Static Anatomic Techniques

CHAPTER 1 Static Anatomic Techniques



Computed tomography (CT) and magnetic resonance imaging (MRI) are the mainstays of anatomic neurologic imaging. CT was first introduced in the early 1970s and MRI in the early 1980s. Since then, CT and MRI have transformed medical diagnosis and proved essential in neuroimaging.



COMPUTED TOMOGRAPHY



Basic Concepts


CT relies on the differential attenuation of x-ray beams passing through tissues to produce an image. The patient lies on the CT table, with his or her long axis aligned along the longitudinal (z) axis of the scanner. The x-ray tube and detector, housed in a gantry, rotate 360 degrees around the patient so the x-ray beam strikes the patient in the transverse (x/y) axis. Conceptually, the slab of tissue imaged can be divided into many small volume elements (voxels), each with x, y, and z dimensions. The degree to which each of these voxels attenuates the x-ray beam is derived by analyzing the data from all the different angular projections, using a reconstruction method known as convolution-backprojection. The computed attenuation value of each voxel is then converted into a gray-scale value of Hounsfield units (HU) and displayed. The attenuation of distilled water at 0° Celsius and 1 bar of pressure is defined as 0 HU. The attenuation of air at the same standard pressure and temperature conditions is defined as −1000 HU.


The spatial resolution of the CT image depends in part on voxel size. Ideally, each voxel of data would be very small to provide high spatial resolution. Each voxel would also ideally be isotropic (having equal dimensions in all three planes) to provide for excellent image reconstructions in any arbitrary plane. It has been relatively easy to achieve high in-plane resolution (along the x/y axes), to the order of 0.5 to 0.7 mm.1 It has proved difficult to achieve high resolution in the longitudinal or z-plane, because longitudinal resolution is determined by the slice thickness. Use of thin submillimeter slices reduces the length of tissue that can be scanned in a reasonable time or increases the scan time for equal lengths of tissue imaged. Evolution of CT technology over the years can be seen in part as the pursuit of this isotropic resolution.




Spiral CT


Spiral CT was developed in the early 1990s to improve scan speed and flexibility. In spiral CT, the x-ray tube and detectors rotate continuously about the patient while the scan table advances the patient continuously through the gantry. As a result, the x-ray beam traces a helical path through the patient and provides a “spiral” of image data. Because the patient is intentionally moved through the gantry during scanning, there is significant motion artifact. However, computational methods known as z-interpolation were specifically developed to manage the spiral dataset and to eliminate the motion artifact caused by patient translation. For any image position along the z-axis of the patient, z-interpolation re-forms the spiral data to fit on a single plane. The conventional convolution-backprojection algorithm for data analysis can then be applied.


Spiral CT does not depend on direct, one-to-one correspondence between scan position and image slice, so image slices can be reconstructed anywhere along the z-axis at different slice thicknesses and varying intervals. This flexibility is an important advantage of spiral CT over conventional CT. Overlapping slices can be acquired with no increase in radiation dose to the patient, resulting in high-quality multiplanar reconstructions. Because scan time is fast, spiral CT examinations can be performed in a single breath-hold to reduce respiratory misregistration and motion artifact, and injected contrast agents can be imaged more quickly over greater lengths of tissue to perform CT angiography (CTA).


One technical factor unique to spiral CT is pitch. Pitch is the ratio of table displacement per 360-degree gantry rotation to slice collimation or thickness (table speed × rotation time/slice collimation). A small pitch gives finer spatial resolution along the z-axis of the patient but covers less tissue in a given time and delivers a higher radiation dose to the patient. A large pitch reduces the radiation dose to the patient but also reduces spatial resolution in the z-axis.



Multislice Spiral CT


The next significant milestone in CT evolution was the introduction of scanners with multiple detector rows. In 1998, all major vendors introduced 4-slice CT scanners capable of acquiring up to four slices per gantry rotation. Instead of a single detector row, multiple detector rows were stacked in the gantry along the z-axis of the patient (Fig. 1-1). The time needed for the gantry to complete a 360-degree revolution (gantry rotation time) was also cut in half from 1 second to 0.5 second. For the same slice thickness, pitch, and scan time, a 4-slice CT scanner could image eight times the distance of a single-slice scanner. Alternatively, the 4-slice scanner could acquire four 1.25-mm slices in half the time that single-slice spiral CT acquired one 5-mm slice. Four-slice CT made higher z-axis resolution feasible for a reasonable anatomic length and scan time.



