Discovered by Wilhelm Roentgen in 1895, x-rays are photons carrying electromagnetic energy which are created by an anode-cathode system within a vacuum.³ These photons are of higher energy and shorter wavelength than visible light. When photons collide with atoms of varying sizes, they either pass through or are absorbed. Larger (heavier, radiopaque) atoms, such as calcium or metals, are more likely to absorb the energy of the photons than smaller (lighter, radiolucent) atoms and small molecules, such as water or air. When a patient is positioned in a beam of x-rays, the x-rays will be differentially absorbed based on the tissue components (bone, soft tissue, aerated sinuses). Photons that pass through the patient strike a detector and create an x-ray image, producing a two-dimensional projected image of the different attenuation properties of body tissue.

Fluoroscopy is a variation of radiography in which images are obtained in rapid succession and displayed in real time on a screen. In angiography, intravenous contrast material is injected into vessels during continuous fluoroscopy. Digital subtraction angiography (DSA) is a technique in which a baseline or mask image is initially obtained of the area of interest. This baseline image is subtracted from subsequent images obtained during intravascular contrast injection, optimizing visualization of the contrast agent itself within opacified vessels. With modern angiography systems, images can be acquired from multiple different projections during a single contrast injection by rapidly rotating the fluoroscopy unit around the patient, allowing reconstruction of a three-dimensional image of the vascular structures.

Computed Tomography

Sir Godfrey Hounsfield and Dr. Allan Cormack invented the first computed axial tomographic scanner in 1972, which earned them the Nobel Prize for Medicine in 1979.⁴ Computed tomography (CT) scanners have advanced significantly since that time, rapidly increasing in speed and resolution. Modern scanners use a rotating x-ray tube and detector array that revolve around the body, obtaining tissue attenuation information from beams or rays of tissues within a slab. Standard axial images are obtained by applying a reconstruction algorithm, typically filtered back projection, to reconstruct the two-dimensional image.⁵ Sagittal, coronal, or oblique imaging planes can be reconstructed from the axial sequences by computer reformatting. Radiodense contrast material administered intravenously or parenterally can outline hollow structures such as blood vessels or the digestive system.

CT density is quantitatively measured in Hounsfield units (HU).⁵ Hounsfield units describe a linear scale of attenuation that is constant across scanner platforms, with water and air given arbitrary values of 0 and −1000, respectively. Materials with increased x-ray attenuation with respect to water have a positive HU value, and those with less x-ray attenuation than water have a negative HU value (Table 3.1).^4,⁶

TABLE 3.1 Computed Tomography Hounsfield Unit Values

Tissue	Hounsfield Units
Air	−1000
Fat	−(60-100)
Water	0
White matter	35
Gray matter	45
Blood—acute hemorrhage	50-70
Calcium	>150
Dense bone	1000
Metal	>>1000

CT images can be viewed in different ways to accentuate different tissues. Window level describes the center point of the gray scale, and window width describes the range of CT values displayed.⁶ For example, gray matter has an attenuation of approximately 35 HU, and white matter attenuation is approximately 45 HU. In order to differentiate gray matter from white matter, a narrow window is needed to highlight small changes in HU values. On the other hand, if detailed evaluation of dense material such as bone is desired, a wide window better delineates the margins. Window level and width are easily manipulated using most imaging viewing software.

In addition to window width and level, CT scans are generally processed using different reconstruction filters, frequently referred to as bone and standard algorithms.⁵ Both filters can be applied to a single acquisition of data, allowing accentuation of different structures. Standard algorithm is a method of averaging adjacent pixels to accentuate soft tissue detail. Standard algorithm images are useful for evaluating gray-white matter differentiation and for detecting blood. Bone algorithm images are processed to maximize edges, thus accentuating high-density materials such as calcium and metal. Bone algorithm images are also useful for evaluating lung parenchyma due to the differences in attenuation between aerated lung and small soft tissue attenuation structures such as blood vessels and pulmonary nodules.

