Key words breast - mammography - technical aspects - breast radiography
Introduction
Adequate compression of the breast is an absolute prerequisite for a good mammogram.
However, the procedure for breast compression has not been standardized with respect
to compression force. The European guidelines do not provide any indication regarding
the required compression force [1 ]. The guidelines of the German Medical Association require compression of at least
10 Kp, but do not offer advice on how to respond to complaints about the resulting
pain [2 ].
Compression-related reduction of the irradiated tissue allows a decrease of radiation
dose and thus a diminution of scattered radiation with an exponential relationship
between breast thickness and average glandular dose (AGD) [3 ]. In addition, geometric blur is diminished since compression reduces the distance
to the detector plate of the remote gland portions [4 ]. In addition to avoiding motion blur, adequate compression can also effect a reduction
of superimposed tissue structures and thus improve the diagnostic distinction between
tumors and artifacts [1 ]. Improper compression of the breast can cause pain in the woman, making acceptance
of mammography more difficult [5 ]. Between 25 and 46 % of women participating in initial mammographic screening and
not participating again, cited pain as the primary reason [6 ]. Previous surgery and radiation can amplify mammography-related pain [7 ]
[8 ]. Numerous studies mention the influence of psychological factors influencing pain
during mammography [9 ]
[10 ]
[11 ]
[12 ]
[13 ]
[14 ]
[15 ]
[16 ]
[17 ]
[18 ]
[19 ]. Recent investigations emphasize the significance of intramammary pressure in the
guidance of breast compression during mammography [20 ]
[21 ]
[22 ]. The authors recommend adapting compression force to the size of the breast, and
thus adjusting intramammary pressure.
The aim of the study was to investigate the relationships among compression pressure,
the surface area of the compressed breast, breast density according to the classification
of the American College of Radiology (ACR) [23 ], prior surgery as well as the pain indicated by the examined women. In addition,
the influence of the compression force on the thickness of the compressed breast and
average glandular dose (AGD) was analyzed.
Materials and Methods
Relationship between breast thickness while compressed and compression force
To determine the relationship between compression force and breast thickness under
compression, a pilot study was conducted with 30 women. A digital display of these
parameters by the mammography unit was recorded using a video camera and subsequently
assessed in slow motion. The average values of breast thickness as a percentage of
initial uncompressed thickness was graphically displayed in relation to the respective
compressive force for the four projections.
Relationship among pain indications during mammography and compression force, the
surface area of the breast, breast density according to ACR [23 ] as well as previous surgery.
Patients
765 mammographic images (Mammomat, Siemens Healthcare GmbH, Erlangen, Germany) were
obtained in craniocaudal and mediolateral-oblique projections of 199 symptomatic patients
(average age 58.2 years, standard deviation 13.7 years, maximum 90 years, minimum
30 years). The study did not include asymptomatic women having early detection examinations
(population-related mammographic screening). The patients were accepted into the study
without exclusionary criteria in the order of their appearance for the examination.
In their medical history, 52 of the 199 patients indicated breast surgery (17 patients
with a biopsy, 25 with a lumpectomy (24 of whom had radiation), and 10 with ablation).
Technical procedure for breast compression
The sequence of positions was the same for all patients. For bilateral examination:
1. right craniocaudal (RCC); 2. left craniocaudal (LCC); 3. right mediolateral-oblique
(RMLO); and 4. left mediolateral-oblique (LMLO). For unilateral examination: craniocaudal
before mediolateral-oblique. Compression force greater than 10 daN was attempted,
depending on the patient’s individual pain tolerance. The “OpComp” function (device-controlled
automatic optimization of compression force) was not used when determining force [24 ]. The 18 × 24 cm table was routinely used. The 24 × 30 cm table was used in the case
of very large breasts. The mammographic settings were performed by three trained and
very experienced technicians. Compression force (in kilopond, kp) and breast thickness
(in cm) under compression were taken from the display of mammography unit. The compression
force values in kp displayed by the mammography unit were calibrated with an electronic
scale (linear of regression: compression force(corrected) = 1.01156 × compression force(unit display in kp) – 0.38140; correlation coefficient = 0.99997) and subsequently converted to decanewton
(daN) (1 daN = 1.0197 kp).
Quantification of pain indication
The Numeric Rating Scale (NRS) criteria were used to quantify the pain level [25 ]. This scale allows standardized assessment of pain perception. After each mammographic
image was acquired, the patient was asked to describe her pain using a scale of 0 – 10
(0 = no pain; 10 = unbearable pain).
