X-ray Imaging, Medical: Tomography and Radiog-

raphy
William R. Hendee
Medical College of Wisconsin, Milwaukee, WI

Keywords

x-ray imaging, medical-tomography and radiography; x-ray na-

ture and production

Medicine is largely devoted to preventing disease or injury, diag-

nosing what is wrong when disease or injury may be present, and
treating the patient to alleviate pain and, when possible, to cure
the problem. Diagnosis is often a major challenge, as the patient’s
symptoms are frequently relatively unspecific and might be caused
by a variety of conditions. Meeting this challenge usually requires
conducting various tests, including laboratory analyses and imaging
procedures. These tests are all products of technological advances
in the practice of medicine that have occurred in the 20th century. A
century ago, physicians were severely limited in the tools available
to help them arrive at a correct diagnosis. Most diagnoses were
made from a verbal description of the patient’s symptoms, combined
with the patient’s history and supplemented by use of the physi-
cian’s senses to detect unusual odors, visual appearances, abnormal
sounds, physical sensations, and, at times, an odd taste. Although
significant problems may have been present within the patient’s
body, they frequently provided few external clues, and the physician
was guided toward a diagnosis largely by his (or, in a few cases, her)
intuition and experience. Needless to say, the final diagnosis was
uncertain or incorrect in more than a few cases.
A major advance in the physician’s ability to detect and diagnose
disease and injury occurred in 1895 with the discovery of x rays by
the physicist Wilhelm Röntgen in Würzburg, Germany. See also X-
Ray. Today, x-ray imaging systems are used in hospitals, clinics, and
physicians’ offices worldwide for the detection and diagnosis of a
wide spectrum of illnesses and injuries. In general, an x-ray imaging

Van Nostrand’s Scientific Encyclopedia, Copyright © 2006 John Wiley & Sons, Inc. 1
system consists of several components. An x-ray tube serves as the
source of x rays that penetrate the patient and impinge on an image
receptor such as a fluorescent screen or digital detector. The image
is “captured” photographically or electronically and processed to
accentuate anatomic features of interest in the patient. This image
is interpreted by the physician to arrive at an opinion concerning the
health status of the patient. The image may be stored physically or
electronically for future referral.

1 Projection Radiography

1.1 Nature and Production of X-Rays

X rays are a form of electromagnetic radiation similar to infrared,
visible, and ultraviolet light except higher in energy. They are widely
used in medicine to produce images (radiographs) of virtually every
region of the body. To achieve acceptable images, a source of x rays
is required that
1. provides a sufficient number of x rays in a short interval of time,
2. allows the user to vary the x-ray energy,
3. produces x rays in a reproducible fashion, and
4. meets current standards of safety and operational efficiency.

The source almost universally employed for imaging is the x-ray

tube. In this device x rays are released as high-speed electrons
impinge on a target of a heavy metal such as tungsten.
The main components of a modern x-ray tube are shown in Fig.
1. As electrical current heats the tungsten filament in the cathode,
electrons are boiled off (thermionic emission) and accelerated to-
ward the target in the anode through a potential difference of from
20 000 to 150 000 V (20–150 kVp, where kVp denotes the peak
kilovoltage across the tube). A focusing cup around the filament
helps direct the electrons onto a small area (the focal spot) of an
annular target that is embedded in a larger disk (the anode). X rays
are released in all directions as the electrons interact in the target.
The x rays are restricted by collimators outside the x-ray tube to
confine the x-ray beam to the anatomic region of interest in the
patient. Although intense bursts of x rays can be produced in this
manner, most (>99%) of the electron energy is dissipated as heat
in the target (Hendee and Ritenour, 1992). The anode rotates at

2
several thousand revolutions per minute to spread the heat over a
larger volume of target metal to keep the temperature of the metal
from reaching its melting point. [Figure 1]
As the high-speed electrons penetrate the target metal, they interact
with the electrons and nuclei of the target atoms through Coulombic
(charged particle) interactions. During interactions with electrons,
target atoms are ionized by ejection of inner-shell electrons, and x
rays are released as the affected atoms regain stability by reordering
of the electron shells. These x rays are termed “characteristic” x rays
because their energy is characteristic of the type of metal employed
as the x-ray target. For example, characteristic x rays of 58 and 67
keV are released from a tungsten target, where keV stands for kilo-
electron volt, and the electron volt is a unit of energy equal to that
gained by an electron accelerated through a potential difference of 1
V.
As electrons interact with nuclei of the target atoms, they are
slowed down, and energy is released in the form of x rays termed
bremsstrahlung (braking radiation). Bremsstrahlung provides a con-
tinuum of x-ray energies with a maximum value defined by the
energy of the impinging electrons. For most x-ray beams used for
medical imaging, bremsstrahlung is a more important contributor
compared with characteristic x rays. A notable exception is the sig-
nificant role played by characteristic x rays in beams employed for
mammography (imaging of breast tissue), an application discussed
later in this article.
To prevent the accelerated electrons from interacting with gas
molecules on their way to the target, components of the x-ray tube
are housed in a glass tube that is evacuated to very low pressure. The
tube is contained within an oil-filled shielded housing that provides
electrical insulation and prevents x rays from emerging in unwanted
directions. During an x-ray exposure, the technique factors [tube
voltage (kVp), tube current (mA), and exposure time (ms)] are
monitored carefully to prevent overheating of the target and other
components of the x-ray tube.