Issues with Computed Tomography

Radiation Exposure

There is increasing awareness among health care providers and the general public of radiation exposure from medical imaging and the carcinogenic potential of x-rays. This is in part related to the dramatic increase in the utilization of CT over the past few decades. According to the American College of Radiology White Paper on Radiation Dose in Medicine ⁷ approximately 3 million CT studies were performed in 1980, compared to approximately 60 million in 2005. Although CT has undoubtedly contributed positively to the care of patients, the cumulative radiation dose may have increased the risk of cancer in exposed patients, and up to 1% of U.S. cancers may be related to medical exposures. Based on studies of Japanese atomic bomb survivors, cancer risk increases with exposures as low as 50 mSv (millisieverts). Millisieverts are a measure of effective radiation dose, which is weighted for tissue sensitivity to the negative effects of radiation.⁸

Cancer risk depends on tissue type, and neural tissue is relatively resistant. Exposure to more radiosensitive tissues may also occur with neuroimaging. For example, exposure of the cornea may lead to cataracts in a dose-dependent fashion. The lens of the eye receives a dose of 40 to 50 mGy (milliGray, a measure of absorbed dose) per head CT,^9,¹⁰ which can be reduced by eye shielding. Lens opacities have been seen with as little exposure as 500 mGy (10 CT scan equivalents), with vision-limiting cataracts forming at doses greater than 4 Gy (approximately 80 CT scan equivalents). Children are more susceptible than adults.

The ALARA principle (As Low As Reasonably Achievable) of keeping exposure to a minimum is an important guideline to follow when imaging patients, particularly when ionizing radiation will be used. This principle aims to balance the clinical benefit of the imaging study with the risks, however low they may be. Additionally, physicians should take care to protect themselves when using fluoroscopy for angiography and during implant and spine procedures. Specific questions about radiation exposure and protection can be addressed to the local staff radiologist or medical physicist.

Iodinated Contrast Agents

Iodinated contrast material may be associated with contrast-induced nephropathy (CIN) in patients with renal failure, particularly those with diabetes mellitus. Strategies to reduce the risk of CIN include volume expansion through intravenous or oral fluids. Sodium bicarbonate infusion and prophylactic N-acetylcysteine have also shown efficacy compared to normal saline infusion.¹¹ In patients with diminished renal function, reduced doses of contrast agent or iso-osmolar nonionic contrast agent can be considered. Of course, the best prevention of CIN is the avoidance of intravenous contrast material altogether, although this is not always feasible.

Another potential complication of iodinated contrast agent is contrast reaction. The incidence of contrast reaction after CT contrast scan is 0.2% to 0.7% (approximately 1/225).^12,¹³ Contrast reactions may be mild or severe. Most reactions are mild, including nausea, vomiting, or rash. Severe reactions occur infrequently, with an incidence of approximately 0.05% (1/2000) for low osmolar iodinated contrast agent.^12,¹³ Severe contrast reactions include bronchospasm, laryngeal edema, and cardiovascular collapse.

In patients with a history of moderate or severe contrast allergy, premedication strategies decrease, but do not eliminate, the risk of recurrent contrast reaction. Premedication strategies include prednisone (50 mg by mouth at 13, 7, and 1 hour prior to contrast injection) or methylprednisolone (32 mg by mouth at 12 and 2 hours before contrast injection), along with diphenhydramine (50 mg 1 hour prior to injection).¹⁴

Magnetic Resonance Imaging

Magnetic resonance imaging (MRI) was developed by a host of innovative scientists over multiple decades of development from a scientific tool to a medical imaging necessity. MRI is based on the principles of nuclear magnetic resonance (NMR), first discovered by Felix Bloch and Edward Purcell, for which they were awarded the Nobel Prize in physics in 1952. NMR can be utilized to characterize and differentiate tissues based on their intrinsic NMR signal. Using NMR techniques, chemist Paul C. Lauterbur and physicist Sir Peter Mansfield developed the gradients and mathematical formulations required for rapid 2-dimensional MR images, publishing the first images in 1973 ¹⁵ and 1974.¹⁶ Drs. Raymond Damadian, Larry Minkoff, and Michael Goldsmith were also instrumental in the development and refining of this technology for use in humans.¹⁷ For their pioneering work in MRI development, Drs. Lauterbur and Mansfield shared the Nobel Prize in physiology or medicine in 2003.

