Downloaded from www.ajronline.org by 58.42.232.6 on 11/04/15 from IP address 58.42.232.6. Copyright ARRS. For personal use only; all rights reserved
359
Improved Imaging Scan Equalization
J. C. Wandtke1 D. B. Plewes1’2
of Bone with Radiography
Use of scan equalization radiography (SER) for imaging bone in the head, face, neck, and shoulder was evaluated in a clinical comparison study with conventional radiographs of randomly selected patients. Two hundred nine pairs of normal and abnormal SER images and conventional radiographs were compared by four radiologists in a side-byside viewing situation. The radiologists compared the visibility of specific anatomic features and rated the SER images as better than, equal to, or worse than the conventional
radiographs.
much
preferred
detail
and
images
In the the
marked
SER
images
in detail
the radiologists
the conventional
of the images
of the
of the
63% of the time because
improvement
of the shoulder,
preferred
evaluation images
radiograph
preferred
soft
spine,
the
radiologists
of slight improvement
tissues.
In the
the SER image
only 5% of the time
(p
in bone
evaluation
58%
< .05).
of the
of the time and In the
evaluation
the radiologists preferred the SER images 62% of the time and the conventional images 4% of the time (p < .05). The superior image quality with the SER technique was recognized by all radiologists in the study and was the overwhelmingly preferred way of imaging the shoulder, neck, head, and face. AJR
of the skull, face,
of the
cervical
157:359-364,
August
and mandible,
1991
Conventional plain film radiography portions of the body that exhibit large
areas
of underexposure
is often variations
or overexposure
suboptimal when in tissue thickness.
on the radiograph.
used to image This results in
Nowhere
problem more evident than in conventional chest and bone radiography. radiography, the thickest part is optimally exposed but the thinner
overexposed. equalization of radiation
In response
to these
limitations
of conventional
is this
With bone areas are
radiography,
scan
radiography (SEA) was developed. SEA generates a nonuniform field on the patient that is modulated in response to transmission variations
of the particular positioning of the patient during film exposure [i -31. The application of SEA to chest radiography has been studied, and the results
Received August 9, 1 990; sion March 26, 1991.
This work was supported
accepted
after
revi-
by grants from the
National Cancer Institute (CA-27823) and the Eastman Kodak Co. 1 Department of Diagnostic Radiology, University
of Rochester Medical Center, 601 Elmwood Ave., Rochester, NY 14642. Address reprint requests to J.C. Wandtke.
are a significant improvement in image quality [4, 5] and diagnostic accuracy [6, 7]. Similar studies with another equalization radiography unit (AMBER) have shown improvement in chest radiography [8]. The problem of large variations in transmission through different body tissues is a common problem associated with
bone radiography also, especially radiography of the head, neck, shoulder, and spine. Most attempts to overcome these limitations have been studies of chest radiography with the use of specialized X-ray filters [9, i 0]. Other approaches include the use of image processing of digitized radiographs [i i , i 2], photostimulable storage phosphor digital radiographic systems [i 3, 14], and direct digital scanning
detector
North York, Ontario, Canada M4N 3M5.
Subjects
and Methods
0361-803X/91/1572-0359
In this investigation, the application mandible, and shoulder was studied.
2
Present address:
physics,
©
University
American
Department
of Toronto,
Roentgen
2075
Ray Society
of Medical BioBayview
Ave.,
systems
[1 5, 16].
of SEA to imaging This was evaluated
of the cervical spine, head, face, by a subjective assessment of
WANDTKE
360
AND
Downloaded from www.ajronline.org by 58.42.232.6 on 11/04/15 from IP address 58.42.232.6. Copyright ARRS. For personal use only; all rights reserved
image quality by a panel of radiologists who reviewed matched conventional and SEA images. The details of the configuration and the operation of the SEA system used in this study have been previously reported [i -31. This SEA system uses a 2.5- by 4.0-cm beam of radiation scanned in a
PLEWES
ventional
AJR:157,
and SEA radiographs.