1.2 Interactions of X-Rays

For a parallel, monoenergetic beam of x rays, the intensity I of x rays
transmitted through a material of thickness z is

3

 
I = I0 exp − µ(z)dz ,

where the line integral is taken over all tissues along the x-ray beam,
I0 is the intensity with no material present, and µ is the linear x-ray
attenuation coefficient of the material, which may vary depending
on the material’s composition at each depth. If the material varies
in composition laterally from one point to the next, the transmitted
x rays also vary in intensity across the x-ray beam. By capturing the
transmitted x-ray beam on a film or fluorescent screen, the pattern
of variation in the material’s composition is reflected in the image.
As mentioned earlier, this image is termed a projection radiograph
because it is formed through projection of a material into an image
by transmitted x rays.
The linear attenuation coefficient µ is the sum of coefficients for
various types of x-ray interactions that may occur in the material.
For the energy range (20–150 keV) of x-ray energies used to produce
medical images, two types of interactions are dominant. These
interactions are the photoelectric effect, and the Compton Effect.
The dependence of these coefficients on x-ray energy is shown in
Fig. 2 for interactions in human soft tissue. [Figure 2]
In a photoelectric interaction, the x ray interacts almost always
with a tightly bound electron in an inner shell of an atom of the
absorbing material. Because of the importance of tightly bound
electrons, photoelectric interactions are strongly influenced by the
atomic number Z, with the probability of photoelectric interaction
per unit mass of material varying directly with Z3 of the material.
All of the energy of the x ray is transferred to the electron, which is
ejected from the atom with a kinetic energy Ek of

Ek = Ex−ray − Eb ,

where Eb is the binding energy of the ejected electron and Ex-ray

is the x-ray energy, which can also be expressed as hv, where h
is Planck’s constant (6.62 × 10–34 J-s) and v is the frequency of
the x ray. Since the x ray is only energy, it disappears during

4
a photoelectric interaction and is essentially replaced by a high-
speed electron and an ionized atom. Photoelectric interactions occur
predominantly with low-energy x rays, and decrease in likelihood as
(hv)–3 . A photoelectric interaction is depicted in Fig. 3(a). [Figure 3]
A Compton interaction results in the scattering of an x ray in a new
direction at reduced energy, rather than complete absorption of the
x ray. Compton scattering produces most of the scattered radiation
that degrades the quality of x-ray images. In this type of interaction,
the x ray interacts with a loosely bound electron in an outer shell
of an atom of the absorbing material. The electron is ejected with
an energy equal to the difference in energy of the incoming and
scattered x rays, with the assumption that the binding energy of
the ejected electron is so small it can be neglected. Because the
interaction occurs almost always with outer, loosely bound electrons,
the Compton interaction is independent of the atomic number Z of
the absorbing material. Compton interactions occur predominantly
for the higher range of x-ray energies employed for medical imag-
ing. The likelihood of these interactions decreases gradually with
increasing x-ray energy. A Compton interaction is depicted in Fig.
3(b).
For both photoelectric and Compton interactions, the linear attenu-
ation coefficient varies with the physical density (kg/m3 ) of the ab-
sorbing material. Hence, the intensity of x rays transmitted through
a material reflects the density of the material and, to varying degrees,
the atomic number of the material, depending on the energy of the
x rays and, therefore, the degree to which photoelectric interactions
are responsible for reducing the intensity. By choosing the x-ray
energy through selection of the voltage across the x-ray tube, an x-
ray beam may yield a projected radiograph that accentuates or sup-
presses differences in atomic number among different constituents of
an absorbing material. This characteristic is used to great advantage
in medical imaging. For example, low-energy x rays may be used
to amplify small differences in tissue composition that may be im-
portant for the detection of subtle abnormalities in soft tissue. Breast
imaging is one example where low-energy x rays are used to amplify
subtle soft-tissue differences. On the other hand, high-energy x
rays may be needed to suppress relatively large differences in tissue
composition so that more subtle features of tissue are not hidden by
“shadows” cast by strongly absorbing structures. Chest imaging is

5
one example where high energy x rays are used to penetrate the ribs
and other strongly absorbing structures to reveal more subtle features
of the underlying lung parenchyma. For x rays at any energy, those
that are not transmitted through the patient are absorbed or scattered
in the patient. These processes result in energy deposition in the
tissues of the patient, where the energy deposited per unit mass of
tissue is known as the “absorbed dose.”