The detailed physics principles underlying MRI are highly complex and beyond the scope of this chapter. In brief,¹⁸ a powerful electromagnetic field is created within the bore of an MRI machine, typically along the cranial-caudal (z) axis. Protons that are present in the human body predominantly as hydrogen atoms in water molecules reach equilibrium aligned along the direction of this magnetic field (longitudinal magnetization). A radiofrequency (RF) pulse is applied at a resonance frequency specific to the protons within the main magnetic field (B-zero), causing them to absorb energy and change their alignment toward the horizontal/vertical plane (x-y axes), called transverse magnetization. When the RF pulse ends, the protons first dephase in the x-y direction (free-induction decay, the basis of T2 signal) at a rate dependent on the molecular structure of the sample. The protons then realign along the z-axis (spin-lattice relaxation, the basis of T1 signal) at a slower rate, which is dependent on the molecular structure surrounding the proton. As the protons realign toward equilibrium, they emit RF energy that is detected by antennas (receiver coil) surrounding the patient in the scanner. By inducing small changes in frequency and phase of the proton resonance frequency that vary as a function of proton position, the MRI system reconstructs the precise location of each signal within the patient. The MRI system thus produces cross-sectional images through the patient where each pixel (corresponding to a defined volume of tissue, or voxel) depends upon the magnetic microenvironment of the corresponding tissue.

From this general principle, a variety of pulse sequences have been developed to emphasize different tissue characteristics.¹⁸ A pulse sequence refers to a specific pattern of RF pulses that may vary in timing, order, repetition, and direction. Basic pulse sequences include spin echo, inversion recovery including short tau inversion recovery (STIR) and fluid attenuated inversion recovery (FLAIR), and gradient echo imaging. The clinical applications of those sequences most pertinent to neuroradiology are described in this chapter.

MR angiography (MRA) can be performed by several techniques, including time-of-flight, phase-contrast, and gadolinium-enhanced MRA. In time-of-flight angiography, protons in moving blood are tagged in one tissue slab by applying an RF pulse to change their longitudinal magnetization.¹⁸ Tagged protons are subsequently detected in a different tissue slab that has not experienced the RF pulse. The direction of blood flow can be selected by applying a saturation pulse to null the longitudinal magnetization from protons traveling in the opposite direction. For example, to selectively visualize tagged blood protons moving superiorly within the cervical arteries, a saturation pulse is applied superior to the scan volume (e.g., within the head) to neutralize the longitudinal magnetization of the tagged protons within the intracranial compartment before they travel inferiorly in the cervical veins.

Phase contrast angiography is another method to detect moving protons such as those in blood or CSF.¹⁸ Phase contrast imaging depends on applying bipolar gradients to protons, so that stationary protons experience both positive and negative gradients, with no net phase change. Moving protons experience only one gradient before moving out of the field, resulting in positive or negative excitation.

Contrast-enhanced angiography uses gadolinium-based contrast agents to highlight blood vessels. Gadolinium is strongly paramagnetic, resulting in significant T1 shortening (high signal on T1-weighted sequences). After intravenous administration of gadolinium, T1-weighted sequences can be obtained during the arterial or venous phase to highlight vascular anatomy. Three-dimensional reformatted sequences can be constructed after any of the angiographic techniques.

