic principles. Both techniques obtain images at the same level before contrast arrival, and during the passage of contrast through the brain parenchyma capturing the arterial wash in phase and the venous washout phase. The contrast on CT is displayed as bright attenuation, while the MRI perfusion sequence results in dark signal of the Gadolinium contrast agent. The number of images that can be simultaneously obtained depends on the scanner. The majority of currently utilized CT scanners allow only 4 slices to be performed concurrently, while the 320-slice CT scanner and majority of MRI scanners are able to cover the entire brain at once. When only a limited number of slices are available, the area of interest needs to be communicated to the radiologist because not all vascular territories may be adequately covered due to the technical limitations of perfusion imaging. Subsequently, a contrast attenuation curve is calculated for each voxel that is translated into different color maps. The ‘time to peak’ (TTP) and the ‘cerebral blood volume’ (CBV) maps are the most important maps for image interpretation. The TTP reflects the time it takes the contrast to reach maximum concentration (equivalent to the peak of the curve) in a given voxel. The CBV map represents the area under the contrast curve and denotes the amount of contrast that reaches a specific voxel. Normally, the TTP and CBV maps are symmetric in appearance (Figure 1.21). With an ischemic infarction, a significant delay in the TTP is expected in the territory supplied by the occluded vessel (Figure 1.22a). The appearance of CBV, however, varies depending on available collateral flow to the territory of the occluded vessel. If sufficient collateral flow is present, the CBV might be completely normal in appearance or show increased CBV due to autoregulatory-induced vasodilatation. This is reflected as red color 26on the CBV map. In contrast, when the infarction is complete, marked reduction in the CBV is noted in the involved vascular territory. This is displayed as a dark blue-purple to black color (Figure 1.22b).
In primary brain tumor imaging, the CBV map reflects tumor grade. High-grade tumors display markedly increased CBV, while low-grade tumors show decreased to absent CBV (Figure 1.23). Since many brain tumors begin as low-grade tumors that have the potential to degenerate over time into a higher grade, perfusion imaging can be used to demonstrate this conversion, potentially earlier than appreciable with any other sequences. In addition, many brain tumors are heterogeneous in nature and may display different tumor grades simultaneously. In such cases, the perfusion imaging can guide the biopsy to the area with the highest tumor grade.
An enhancing lesion in a patient following radiation therapy for a primary brain tumor poses a diagnostic dilemma because it may represent either radiation necrosis or recurrent tumor. Conventional radiological observation over time is often not helpful because radiation necrosis may show progression similar to a recurrent tumor. In such cases, perfusion imaging might lead to the correct diagnosis as radiation necrosis shows significantly reduced CBV in contrast to a recurrent high-grade tumor (Figure 1.24). A lower grade tumor displays similar CBV characteristics to radiation necrosis but does not enhance on the Gadolinium-enhanced T1-weighted images.
CT and MRI perfusion images are significantly hampered by motion degradation as the contrast attenuation of different voxels is plotted in the same curve. The majority of CT scanners allow for some correction of the motion degradation. This is not directly possible with the currently available MRI scanners. In addition, MRI perfusion imaging is unreliable when susceptibility artifacts caused by hemosiderin deposition or metallic surgical clips are present.
 
Vascular imaging
Currently, ultrasound, CT, and MRI are the most commonly used vascular imaging modalities in ophthalmology. While all three techniques are capable of imaging the cervical vessels, the intracranial vasculature can only be demonstrated with CT and MRI.
zoom view
Figure 1.21: Normal brain perfusion color maps. (a) Axial time to peak (TTP) and (b) cerebral blood volume (CBV) images show normal and symmetric appearance of brain perfusion at the level of the lateral ventricles.
27Each modality is based on specific radiological principles with their own unique shortcomings.
 
Ultrasound
Ultrasound serves as a great screening tool for the detection of significant atherosclerotic disease within the cervical portions of the carotid arteries but has limited capability in evaluating the vertebral and the subclavian arteries. In addition, it is not able to evaluate the aortic arch and the great vessels within the thorax including the proximal aspects of the left common carotid artery (CCA) and of the brachiocephalic artery. A carotid ultrasound is typically performed with the patient in a supine position with the neck slightly hyperextended. There are three components to an ultrasound examination:
  1. Gray scale imaging: The gray scale imaging demonstrates the extent and the composition of the atherosclerotic plaque. It can distinguish calcified (hard) from a soft plaque (containing lipid, fibrin, or intraplaque hemorrhage). It is also able to detect ulcerations and occasionally dissections. On gray scale images, calcified plaque is hyperechoic and causes marked shadowing, while soft plaque is hypoechoic in appearance (Figures 1.25 and 1.26). Hypoechoic, mixed echogenicity, and ulcerated plaques are more unstable and more likely to cause transient or permanent ischemic symptoms
    zoom view
    Figure 1.22: Abnormal perfusion color maps in acute middle cerebral artery (MCA) territory stroke. The time to peak (TTP) image (a) reveals severely delayed perfusion in the right MCA territory reflected as red color. On the CBV image (b), there is a marked reduction in the color in the same territory indicating severe decrease in blood volume. The combination of imaging findings is consistent with a completed MCA infarction that was caused by occlusion of the right MCA (arrow) as seen on the computed tomography angiography (CTA) image (c). Follow-up CT imaging (d) demonstrates marked cytotoxic edema (between arrowheads) in the right MCA territory with sulcal effacement confirming the irreversible damage to the brain parenchyma.