1.3 Image Production on Screens and Film

Image receptors employed for x-ray imaging include film, usually
used in combination with fluorescent screens; fluorescent screens
contained within an image intensifier for fluoroscopic imaging of
dynamic processes; and solid-state detectors such as semiconductor
detectors and photostimulatable phosphorescent plates for genera-
tion of electronic signals. Photographic emulsions are sensitive to x
rays, and an x-ray image of exceptionally high quality can be formed
by direct exposure of photographic film to an x-ray beam. This
approach to medical imaging is occasionally used when exquisite
detail is required in the image, as might be the case when hair-
line fractures in the bones of extremities are suspected. However,
film by itself is an inefficient absorber of x rays, and for most
applications it would require long exposure times that would yield
images with unacceptable amounts of motion blurring. In almost all
cases, permanent x-ray images are produced by sandwiching the film
between “intensifying screens” that absorb x rays with reasonable
efficiency (up to 50%) and emit visible light that in turn exposes the
photographic film.
Most x-ray images are produced by sandwiching an x-ray film
between intensifying screens in an imaging cassette that is tightly
sealed (sometimes even evacuated) to produce intimate contact be-
tween the film and screens, thereby reducing geometric blurring.
Cassettes are available to hold film of different sizes from 4 × 4
inches to (10 × 10 centimeters) 14 × 17 inches (35.6 × 43 centime-
ters) used for chest radiography. Once the film has been exposed to
x rays or light from the intensifying screens, it contains a “latent
image” that can be made visible by chemical processing. In the
visible image, darker areas represent regions penetrated by a larger
number of x rays, whereas lighter areas depict regions where more
x rays are absorbed. In many cases, the lighter areas reveal bony

6
structures because the higher density and atomic number of bone
results in greater absorption of the impinging x rays.

1.4 Fluoroscopy
Projection radiography yields exquisite images that represent a
“snapshot in time” of the patient’s anatomy. In many cases these
images are sufficient to detect an abnormal condition and to arrive
at a diagnosis of the cause of the abnormality. But this is not true
in all cases. Sometimes a continuous image is required that yields
information about how the region of anatomy is changing or func-
tioning over the course of a few seconds or minutes. The technique
that yields such a continuous image is referred to as “fluoroscopy”
because it captures the image instantaneously on a fluorescent screen
and displays it in real time to the viewer.
Early applications of fluoroscopy employed a fluorescent screen
that was viewed directly by the physician. The image was so
dim, however, that the observer’s vision had to be “dark-adapted,”
and even then only gross features could be detected. The viewing
conditions resembled those present in a darkened movie theater.
Although this approach was highly unsatisfactory, it was the only
method available for fluoroscopy until the 1950s when the image
intensifier was developed.
An image intensifier is depicted in Fig. 4. X rays impinging on
the CsI input screen produce light that dislodges electrons from the
photocathode. These electrons are accelerated toward the anode and
are focused to pass through the aperture in this structure and onto the
output screen. Here they produce a minified image that is thousands
of times brighter than the image on the input screen. This image may
be captured optically and projected through a system of lenses and
mirrors for direct viewing by the observer. More often it is projected
into a television camera and converted into an electrical signal that
can be transmitted to a remote television monitor for reconfiguration
and viewing. Since the image is converted from light to an electrical
signal during this process, it can be converted easily to a digital
image for integration into a digital imaging network. [Figure 4]
Fluoroscopic methods are an essential part of medical x-ray imaging.
They are widely employed in many types of x-ray studies, includ-
ing those of the gastrointestinal tract and angiographic studies of

7
the central and peripheral circulatory system. They also are used
for image-guided therapy (e.g., coronary angioplasty, interventional
neuroradiology, and minimally invasive surgery), a therapeutic ap-
proach that is growing rapidly in application because of its reduced
morbidity and, often, lower cost compared with alternative therapeu-
tic approaches.

1.5 Digital Imaging

The combination of intensifying screens and photographic film has
many advantages for capturing and recording x-ray images. It is sim-
ple, portable, and inexpensive, and it yields images with excellent
anatomic detail. This approach, however, has some disadvantages as
well. It is limited to a rather narrow range of acceptable exposures
and offers little flexibility for image enhancement or data compres-
sion. Film images are bulky to store and easy to misplace, and they
must be transported physically from one location to another. Replac-
ing this analog method of image production with an approach that
yields digital images would eliminate these disadvantages. Hence,
considerable interest in digital approaches to projection radiography
has surfaced in recent years. In all likelihood, digital methods will
capture an increasing fraction of the projection radiography market
over the next several years.
In the digital radiographic unit the x-ray source and receptor are
computer controlled to yield digital images that are displayed in real
time on a video screen. Digital images can be stored on magnetic
media or optical disk, and a film-writing device can be used to
produce permanent analog copies of the images if they are needed.
Two types of image receptors currently are used in digital radio-
graphy. One type records the entire image simultaneously, while
the other acquires a complete image by scanning the x-ray beam
and receptor in synchrony across the patient, a technique known
as scan-projection radiography. Various methods are employed for
digital data acquisition, including the use of image intensifiers with
frame-grabbing television cameras, ionographic chambers, solid-
state detector arrays, and photostimulated-luminescence plates.
Conventional projection radiographic films can be digitized with a
scanning micro-densitometer to yield digital radiographs. As with
any other digital image, a digital radiograph can be transmitted

8
electronically to distant sites for interpretation and consultation,
a technique known as teleradiology. Digital radiographs can be
integrated with other digital images into an electronic image network
for correlation with other digital images such as those produced by
computed tomographic techniques, as discussed later.
Digital x-ray images offer several advantages over screen and film
radiographs. They can be computer-processed to highlight selected
features (edge sharpening, contrast enhancement, and noise smooth-
ing), and the data can be compressed for efficient transmission
electronically to distant sites. They require less storage space, and
copies can be easily presented for viewing without risk of misplac-
ing the original data. Digital x-ray images sometimes require less
radiation exposure compared with analog projection radiographs.
Finally, computer subtraction of digital images taken at different
times can often reveal changes that are not visible in the original
images. Disadvantages of digital radiographs include the cost of the
equipment required for their production compared with simple film-
screen techniques and possible losses in image detail occasioned by
the finite number of picture elements (pixels) used for image display.

