CHAPTER

Structural Neuroimaging

Matthew W. Luedke and William B. Gallentine

Clinical localization is one of the hallmark skills of a neurologist. Seizure semiology is an elegant method of localization, and the positive symptoms of an epileptic lesion are focal cortical functions made manifest. EEG is a physiologic tool of localization, permitting a skilled neurophysiologist to identify functional abnormalities and their rough anatomic locus.

Despite the utility of clinical and physiologic localization, anatomic localization is a critical tool for the diagnosis and management of epilepsy, at all stages of patient management. Neuroimaging by CT and MRI has revolutionized the physician’s ability to identify both normal and abnormal anatomy. Whether it is a CT scan to rule out an acute neurologic emergency after a first seizure or a high-definition MRI to demonstrate a subtle cortical dysplasia, neuroimaging can play a vital role in the assessment of epilepsy. Particularly in complex epilepsy cases, the presence and location of anatomic abnormalities can govern the location and size of a surgery, or even whether a surgery can occur. The mere absence of a lesion in otherwise clinically and physiologically focal epilepsy can have dramatic prognostic implications (1).

This chapter will examine anatomic imaging in epilepsy, beginning with the basic history of neuroimaging. Next, the physics of CT and MRI and the limitations this imposes on their clinical use will be discussed. Thereafter, the clinical utility of CT versus MRI in general will be reviewed. The value of neuroimaging in evaluation of the acute seizure and in established epilepsy will then be noted. Finally, the common abnormalities seen on neuroimaging in patients with epilepsy will be discussed.

HISTORY

The history of medical imaging can be traced back to Wilhelm Rontgen. In 1895, Rontgen identified the x-ray, a high-frequency band of electromagnetic radiation lying between the ultraviolet and gamma range. During his investigations, he discovered that x-rays could be captured on film, and moreover, a limb interposed between the x-ray emitter and the film would generate a shadow image of underlying bones. This discovery rapidly found its way into the hands of physicians, leading to the birth of both diagnostic and therapeutic radiology. For his efforts, Rontgen was awarded a Nobel Prize, the first of many awarded to pioneers in medical imaging.

Early radiographers faced a challenge when it came to neuroimaging: the brain is radiolucent, and of a relatively uniform density. Unlike bone imaging, where the organ of interest is radiopaque, or thoracic imaging, where air and fluid levels and differential organ densities allow for resolution of adjacent structures, the plain-film of the head is limited to an evaluation of the skull (2).

In 1918, Walther Dandy adapted the air-contrast abdominal film to x-ray imaging of the human brain. He developed a technique to insufflate air into the thecal sac, and then distributed it throughout the CNS. The introduction of air into the ventricles and the subarachnoid space permitted rudimentary resolution of large brain structures. If sufficiently outlined by air, and if adjacent to contrasting structures, pneumoencephalography could identify large brain lesions, though the sensitivity and specificity were poor. Nonetheless, some sources suggested that CNS tumor diagnosis rates increased by as much as 33% with the use of pneumoencephalography (2).

Advances in pneumoencephalography occurred through the 1960s, with the ultimate evolution involving chairs that flipped patients after insufflation, allowing for the even distribution of air into the ventricles, cisterns, and subarachnoid space. Nevertheless, resolution remained poor. Beyond the limited resolution, the pneumoencephalogram was notoriously painful; its reputation was so frightening, it made its way into popular culture as an act of medical futility in the 1973 horror film, The Exorcist (2).

Shortly after Oldendorf’s rejection, Cormack devised an algorithm for tomographic reconstruction, and by the early 1970s, Hounsfield helped devise the first commercially successful CT scanner for the British device company, EMI, Ltd. Despite their similar efforts, Cormack and Hounsfield were unaware of Oldendorf’s creation. In 1979, Cormack and Hounsfield won the Nobel Prize in Medicine, leaving Oldendorf unrecognized, and largely forgotten (2).