    28
    zoom view
    Figure 1.23: High-grade versus low-grade glioma on magnetic resonance perfusion. Axial Gadolinium enhanced T1-weighted image (a) shows an ill-defined, markedly enhancing mass (arrows) in the right posterior hemisphere with surrounding edema and mild right to left subfalcine herniation. This mass demonstrates increased blood volume on the cerebral blood volume (CBV) map image (b) reflected as green to red colors (arrows). Such appearance on CBV images is consistent with a high-grade tumor. Glioblastoma multiforme was pathologically confirmed. In another patient, a well-defined lesion is seen in the posterior insular region (arrowheads) on the fluid-attenuated inversion recovery-weighted image (c). This lesion showed no significant enhancement on the Gadolinium-enhanced T1-weighted image (not shown). The CBV map image (d) at the same level demonstrates that the vast majority of the lesion has lower perfusion than the contralateral brain parenchyma reflected as light blue color (white arrowheads). Only small portions laterally and posteriorly reveal slightly higher blood volume (black arrowheads) than the contralateral cortex. Such appearance on the CBV images is consistent with low-grade glioma, which was confirmed with biopsy.
  2. Color Doppler imaging: These images capture the direction and homogeneity of the flow within a vessel. Laminar and homogenous flow directed toward the head (reflected as red color) is typically seen in the majority of the carotid artery system. Only at the carotid bifurcation, there is mild turbulent flow visualized reflected as red color intermixed with small amount of blue. Areas of markedly turbulent flow represented by red and marked amount of blue intermixed may indicate significant underlying stenosis and require further evaluation with spectral Doppler analysis (see below) (Figure 1.27). In CCA occlusion, the internal carotid artery (ICA) might also be occluded or may remain patent through collateral flow provided from the external carotid artery (ECA). In the latter instance, reversal flow (reflected as blue color) is observed within the EAC and normal, antegrade flow (expressed as red color) is seen in the patent ICA (Figure 1.28)29
    zoom view
    Figure 1.24: Recurrent tumor versus radiation necrosis on magnetic resonance perfusion. Axial Gadolinium-enhanced T1-weighted image (a) demonstrates an enhancing mass (arrows) in the right medial frontal lobe. The mass reveals markedly increased blood volume (arrows) on the cerebral blood volume (CBV) map image (b) consistent with recurrent glioma confirmed during surgery. Axial Gadolinium-enhanced T1-weighted image (c) of a different patient shows ill-defined enhancement (arrowheads) adjacent to the right lateral ventricle. The CBV map image (d) reveals no abnormal blood perfusion to this area consistent with radiation necrosis. Subsequent follow-up studies (not shown) demonstrated complete resolution of the enhancement confirming the diagnosis of radiation necrosis.
    zoom view
    Figure 1.25: Obscuration of the vessel lumen by calcified plaques on ultrasound examination. Gray scale image (a) along the common carotid artery demonstrates irregular appearance of the vessel walls with focal areas of echogenicity (arrowheads) and marked black shadowing (arrows) posterior to the vessel consistent with calcified plaques. Calcified plaques obscure the vessel lumen seen as lack of red column on the color flow image (b) making it impossible to measure the velocities at the affected vessel level.
    30
    zoom view
    Figure 1.26: Soft plaque in the internal carotid artery on ultrasound. Gray scale image of the internal carotid artery (*) demonstrates soft plaque (arrowheads) on both sides of the vessel lumen. In contrast to calcified plaques, there is no shadowing associated with this soft plaques.
    zoom view
    Figure 1.27: Physiological versus pathological turbulent flow. Color Doppler image (a) demonstrates normal laminar color Doppler flow in the distal common carotid (*) and mild turbulent flow reflected as focal area of blue color (arrows) at the carotid bulb level. This is physiological and typically located along the posterior wall of the bulb. In contrast, the turbulent flow seen in image (b) is more pronounced (…) and primarily central in location suggesting an underlying stenosis. Mild wasting of the color Doppler column (between arrowheads) is seen just proximal to the turbulent flow supporting this suspicion. The degree of stenosis needs to be determined using velocity measures as outlined in Table 1.3.
    zoom view
    Figure 1.28: Common carotid artery (CCA) occlusion with collateral flow to the internal carotid artery (ICA) from the external carotid artery (ECA). The color Doppler ultrasound image in the longitudinal (a) plane shows lack of flow in the visualized common carotid artery (**) and normal laminar flow in the ICA (red color). Markedly turbulent flow [blue in image (a)] is seen at the bulb of the ICA caused by collateral blood flow jet from the ECA to ICA. The reversal of flow in the ECA is confirmed on the axial color Doppler flow image (b).
    31
  3. Spectral Doppler imaging: Spectral Doppler images provide characteristic flow velocity curves of the different cervical vessels. Low resistance waveform with a sharp upstroke and moderate diastolic flow is identified in the ICA and the vertebral arteries as the brain requires perfusion throughout the entire cardiac cycle (Figure 1.29a). In contrast, high resistance waveform with little diastolic flow is seen in the ECA (Figure 1.30). A mixed waveform with some diastolic flow is normally present in the CCA. Abnormal appearance of the spectral waveform might be related to inadequate ultrasound quality or underlying pathology:
    1. Bilateral depressed waveforms within the CCAs are concerning for low cardiac output from cardiomyopathy, extensive myocardial infarction, or cardiac valve disease
    2. Unilateral depressed waveform is often caused by significant stenosis in the proximal CCA or in the brachiocephalic artery on the right side
    3. Bilateral elevation of the systolic peak may be related to high cardiac output such as seen in hypertensive patients or young athletes
    4. Double peak leading to a ‘bisferious’ waveform indicates aortic valve regurgitation (Figure 1.31). The second peak is usually of the same height or higher than the first
      zoom view
      Figure 1.29: Normal versus abnormal low resistance velocity curves. The normal low resistance flow velocity curve (a) shows a steep upslope during systole (arrowhead) and continues flow during diastole (between arrows) with the area under the curve being primarily 'black’ in appearance. With significant stenosis (b), the flow velocity curve widens, referred to as spectral broadening. This is reflected by filling in of the space underneath of the flow velocity curve with flow signal (b). Therefore, the area under the curve assumes ‘white’ echogenicity.