1.6 Image Detail

The level of detail visible in an x-ray image is a compromise among
four kinds of resolution: spatial, contrast, temporal, and statistical.
Spatial resolution is related to the geometry of the image-forming
process and the resolving capacity of the imaging device (often
called the “image receptor”). Contrast resolution is affected by the
differences in the absorption and scattering characteristics of the
various tissues depicted in the image and by the impact of scattered
x rays on the receptor. Temporal resolution refers to the blurring
caused by the patient’s voluntary or involuntary movement (e.g.,
breathing, peristalsis, cardiac motion, or vascular pulsatile motion)
during exposure to x rays. Statistical resolution is related to the
number of x rays absorbed in the receptor to produce the image;
the noise is inversely proportional to the square root of the number
of x rays used to form the image.
The challenge of x-ray imaging is to capture subtle differences
(i.e., modulation) in tissue composition by passing an x-ray beam
through the tissue, and to transfer these captured differences onto

9
an image receptor where they can be seen by the human eye. That
is, the purpose of medical imaging is to transfer modulation in the
tissue (object) into modulation in the image. The degree to which
this modulation is transferred faithfully by an imaging system is
described as its modulation transfer function (MTF). With most
imaging systems, object modulation can be transferred to the image
with high fidelity if the modulation is of relatively low frequency
(i.e., the object is composed of relatively large structures that are
well separated from adjacent structures). For smaller structures that
are closer together, however, the integrity of modulation transfer
may be less exact. Hence, the MTF is frequency dependent, as
shown in Fig. 5 for two imaging systems. In this illustration, the
fidelity of system 1 is superior to that of system 2, as evidenced by
the higher values of MTF at higher frequencies. [Figure 5]
The MTF is a relatively simple (and somewhat simplistic) measure
of the performance of an imaging system. To determine the MTF,
one obtains an x-ray image of a very thin transparent slit in an
otherwise opaque object. The image of the slit is scanned with a
microdensitometer to determine the spread of the slit in the image.
The resulting curve is described by a “line spread function” that
can be characterized in frequency space through the process of
Fourier transformation. The result is the MTF, expressed as a
function of spatial frequency, as demonstrated in Fig. 5. The spatial
frequency, expressed in units such as line pairs per millimeter or
cycles per millimeter, is a measure of the frequency of variation
per unit distance. It may be compared with the temporal frequency,
expressed in units such as cycles per second or hertz, which is a
measure of the frequency of variation per unit time.

2 Image Interpretation
Interpretation of medical images is a complex process that is as
much art as science. It requires years of training, much of which is
provided in a “master and apprentice” style since it has not been pos-
sible to model the interpretive process quantitatively. On the other
hand, computer programs have been written to perform selected
analyses of the patterns of information present in certain types of
images such as chest radiographs and mammograms. By comparing
these analyses to databases compiled from a large number of normal

10
and abnormal images, a tentative diagnosis can be arrived at for pre-
sentation to the interpreter for confirmation. Such a decision-support
system is referred to as “computer-aided diagnosis.” Studies of this
approach have shown that it increases the accuracy of diagnosis of
the typical radiologist (Doi et al., 1993).
Computer-aided diagnosis is a far step from the use of expert
artificial-intelligence systems to provide a primary diagnosis from
medical images without human intervention. Such systems would
essentially have to mimic the thought processes of physicians inter-
preting images, including the incorporation of ancillary information
such as the patient’s history and data acquired by physical examina-
tion. These thought processes are largely intuitive, acquired through
apprenticeship training and experience, and are almost impossible to
articulate clearly. Replacement of the human interpreter of images
with a computer algorithm for image interpretation would be a major
challenge.
It is important to recognize, however, that an expert system for image
interpretation does not have to be perfect. Instead, it must function
only at a level at least comparable to the accuracy of the average
radiologist. With this criterion in mind, exploration is occurring
of expert systems for image interpretation in medicine. Some of
these explorations employ “neural networks,” algorithms that use
feedback information from humans to build decision rules based on
the successes and failures of past decisions. Medical imaging is still
a long way from automated diagnosis, as contrasted with computer-
aided diagnosis, which relies ultimately on human confirmation or
rejection of a proposed diagnosis. Still, progress in expert systems in
other imaging disciplines, including military surveillance and space
exploration, has been spectacular over the past two decades, and the
application of such systems to medical imaging remains promising.
See also Artificial Intelligence: Neural Networks.

3 X-Ray Tomography
In a conventional radiograph, all structures within a three-
dimensional volume of tissue exposed to X rays are projected
“in focus” onto the two-dimensional image receptor. As a result,
an anatomic structure of interest is often obscured in the image
by the shadow of objects that lie above or below it in the patient.

11
Removing the presence of these obscuring shadows often helps
reveal the structure of interest. Two approaches, analog tomography
and computed tomography, are available for removing the image of
obscuring structures. As computed tomography has grown over the
past two decades, analog tomography has diminished in popularity.