Planimetry and ACR classification of the mammographic images
Plane measurements (in cm2 ) were made of each of the four projections with respect to the surface areas of the
compressed breasts using a polygon function of the viewing software (Osirix PRO, aycan
Digitalsysteme GmbH, Würzburg, Germany). In addition, tissue thickness was evaluated
visually based on the mammographic images in accordance with the classification of
the American College of Radiology (ACR) [23 ].
Radiation dose
The automated system of the mammography unit, using the device-specific dosage optimization
program (“Opdose”), selected the exposure program as a function of breast thickness
under compression [24 ]. The average glandular dose (in milligray) was taken from the visual display of
the mammography unit.
Statistics
The study design was concomitant prospective. The values of the categories pain, compression
force and surface area for the 765 mammograms were broken down into three classes
with the greatest similar number of observations ([Table 1 ]). The variables for pain sensation and compression force are discrete. This explains
the greater variation of number of examinations classified into the three respective
categories (“low”, “medium” and “greater” for pain, and “low”, “medium” and “great”
for compression force). Statistical evaluation was performed descriptively using contingency
tables and stacked columns (to 100 %) (Excel, Microsoft Cooperation, Redmond, WA,
USA), as well as using a statistical procedure for testing for the independence of
two attributes (chi square test of unrelated samples and the Dixon and Mood staircase
method for related samples). The ratio of radiation dose and breast thickness under
compression was set as a scatter plot for 752 mammograms (data regarding dose or breast
thickness was not documented for 13 of the 765 mammograms) and descriptively represented
as a 4th order polynomial; the correlation coefficient was determined using Excel
(Microsoft Cooperation, Redmond, WA, USA).
Table 1
Definition of the particular classes and distribution of the number of mammograms
in relation to the different categories (projections, pain, compression force, surface
area of the breast, density of the breast (ACR) and previous surgery).
patients 199
mammograms 765
projections
RCC
LCC
RMLO
LMLO
number of mammograms (total = 765)
188
192
191
194
pain (x in 0 – 10 according to the numeric rating scale)
low
medium
strong
x < = 3
3 > x < = 5
x > 5
number of mammograms (total = 765)
265
245
255
compression force (F in daN)
low
medium
great
F < 9
9 > = F < 11
F > = 11
number of mammograms (total = 765)
256
243
266
surface area (A in cm2 )
small
medium
large
A< 143
143 > = A< 206
A> = 206
number of mammograms (total = 765)
255
255
255
ACR classification
ACR 1
ACR 2
ACR 3
ACR 4
number of mammograms (total = 765)
137
168
372
88
previous surgery (with/without radiation)
PE
lumpectomy
ablation
number of patients (total = 52)
17
25 (24 with radiation)
10
Topographical distribution of compression-related pain
After each mammogram, the last 52 of the 199 patients were asked to indicate the site
on their body where pain during the mammogram was the greatest. Four regions on each
side were differentiated: breast, upper thoracic wall, lower thoracic wall and axilla.
Results
Breast thickness while compressed and radiation dose
The ratio of AGD to breast thickness under compression is shown in [Fig. 1 ] for 752 of the 765 mammograms as a scatter plot and a 4th order polynomial as a
trend line (y = 0.0029x4 – 0.0486x3 + 0.301x2 – 0.6659x + 1.1137, n = 752). The trend line points to an exponential relationship
between the average glandular dose displayed in the mammogram and breast thickness
while compressed. The correlation coefficient was 0.41 (p < 0.001).
Fig. 1 Average glandular dose (AGD in mGy) in relation to the thickness of the compressed
breast (cm) as a scatter-plot with a 4th degree polynomial as a trendline.
Breast thickness while compressed and compression force
[Fig. 2 ] shows the percentage thickness of the breast under compression relative to the initial
uncompressed value for 30 patients and four projections. The course of compression
was broken down into three phases. After an initial phase with a steeper progression
to a compression force of 4 daN to 78.4 % of the initial value, the curve becomes
shallower, to flatten again after 10 daN. Likewise, using compression with more than
10 daN compression force, additional reduction of breast thickness was possible. This
applied particularly to the first RCC projection. Thus with a compression of 10 daN,
the breast could be reduced to an average thickness of 65.2 % of the initial value
for all four projections. If compression was increased to 15 daN, breast thickness
could be further reduced to 57.8 % of its initial value.