Nonetheless,
August
the radiologists
1991
in that
study said that they preferred the side-by-side comparison. A sideby-side comparison permits easier selection of a preferred image when the differences are small.
series of 39 vertical lines. The X-ray output varies up to 2000 times
The radiologists were asked to compare the visibility of all anatomic features and overall image quality and then indicate on a five-point
during
scale whether
a 2-sec scan, and this is achieved
by computer-controlled was designed to be but was adjusted to
than
provide artifact-free
kVp with a maximum
with the patient’s
equalized
of 600 mA. The conventional
images at 80-90 radiographs
were exposed
radiation by using a standard phototimed technique, the total scanning time ranged posteroanterior,
lateral,
or oblique
views
with 75-kVp
system. For the SEA from i .2 to 4.5 sec for
depending
on the thickness screens
of the body part. Kodak T Mat-G film and Lanex regular were used for the conventional
radiograph.
This is the screen-film
system most often used in our department. Because this prototype SEA system is limited by tube output, a faster combination of Kodak T Mat-G film and Lanex fast screens was used for the SEA images.
The source-to-image distance for the SEA images was fixed at 180 cm, but the conventional bone radiographs were obtained at i 22 cm. The conventional radiographs were obtained with a wall-mounted Bucky grid assembly (grid ratio, 8:1). No grid was used for the SEA images because scatter rejection was achieved by a 3.0-cm slit mounted in the aft collimator assembly. All the comparison examinations were done in the upright position
better
the SEA images were much better than (+2), slightly
modulation of pulse width. The SEA system operated at 1 20 kVp for chest radiography,
than (+i), (-2)
the
equal to (0), slightly conventional
images.
worse The
than (-1), radiologists
age and sex but not the patient’s
or much worse were
provided
clinical history.
There were no time constraints for viewing the radiographs. Before the study, the comparative features of the SEA and conventional radiographs were illustrated in a 30-mm training session with four demonstration cases. A brief technical explanation of the equalization technique was also given. The training sessions consisted of individual meetings with verbal instructions and the opportunity for discussion.
Results The 175 patient examinations resulted in 209 pairs of comparison images on which each of four radiologists made 501 observations. A summary of the radiologists’ opinions of the neck, shoulder, and head shown in Table i The radiologists .
radiologic examinations is had a marked preference
because this was a design limitation of the SEA prototype. For the head and neck radiograph obtained with the SEA system, the patient was positioned as close to the film cassette as possible, but, because of the size of the aft collimator assembly, it was not feasible to place the film cassette in direct contact with the exit side of the body part examined; however, only about 5% magnification resulted. During some examinations of the cervical spine and shoulder, there were
for the SEA images. The average improvement in score (on the scale of -2 to +2) of the 2004 SEA images compared with conventional bone radiographs was +0.75. In 690 comparisons, no difference was seen. In i 233 comparisons, the SEA image was better, whereas in 8i comparisons the
slightly different degrees of magnification in positioning of the patients. Positioning
was a significant
SEA system aminations
as with the conventional were
examination. nologist
in a different
system. room
The comparison
5-1 5 mm after
In about half of the cases, a different
positioned
tional radiographs film density
done
owing to slight differences was as convenient with the
the patient
for the SEA
were obtained
and on-line
monitoring
examination.
The
ex-
the initial
radiologic
tech-
conven-
in the usual fashion with the usual of film quality
by a radiologist.
The
SEA system was set by using a phantom to provide an excellent image and was not readjusted during the study. Phantom studies have shown that for SEA head and shoulder examinations, radiation
exposure is less than for conventional elimination
of overexposure
around
studies. This is due to the
the thin edges
of these
anatomic
structures
and the film-screen used for SEA, which is twice as fast. Comparison SEA and conventional radiographs were collected from 175 volunteers (38-76 years old) during a 3-month period after informed consent was obtained. Three or four volunteer patients were recruited on each of 3 days a week. Patients were recruited without bias as to type or severity of bone disease. Most of the patients asked to volunteer did participate. A variety of abnormal
conditions suIted
shoulder
in
were included
as well as normal examinations.
209 pairs of comparison
radiographs
(65), and head (63). The comparison
within 1 5 mm of each other. The images were reviewed
This
re-
of the neck (81),
images were obtained
of the radiologists
Cervical
lateral,
by four
radiologists
about
the relative
quality
image was better. By chi-square (p < .05) improvement.