3.1 Analog Tomography

Analog tomographic images are produced by keeping structures in
focus in a selected image plane in the patient, while blurring the
image of structures above and below the imaging plane. The blur-
ring is accomplished by moving the X-ray tube and image receptor
during exposure through an angle with a pivot (fulcrum) in the image
plane. In actuality, the image plane has a finite thickness (and is
more properly referred to as an image section) that can be reduced
by increasing the angle of motion. The location of the section within
the patient may be changed by positioning the patient appropriately
with respect to the pivot of the tomographic angle. Although the
shadows of overlying and underlying structures are blurred, they are
not completely removed from the image, and they frequently appear
as “ghostlike” artifacts that interfere with the clarity of structures
of interest. In addition, analog tomography has the limitations that
conventional image receptors are unable to resolve small differences
(1–2%) in the intensity of incident radiation, and the wide-area x-
ray beams employed produced considerable scattered radiation that
interferes with the visualization of subtle detail in the image. Analog
tomography is still used for certain applications such as a supple-
ment to chest radiography. For other tomographic needs, however, it
has been replaced by computed tomography because this approach
reduces the limitations often encountered with conventional tomog-
raphy.

3.2 Computed Tomography

The mathematical model used in computed tomography (CT) was
described in 1917 in a remarkable work by the Austrian mathemati-
cian Radon (Radon, 1917). The model was first used in applications
of image reconstruction to radio astronomy, electron microscopy,
and optics. In 1972, the first commercially available x-ray trans-
mission CT unit for studies of human anatomy was announced. In
1979, the developer of this unit, G. Hounsfield (1973), shared the

12
Nobel Prize in Medicine with a physicist, A. Cormack (1963), who
did early work on image reconstruction. The first commercial CT
unit employed a narrow beam of x rays scanned across the patient in
synchrony with a scintillation detector moving on the opposite side
of the patient. The intensity I of x rays measured by the detector is
given by

  
I = I0 exp − µi xi ,

where µi represents the linear attenuation coefficient of each of

several structures in the path of the narrow x-ray beam and xi
represents the thickness of each of the structures.
With a single measurement of X-ray transmission, the separate atten-
uation coefficients cannot be determined. With multiple transmis-
sion measurements obtained at different orientations, however, the
separate coefficients can be distinguished to yield a two-dimensional
distribution of linear attenuation coefficients across the imaging sec-
tion defined by the thickness of the scanning X-ray beam. This
distribution can be mapped as CT numbers determined from the
formula

CT number = 1000(µ − µw )/µw ,

where µ is the coefficient at a specific location in the section and

µw is the linear attenuation coefficient of water for the x-ray energy
employed for CT scanning. From this expression it can be appre-
ciated that the CT number for water is zero and the CT number
for air, with a very small value of µ, is close to –1000. The CT
number would be about +1000 for a structure such as compact bone
or colon containing barium contrast medium, because the physical
density would be almost twice that of water. The CT numbers can
be presented as various shades of gray to provide a visual display of
the distribution of tissues across the image section. This display is
referred to as a CT image.
The X-ray tube and scintillation detector of early CT units scanned
the patient along a linear path perpendicular to the axis of rota-
tion of tube and detector. Measurements of x-ray transmission
were obtained every 2.5 mm (0.09843 in) over the linear path of

13
40 cm (15.75 inches) length to yield a total of 160 transmission
measurements. Then the angular positions of the x-ray tube and
detector were advanced 1°, and another scan of 160 measurements
was obtained. This process was continued over 180° to provide 28
800 transmission measurements. These data were then transmitted
to a computer equipped with a mathematical package for recon-
structing a cross-sectional image of the scanned section. Initially,
reconstructed images were computed by a time-consuming iterative
mathematical model. More recently, faster algorithms using a con-
volution (filtered back-projection) model for image reconstruction
have been employed (Kak and Slaney, 1988). These algorithms rely
on a Fourier transformation of x-ray projection data into frequency
space, permitting use of ramp and cutoff-frequency filters to improve
image quality and enhance subtle features in the image. See also
Fourier Transform.
The combination of translational and rotational motion employed in
early CT units yielded satisfactory CT images, but several minutes
were required for data acquisition, limiting the method to studies
of structures such as the head and extremities where voluntary and
involuntary motion of the patient is not a problem. There was great
interest in applying CT to thoracic and abdominal studies, but the
scan time had to be decreased from minutes to seconds if these
areas were to yield satisfactory images. Major efforts to reduce
scan time led to the development of successive generations of CT
scanners as depicted in Fig. 6. The initial translate-rotate scanner
(first generation) is shown in Fig. 6(a), whereas an array of detectors
(second generation) that tracks the linear motion of the x-ray beam
is shown in Fig. 6(b). A purely rotational design (third generation)
for CT scanning is depicted in Fig. 6(c), and a fourth-generation
design is shown in Fig. 6(d) in which an X-ray tube rotates around
the patient within a stationary ring of X-ray detectors. With the
rotational scanners, data for a single section can be acquired in just a
few seconds. By combining these geometries with the patient gantry
moving continuously along its long axis, cross-sectional images of
many slices can be obtained in minimum time, a process known as
“spiral” or “helical” scanning (Heath et al., 1995). Finally, images of
vertical (sagittal or coronal) slices through the body can be produced
by compiling arrays of attenuation coefficients along axes parallel to
the patient’s long axis. [Figure 6]