Fig. 2 Thickness of the breast under compression in percentage of the thickness of the uncompressed
breast relating to compression force in 30 patients.
The average baseline value of the thickness of the non-compressed breast for these
30 patients and four projections was 8.2 cm. Accordingly, at a compression force of
10 daN, the average breast thickness was reduced to 5.4 cm (65.2 % of the initial
thickness of 8.2 cm). Compression force of 15 daN resulted in a reduction to 4.7 cm
(57.8 % of the initial thickness of 8.2 cm). Comparing these breast thickness values
after compression with the breast thickness value-dependent average glandular dose
(AGD) values of the 752 mammograms documented in our study, we obtained average dose
values per mammogram of 1.2 mGy for 5.4 cm breast thickness and 1.0 mGy for a breast
thickness of 4.7 cm. An increase of compression force from 10 to 15 daN resulted in
an average dose reduction of 17 % (0.2 mGy from 1.2 mGy).
Pain indications, compression force, surface area of the compressed breast, breast
thickness (ACR), projection and prior surgery
Compression force and breast surface area
The compression force tolerated by the patients correlated positively with the surface
area of the compressed breast ([Fig. 3 ]). In the course of the individual mammography, women with a small breast surface
had a decreasing acceptance of great compression force (p < 0.001, [Table 2 ]). This applied particularly to the LMLO projection which was performed last. On
the other hand, women with a large breast surface area more frequently tolerated great
compression force.
Fig. 3 Relationship between breast-surface area and compression-force. The numbers represent
mammograms, where 100 % corresponds to the sum of mammograms with small, medium or
great surface area respectively in each of the four projections.
Table 2
Overview of the results of statistical tests regarding the relationship of the parameters
(projections, pain, force of compression, surface area of the breast, breast density
(ACR) and previous surgery).
projections versus pain
n.s. (not significant)
projections versus force
p < 0.001
craniocaudal projection (CC)
mediolateral-oblique projection (MLO)
surgery versus pain
n.s.
n.s.
RCC
LCC
RMLO
LMLO
force versus pain
p < 0.001
p < 0.001
p < 0.001
p < 0.001
surface area versus force
p < 0.001
p < 0.01
p < 0.001
p < 0.001
pain versus surface area
n.s.
n.s.
n.s.
n.s.
ACR versus pain
n.s.
n.s.
n.s.
n.s.
ACR versus force
p < 0.05
p < 0.05
n.s.
n.s.
Pain and compression force
The patients indicated greater pain in more than half of mammograms with low compression
force ([Fig. 4 ]). The results were highly significant (p < 0.001, [Table 2 ]). There was no recognizable positive correlation between tolerated compression force
and indicated pain.
Fig. 4 Relationship between pain and compression-force. The numbers represent mammograms,
where 100 % corresponds to the sum of mammograms with lower, medium or greater compression-force
respectively in each of the four projections.
Pain and previous surgery
In their medical history, 52 of the 199 patients indicated breast surgery (24 patients
with radiation). In 39 of these 52 patients (18 with radiation), compression-related
pain on the operated side could be compared to the non-operated side (patients with
bilateral lumpectomy and those with ablation were therefore not included in the analysis).
Six patients (craniocaudal projections) and six patients (oblique projections) indicated
a stronger experience of pain during compression of the operated side compared to
the non-operated side ([Table 3 ]). A lower sensation of pain on the operated side compared to the non-operated side
was reported for one craniocaudal projection and for 3 oblique projections. In the
majority of patients (32 craniocaudal and 30 oblique projection images), no change
in pain perception resulting from prior surgery with or without radiation could be
observed. Statistical significance with respect to the influence of previous surgery
on pain indication was not evident (p > 0.05, Dixon and Mood staircase method for
related samples, [Table 2 ]).
Table 3
Comparison of compression pain: operated breast vs. contralateral side without operation
in 39 patients (p > 0.05).
projection
craniocaudal
mediolateral-oblique
compression-related pain same on both sides
32
30
compression-related pain greater in operated breast
6
6
compression-related pain less in operated breast
1
3
Pain and projection, breast surface area and ACR classification
Pain perception was unrelated to the projection (RCC, LCC, RMLO and LMLO) as well
as breast area ([Table 2 ]). In addition, radiologically-measured breast density following classification of
the American College of Radiology (ACR) [23 ] did not affect pain sensation. Patients with radiopaque glandular tissue (ACR 4)
more frequently tolerated only lower force in craniocaudal projections (RCC, LCC)
(p > 0.05, [Table 2 ]).