      32
      zoom view
      Figure 1.30: Normal velocity curve in the external carotid artery (ECA). The velocity curve of the ECA shows markedly lower diastolic flow (between arrows) than seen in the internal carotid artery. At the beginning of the diastolic phase, the flow is even extending below the baseline (arrowheads). This is characteristic for high resistance velocity curve as seen in all vessels leading to the face.
      zoom view
      Figure 1.31: Cardiac causes of abnormal flow velocity curves. Bisferious flow velocity curve is seen in image (a) depicted as two systolic peaks (arrows) in each cardiac cycle. This is concerning for aortic valve regurgitation. Irregular heart rhythm is seen in image (b) with a long diastolic phase (between arrows) in the first cardiac cycle and a severely foreshortened diastolic phase (arrowheads) during the second cardiac cycle.
    5. High resistance waveform in the CCA suggests high-grade stenosis or occlusion of the ipsilateral ICA. This is also referred to as ‘externalization of the CCA’
    6. ‘Spectral broadening’ of a waveform is caused by turbulent flow observed with significant underlying vascular stenosis. This is reflected by a wider appearance of the curve and filling of the space below the waveform with bright velocity signal (Figure 1.29b)
    7. Severely depressed and very short peak, also referred as ‘thud flow,’ is classically seen with near total to total occlusion of a vessel (Figure 1.32)
    8. Mid systolic depression of the waveform to baseline or below baseline within the vertebral artery is concerning for partial subclavian steal phenomenon (Figure 1.33). Exercising of the ipsilateral arm during the ultrasound examination often deepens the systolic depression. With complete subclavian steal syndrome, reversal flow is observed in the vertebral artery with the entire waveform being below the baseline33
      zoom view
      Figure 1.32: Complete occlusion of the internal carotid artery on ultrasound. Color ultrasound image demonstrates lack of any flow in the internal carotid artery (*). The blue flow is in the internal jugular vein. The velocity curve demonstrates short and depressed peaks (arrowheads) consistent with ‘thud’ flow that is characteristic for vessel occlusion. Confirmation with computed tomography angiography is required as near total occlusion of a vessel has similar appearance to complete occlusion on ultrasound imaging.
      zoom view
      Figure 1.33: Subclavian steal syndrome on ultrasound. Color Doppler image with velocity curve demonstrates reversal of the flow in the left vertebral artery that is reflected as blue color on the color Doppler part of the examination. The velocity curve is very abnormal with lack of a sharp peak during the early systole (arrowhead) and marked blood flow below the baseline (arrows) during the diastolic phase of the cardiac cycle.
    9. Reversal of flow in the ECA is visualized in near complete to complete occlusion of the CCA when collateral flow to the ICA is provided from the ECA (Figure 1.28)
    10. With complete occlusion of the ICA, it is sometimes difficult to determine if the ICA or the ECA is occluded. Tapping along the temple close to the superficial temporal artery transmits sharp peaks to the spectral waveform of the ECA only (Figure 1.34). The waveform of a patent ICA will not be affected by the tapping34
zoom view
Figure 1.34: Differentiation of the external carotid artery (ECA) from internal carotid artery (ICA). The ultrasound examination reveals only one vessel distal to the carotid bifurcation without characteristic low or high resistance flow pattern (between arrows). Therefore, temporal tapping was applied with transmission of the tapping to the flow velocity curve reflected as multiple peaks in systole and diastole (arrowheads) indicating that the visualized vessel corresponds to the ECA. The ICA is not affected by temporal tapping (not shown).
The spectral Doppler curves also provide peak systolic (PSV) and end-diastolic (EDV) velocities of the arterial flow within a vessel. Specific velocities have been established to correlate with significant stenosis within the proximal ICA, the most common location of atherosclerotic disease, and are listed in Table 1.3.
All three components (gray scale imaging, Doppler color imaging, and spectral Doppler imaging) are used in the interpretation of the ultrasound examination. Each of these elements provides complementary information about the disease process. Significantly calcified plaque formation is the most commonly encountered limitation of ultrasound. The associated shadowing posterior to the calcified plaques can obscure an entire vessel segment and prevent velocity measurements at the affected level (Figure 1.25). The velocities can be measured distal to the calcified vessel segment but might underestimate the degree of stenosis as the noncalcified portion of the vessel may be normal or narrowed to a lesser degree than at the plaque formation. Falsely low velocity values can also be observed in patients with depressed waveform secondary to poor cardiac output or proximal high-grade stenosis.
Table 1.3   Doppler ultrasound criteria for detection of internal carotid artery (ICA) stenosis.
Extent of internal carotid artery stenosis
Internal carotid artery PSV
Internal carotid artery EDV
< 50%
< 125 cm/s
< 40 cm/s
50–69%
125 – 230 cm/s
40 – 100 cm/s
≥ 70 to near occlusion
> 230 cm/s
> 100 cm/s
Near occlusion
Variable
Variable
Complete occlusion
Untraceable
Untraceable
Adapted from: Grant EG, Benson CB, Monetat GL, et al. Carotid artery stenosis: gray-scale and Doppler US diagnosis - Society of Radiologists in Ultrasound Consensus Conference. Radiology 2003; 229: 340–346. PSV; peak systolic velocity; EDV, end diastolic velocity
35In addition, ultrasound is not able to distinguish near total occlusion (also called ‘string sign’) from total occlusion of a vessel. This is a critical clinical question as patients with near total occlusion of the ICA might qualify for surgical intervention that is not an option for patients with complete occlusion. In such patients, additional imaging, preferably with CTA, is required.