14
Although the evolution of CT scanning geometries shortened the
time to acquire images, none of the geometries permitted data ac-
quisition in a time (less than 0.1 s) short enough to capture images
of the heart and other blood-perfused organs without significant
degradation caused by motion. For CT images of these organs, a
way is needed to produce x-ray projection data without mechanical
motion in the scanner. A scanner of this sort is known as a fifth-
generation scanner (Hendee, 1983). It employs an electron-beam
gun that generates x-ray beams in different directions by scanning
over a stationary concave metal target. The resulting scan times are
as fast as 50–100 ms.
CT images are most commonly displayed as cross-sectional rep-
resentations of patient anatomy. By examining continuous cross-
sectional images, radiologists can intuitively integrate the imaging
information into a three-dimensional model of anatomy. This inte-
gration can also be accomplished physically by recognizing that a
data set of contiguous two-dimensional sections in computer mem-
ory is in actuality a three-dimensional database. By mathemati-
cally compiling data across planes that cut across this database in
preferred directions, coronal and sagittal images, as well as im-
ages at oblique angles, can be produced. Furthermore, the three-
dimensional database can be analyzed to yield images that give
the distinct impression of being three-dimensional. They also can
be windowed to yield images of specific ranges of CT numbers
corresponding to selected types of tissue (Fishman et al., 1989).

4 Image Networking
Medical imaging includes several systems, including x-ray com-
puted tomography and magnetic resonance imaging, that present
imaging data in a digital format. Linking these systems electron-
ically in a picture archiving and communications system (PACS)
[referred to as an image management, archiving and communication
system (IMACS) when connected into other information networks
such as the radiology and hospital information systems] offers sev-
eral advantages, including these: images can be stored electronically
in a central location, images from different systems can be called
up for display and comparison in a viewing center, and images
can be transmitted to remote locations for viewing by specialists
without loss of the central file of data. A PACS has the potential

15
of expediting diagnoses, improving diagnostic accuracy, enhancing
information to other physicians, and eliminating misplaced and lost
films. With the availability of digital x-ray imaging methods, PACS
can be expanded to encompass essentially all imaging systems em-
ployed in medicine. Many leaders in medical imaging believe that
a fully integrated PACS represents the radiology department of the
future. Others are convinced that the expense of converting fully to
a PACS (several million dollars for a typical imaging facility) will
prevent most departments from converting completely to PACS, at
least in the near future. Instead, they believe their conversion will
occur incrementally, with film remaining as the preferred detector
for some applications (e.g., mammography and chest imaging) for
some time to come. See also Nuclear Magnetic Resonance (NMR)
and Magnetic Resonance Imaging (MRI).

5 Applications
The use of X-rays is so ubiquitous in medicine that almost everyone
has had an x-ray examination. Most examinations are for routine
purposes such as to detect dental caries, confirm that a fractured
bone has been set properly, provide a routine examination of the
chest to rule out pneumonia, or perform a barium swallow or enema
to study the integrity of the gastrointestinal tract. Many more
applications of X-rays to medicine exist, however. Even though
these applications may be encountered less frequently, they are
critically important to the patients who need them. Two specific
applications are described here as examples of the many interesting
challenges presented to physicists and engineers working in the
discipline of medical physics.

5.1 Mammography
One of every eight women will develop breast cancer in her lifetime,
and many will die as a result. The most effective measure against
this disease is periodic screening of the breasts with x rays for
detection of cancer before it has a chance to spread. Many voluntary
and professional organizations recommend annual screening with X-
rays for women 50 years of age and older, and some recommend
screening every 1 to 2 years for women between 40 and 50 years of
age. X-ray imaging of the breasts is referred to as mammography.
See also Cancer and Oncology.

16
A mammographic X-ray image must be able to reveal very subtle
differences in the composition of soft tissues of the breast, as well
as the presence of tiny (<100 µm) specks of calcium hydroxyapatite
(calcifications) that suggest the presence of malignant cells. That
is, mammographic images must bring out very small variations in
contrast and must also yield exquisite detail within a background of
relatively low noise. These images must depict the relatively thick
tissues near the chest wall without losing detail in thinner regions
of the breast. These demands place inordinate requirements on
systems for production of satisfactory mammographic images. They
also necessitate periodic evaluation of image quality and radiation
dose by a physicist experienced in inspection and quality control of
mammographic units (American College of Radiology, 1993).
Subject contrast is enhanced in mammography by the use of low-
energy x rays that interact strongly by photoelectric processes. Re-
liance on these interactions yields the greatest difference in trans-
mission among structures that are similar, although not identical, in
chemical composition. The use of low-energy X-rays also aids in the
identification of calcifications. In most cases today, mammography
is performed with special x-ray units that employ an X-ray tube
operated at 20–35 kVp and a film and single-screen image recep-
tor designed specifically for mammography. The x-ray tube has a
molybdenum target that produces characteristic K x rays of 17.4 and
19.8 keV superimposed on a bremsstrahlung x-ray spectrum with a
maximum energy equal numerically to the peak voltage expressed
as kVp. The x-ray beam is filtered with a thin sheet of molybdenum
that is relatively transparent to the characteristic x rays but absorbs
higher-energy x rays with high efficiency because of the existence of
the K absorption edge above 20 keV.
Screen–film mammography is an exquisitely sensitive and useful
imaging method. Still, it may be possible to improve breast imaging
by replacing screen-film receptors with digital detectors. Unlike
film, digital systems acquire, display, and store images indepen-
dently, so that each of these functions can be optimized without
interfering with the others. Also unlike film, the response of digital
detectors varies linearly over a wide range of exposures. In addition,
digital images can be “windowed” to enhance the contrast of spe-
cific regions of tissue. Both of these attributes of digital detectors
improve the contrast sensitivity of digital mammographic methods