analysis,
SER
Spine
Cervical
spine
of 47 patients resulted in 47 and i 4 anteropostenior comparison ex-
examinations
20 oblique,
aminations. For each of these groups the radiologists were asked to comment on the clarity of the anatomy at the CiC2 level, at the C3-C7 level, and of the soft tissues. Table i
shows that for cervical spine examinations there was a preference for the SEA images, with 209 (22%) regarded as much better, 395 (41 %) as slightly better, and 333 (34%) as equivalent. Only 35 (3%) of the SER images were classified as inferior to the conventional images. Comparison of an SEA and conventional cervical spine examination is shown in Figure i , where all four radiologists preferred SEA. A markedly improved soft-tissue detail in the upper pharynx and larynx is usually seen. Most of the cases included in the study had normal soft tissue. The bones were frequently abnormal. The improved contrast in the bony structunes and better visibility of C7 was due to the adjustments in exposure
independently
possible
with
SEA.
in
viewing sessions that comprised approximately 50 comparison pairs each. Side-by-side comparison was used because the radiologist could not be blinded since it was obvious which radiograph was the conventional and which was the SEA. A previous side-by-side and independent comparison study [4] showed no significant difference in the opinions
conventional
of con-
Shoulders
For the 65 comparison shoulder examinations, the radiologists were asked to compare the visibility of all anatomic detail of the humerus, scapula, and clavicle. The SER images
AJR:157,
August
SCAN
1991
TABLE
1:
Ratings
of Scan
EQUALIZATION
Equalization
Radiography
BONE
RADIOGRAPHY
(SER)
and
36i
Conventional
Radiography
by
Four Radiologists Rating
Imaged
Region
+2
spine Anteroposterior
+i
0
-2
-i
Downloaded from www.ajronline.org by 58.42.232.6 on 11/04/15 from IP address 58.42.232.6. Copyright ARRS. For personal use only; all rights reserved
Cervical
(n
=
C1-C2 C3-C7
23 7
28 19
5 26
0 4
0 0
19
33
4
0
0
17 0 20
37 20 30
25 51 29
1 9 1
0 0 0
C1-C2 C3-C7
31 25
75 63
79 83
2 16
1 i
Soft tissues
67
90
3i
0
0
209
395
333
33
2
20 23 47
i07 131 1 27
117 95 73
i6 ii 12
0 0 1
90
365
285
39
1
Soft tissues Oblique (n = 20) Ci-C2 C3-C7 Soft tissues Lateral (n = 47)
Subtotal Shoulder (n Humerus
=
65)
Scapula
Clavicle
Subtotal Head
14)
(n = 63) Skull (n = i 1) Face (n = 22) Mandible (n = 30) Subtotal Total
9
19
i4
2
0
24
42
20
2
0
20
60
38
2
0
72
6
0
690
78
3
53 (n
=
209)
352
i2i 881
Note-A rating of +2 means SEA is much better than conventional radiography, +1 means SEA is slightly better than conventional, 0 means SEA is equal to conventional, -1 means SEA is slightly worse than conventional, and -2 means SEA is much worse than conventional.
Fig. 1.-A, of cervical
Conventional spine
vere degenerative B,
in a patient
radiograph with se-
arthritis.
Scan equalization radiograph shows better detail, especially of soft tissue, base of skull, and C7.
WANDTKE
Downloaded from www.ajronline.org by 58.42.232.6 on 11/04/15 from IP address 58.42.232.6. Copyright ARRS. For personal use only; all rights reserved
362
AND
were regarded as being either much better or slightly better than the conventional images 58% of the time. The SER images were equivalent 37% of the time, and 5% were found to be inferior. Examinations of the shoulder showed marked differences between the SEA and conventional radiographs. With the SER technique, the soft tissues were seen uniformly. In many cases, the bony structures were seen optimally without the use of a bright light. An example in which all four radiologists thought the antomic detail was better with SEA is shown in Figure 2. Head
Examinations
1).
of the head resulted
in improved
visibility
of
AJR:i57,
August 1991
anatomic detail when examinations were performed not only of the skull but also of the face and mandible. The uniformity of exposure resulted in improved contrast and visibility, especially around the edges of the calvaria. The underexposure commonly seen in the thicker parts of the face and skull also was reduced. This is seen well in an example of the anteroposterior examination of the face (Fig. 3). Table 2 is a summary of Table 1 , grouped so that it is easier to compare the ratings of visibility in terms of percentages. The preference of the individual radiologists for either the SER or conventional study is listed in Table 3; the categories of much better and slightly better and of much worse and slightly
The 63 radiographic examinations of the head area included 1 1 radiographs of the skull, 22 of the face, and 30 of the mandible. The SEA images were judged to be superior in i 74 (69%) of the images, equivalent in 72 (29%), and inferior in six (2%) (Table
PLEWES
worse
were
grouped.