The first clinical EMI scanner was used for clinical trials in 1971 to 1972 at the Atkinson Morley Hospital of London, England. In 1973, the first North American scanner was installed at the Mayo Clinic in Rochester, Minnesota. The first units had a resolution of 80 × 80 pixels of approximately 3 mm × 3 mm resolution, with slice widths of between 8 mm and 13 mm. The scans were time consuming and radiation intensive. But even with these limitations, the scanner was revolutionary. A contemporary of the first Mayo Clinic EMI scanner wrote, “As I saw the images it was obvious that, despite some streaking on certain sections caused by patient motion, the system was capable of displaying with remarkable clarity many pathologic processes involving the brain, including tumors, infarcts, hemorrhages, and infectious processes” (3–5).

In the ensuing four decades, CT resolution has improved, acquisition time has decreased, and radiation exposure has decreased. Moreover, advances in computer technology allow clinicians to create complicated reconstructions.

In contrast to the CT, the MRI is based off of a younger technology. Nuclear magnetic resonance (NMR) was developed in the 1940s as a tool for chemists, using powerful magnets to generate characteristic radio-frequency emissions from atomic nuclei. Initially a laboratory tool used on small chemical samples and requiring small and powerful magnets, its clinical application was less apparent and more difficult than that of early x-rays. While it took Rontgen a matter of weeks to start experimenting with x-rays for medical use, it took over 20 years for the first attempt to use NMR scans on laboratory animals. The first human imaging was performed in 1977, and the first brain imaging was published in 1980, the same year the first clinical scanners were released. Early scanners were limited to T1 and T2 sequences, and contrast by gadolinium was only FDA approved in 1988. Yet, even the early 1.5T scanners could provide better contrast between gray–white matter structures than contemporary CT scanners, and rapidly became a staple in the evaluation of neurologic disease. As with the CT scanner, the impact of MRI was so profound that Paul Lauterber and Sir Peter Mansfield, early pioneers in the clinical application of NMR, were awarded the Nobel Prize in medicine in 2003 (2).

As with CT scanners, MRI has undergone an evolution since its initial clinical debut. Clinical 3T scanners are now common, and several institutions have 9.4T MRI scanners for clinical research purposes. In addition to advances in magnet power and resolution, different pulse sequences have been developed to assess for different anatomical structures.

PRINCIPLES

CT and MRI rely on different technology to render anatomic imaging. An appreciation for the basic physics involved in these systems, along with the technology, nomenclature, and potential pitfalls, is vital to their appropriate implementation in the clinical setting.

The CT scan relies on x-ray spectrum electromagnetic radiation. X-rays consists high-energy ionizing photons, with frequencies faster than that of ultraviolet radiation, but still lower than that of gamma radiation. Shorter wavelength x-rays are considered “hard,” and can penetrate objects, and these are the x-rays that are used in medical imaging.

Clinical x-rays were traditionally resolved on film. As photons from an emitter passed through a patient, tissues of differing density and composition would either absorb or diffract x-rays in transit. Radiopaque objects like bone or dense organs absorb or diffract more photons, whereas radiolucent structures, such as fluid or air, would allow greater pass-through. The photons would then strike a plate of film, with areas behind radiolucent structures receiving a greater exposure. When developed, areas of lower photon exposure would appear brighter, and higher photon exposure would be dark, creating a two-dimensional shadow of a three-dimensional object (6).

Instead of creating a two-dimensional shadow of a three-dimensional form, CTs generate tomograms. A tomogram is two-dimensional cross section of a three-dimensional object. The advantage is twofold. First, a cross-sectional slice is, by nature, a two-dimensional image and there is less data lost than there is in the rendering of a three-dimensional object into a two-dimensional shadow. Second, by combining a series of cross-sectional samples, an imager can reconstruct three-dimensional relationships in a subject.

Because CT scans rely on the differential attenuation of x-rays to generate images, they best resolve adjacent structures with dissimilar radiopacity. It is possible to create dissimilar opacity by introducing contrast agents—radiopaque fluids that can magnify the density of a natural fluid space. Typically iodinated contrast agents are used in contemporary CT scans, and they can be injected into the blood stream to contrast arterial or venous systems, or, less frequently, into the CSF. The contrast can highlight vascular structures and regions of increased vascular permeability.