Issues with Magnetic Resonance Imaging

MRI offers many advantages over CT. MRI provides excellent soft tissue detail and does not use ionizing radiation, but it does have several drawbacks. Study times for MRI are significantly longer than those for CT. MR images are significantly degraded by motion artifact, which coupled with the longer scan times becomes problematic in acutely ill patients and children. MRI coils must be in close approximation to the body area being imaged, and the bore of the MRI scanner is generally both smaller and more enclosed than CT. This can be a significant problem for patients with claustrophobia or those with a large body habitus.

Because of the strong magnetic fields used for MRI, it is not safe to scan patients with certain metal implants or ferromagnetic foreign bodies. Some metal implants or foreign bodies can move because of the influence of the external magnetic field, with potentially devastating consequences depending on the nature of the object (e.g., older ferromagnetic aneurysm clips). Certain metallic objects can heat during scanning, which can be uncomfortable to the patient or cause tissue damage. These effects of the magnetic field are more pronounced at higher field strength magnets (that is, 3-tesla compared to 1.5-tesla scanners). In addition, certain implants, such as vagal nerve stimulators, deep brain stimulators, and cardiac pacemakers, can malfunction during exposure to strong magnetic fields. It is worth remembering that implanted devices or other foreign bodies do not need to lie within the area of interest (scan volume) in order to be affected by the magnetic fields of the MR scanner, and that in most modern scanners the main magnetic field is always present.

There are several resources available to determine whether a particular implant is compatible or considered safe to scan with MRI, including a website maintained by the Institute for Magnetic Resonance Safety, Education, and Research (http://mrisafety.com).¹⁹ This website includes a free searchable database of the safety profile of many different implants and devices. The radiologist or MRI technologist can also provide valuable advice on MR safety.

Gadolinium Contrast Agents

Gadolinium-based contrast agents used in MRI have associated risks. The risk of contrast reaction after gadolinium-based contrast agent injection is lower than for iodinated contrast, with a reported incidence of 0.04% to 0.07% (approximately 1/2000).^12,¹³ In one large series of contrast reactions after gadolinium-based contrast agent, 88% were considered mild. The overall incidence of severe reactions (those requiring epinephrine for treatment) was 0.001% to 0.01% of injections (approximately 1/20,000).

A relatively recent development in the use of gadolinium-based contrast agents has been the recognition of the link between contrast administration and nephrogenic systemic fibrosis (NSF) in patients with renal failure. NSF is a progressive fibrosing disease affecting the skin and soft tissues, often of the extremities. NSF may also affect striated muscle and the diaphragm.^20,²¹ There is no clearly effective treatment. This entity is believed to be associated with gadolinium deposition in tissues and is not prevented by dialysis.²² The risk of NSF may vary depending on the particular contrast agent, but this is still under investigation.

NSF is rare, with an incidence of 1% to 7% in patients with renal dysfunction ^12,²¹ who receive gadolinium. It is associated with severe acute renal failure or chronic renal failure with an estimated glomerular filtration rate (GFR) of less than 15 to 30 mL/minute. It is also associated with renal or liver transplantation.²⁰ Because of this association, gadolinium agents should be used with caution in patients with compromised renal function and should be avoided if possible in patients with GFR less than 30 mL/minute. As research in this area continues to evolve, discussion of these cases with the local radiologist is recommended.

Clinical Imaging

Radiography

Once the mainstay of neuroradiology, skull radiography and its permutations have been largely replaced by cross-sectional imaging modalities such as CT and MRI. Radiography is still used in the evaluation of shunts and other neurosurgical implants such as intrathecal pumps and deep brain stimulators.

Shunt series are the most commonly encountered of these studies and include radiographs of the shunt components in two planes. Shunt series are used to evaluate the nature of the shunt, including the location of the ventricular and distal catheters, drainage location (e.g., atrial, peritoneal, pleural), and the type and setting of the shunt valve, as well as to identify causes of shunt dysfunction.²³ Shunt catheters, valves, and tubing can be quickly examined for kinks or disruption, without the cost or radiation exposure of CT or the time and cost of MRI. The type and setting of most implanted shunt valves can also be determined based on radiographic appearance.

