Radiography: Initial Clinical
William A. Brody1
Douglas M. Cassel1
F. Graham Sommer1
Leonard A. Lehman&
Robert E. Alvarez1
Norbert J. Pelc
Stephen J. Riederer2
Anne L. Hall2
Received March 26, 1981; accepted May 12,
Presented at the annual meeting of the Ameri-
can Roentgen Ray Society, San Francisco, March
This work was supported by the General Elec-
tric Company (Medical Systems Division), the Na-
tional Science Foundation, and Established Inves-
tigator Award (W. R. Brody) from the American
Heart Association and the Central Mission Trails
‘Department of Radiology S-072, Stanford
University School of Medicine, Stanford, CA
94305. Address reprint requests to W. R. Brody.
2 Medical Systems Division, General Electric
Co., Milwaukee, WI 53201.
AJR 137:201-205, August 1981
0361 -803X/81 I 1372-0201 $00.00
© American Roentgen Ray Society
Selective removal of substances of a particular mean atomic number in projection
radiographs has been accomplished using an experimental system for line-scanned
digital radiography. X-ray images were produced with two different x-ray spectra by
modulating the kilovoltage applied to the x-ray source and the filtration between the
source and the patient. A Compton/photoelectric decomposition algorithm provided
subtraction images with bone or soft-tissue (water) shadows removed. Applications to
chest, abdominal, and skeletal imaging in vivo are demonstrated.
Existing radiographic systems do not make explicit use of the energy infor-
mation present in the transmitted x-ray spectrum. Manipulation of the energy
spectrum in film radiography has been quite limited, for example, high kilovoltage
(kVp) for minimizing bone shadows in chest radiographs, and low kilovoltage
spectra for maximizing soft-tissue contrast in mammography. True energy-de-
pendent radiographic imaging methods have been applied to projection radiog-
raphy for dual-beam and three-beam K-edge detection of iodine [1 , 2] using
quasi monoenergetic x-ray spectra with subtraction fluoroscopy.
The electronic x-ray detectors for computed tomography (CT) have made
possible the use of dual-energy methods for identification of material atomic
number and density. A general method for dual energy CT developed by Alvarez
and Macovski  is based on a Compton/photoelectric decomposition of x-ray
attenuation. While the concept of dual-energy CT has become familiar to radiol-
ogists, it has not been generally recognized that these same dual energy methods
can be applied to projection x-ray imaging as well . For projection radiography,
these dual energy techniques provide selective subtraction or cancellation of
materials of certain average atomic number. This report describes the initial
application of a prototype imaging system for dual-energy line-scanned radiog-
raphy for chest and abdominal radiography in vivo.
Materials and Methods
An experimental system for scanned projection radiography was developed using a
General Electric 8800 CT/T scanner extensively modified for dual-energy imaging. A line-
scanned radiograph is produced by mechanically translating the patient through the
narrowly collimated (1 .9 mm) x-ray fan beam. A 511-element linear-array xenon detector
collects the x-rays transmitted through the patient and produces an electrical signal which
is converted into digital format. A minicomputer performs the subsequent normalization,
calibration, and logarithmic compression of the intensity signal similar to CT.
As the patient is translated through the gantry, the x-ray generator is pulsed, each pulse
yielding one line of x-ray transmission information. An image is constructed by the
juxtaposition of such lines. For dual-energy imaging, the generator produces, at 60 Hz, x-
ray pulses that alternate between 135 kVp, 250 mA, 5.5 msec pulse width and 85 kVp,
1000 mA, and 3.3 msec. In synchrony with the x-ray pulses, a rotating filter assembly
placed between the tube and the patient filters the low and high kilovoltage spectra with
Fig. 1 -Dual-energy radiography for oral cholecystography. A, 85 kVp effectively removes bowel gas, allowing higher contrast image of gallbladder.
(unsubtracted) line-scanned radiograph. Gallbladder opacified, but partially Extrahepatic common bile duct faintly seen.
obscured by superimposed bowel gas. B, Soft-tissue subtraction image
BRODY ET AL.
AJR:137, August 1981
Fig. 2.-Dual-energy excretory urogram. A. Line-scanned radiograph at
85 kvp. Kidneys and collecting systems faintly visualized. B, Soft-tissue
subtraction image removes superimposed bowel gas, allowing enhancement
of kidneys, collecting systems, and descending colon, which contains dilute
0.1 mm erbium and 1 .52 mm bronze, respectively. The combination
of kilovoltage and filtration yields x-ray spectra with average ener-
gies of about 45 and 85 keV.
Because the x-ray table is translated continuously and alternate
lines are generated at low and high kilovoltage, subtraction imaging
requires the low kilovoltage image to be shifted by one line with
respect to the high kilovoltage image. Before subtraction, the low
and high energy images are interpolated so that they will be exactly
water-soluble contrast medium from previous CT scan. Retroperitoneal fat
planes, which appear as low density (negative) shadows, are enhanced by
soft-tissue subtraction process.