17
compared with analog screen–film methods. These advantages may
well outweigh the disadvantage of a slightly lower spatial resolution
with digital methods. Although digital detectors have high potential,
the practicality, dependability, and low cost of film continue to make
it the preferred detector for mammography in clinical settings.
Radiation can induce cancer as well as detect it. (Radiation can also
be used to treat cancer, which is the medical specialty of radiation
oncology.) In the past, concern has been expressed that screening
large numbers of asymptomatic women for breast cancer could in-
duce as many cancers as are cured through the process of early detec-
tion. This concern was primarily a result of the relatively high doses
delivered to patients with earlier, less sensitive methods of image
production. With today’s ultrasensitive screen–film combinations,
together with the use of low-energy, highly filtered x-ray beams, the
induction of cancer is much less likely. The overwhelming balance
of benefit over risk in screening programs for older women has
largely dispelled concern over the possible induction of cancer in
the screened women (Kimme-Smith, 1992). Nevertheless, efforts to
reduce radiation dose further in mammography should be continued
until each examination uses no more than the essential number of x
rays to reveal underlying pathology.

5.2 X-Ray Treatment Planning

Cancer is characterized by an abnormally rapid proliferation of
cells. The successful treatment of cancer with X-rays requires that
the cancer is localized and that that cancer and its microscopic
extensions into normal tissue receive a dose of radiation sufficient
to kill the cancer cells. At the same time the dose to nearby normal
tissues must be kept low enough to avoid serious complications. The
balance between these doses is depicted in Fig. 7 (Wolbarst, 1988).
The probability Ptumor (D) that the cancer will be controlled by killing
tumor cells increases with D, the radiation dose. The probability
Porgan (D) that life-threatening complications from radiation damage
can be avoided is, of course, a decreasing function of D. The
probability Ppatient (D) that the patient will survive both the disease
and the treatment is

18
 Ppatient (D) = Ptumor (D) × Porgan (D).

[Figure 7]
The patient’s survival probability as a function of radiation dose
is the lower curve in Fig. 7. If the dose is low, the tumor is
likely to recur. At high doses, the therapy itself is likely to cause
unacceptable complications, perhaps threatening the patient’s life.
Between these extremes the physician seeks to maximize the pa-
tient’s survival of both the cancer and the treatment. If the radiation
dose that optimizes the balance between curative treatment and com-
plications is so high that the patient experiences extreme discomfort,
the physician may be forced to settle for a lower dose even though
the patient’s chance of surviving the cancer is reduced.
Radiation-dose optimization demands accurate delineation of the
margins of the cancer. That in turn demands good medical imaging.
Before CT, cancer localization and treatment planning depended
principally upon projection radiography combined with clinical and
surgical (if available) information about the patient, together with
knowledge of the natural course of the disease. But projection
radiography has many disadvantages. It is frequently hard to see the
cancer clearly, and it is often difficult to transcribe the radiographic
information onto a cross-sectional plane through the patient for
designing the treatment plan describing the orientation and relative
weighting of radiation beams used for treatment.
CT and, more recently, magnetic resonance imaging have largely
overcome these problems. Many treatment centers have a CT unit
dedicated to treatment planning and monitoring. The images pro-
vide not only a cross-sectional picture of the patient’s anatomy,
including the cancer and surrounding tissues, but also an accurate
representation of the body contour and organs that are particularly
sensitive to radiation. Because the data are digital, they can be
entered immediately into the treatment-planning computer so that
proposed treatment plans can be superimposed directly onto the CT
images. Once an acceptable plan is accepted, it can be implemented
on the x-ray treatment machine. This entire procedure must be care-
fully monitored to ensure that no misalignment occurs and that the
treatment plan accurately reflects the radiation administered during

19
treatment. Monitoring of this process is the responsibility of the
medical physicist.
In most radiation therapy centers, commercial imaging units are
used to generate CT information for treatment planning. In an
effort to improve alignment of the CT information with the treat-
ment geometry, a few medical physicists have built CT scanners
on treatment-simulating gantries. Although these units yield spatial
resolution inferior to that with commercial CT units, they duplicate
the treatment geometry and provide images good enough to be useful
for treatment planning.
Most X-ray treatments involve the use of multiple fixed radiation
fields converging on the cancer from different directions. In this
manner the dose is concentrated in the cancer while much lower
doses are administered to surrounding normal tissues. In some
cases a better dose distribution can be achieved by continually
rotating the gantry of the treatment machine around the patient
during treatment so that only the cancer is constantly in the path
of the x-ray beam. One complication is that cancers are almost
always asymmetrical. Therefore, the size of the x-ray beam must
be expanded or contracted continuously during rotation. The dose
distribution could be improved even more by simultaneously varying
the dose rate during rotation. This approach, termed “conformal
therapy,” is being pursued in several centers in the United States
and overseas (Mageras et al., 1994). It requires detailed three-
dimensional knowledge of the anatomy of the irradiated tissue,
together with exquisite computerized control of the x-ray unit itself,
as well as the x-ray gantry and patient couch position.
See also X-Ray Scan and Other Medical Imagery.