Subsequently, 16-, 40-, and 64-slice scanners were introduced widely for clinical use, the latter in 2004. As a result, slices as thin as 0.5 mm can now be acquired very quickly and over long distances to provide submillimeter resolution in the z-axis, truly isotropic voxels, and isotropic resolution. The advantages of multi-slice CT over single-slice imaging can be summarized as better spatial resolution in the z-axis, faster imaging time, and longer anatomic coverage.



Imaging




Reconstruction Parameters


The data acquired during the scan are processed through convolution-backprojection algorithms to provide the CT images. Different algorithms or convolution kernels can be applied during convolution-backprojection to emphasize different tissues. Soft/smoothing or sharp/edge-enhancing algorithms will highlight different tissues such as soft tissue or bone, respectively.


Spiral CT slices can be reconstructed at different thicknesses. Images acquired at 1.25-mm collimation can be reconstructed at 2.5 mm, 3.75 mm, or 5.0 mm. However, slices cannot be reconstructed at thicknesses smaller than the original collimation. Slices can also be reconstructed with varying degrees of overlap, or reconstruction intervals. For a 1-mm thick slice, a reconstruction interval of 0.8 mm signifies 20% slice overlap, which is approximately the amount of overlap desired if slices are to be reformatted into other planes.


The CT data can be reprocessed in a number of useful ways. CT images obtained in the axial plane can be reformatted into coronal, sagittal, or oblique sections with multiplanar reformation (MPR), a two-dimensional (2D) technique that preserves all the data in the original source images. Maximum intensity projection (MIP) processing collects only the brightest voxels from a predefined volume and collapses this information onto a single slice. In this 2D technique, depth information is lost but attenuation data are retained. Shaded surface display (SSD) is a three-dimensional (3D) method for displaying the surfaces and shapes of objects, but with significant loss of attenuation information. Volume rendering (VR) is a superior 3D method to SSD and assigns color and opacity to each CT value.




Normal Appearance of Images


Attenuation is represented in Hounsfield units on a gray scale in which distilled water is set at 0 HU for standard temperature and pressure, and air is set at −1000 HU. Tissues such as bone, which attenuate the x-ray beam more than water, have positive HU values (approximately 1000 HU for bone) and appear very white. Tissues such as fat, which attenuate the x-ray beam less than water, have negative HU values and appear darker than water (−30 to −100 HU for fat).


The human eye can typically differentiate only 60 to 80 different levels of gray. In practice, therefore, the Hounsfield scale must be narrowed to illustrate specific structures of interest. This is achieved by selecting a gray-scale window of displayed Hounsfield units and arbitrarily making all structures above the chosen window white and all structures below the window black. The window width describes the range of Hounsfield values displayed as shades of gray. The window level gives the center value of that gray-scale window. A head CT is typically viewed at window width of 80 HU and window level of 40 HU, which means that 0 HU and 80 HU are the lower and upper limits of the window, respectively, with 40 HU in the center. This relatively narrow window width successfully displays the small differences in attenuation values of the brain. Figure 1-2 emphasizes the importance of choosing appropriate windows to properly display structures of interest and to detect clinically important pathologic processes.




Artifacts


Common artifacts encountered in CT include patient motion, beam hardening, partial volume effects, and metallic object streak artifacts. Patient motion during scanning creates extensive blurring and misregistration of images. This can be partly mitigated by reducing scan times as much as possible. Beam hardening occurs because the energy profile of the x-ray beam changes as it passes through dense objects such as bone. The softer (lower energy) x-rays are absorbed and filtered out by the bone, leaving a beam composed of only harder (higher energy) x-rays. On head CT, beam hardening typically occurs in the posterior fossa between the petrous apices, causing dark horizontal lines across the brain stem and limiting the utility of CT for assessing pathologic processes in this area. Partial volume artifacts ensue when an imaging voxel contains different types of tissue. The attenuation value of the voxel is a numerical average of the attenuation of all the tissues contained within that voxel. If a portion of the voxel has a very high (or low) Hounsfield unit value, that portion may influence the net attenuation of the voxel disproportionately and obscure the presence of other tissues. Like beam hardening, partial volume effects are most troublesome in the posterior fossa, where they cause streaks or bands of light and dark. Reducing scan thickness produces smaller voxels and helps to reduce partial volume effects. Metallic objects such as aneurysm clips or dental hardware generate intense streak artifacts because their exceptionally high density causes beam hardening and partial volume artifacts. The streaks can completely obscure adjacent structures and prevent their evaluation. Figure 1-3 illustrates these typical artifacts.