The manner in which the low and high kilovoltage images are
combined has been discussed in detail elsewhere [5, 6]. Briefly,
the algorithm converts the high and low kilovoltage images into the
aluminum and Lucite components of the scanned object. These
aluminum and Lucite ‘basis” images can in turn be combined
linearly to form a tissue cancellation image. Images in which the
bone is cancelled, or the bone shadows are removed by substituting
the bone with an equivalent amount of soft-tissue, can be similarly
produced. In fact, material of any desired atomic number can be
Fig. 3.-Metastatic prostate carcinoma. A, Diffuse rib metastases from (bone mimic tissue subtraction) shows pleural or extrapleural soft-tissue
prostate carcinoma. B, 85 kVp scanned radiograph (unsubtracted). C, Soft- shadows associated with rib metastases. This image enhances airlsoft-tissue
tissue subtraction shows diffuse metastases to ribs and spine with complete interfaces: pleura, airways, pulmonary vessels.
removal of heart, pulmonary vascularity, and airways. D, Bone subtraction
AJR:137, August 1981
DUAL-ENERGY RADIOGRAPHY 203
cancelled by an appropriate linear combination of the two basis
images. With monoenergetic low and high energy beams the algo-
rithm would simply perform a weighted linear subtraction to produce
the aluminum and Lucite images (or any other desired subtraction
image). However, with the broad spectra resulting from x-ray tubes,
the effects of beam hardening introduce significant nonlinearities
that require more complex image combination [5, 6].
A scanning speed of 35 mm/sec was used for these dual energy
radiographs. The radiation exposure, measured at the surface of a
chest phantom using thermoluminescent dosimeters, was 130 mR
(33.5 x 106 C/kg).
Dual-energy technique has been used to image the chest, skel-
eton, kidney, liver, gallbladder, bowel, and vascular systems. De-
tailed evaluation of the diagnostic effectiveness of dual-energy
radiography for different disease states is underway. Informed
consent was obtained for all examination under protocols approved
by the Stanford Human Subjects Committee.
For abdominal imaging, selective soft-tissue cancellation
provides effective removal of bowel gas shadows as dem-
onstrated in figures 1 and 2 of the gallbladder and kidney.
Visualization of the gallbladder in the unsubtracted image is
limited by the superimposed bowel gas shadows as shown
in the 85 kVp line-scanned radiograph (fig. 1A). With the
use of soft-tissue subtraction, the bowel gas shadows are
removed, allowing the gallbladder to be displayed at much
BRODY ET AL. AJR:137, August 1981
higher contrast (fig. 1 B). The extrahepatic common duct
can be seen on this subtraction image.
An excretory urogram (fig. 2) of a patient with lymphoma
shows how soft-tissue subtraction (removal of water-density
variations) eliminates the confusing bowel gas shadows that
obscure the contrast-filled kidneys and collecting systems.
This particular patient had water-soluble contrast in the
gastrointestinal tract for a prior CT scan. While this contrast
is not particularly apparent on the unsubtracted image, the
distal colon is well demonstrated on the soft-tissue-can-
celled image (fig. 2B). Also the perirenal and other retro-
peritoneal fat planes are enhanced by the subtraction pro-
cess. This enhancement is achieved because of the differ-
Fig. 4.-Soft-tissue subtraction provides improved skeletal detail by re-
moving interfering superimposed structures. Soft-tissue-cancelled left ante-
rior oblique projection shows sternum, ribs, and other skeletal structures.
ence in chemical composition (average atomic number)
between fat and lean body tissues.
In the chest, dual-kilovoltage imaging allows selective
cancellation of either bone or soft-tissue shadows. A film
chest radiograph on a patient with prostate carcinoma and
known bony metastases shows some of the rib lesions (fig.
3A), but obscures the diffuse involvement of the spine. The
unsubtracted scanned projection radiograph at 85 kVp (fig.
3B) shows similar features. However, the selective soft-
tissue subtraction image (fig. 3C) shows the bony metas-
tases clearly with rib and spine involvement. The alternative,
the bone-cancellation image, highlights the extraosseous
soft-tissue shadows associated with some of the rib metas-
tases (fig. 3D). It displays the soft tissues of the lungs and
mediastinum without the usual thoracic cage image.
The improved skeletal detail that results from soft-tissue
subtraction is illustrated by figure 4, a soft-tissue-cancelled
image of the human chest obtained with a shallow left
anterior oblique projection. Note the clarity with which the
sternum is demonstrated.
Finally, application of soft-tissue subtraction to determine
calcification of pulmonary nodules is shown in figure 5. A
large pulmonary nodule located adjacent to the left hemidia-
phragm is seen with the line-scanned system at 85 kVp (fig.