Radiography is also sometimes used in the evaluation of the bony calvarium. Although CT has largely replaced radiography for evaluation of facial fractures or sinusitis, skull films are occasionally obtained for these purposes. Linear nondisplaced skull fractures that may be missed on axial CT can at times be detected by radiography. Such occult fractures can sometimes be visualized by either examining the scout tomographic view obtained as part of a routine CT scan, or by constructing surface rendered reformats of CT images. In addition, skull films are sometimes obtained to evaluate for radiolucent bone lesions prior to the placement of stereotactic frames for gamma knife treatment or stereotactic biopsy.

Computed Tomography

The nonenhanced head CT has become the workhorse of acute neuroimaging, due to its wide availability, speed, and relatively low cost. CT provides rapid imaging of many intracranial processes and is excellent for the detection of intracranial hemorrhage, mass lesions, and evaluation of the ventricular system. CT is highly sensitive for calcifications, fat, and air, as well as metallic foreign bodies. In addition, CT allows rapid evaluation of the sequelae of intracranial pathology, including mass effect and brain herniation. Potential drawbacks of head CTs include radiation exposure and cost. In addition, MRI is more sensitive at detecting many parenchymal processes such as stroke, subtle enhancement, and small masses.

Noncontrast Imaging

Because of the importance of the noncontrast head CT to the practice of neurosurgery, clinically relevant findings commonly found on head CT are described here.

Mass Effect and Herniation

Space-occupying lesions within the intracranial compartment distort normal cranial anatomy and can cause mass effect and brain herniation. Mass effect can affect local structures, distorting the ventricles or narrowing the cerebral sulci. Focal mass effect can also result in herniation of brain across fixed structures such as the falx cerebri and tentorium. Brain herniation is an important predictor of severe neurological injury.

There are several types of brain herniation, including subfalcine, uncal, transtentorial (upward and downward), and tonsillar herniation (Fig. 3.1). Brain can also herniate extracranially through a skull defect. CT is excellent at detecting all types of brain herniation.

FIGURE 3.1 Brain herniation patterns: A, B, and C, Normal head computed tomography (CT) scan in a 21-year-old showing normal appearance of the perimesencephalic and ambient cisterns (black arrowheads, A), uncus (white arrow, B), and falx cerebri and septum pellucidum (white arrowheads, C). D and E, Downward transtentorial herniation in a 50-year-old woman with rapid neurological decline. There is complete effacement of the perimesencephalic and ambient cisterns (black arrowheads, E) with a Duret hemorrhage in the pons (black arrow, D). F and G, Uncal and subfalcine herniation in a 71-year-old confused man with an acute subdural hemorrhage (SDH). Note the medial placement of the uncus and parahippocampal gyrus (white arrows, F), enlarged temporal horn of the right lateral ventricle (red arrows), and rightward shift of the septum pellucidum under the falx cerebri (white arrowheads, G).

Subfalcine herniation occurs from a space-occupying lesion in the cerebral hemisphere causing the cingulate gyrus to be pushed under the rigid falx cerebri into the contralateral cranial vault ²⁴ (Fig. 3.1G). Subfalcine herniation commonly occurs with frontal lobe and parietal lesions. If midline shift is severe, the anterior cerebral arteries may be compressed, potentially leading to infarction. Subfalcine herniation is often measured as midline shift at either the level of the septum pellucidum, foramen of Monro, or third ventricle. It is important to measure midline shift at the same location to assess for changes between scans.

Uncal herniation is a type of unilateral descending transtentorial herniation involving the uncus, a component of the mesial temporal lobe that appears as a focal convexity at the anterior margin of the parahippocampal gyrus (Fig. 3.1B). Uncal herniation occurs when the mesial temporal lobe herniates medially and inferiorly, usually because of a temporal lobe or middle cranial fossa mass (Fig. 3.1F). To diagnose uncal herniation, find the suprasellar cistern and look for medial displacement of the uncus with compression of the ipsilateral perimesencephalic cistern. In severe cases, the herniated uncus will compress the ipsilateral cerebral peduncle²⁴ and the brainstem may be shifted to the opposite side. There may be entrapment of the contralateral temporal horn of the lateral ventricle as CSF outflow is compressed. Recognizing and alleviating uncal herniation before progression to brainstem compression are important to minimize severe neurological consequences.