 
CT angiography (CTA) and CT venography (CTV)
CTA and CTV demonstrate the vasculature during the maximum contrast opacification of the arteries and veins, respectively. CTV is rarely requested by the clinician and is typically performed when the patient has a contraindication to MRI. The acquisition of CTV images is more forgiving than with CTA as the intravenous contrast tends to stay longer in the venous structures than in the arteries. Therefore, the injection rate and scan delay (see below) do not play such a critical role as compared with CTA. The majority of routinely obtained contrast-enhanced head and neck CT examinations demonstrate sufficient contrast opacification of the venous structures to exclude a major thrombosis.
CTA is based on the same imaging principle as contrast-enhanced CT but has a shorter delay between contrast injection and the start of scanning. The optimal time point of scanning is primarily patient dependent and can be individualized by the use of automatic bolus tracking techniques. The most widely used tracking method places a region of interest over a large vessel such as the aortic arch. Low-radiation images are then acquired through the region of interest during the contrast injection automatically triggering the start of CT scanning when a specific contrast density is reached. This leads to optimal CTA images in the majority of patients. Only when a very high-grade stenosis is present distal to the region of interest, the vessel segment distal to the narrowing may not be sufficiently opacified with contrast to be adequately visualized on the CTA images.
The contrast injection rate is also critical for CTA imaging. A high injection rate of 4–5 mm per second is optimal because the arteriovenous transit time in the head is very short. Therefore, the tighter the bolus, the lower the venous contamination on the CTA images. Faster CT scanning techniques also facilitate reduction in venous contamination as 64 or higher multislice CT scanners are able to image the entire head within 4 seconds or less. Reduction in the venous contamination might be less optimal when the head and neck CTA is obtained together that requires an acquisition time of 15 seconds or less.
The site of intravenous contrast administration can also influence the quality of CTA images and is especially critical in patients with suspected vessel stenosis or occlusion at the thoracic inlet. Typically, contrast accumulates in the subclavian vein ipsilateral to the injected arm resulting in significant beam hardening artifacts at the thoracic inlet that might obscure the ipsilateral arterial vasculature (Figure 1.35a). 36This contrast stasis can be minimized by flushing of the contrast out of the venous system through injection of saline following the contrast bolus. Usually, this is only partially successful as the contrast injection is often not completed when the scan is started. Therefore, there is typically not sufficient time for the accumulated contrast to be cleared out. A more robust solution is to inject the contrast into the arm contralateral to the side of suspected stenosis (Figure 1.35b and c).
For CT scanners that are capable of simultaneously obtaining perfusion images through the entire brain, both CTA and CTV can be extracted from the perfusion CT data set and no additional contrast administration and radiation exposure are needed to visualize the intracranial vasculature.
zoom view
Figure 1.35: Obscuration of subclavian artery stenosis by beam hardening artifacts from venous contrast stasis. Axial source image of a computed tomography angiography (CTA) performed with left-sided contrast injection (a) shows marked beam hardening artifacts caused by a contrast column (CC) in the venous system at the thoracic inlet on the left that is obscuring the left-sided arteries while the right common carotid artery (CCA) (ca) and subclavian artery (sa) are well visualized. Repeated CTA examination with right-sided contrast administration was done. The source images (b and c) of the repeated examination show normal patent lumen of the left subclavian artery proximally [arrow in image (b)] and 75% stenosis [arrow in image (c)] just distal to it. The coronal reformations (d) reveal an area of severe stenosis (arrowhead) in the left subclavian artery just proximal to the origin of the left vertebral artery (…). The curved reformation (e) along the left subclavian artery is better able to capture and evaluate the entire left subclavian artery (*) including the lateral portion that is not visible on the presented coronal reformation (d).
37In addition, the noncontrast-enhanced CT images can be subtracted from the CTA or CTV source data set leading to removal of the skull and skull base as well as of calcified plaques. This allows better evaluation of heavily calcified vessel segments and of vasculature coursing through the skull base (Figure 1.36). The subtracted images can, however, be falsely positive when significant motion is present between the two data sets. Therefore, it is advisable to review both data sets (nonsubtracted and subtracted) to avoid such an error.
 
MR angiography (MRA) and MR venography (MRV)
MRA and MRV can be obtained with or without Gadolinium administration. In general, Gadolinium-enhanced methods are less prone to artifacts and are more accurate as they do not rely on flow direction and velocities to generate the optimal signal intensity within a vessel.
 
Noncontrast-enhanced MRA
Time of flight (TOF) is the most commonly used technique for MRA of the circle of Willis. It is also applied to the cervical vasculature when the patient has a contraindication to Gadolinium administration. TOF maximizes the differences in signal intensity between flowing blood and stationary tissue.
zoom view
Figure 1.36: Nonsubtracted versus subtracted computed tomography angiography (CTA). Conventional maximum intensity projection (MIP) image of a CTA (a) is not able to demonstrate the internal carotid artery as it courses through the central skull base. The MIP image of a CTA with subtracted bone (b) at the same level reveals marked narrowing of the right petrous segment of the internal carotid artery (ICA) (arrows) when compared with the normal left ICA (arrowheads).