5A). Calcification was suspected on plain film chest radi-
ography, but could not be confirmed with this unsubtracted
image, even when edge enhancement algorithms were ap-
plied. On the other hand, the soft-tissue subtraction image
(fig. SB) shows unequivocally the crescentic calcification
around the periphery of the nodule.
Dual-energy scanned projection radiography provides se-
lective cancellation (or enhancement) of materials of any
specified average atomic number. While non K-edge dual-
energy techniques have been proposed before, attempts to
produce subtraction images have been impaired for two
Fig. 5.-Calcification in chest nodule
can be detected using dual-energy sub-
traction. A. Large nodule adjacent to left
hemidiaphragm on unsubtracted 85 kVp
film; calcification cannot be demon-
strated. B. Soft-tissue subtraction image
unequivocally shows rim of calcification
surrounding periphery of lesion (arrows).
AJR:137, August 1981
reasons: (1 ) the film/screen systems for x-ray detection
may have been limited by the nonlinearities and limited
dynamic range of film detectors, and by the energy depen-
dent scatter accepted by these large area detectors [7, 8];
and (2) the lack of rapid energy switching makes the sub-
traction method highly susceptible to motion artifacts for
application in vivo . With the rapid switching generator
used in the current study, the low and high kilovoltage
images are in perfect spatial registration, so that motion
effects are minimized.
While any desired atomic number material may be can-
celled by this technique, in our limited experience with the
method, the selective soft-tissue (water) subtraction and the
bone subtraction were most useful. The soft-tissue subtrac-
tion images remove soft-tissue variations and, in the chest
or abdomen, isolate the skeletal structures from confusing
superimposed shadows. Abdominal and retroperitoneal fat
planes, having a lower average atomic number than water,
become negative shadows on the tissue-cancelled image.
The soft-tissue subtraction also seems useful for the detec-
tion of calcification in soft-tissue lesions, such as pulmonary
With the administration of urographic or cholecysto-
graphic contrast media, the soft-tissue subtractions en-
hance kidney and gallbladder visualization. The removal of
confusing bowel gas shadows may alleviate the need for
conventional tomography which is so often required for
satisfactory studies. In addition, soft-tissue subtraction may
be useful in small and large bowel imaging either with water-
soluble contrast media or with barium.
The bone subtraction images, especially those with the
bone space filled by an equivalent thickness of soft tissue
(bone mimic tissue subtraction), were found most useful in
the chest, where the confusing rib shadows are eliminated
for unimpaired view of the pulmonary parenchyma. Because
of the obscuration of pulmonary nodules lying under the ribs
, we postulate that the bone removal will improve nodule
detection. In addition, we learned from these studies that
the bone mimic tissue image enhances the visualization of
air-tissue interfaces, such as the mediastinal pleural reflec-
tions, the pleural surfaces, the larynx, trachea, and bronchi.
Although the x-ray exposure for dual energy imaging may
be higher than that of a single energy image, this is not a
requirement. Moreover, the use of tissue or bone subtraction
with dual energy leads to images in which lesion detection
is limited by signal-to-noise ratio and hence by dose. In the
unsubtracted images as with conventional radiography, le-
sion detection is more often limited by interference from
superimposed high contrast structures than by dose. With
subtraction, these interfering tissues can often be removed,
allowing one to see low contrast lesions by increasing the
contrast of the displayed image, up to the limits imposed by
1 . Jacobson B. Dichromatic absorption radiography. Dichroma-
tography. Acta Radio! (Stockh) 1953;39:437
2. Houk TL, Kruger RA, Mistretta CA, et aI. Real-time digital k-
edge subtraction fluoroscopy. !nvest Radio! 1 979; 1 4 : 270-
3. Alvarez RE, Macovski A. Energy-selective reconstructions in
x-ray computed tomography. Phys Med Bio! 1976;21 : 733-
4. Chan JLH, Alvarez RE, Macovski A. Measurement of soft tissue
overlying bone utilizing broad band energy spectrum tech-
niques. IEEE Trans Nuc! Sd 1976;NS-23:551 -554
5. Brody WR, Butt G, Hall A, Macovski A. A method for selective
soft tissue and bone visualization using scanned projection
radiography. Med Phys 1981 ;8:353-357
6. Lehmann LA, Alvarez RA, Macovski A, et al. Generalized image
combinations in dual kvp digital radiography. Med Phys 1981 ;8
7. Yeh PS, Chan JLH, Macovski A. Applications of filtered
bremsstrahlung spectra in radiographic studies. II. Measure-
ments of iodine in tissue and separation of bone and water.
IEEE Trans Nuci Sci 1979; NS-26:431 1-4319
8. Kruger RA, Sorenson JA, Nikalson L, Armstrong JD. Dual
energy film subtraction technique for detection of calcification
in pulmonary nodules. Presented at the annual meeting of the
Radiological Society of North America, Dallas, November 1980
9. Breckenridge JW, Bird GC. Errors of omission in pulmonary
nodule detection. App! Radio! 1977:6:51-54