Specific Uses


Brain CT is most useful in acute settings, especially emergency departments, because of its fast acquisition time, ready accessibility, and lower cost compared with MRI. As the first-line examination after trauma, CT is more sensitive than MRI for detecting skull fractures and radiopaque foreign bodies such as metal or glass.2 CT readily identifies acute subdural/epidural and parenchymal hematomas and hemorrhagic contusions and is superior to MRI for detecting acute subarachnoid hemorrhage.3 CT is particularly helpful for identifying calcification and assessing pathologic processes of bone, both of which may narrow a differential diagnosis. CT is indispensable for studying patients with cardiac pacemakers, defibrillators, intra-orbital metal, or other implants that contraindicate the use of MRI.


CT angiography (CTA) has become important in the initial evaluation of subarachnoid hemorrhage, achieving 90% to 93% sensitivity for detecting aneurysms according to meta-analyses of older studies.4,5 The faster scan times available with 16- and 64-slice scanners permit selective capture of the arterial phase of contrast opacification without venous contamination and provide images close to true angiograms. The fast, thinly collimated multi-slice acquisitions now permit CTA to be performed over long distances in short periods of time, so CTA can image the entire region from the base of the heart to the vertex of the skull to evaluate stroke patients for left atrial thrombi and potential occlusions in the cervical and intracranial circulations. Although digital subtraction angiography (DSA) remains the gold standard for angiography at present, the sensitivity and speed of CTA are constantly improving, so CTA will come to rival DSA in the near future.6



Analysis


In any acute setting, noncontrast head CT can be used to quickly assess for the three Hs—hemorrhage, herniation, and hydrocephalus—which may necessitate immediate neurosurgical intervention. Figure 1-4 illustrates the utility of CT in the acute setting, as well as its importance in the evaluation of bony lesions.



A sample report is shown in Box 1-1.



BOX 1-1 Sample Report: CT and CT Angiography of the Head(Fig. 1-5)









Pitfalls and Limitations


Several important problems do limit the utility of CT. In patients with renal impairment, the use of iodinated intravenous contrast is limited by concerns about contrast-induced nephropathy, generally identified as an increase in serum creatinine concentration after administration of a contrast agent, without an alternative explanation. Although there are no uniform diagnostic criteria (because creatinine levels are not necessarily precise), the two most important risk factors for developing nephropathy are preexisting renal impairment and diabetes.7,8 Adequate hydration, acetylcysteine, and sodium bicarbonate may help prevent nephropathy in patients with borderline renal function.9,10


Radiologists are frequently asked what to do with patients who are “allergic” to shellfish or iodine. There is a mistaken assumption that iodine in each of these compounds confers cross-reactivity to iodinated contrast agents. However, there is little to no evidence to indicate that the iodine itself triggers adverse reactions to contrast, seafood, or topical povidone-iodine.11 In patients with a history of significant prior contrast reaction, premedication with histamine blockers and corticosteroids can be performed. Patients describing allergies to seafood should be questioned about the nature of the reaction but only insofar as a history of severe allergy to any food increases the risk of contrast reaction.