The use of x-rays carries risk, and CT scanners can emit significant doses of ionizing radiation. While a typical plain-film x-ray can emit 0.01 to 0.15 Gy of radiation, a CT scan can emit 10–20 Gy. An average annual exposure for a human in the United States is 2.4 Gy; clearly, CT scans represent an exposure beyond baseline (7). Repeated CT scans, which can occur in prolonged hospitalizations during critical illness, multiply that exposure. This cumulative radiation dose represents a cancer risk, and this risk is magnified in younger patients. While the overall lifetime risk of developing a terminal cancer from a single CT scan is estimated at 1:10,000, the risk increases to 1:1000 for a 1-year-old (8). Taken individually, the risk is small, but repetitive imaging carries a nonnegligible risk.

Professional organizations have created campaigns to encourage strategies for generating high-quality imaging while minimizing radiation exposure. The Image Gently Campaign, directed toward pediatric patients, and Image Wisely Campaign, for adults provide best practice guidelines to reduce radiation exposure.

In addition to the radiation, the use of iodinated contrast agents can present a risk. Acute kidney injury can be a side effect of iodinated contrast, and is most likely to occur in patients with preexisting kidney injury. Contrast agents can also lead to allergic reactions, ranging from fevers and flushing to overt anaphylaxis. Modern low-osmolar contrast agents are less provocative than older agents. However, caution must be used whenever contrast is considered (6).

MRI is a larger-scale implementation of the analytic chemistry technique of NMR spectroscopy. When exposed to a powerful oscillating magnetic field, the protons of certain atomic nuclei can be brought into alignment (for the purposes of MRI, the atoms of interest are ¹H). When the magnetic field is removed, those protons relax into their equilibrium state, releasing the energy as a photon with a characteristic radio-range frequency at a predictable rate of decay. The relationship of the atom to its surroundings, or lattice, can also change that frequency and its emission time (6).

MRI, overall, is considered a safe imaging modality. The most significant danger is posed by the magnetic field. Patients have been injured and killed when ferromagnetic objects were captured by the MRI magnetic field and propelled into the scanner’s aperture. Likewise, patients with ferromagnetic implants or devices may not be able to receive MRI scanners. With appropriate safety measures, this risk can be minimized, and device makers are making many of their implants, such as pace makers, MRI compatible (12).

NOMENCLATURE

The language of CT imaging is that of relative density. Brighter structures, those with higher relative attenuation of the x-ray beam, are described as hyperdense. Darker structures, those with a lower relative attenuation, are hypodense. Structures that are of similar density are labeled isodense. Structures with very high Hounsfield unit values are sometimes referred to as radiopaque, while very low value structures are infrequently referred to as radiolucent. If a structure demonstrates a higher density after the introduction of a contrast agent, that structure can be described as enhancing. In the authors’ experience, the term “enhancing” is often used in error to mean “hyperdense”; this is a potentially confusing error, as true contrast enhancement can have significant clinical implications.

MRIs are described in terms of relative signal intensity in a given sequence. Bright spots on an image are hyperintense, dark spots are hypointense, and regions of comparable brightness are isointense. However, it is critical to note the sequence when making these descriptions, because signal intensity can vary dramatically. For example, a ventricle on T2 will appear hyperintense, but the same fluid space on T1 or T2-FLAIR will appear hypointense relative to its surroundings. Hyperintensity is sometimes referred to as signal prolongation. As with CT, it is important to reserve enhancement to describe regions of a given sequence that are brighter after the addition of contrast than on a precontrast scan.

CT and MRI scans have their strengths and weaknesses, and together are complementary technologies. CT scans are rapid and clearly resolve tissues with very different densities. This makes them excellent for emergent imaging. CTs can, for example, clearly demonstrate acute extravascular blood in the brain, as it has a markedly higher attenuation coefficient than surrounding tissues or CSF. Likewise, boney injuries or regions of severe edema can be identified easily. CTs are also a good gross anatomy scan, able to identify mass effect, herniation syndromes, or large tissue deficits like old infarctions or regions of sclerosis and atrophy. Calcifications are also very visible on CT scans. They are poor, however, at differentiating similar tissues, such as gray and white matter, and they do not pick up the subtleties of cortical or nuclear structure. CTs can easily miss cortical dysplasia, small tumors without significant mass effect, and areas of subtle sclerosis or scarring. The addition of contrast can help identify blood vessels, vascular tumors, and inflammatory lesions, but it cannot improve gray–white matter differentiation or add significant detail to the overall brain anatomy. CT scans also carry with them the risk of radiation exposure, which is a particular consideration in the young (6,9).