Transtentorial herniation is assessed by evaluating the basilar cisterns around the midbrain, including perimesencephalic cistern, ambient cistern, and quadrigeminal plate cistern.²⁴ In the case of severely increased intracranial pressure or focal mass effect, the brain herniates downward through the tentorial incisura, first resulting in narrowing, then effacement of the basilar cisterns. If intracranial pressure continues to increase, the herniated brain will compress the brainstem (Fig. 3.1D and E), with narrowing of the transverse diameter of the midbrain. Transtentorial herniation may lead to compression of branches of the posterior cerebral artery against the tentorium with temporal or occipital infarction. Hydrocephalus may occur if the cerebral aqueduct is compressed. Severe transtentorial herniation may result in hemorrhages within the brainstem (Duret’s hemorrhages), which are a sign of grave prognosis.

If intracranial mass effect arises in the posterior fossa, cerebellar contents can herniate upward through the tentorial incisura.²⁴ This displacement is usually accompanied by tonsillar herniation, the downward herniation of the cerebellar tonsils through the foramen magnum. Upward tentorial herniation appears similar to downward tentorial herniation on axial images at the level of the incisura, but a mass lesion in the posterior fossa will be present. Tonsillar herniation appears on axial CT as crowding of the contents at the foramen magnum with effacement of the perimedullary cistern. Sagittal reconstructions may be particularly helpful in delineating upward transtentorial herniation and cerebellar tonsillar herniation.

Hemorrhage

CT is highly sensitive for detecting intracranial hemorrhage. In general, acute hemorrhage (within hours of injury) is hyperdense to brain, with Hounsfield units in the range of 50 to 70 ²⁵ (Fig. 3.2A). As the blood products break down and are reabsorbed, the density of the hematoma decreases. Subacute blood products (1-6 weeks) may appear isodense to brain (Fig. 3.2B). As the hematoma becomes chronic, the density approaches that of CSF. In the case of chronic subdural hematoma, there may be blood products of different density or fluid-fluid levels due to hemorrhages of different ages (Fig. 3.2C).

FIGURE 3.2 Blood of various ages: A 67-year-old woman on heparin with acute decline in mental status: A, Noncontrast head computed tomography (CT) scan shows hyperdense acute hemorrhage within the pons (black arrows). Note hypodense areas with fluid-fluid levels reflecting hyperacute hemorrhage. B, Postcontrast CT image in the same patient shows layering contrast material (arrowhead) within the hemorrhage, indicating active extravasation. C, A 50-year-old with 2 weeks of severe headaches. Noncontrast CT image shows isodense subdural hemorrhage (SDH) (black arrows). Hyperdense focus in the SDH (arrowhead) indicates more recent hemorrhage. D, A 72-year-old who fell. Noncontrast CT image shows hypodense chronic SDH (black arrows). Layering isodense component on the right and fluid-fluid level on the left (arrowheads) represent more recent blood products.

Hyperacute blood (ongoing bleeding or imaging immediately after hemorrhage) is heterogeneous in appearance (see Fig. 3.2A). If contrast agent is given as part of the study, active hemorrhage is evident as contrast extravasation. Acute hemorrhage may occasionally appear isodense in the setting of anemia or coagulopathy.

CT is also excellent for determining the location of hemorrhage (Fig. 3.3). In general, different compartments within the cranium include the epidural space, between the dura and inner table of the calvarium; the subdural space, between the dura and arachnoid membranes; the subarachnoid space, between the arachnoid membrane and brain surface; within the brain parenchyma; and within the ventricles. The location and pattern of bleeding can yield vital clues to the underlying cause of hemorrhage.