38All stationary tissues are suppressed and displayed in shades of dark gray to black. In contrast, flowing blood assumes a bright signal (Figure 1.37a and b). The highest signal intensity is produced by blood that courses perpendicular to the imaging plane. Any oblique orientation of the blood or turbulent flow will decrease the intensity of the bright signal potentially mimicking areas of stenosis or thrombosis (Figure 1.37a and b). Reverse or sluggish flow will result in more substantial drop in signal intensity that may even imitate complete vessel occlusion or underestimate the size of an aneurysm or arteriovenous malformation (AVM) (Figure 2.27). The venous flow is removed by placement of a so-called ‘saturation band’ over the high convexities of the brain hemispheres. It essentially nulls the signal from flowing venous blood. If the band is not applied or placed to high, venous contamination on the MRA images is seen. In addition, venous contamination might be present in cases of venous shunting secondary to an underlying arteriovenous fistula or AVM (Figure 2.27).
TOF technique can be obtained as two-dimensional (2D) or three-dimensional (3D) acquisition. For the 2D TOF MRA, contiguous but separate slices are obtained that are subsequently stacked together. The 2D technique is more sensitive in detection of slow flow and is therefore more likely to distinguish near total from total occlusion. In contrast, the 3D TOF MRA acquires a certain volume through the head or neck at the same time that is subsequently reconstructed into multiple images.
zoom view
Figure 1.37: Turbulent flow at the bulb of the internal carotid artery (ICA) mimicking atherosclerotic plaque or thrombus. The maximum intensity projection image (a) of the left cervical vasculature shows focal narrowing (white arrowheads) at the proximal bulb of the ICA with a distal step off formation suggesting atherosclerotic plaque or thrombus. This area is, however, prone to turbulent flow because of the physiological widening of the bulb. Turbulent flow is known to cause signal dropout on noncontrasted magnetic resonance angiography (MRA), which is also appreciated on the axial MRA source image (b) reflected as decreased attenuation (arrows) in the posterior bulb of the left ICA. The ill-defined appearance of the signal dropout (arrows) on the MRA source image (b) is more typical for turbulent flow than atherosclerotic plaque formation that usually shows more distinctive margins. Confirmation of the artifactual nature of this narrowing was obtained with Gadolinium-enhanced MRA (c) that demonstrates a normal physiological widening of the bulb (*) without evidence of plaque formation or thrombus. Notice that the Gadolinium-enhanced MRA (c) is able to display the entire carotid artery system from the aortic arch (arch) to the cavernous segment (…), while the noncontrasted MRA (a) covers only the carotid bulb and adjacent vascular segments. The faint black bands (black arrowheads) extending in regular distances through the vertebral and common carotid arteries on the noncontrasted MRA (a) represent the overlapping bands of multiple overlapping thin slab acquisition MRA technique.
39The 3D technique has a higher spatial resolution compared with the 2D TOF acquisition. It is, however, hampered by inherent T2 weighting that can mistake a recently formed thrombus for a patent lumen or a focus of intracranial subacute hemorrhage for an aneurysm (Figure 1.38). Therefore, the 3D TOF technique should never be performed without supplementation with conventional MRI images.
Multiple overlapping thin slab acquisition (MOTSA) is a hybrid technique that provides the resolution of the 3D TOF technique and the flow-related advantages of the 2D method. It is the most commonly utilized technique in the clinical practice. The different thin slabs are obtained with some overlap and subsequently stacked together to display a longer vessel segment in one data set. The overlapping segments of the two adjacent slabs usually create a thin black line that extends through the post-processed images (Figure 1.37a).
All three noncontrast MRA methods are prone to artifacts. As outlined above, turbulent flow and reversal of the flow direction in a tortuous vessel can mimic areas of stenosis or even occlusion. The carotid bulb represents an area of natural widening that predisposes to turbulent flow. Therefore, the bulb frequently demonstrates areas of mixed signal attenuation that can be mistaken for a focal thrombosis (Figure 1.37a and b).
zoom view
Figure 1.38: Normal magnetic resonance angiography (MRA) appearance in vessel occlusion. MRA source image (a) at the level of the inferior posterior fossa shows normal flow within the internal carotid arteries (white arrows) and vertebral arteries (arrowheads) bilaterally. The subsequently performed computed tomography angiography (b) reveals complete occlusion of the left vertebral artery (black arrow) when compared with its normal enhancement on the right (black arrowhead). Note that on the noncontrasted MRA image (a), the signal intensity within the left vertebral artery (left arrowhead) is artifactural in nature and caused by T2 signal of acute thrombus that mimics a patent lumen.
40Variable degrees of turbulent flow can also be seen in aneurysms and in AVMs causing underestimation of the size and extent of the underlying disease (Figure 2.27). Strong bony interfaces may also cause artifacts and make the vessels that course through them appear more narrow or even occluded. This artifact is most prevalent at the skull base and typically affects the petrous and proximal cavernous segments of the ICA (Figure 1.39a–c). Comparison to Gadolinium-enhanced MRA (Gd MRA) of the neck, when available, is usually helpful in deciding if this is related to artifact or not (Figure 1.39d).