Additional Reading

Cormack, A.: Representation of a Function by Its Line Inte-

grals, with Some Radiological Applications, J. Appl. Phys. 34,
2722–2727 ( 1963).
Doi, K., M. Giger, R. Nishikawa, H. MacMahon, and R. Schmidt:
Artificial Intelligence and Neural Networks in Radiology Appli-
cation to Computer-Aided Diagnostic Schemes, pp. 301-322,
in Hendee, W. R., and J. Trueblood (Eds.), Digital Imaging,
Medical Physics Publishing, Madison, WI, 1993.

20
E. Fishman, D. Magid, D. Ney, K. Macrae, J. Kuhlman, and
D. Robertson,: Three-Dimensional Imaging: Advanced Medical
Applications, in: J. Anderson (Ed.), Innovations in Diagnostic
Radiology, Springer-Verlag, New York, NY, pp. 47–68, 1989.
Heath, D., et. al: Three-Dimensional Spiral CT during Arterial
Portography: Comparison of Three Rendering Techniques, Ra-
diographics 15, 1001–1012, ( 1995).
Helms, C. A.: Fundamentals of Skeletal Radiology, 3rd Edition,
Elsevier Health Sciences, New York, NY, 2004.
Hendee, W. R., G. S. Ibbott, and E. G. Hendee: Radiation Ther-
apy Physics, 3rd Edition, John Wiley & Sons, Inc., Hoboken, NJ,
2004.
Hendee, W. R., and R. Ritenour: Medical Imaging Physics, 4th
Edition, John Wiley & Sons, Inc., New York, NY, 2002.
Hendee, W. R.: The Physical Principles of Computed Tomogra-
phy, Little, Brown & Company, Boston, MA, 1983.
Hounsfield, G.: Computerized Transverse Axial Scanning (To-
mography). I. Description of System, Br. J. Radiol. 46,
1016–1022, ( 1973).
Kak, A., and M. Slaney: Principles of Computerized Tomo-
graphic Imaging, IEEE Press, New York, NY, 1988.
Kimme-Smith, C.: New and Future Developments in Screen-
Film Mammography Equipment and Techniques, Radiol. Clin.
North Am. 30, 55–66 ( 1992).
Mettler, F. A.: Essentials of Radiology, 2nd Edition, Elsevier
Health Sciences, New York, NY, 2004.
Radon, J.: Über die Bestimmung von Funktionen durch ihre Inte-
gralwerte laengs gewisser Mannigfaltigkeiten (on the Determina-
tion of Functions from Their Integrals along Certain Manifolds),
Ber. saechs. Akad. Wiss. Leipzig, Math.-Phys. Kl. 69, 262–277 (
1917).
Staff: American College of Radiology: The Overview of Mam-
mography Accreditation, American College of Radiology, Re-
ston, VA, 1993.
Webb, W. R., and W. E. Brant: Fundamentals of Body CT, 3rd
Edition, Elsevier Health Sciences, New York, NY, 2005.
Wolbarst, A.: Optimization of Radiotherapy Treatment Planning,
Encyclopedia of Medical Devices and Instrumentation, Vol. 4,
Wiley-Interscience, New York, NY, 1988, p. 2488.

21
Figures

Figure 1.

Components of a modern diagnostic x-ray tube.

Figure 2.

The linear attenuation coefficient for x rays in soft tissue is the sum
of two dominating contributions. Attenuation coefficients for these
two contributions, Compton scattering and photoelectric absorption,
are plotted as functions of the x-ray energy.

22
Figure 3.

(a) Photoelectric interaction; (b) Compton scattering.

Figure 4.

X-ray image intensifier.

23
Figure 5.

Modulation transfer function (MTF) for two x-ray imaging systems,

with system 1 providing higher spatial fidelity compared with system
2.

Figure 6.

Evolution of geometries of x-ray CT scanners. (a) First-generation

scanner, in which the pencil x-ray beam is both translated and

24
rotated to cover the region of the body being imaged; (b) second-
generation scanner, with a diverging fan beam and detector array that
are translated and rotated; (c) third-generation scanner, with a fan-
beam source that rotates around the body, together with its detector
bank; (d) fourth-generation scanner, with a rotating fan-beam source
and a stationary ring of detectors.

Figure 7.

Therapeutic outcome probabilities for cancer patients undergoing x-

ray treatment, plotted schematically as functions of radiation dose D.
The probability of success (Ppatient ) is the product of the probability
that the tumor will be destroyed (Ptumor ) and the probability (Porgan )
that the treatment will not cause life-threatening damage to normal
tissue.

25