Pregnancy and lactation generate additional safety considerations for CT. The radiation dose to the fetus during the mother’s head CT has been estimated at 0 to 1 mGy and is from scattered radiation only. It is generally believed that the risk to the fetus of teratogenesis or childhood cancer is negligible at radiation dosages less than 50 mGy.12,13 Because the uterus lies outside the field of view and the radiation dose to the fetus is negligible, it is not clear that it is necessary to place lead shielding over the abdomen/pelvis. However, placing shielding may provide reassurance to the patient. Iodinated contrast material should be avoided if possible during pregnancy because of potential concern for fetal hypothyroidism. For lactating women, the traditional recommendation is to discontinue breast feeding for 12 to 24 hours after contrast agent administration and discard the milk.14



Current Research and Future Direction


CT scanners capable of up to 64-slice acquisitions are in common clinical use and afford submillimeter isotropic resolution, rapid scan times (<5 seconds for head CT), and good coverage (32 to 40 mm z-axis coverage with a single gantry rotation). Because the imaging parameters are now able to meet most clinical demands, it is not clear that increasing the number of slices acquired simultaneously is particularly useful or warranted. Instead, research has focused on meeting specific clinical needs, such as dynamic imaging for perfusion measurements and faster scan times for cardiac imaging.


Increasing the length of coverage along the z-axis may permit an entire organ to be imaged during a single gantry rotation, opening up the potential for dynamic perfusion imaging of individual organs. Ways of increasing the volume of coverage include using flat-panel detectors, manipulating the detector array, and increasing the number of detector rows. Indeed, 256- and 320-slice scanners have been developed and installed in limited capacity, providing 12 to 16 cm of z-axis coverage, although higher data load and cost burden are important considerations.


Dual-energy source CT is another promising area for future development. In this approach, two x-ray tubes and two detectors are housed in the same gantry and are used to deliver two x-ray beams at different voltages (e.g., 80 kV and 140 kV). Advantages of dual-source CT include much faster scan times and higher temporal resolution, which are invaluable for cardiac imaging. Dual-source scanning also has the potential to differentiate between specific tissues such as calcium and blood. This ability can be used to selectively depict a single tissue or selectively delete one tissue from the image. For example, one can accurately subtract bone from CTA images to clearly evaluate vessels at the skull base, an area that has traditionally been difficult to visualize.



MAGNETIC RESONANCE IMAGING


The U.S. Food and Drug Administration first cleared MRI for commercial use in 1984, and MRI has grown remarkably since that time. Most current MRI scanners have a magnetic field strength of 1.5 tesla (1.5 T), but units employing higher magnetic field strengths of 3 tesla (3 T) are coming into increasing use. Both the 1.5-T and newer magnets offer an unparalleled look at anatomic structures, with relative safety and freedom from the concerns about radiation dose that are inherent in CT.


MRI employs an astonishing array of sequences that are acquired by diverse means, are used for different purposes, and are designated by different acronyms by each manufacturer. Table 1-1 offers an overview of the major sequences commonly used in MRI (including their acronyms), which may be a useful reference during review of this chapter.




Basic Concepts



MR Signal Creation


Clinical MRI relies on the hydrogen nucleus. In their native state, the hydrogen nuclei exhibit random orientation and precess or rotate at varying rates. When an external magnetic field (B0) is applied, the hydrogen nuclei begin to precess at a resonance frequency (designated the Larmor frequency) that is proportional to the magnetic field strength. Additionally, the external magnetic field prompts the hydrogen nuclei to align and precess along the axis of the magnetic field, creating a net magnetization vector. By convention, the direction of B0 is designated the longitudinal or z-axis. The plane oriented perpendicular to the z-axis is designated the transverse or x/y-axis.


The precession of the hydrogen nuclei at the Larmor frequency creates a current, measured as the MR signal. This current cannot be detected in the z-axis; it can only be detected when its magnetization lies in the transverse plane. To measure the current, the net magnetization must be moved from the z-axis (where it cannot be measured) into the transverse x/y-axis (where it can be measured). To accomplish this, a radiofrequency (RF) pulse is applied to “flip” the net magnetization by a certain angle (the flip angle) into the transverse plane. Immediately after the RF pulse, nuclei in the transverse plane are in phase. They precess together at the same frequency and in the same direction, creating a signal known as the free induction decay (FID). However, the FID signal is rapidly lost as inhomogeneities in the magnetic field cause the nuclei to dephase and spin at different frequencies. The FID cannot be measured directly for imaging purposes. Instead, an echo of the FID—either a spin echo or gradient echo—must be produced by rephasing the nuclei. This is the basis for sequence design, as will be discussed later.


Jan 22, 2016 | Posted by in NEUROLOGICAL IMAGING | Comments Off on Static Anatomic Techniques
Premium Wordpress Themes by UFO Themes