 
Gadolinium-enhanced MRA (Gd MRA)
Gd MRA is based on a modified TOF technology. Since Gadolinium is used to generate the signal intensity within the vessel, the TR time can be significantly reduced with Gd MRA that allows for faster imaging time than with noncontrast-enhanced MRA techniques. In addition, the image acquisition can be obtained in any desirable plan as the sequence is not dependent on flow effects. For the neck MRA, the coronal imaging plane requires the thinnest slab to cover the major cervical vessels that is significantly less in thickness than for the noncontrast-enhanced MRA providing additional time savings. Therefore, the typical Gd MRA takes approximately 30 seconds in contrast to 6- to 10-fold longer scanning time for the noncontrasted MRA versions. The faster acquisition time significantly reduces the risk of motion degradation and is better tolerated by the patient.
zoom view
Figure 1.39: Artifacts at the skull base mimicking areas of stenosis in the internal carotid artery (ICA) on magnetic resonance angiography (MRA). The maximum intensity projection (MIP) image of the noncontrasted MRA of the circle of Willis (a) shows areas of severe stenosis just proximal to the petrous segment of the right ICA [arrows in image (a)] and in the proximal cavernous sinus segment of the left ICA [arrowhead in image (a)]. The regions of narrowing are confirmed on the MRA source images just below the skull base [arrow in image (b)] and at the cavernous sinus level [arrowhead in image (c)]. These vessel segments are often affected by artifacts from the adjacent skull base. Therefore, it is unclear if the areas of stenosis are real or not. Composite MIP image of the Gadolinium enhanced MRA (d) performed in the same setting covers the proximal circle of Willis and demonstrates normal caliber of all vessels at the skull base and cavernous sinus levels.
41In addition, Gd MRA allows imaging of a longer vascular segment of the cervical vasculature with inclusion of the aortic arch and the proximal aspects of the circle of Willis (Figure 1.37c). In contrast to the noncontrast-enhanced MRA techniques, it is not prone to artifacts at the carotid bulb level or skull base (Figures 1.37 and 1.39). Therefore, it is more accurate in the assessment of the degree of stenosis and extent of an aneurysm or AVM.
The main disadvantages of Gd MRA are related to the Gadolinium administration that requires an intravenous line placement, high injection rate of 4–5 mm per second and appropriate timing of the contrast bolus. As with CTA, automatic bolus tracking techniques are available to optimize the arterial opacification. Gd MRA is usually not used for scanning of the circle of Willis for two reasons. First, the short arteriovenous transit time of the brain will lead to significant venous contamination that might partially obscure the intracranial vasculature, especially when in proximity to venous structures such as the cavernous sinuses. Second, the time saving is substantially higher in the neck, and therefore, a noncontrasted MRA of the circle of Willis and Gd MRA of the cervical vasculature are typically combined. In addition, a repeated injection of Gadolinium is clinically not desired because of the increased risk of NSF with higher Gadolinium dosage and because of the image degradation of the second MRA acquisition by the venous contamination from the first MRA injection.
 
MR venography (MRV)
MRV is typically obtained without Gadolinium. As with noncontrast-enhanced MRA, it is also based on the concept of the TOF technique optimized to visualize slower flow. In addition, MRV is performed in a different imaging plane than the MRA as the transverse sinuses, the straight sinus, and the majority of the superior and inferior sagittal sinuses course within the axial plane. Therefore, imaging in the axial plane would result in signal dropout that in turn might mimic areas of stenosis or thrombosis. Hence, MRV images are usually obtained in an oblique sagittal or oblique coronal plane to minimize the in-plane course of the venous structures. Even with this method, the posterior aspect of the superior sagittal sinus commonly shows decreased signal intensity that might be mistaken for focal area of thrombosis (Figure 1.40). As with noncontrast-enhanced MRA methods a ‘saturation band’ is placed, but over the neck, to eliminate the signal generation by incoming arterial flow from proximally.
The main limitation of MRV is the frequent presence of anatomical variations within the venous system. Symmetric appearance of the transverse sinuses is seen only in about one-third of patients. In >50% of patients, the left transverse sinus is aplastic (20%) or hypoplastic (39%). Therefore, the nonvisualization of the left transverse sinus on a MRV examination might be simply related to a normal anatomical variation and not to a sinus thrombosis. In cases of chronic venous sinus thrombosis, the thrombosed venous segment might not recanalize.42
zoom view
Figure 1.40: Common flow-related artifacts on magnetic resonance venography (MRV). The maximum intensity projection (MIP) images of a MRV in sagittal (a) and anterior posterior (b) plane demonstrate filling defects in the posterior aspect of the superior sagittal sinus (arrowheads). These represent flow-related artifacts and are commonly observed in this segment of the sagittal sinus.
It also tends to shrink over time mimicking congenital hypoplasia or aplasia of a vessel. In addition, MRV images are usually not able to display small veins because of their suboptimal spatial resolution and their inability to generate adequate flow-related signal. Furthermore, arachnoid granulations often protrude into the venous sinuses reflected as filling defects within different portions of the venous sinuses on the MRV images. Such arachnoid granulations might be mistaken for focal thrombi (Figure 5.10). In contrast to a thrombus, an arachnoid granulation is usually more round in appearance and follows the signal intensity of CSF on all sequences, which helps to distinguish it from a small thrombus (Figure 5.10).
The above-mentioned limitations emphasize that noncontrasted MRV images are not very reliable in the evaluation of a patient suspected to have venous sinus thrombosis. Some authors suggest better performance of Gadolinium-enhanced MRV; however, the lower resolution of MRV and the presence of significant anatomical variations of the venous system cannot be completely overcome by the addition of Gadolinium. More effective solution is to combine the MRV with conventional MRI sequences (FLAIR, T1- and T2-weighted images), volumetric Gadolinium-enhanced T1-weighted acquisition, and a SWI sequence for the following reasons:
  • The conventional sequences and the high-resolution volumetric contrast-enhanced T1-weighted images are able to demonstrate a discrepancy in the size of the venous sinuses between the MRV and the other sequences that is often caused by partial or complete thrombosis of a sinus (Figure 5.5)
  • The high-resolution volumetric Gadolinium-enhanced T1 sequence is able to distinguish sluggish flow from a thrombosed vessel segment. This is not possible with MRV or the other conventional sequences.