Statistical
analysis
with the chi-
square method showed a significant difference (p < .05) for all categories and radiologists except one. Table 4 shows a summary of the data in Table 3, listing each radiologist’s preferences for the images of the head, neck, and shoulder. The five categories were grouped into three because not all the radiologists found it easy to differ-
Fig. 2.-A, Conventional radiograph of shoulder in patient with fracture of humeral head shows overexposure of acromioclavicular joint. B, Scan equalization radiograph shows better detail of soft tissue, ac-
romioclavicular humeral head.
joint, and lateral part of
Fig. 3.-A, Conventional radiograph of facein patientwith bilateral maxillary sinusitis. B, Scan equalization radiograph shows more uniform exposure and improved contrast of all bony and softtissue structures.
Downloaded from www.ajronline.org by 58.42.232.6 on 11/04/15 from IP address 58.42.232.6. Copyright ARRS. For personal use only; all rights reserved
AJA:157,
August
SCAN
1991
EQUALIZATION
BONE
entiate between slightly better and much better. Nonetheless, the difference between individual radiologists is still noticeable. The general trend was a preference for the SER images, ranging from 42% (radiologist 1) to 88% (radiologist 4) being judged as superior to the conventional images. Even though radiologists 1 and 4 had a different frequency of preference for the SER images, there was less of a difference in the opinion that the SEA was rarely worse (2-i 1 %) than the conventional square, p
image.
There
was
a significant
preference
7]. Although radiographs
TABLE
Equal Worse Mandible
deemed
so because
of slight
underexposure
TABLE Quality
that
is characteristic
2: Ratings of Visibility of Anatomic on Scan Equalization Radiography
Conventional
(SER)
Ratings Images
for Scan
Radiography
of all SEA
Worse Facial bones Better
Equal
Worse
or
Better
Equal Worse Scapula Better Equal Worse Clavicle/acromion
with
4 7 0
10 i 0
4 6 ia
iO 0 i
15 7
18 4
12
8
1
0
0
2
0
14 15
23 7
16 4
27 2
1
0
0
1
14 51 0
39 26 0
26 25 14
48 15 2
23 41 1
39 26 0
40 18 7
52 10 3
64 1 0
Better Equal
34
47
61
70
47
32
19
10
Worse
0
2
1
1
17 60 4
22 55 4
38 25 18
61 6 4
48 33 0
56 25 0
74 6 1
78 3 0
C3-C7 Equal
Worse
63
34
3
Better
Shoulder Skull
58 64 74 66
37 32 24 32
5 4 2 2
Equal Worse
Soft tissue
Note-Better
=
equal to conventional radiography. a Difference
4
Detail and Image Radiographs
SER better than conventional radiography; equal = SEA radiography; worse = SEA worse than conventional
not statistically
Quality
significant.
by Individual
No. (%) of Images by Radiologis
Radiologists
t
No.
SEA Rating Better
than
conventional
radiography
Equal to conventional radiography Worse than conventional radiography
21
27 ii
Better
TABLE 4: Ratings of Visibility of Anatomic on Scan Equalization (SER) vs Conventional
4
27
Cervical spine
34
3
24 2
Worse
62
2
39
Equal
Average
No.
23 0
Better
Mandible
Radiologists
42
Co nventional Study ( % of Images)
Face
consist-
Equal Worse Ci-C2
Better
[4,
by Region
Region
the
Humerus
Detail and Image (SER) vs
Quality o f SEA Compared
on conventional light,
Equalization
by Individual
Better Equal
differences in the positioning of the patients. Some of those SEA examinations might have been repeated, but the protocol did not permit it. Some of the comparison conventional radiographic examinations were repeated. The frequency of repeat examinations was not recorded. We think that the primary value of the SEA image is the consistency of image quality uniformity
a bright
Radiologist
The capability of SEA to provide a more uniform exposure over the entire radiograph, regardless of variations in body tissue X-ray transmission, results in improved visibility of normal anatomic detail as judged by this subjective analysis by four radiologists. Despite the faster film-screen combination and higher-kilovoltage X-ray spectrum used for the SEA images, the SEA image quality was regarded as equivalent (34%) or superior (62%), with only a very few cases (4%) regarded as inferior. Most of the 4% of SER cases that were
and exposure
with
significance. Skull Better
were
that is overexposed
be visualized
3: Preference
Radiographic
Discussion
inferior
anatomy can
(chi