Three
Three
CHAPTER
Analysis of Density, Signal Intensity, and Echogenicity
Analysis of Density, Signal Intensity, and Echogenicity
RADIOGRAPHIC/CT DENSITY Standard Against Which to Measure Density
Conventional radiography and computed tomography are based on the differential attenuation of photons by tissues as they pass from an x-ray source on one side of the body to a detector on the opposite side. Mathematically, the measurement at the detector is determined by the sum of the values of the linear attenuation coefficient, U, of each individual tissue along the course of the x-ray beam. At each point within an object, U characterizes the rate at which x-rays are removed by scatter or absorption and thus reflects the biophysical interaction between photons emitted by the x-ray source and the tissue irradiated. For the relatively high photon energies used in diagnostic medical imaging and low atomic numbers of most organic matter, the primary determinant of U is Compton scatter, which results in a magnitude of photon attenuation that is nearly linearly proportional to tissue density (mass per unit volume). There are several additional contributors to x-ray attenuation that depend both on the x-ray source and the object being imaged, but in practice the primary basis of contrast in both radiography and CT can be considered to be tissue density.
Planar projections of linear attenuation, the source of the imaging data depicted on plain films, reliably resolve only five different biologic densities: air, fat, water, soft tissue, and bone. Before the advent of tomographic imaging modalities such as CT and MRI, neuroradiologists went to great lengths to manipulate contrast in plain film radiography to make diagnoses, for example by purposefully introducing air or iodinated media into the subarachnoid space or blood vessels and thereby identify masses in the skull vault, spinal disc herniations, and intracranial aneurysms. Although important diagnoses can still be made from radiographic density abnormalities, conventional radiographs have only limited application in modern neuroimaging and are commonly used at present for the gross evaluation of integrity of medical devices such as cerebrospinal fluid shunts or spinal fusion hardware or in the detection of fractures or malalignment of the skull or spinal column.
Linear attenuation is a useful physical concept for understanding image formation in radiography and CT but is not directly applicable to the visual interpretation of images. Before display and storage, each pixel in a reconstructed CT image is normalized to an integer value termed the Hounsfield unit or CT
number. This normalized attenuation scale arbitrarily assigns water an attenuation value of zero, such that a difference of ten HU reflects approximately a one percent difference in linear attenuation. The maximum and minimum values of the Hounsfield scale depend on the numerical storage scheme of the manufacturer, but the range in attenuation that can be discriminated by most modern scanners is four thousand ninety-six HU (from roughly negative one thousand HU to three thousand HU). Small numbers correspond to relatively radiolucent structures such as air and fat, and large numbers correspond to radiodense structures such as bone and calcium. There is considerable overlap between CT numbers for different tissues, but certain tissue densities can usually be distinguished based on their typical Hounsfield numbers.
The dynamic range of the human visual system, which can reliably discriminate fewer than one hundred shades of gray, is far less than the range in tissue density represented by the Hounsfield scale. To facilitate visual analysis of images on a digital workstation, different display windows are applied to the raw CT numbers to optimally visualize the different tissues of interest. The effect of windowing is to linearly map a subsegment of the Hounsfield scale to two hundred fifty-six shades of gray, a standard range of gray values between black and white depicted on a computer monitor. The central Hounsfield unit of the window is designated the window level. The window width determines the overall contrast of the displayed image, translating the values of the standard Hounsfield scale within the window to various shades of gray that are more easily interpreted by the human eye. Typical window parameters used to evaluate different tissues of interest are given in Table Three-Two.
The recognition of abnormal density on CT images relies foremost on familiarity with the range of normal densities of the anatomic structures of the central nervous system and its supporting structures. Brain regions where neuronal cell bodies are located comprise the gray matter of the cortex and deep gray nuclei, including the basal ganglia and thalami. These structures normally have CT numbers of twenty to forty HU, which is slightly greater than those of white matter (twenty to thirty-five HU), where the neuronal axons and their supporting glia are concentrated. As a consequence, optimal examination of the brain parenchyma requires a narrow display window that allows accurate discrimination between the densities of these two types of tissues. Any interruption of the normally homogenous density within a discrete white or gray matter structure implies a disruption in its normal physiology. The normal brain has sharp, well-defined boundaries between gray and white matter, and any regions where this distinction is lost should be viewed with suspicion.
CSF within the ventricular system and subarachnoid spaces of the brain and spinal cord normally has uniformly low attenuation that is nearly isodense to water. Inhomogeneity or altered attenuation in these regions is invariably abnormal. Similarly, the blood pool of the intracranial and extracranial vasculature is readily visualized within major arteries and veins and normally has homogeneous density that approximates the density of unclotted blood. Focal intravascular hyperdensity may be the only finding of acute stroke or dural venous thrombosis, and densely calcified vessels suggest atherosclerosis or an underlying disorder in calcium metabolism that could predispose to arterial insufficiency.