  • The conventional and Gadolinium-enhanced sequences are more readily able to distinguish an arachnoid granulation from a localized thrombus (Figure 5.10)
  • The SWI images facilitate detection of thrombosis of smaller veins that will appear bright instead of dark on the mIP images43
  • The SWI images are more likely to detect focal areas of intraparenchymal hemorrhage that might be related to hemorrhagic venous infarction. Therefore, the physician's attention is guided to a specific area for more in depth analysis
 
Post-processing techniques
The acquired CTA, MRA, CTV, and MRV source images typically undergo different post-processing techniques to improve the display of the cervical and intracranial vasculature. Each of these post-processing techniques has advantages and disadvantages:
 
Surface rendering
The attenuation of the outer vessel border is used as the threshold for separation of the vessel of interest from the adjacent anatomical structures (Figure 1.41a). This 3D technique underestimates the degree of vessel stenosis as the threshold is along the outer side of the stenotic area. In addition, this method is unable to differentiate calcified plaques or stents from the contrast column in the vessel, leading to an underrating of the degree of stenosis. On CTA, surface rendering is also hampered by difficulty in separating bony structures that are directly abutting the vessel wall from the enhancing vascular lumen. This is most apparent when the vertebral arteries are selected. Often, the bony transverse processes will be partially incorporated in the display of the vertebral artery making its evaluation impossible or unreliable (Figure 1.42). Application of bone elimination techniques or manual removal of the bony structures can avoid this shortcoming but are time consuming.
 
Maximum intensity projection (MIP)
The maximum intensities of the of flowing blood or intraluminal contrast column are used to create the vessel lumen on non-contrast enhanced MRA and contrast enhanced CTA/MRA, respectively. Since less blood flow or contrast accumulates along the vessel walls, these areas are not included in the reformation algorithm.
zoom view
Figure 1.41: Difference between post-processing techniques. In surface rendering (a) the vessel (red) is captured by placing the threshold value along the outer surface of the vessel [thick blue lines in image (a)]. In contrast, the maximum intensity projection technique (b) uses the maximum signal intensity values that are characteristically medial to the actual vessel lumen [thick blue lines in image (b)]. Only the multiplanar reformations images (c) reflect the true lumen of the vessel. Therefore, their reconstruction is exactly along the vessel wall.
44
zoom view
Figure 1.42: Bone elimination shortcomings during post-processing. Surface shaded images (a and b) of the vertebral arteries demonstrate suboptimal elimination of the adjacent transverse processes. In image (a), some of the transverse processes (*) are still included on the post-processed image and obscure the view upon the vertebral artery, while in other areas segments of the vertebral arteries are missing together with the transverse processes resulting in gaps within the vessel (arrows). In image (b), small portions of the transverse processes (arrowheads) are included at each level that is significantly affecting the view upon the vertebral artery only at the most inferior level, while the other levels are barely obscured by the remaining bony fragments.
Therefore, the MIP images overestimate the degree of stenosis (Figure 1.41b). Similar to the surface rendering technique, this method is unable to separate calcified plaques or stents from the patent lumen that may obscure the patent vessel lumen. In contrast to the surface rendering technique, however, MIP images can be displayed with different slice thickness and using different image displaying window and level settings. Therefore, reduction in the slice thickness in combination with the appropriate window and level settings is often able to adequately display the patent vessel lumen (Figure 1.43). As with surface rendering, the MIP method has difficulties in separating bony structures from the enhancing vascular lumen on CTA and requires bone elimination techniques or manual removal of the bone. Nevertheless, MIP is the most commonly used post-processing algorithm. In particular when MRA, MRV, or subtracted CT applications are used, it is able to easily display a large portion of the intracranial or cervical vasculature. The overestimation of an area of stenosis is typically an advantage as the MIP images provide a quick overview to the physician (Figure 1.44a). Areas of concern can be then more accurately assessed using the CT or MR source images or their multiplanar reformations (Figure 1.44b).
 
Multiplanar reformations
Similar to regular CT images, multiplanar reformations can be obtained in any desirable plane including along a curved line (Figure 1.35d and e). They create a 2D view of a vessel without any loss of information as the source images are not post-processed but only displayed in an arbitrary plane (Figure 1.44c). The multiplanar reformations can also be displayed with different slice thickness and therefore are able to capture the entire vessel lumen if desired.45
zoom view
Figure 1.43: Thick versus thin maximum intensity projection (MIP) images. The thick MIP image (a) in a patient with a carotid artery stent (arrows) completely obscures the view upon the vessel lumen at the stent segment in contrast to the thin section MIP image (b) that clearly is able to display the vessel lumen (*) surrounded by the stent (arrowheads).
zoom view
Figure 1.44: Overestimation of stenosis on the maximum intensity projection (MIP) image. The MIP image (a) of the left carotid bifurcation demonstrates focal area of complete occlusion (arrow), while the curved MPR computed tomography angiography image (b) reveals that this vessel segment is only narrowed by approximately 50% (arrow) when compared with the more distal internal carotid artery vessel lumen.
In contrast to the other post-processing techniques, it can also display calcified plaques, and differentiate between the vascular calcifications and the contrast column in the vessel lumen with the appropriate choice of window and level (Figure 1.45). The multiplanar reformations and the source images are the only data sets that accurately display the areas of stenosis and are routinely used by the radiologist in the CTA or MRA assessment of vessels. Bone elimination techniques are usually not applied with this method.46
zoom view
Figure 1.45: Window and level setting influencing the appearance of a vessel. Axial computed tomography angiography (CTA) source images displayed in different window and level settings (a and b) illustrate the potential pitfalls of viewing of CTA images in suboptimal display. In image (a), high contrast level was chosen that is suggesting marked enlargement of the left internal carotid artery (arrow) when compared with the normal caliber on the right (i). This appearance could be misinterpreted as pseudoaneurysm formation. In image (b), a more optimal combination of window and level setting was selected, revealing that the enlargement is caused by calcified plaques (arrowheads) along the outer surface of the vessel. The patent lumen of the vessel (arrow) is markedly smaller than suggested in image (a).
 
Volume rendering
Volume rendering is a 3D technique that transfers the vessel attenuation into different colors and opacities. This technique is heavily operator dependent as different threshold values and different ranges of vessel attenuation can be selected that markedly influence the appearance of the post-processed images (Figure 1.46). For example, inclusion of lower attenuation values will also incorporate smaller caliber vessels, while a narrow range of attenuation values will lead to a ridged appearance of the vessel borders. Therefore, volume rendering is unreliable in assessment of the degree of vessel stenosis. It is most commonly used to provide 3D CTA display of the intracranial vasculature as close as possible to their normal anatomical appearance. This information is valuable to the neurosurgeon for surgical planning purposes when an aneurysm or AVM is present. Bone elimination techniques are routinely applied with this method.
In summary, only the source images and multiplanar reformations accurately display the patent vessel lumen and should always be primarily used for interpretation purposes of CTA/CTV and MRA/MRV studies. One needs to adjust the window and level settings to separate calcified plaques from the contrast column within the vessel (Figure 1.45). Any post-processing technique that uses threshold or attenuation values can result in alteration of the vessel lumen! In particular, surface and volume rendering methods are unreliable because they underestimate the degree of stenosis or are operator dependent. Only MIP images are valuable as their overestimation of the degree of narrowing leads the physician's eye to the areas of concern.47
zoom view
Figure 1.46: Variability of volume rendering images. Volume rendering allows wide range of customization for post-processing of computed tomography angiography images leading to variable appearance of the vasculature. Volume-rendered images of the carotid arteries are shown using three different settings as detailed in the graphs below each image. The carotid arteries assume different appearance with each setting suggesting a high-grade stenosis at the origin of the internal carotid artery (ICA) (arrow) and mild proximal common carotid artery (CCA) stenosis (arrowhead) in image (a). While the ICA stenosis (arrow) becomes less apparent in images (b) and (c), the proximal CCA stenosis is progressive in nature and assumes a moderate grade (arrowhead) in image (c). The wide range of findings makes it difficult to determine which of the volume-rendered images is correct. Only the source and multiplanar reformations (MPR) images are accurate. The curved MPR (d) aligned with the course of the CCA and ICA reveals only minor stenosis at the origin of the ICA (arrow) and 75% stenosis in the proximal CCA (arrowhead). Therefore, the volume-rendered image in c best displayed the severity of the abnormalities.
Suggested Reading
  1. Alley MT, Shifrin RY, Pelc NJ, Herfkens RJ. Ultrafast contrast-enhanced three-dimensional MR angiography: state of the art. Radiographics 1998; 18: 273–285.
  1. Alper, et al. Importance of anatomical asymmetries of transverse sinuses: an MR venographic study. Cerebrovasc Dis 2004; 18: 236–239.
  1. Haacke EM, Mittal S, Wu Z, et al. Susceptibility-weighted imaging: technical aspects and clinical applications, Part 1. AJNR 2009; 30: 19–30.
  1. Morcos SK, Thomsen HS, Webb JAW, and members of the Contrast media Safety Committee of the European Society of Urogenital Radiology (ESUR). Dialysis and contrast media. Eur Radiol 2002; 12: 3026–3030.
  1. Nair JR, Van Hecke W, De Belder F, et al. High-resolution susceptibility-weighted imaging at 3T with a 32-channel head coil: technique and clinical applications. AJR 2013; 195: 1007–1014.
  1. Patay Z. Diffusion-weighted MR imaging in leukodystrophies. Eur Radiol 2005; 15: 2284–2303.
  1. Phillips CD, Bubash LA. CT angiography and MR angiography in the evaluation of extracranial carotid vascular disease. Radiol Clin North Am 2002; 40: 783–798.
  1. Robbin ML, Lockhart ME. Carotid artery ultrasound interpretation using a pattern recognition approach. Ultrasound Clin 2006; 1: 111–131.
  1. Salgado R, Mulkens T, Ozsarlak O, et al. CT angiography: basic principles and post-processing applications. JBR-BTR 2003; 86: 336–340.
  1. Santhosh K, Kesavadas C, Thomas B, et al. Susceptibility weighted imaging: a new tool in 48magnetic resonance imaging of stroke. Clinical Radiol 2009; 64: 74–83.
  1. Sheth S, Branstetter BF, Escott EJ. Appearance of normal cranial nerves on steady-state free precession MR images. Radiographics 2009; 29: 1045–1055.
  1. Tamasebpour HR, Buckley AR, Cooperberg PL, Fix CH. Sonographic examination of the carotid arteries. Radiographics 2005; 25: 1561–1575.
  1. Tomandl BF, Klotz E, Handschu R, et al. Comprehensive imaging of ischemic stroke with multisection CT. Radiographics 2003; 23: 565–592.
  1. Wu Z, Mittal S, Kish K, et al. Identification of calcifications with magnetic resonance imaging using susceptibility-weighted imaging: a case study. J Magn Reson Imaging 2009; 29: 177–182.