Compression force and projection
In the course of each individual mammographic examination, increased compression force
was increasingly less tolerated ([Fig. 5 ]). Although the relative proportion of mammograms in which the patients tolerated
greater force in the initial RCC projection was 47.3 %, the relative proportion declined
to 23.7 % in the final LMLO projection (p < 0.001, [Table 2 ]).
Fig. 5 Relationship between the compression-force and the four projections (RCC, LCC, RMLO
und LMLO). The numbers represent mammograms, where 100 % corresponds to the sum of
mammograms in each of the four projections.
Topographical distribution of compression-related pain
Less than half of the women indicated their breast as the location where the pain
was greatest during the mammogram ([Table 4 ]). During oblique projections, 60 % of the women named the axilla as the site with
the greatest pain. During craniocaudal projections, the upper thoracic wall area was
named as the maximum pain point by more than 40 % of patients.
Table 4
Topographic distribution of point of maximum pain during mammographic compression
(n = number of patients).
projections
breast
axilla
thoracic wall, upper
thoracic wall, lower
row sum
n = number of patients
RCC
n
18
0
22
7
47
%
38.3
0.0
46.8
14.9
100.0
LCC
n
20
0
21
8
49
%
40.8
0.0
42.9
16.3
100.0
RMLO
n
7
30
12
1
50
%
14.0
60.0
24.0
2.0
100.0
LMLO
n
7
31
13
1
52
%
13.5
59.6
25.0
1.9
100.0
Discussion
Breast compression during mammography and average glandular dose
Adequate compression of the breast during mammography reduces the radiation dose with
an exponential relation between dose and breast thickness [3 ]. In the examinations under study, the average glandular dose for a 6 cm breast thickness
with 1.1 mGy was only 55 % of the average dose for a breast thickness of 8 cm with
2 mGy. These results underscore the significance of compression-related breast thickness
reduction with respect to radiation protection.
In our study the effectiveness of breast thickness reduction decreased with increasing
compressive force. Forced compression using 15 daN resulted in an average reduction
of breast thickness of 57.8 % of the original thickness of the uncompressed breast,
thus allowing an average dose reduction of 17 % compared to that achieved using 10
daN. De Groot et al. [26 ] describe similar curve progressions for the mammographic compression process. They
divided breast compression into a “deformation” and a “clamping” phase. Concurring
with our results, the authors describe only minimal reduction of breast thickness
in the clamping phase and recommend shortening this phase in the interest of pain
reduction.
Breast compression, pain sensation and compression force
Contrary to expectations, greater pain was more frequently reported when low compression
force was used. It should therefore be presumed that the pain indicated by these women
was less the result of the physical extent of compression but rather was influenced
by their individual sensitivity to pain [21 ]. Women with heightened pain sensitivity consequently terminated the compression
procedure earlier. This concurs with studies investigating women’s psychological experience
of pain during mammography [9 ]
[10 ]
[11 ]
[12 ]
[13 ]
[14 ]
[15 ]
[16 ]
[17 ]
[18 ]
[19 ]. Pain is thus less suitable as a parameter for inter-individual optimization of
breast compression during mammography since individual factors independent of the
breast have significant influence on feeling pain. This likewise explains why in our
study, the projection, surface area of the breast as well as relation of glandular
and fat tissue according to the classification of the American College of Radiology
(ACR) [23 ] exhibited no significant influence on experienced pain. Markle et al. could also
demonstrate no relationship between breast tissue composition and compression-related
pain [27 ]. On the other hand, Kornguth et al. have described a corresponding correlation [28 ].
Our larger-breasted patients tolerated greater compression force. This applied particularly
to both oblique projections. If, therefore, the same compressive force were used as
the criterion for optimal compression of all breasts, large breasts would tend to
be insufficiently compressed, whereas smaller breasts would be subjected to excessive
compression. Using the same compression force, higher intramammary pressure is produced
in a smaller breast compared to a large breast as a function of the compressed breast
surface area [21 ]. Our results suggest that during breast compression, intramammary pressure as a
quotient of compressive force and breast surface area is a better measure of compression
tolerance than the patients’ pain indication. De Groot et al. likewise refer to the
significance of intramammary pressure as a parameter for applying individualized compression
independent of breast size. Consequently they developed a device for mammography units
to continuously display intramammary pressure during compression [20 ]
[21 ]. This allowed standardization of the compression procedure as well as a reduction
of compression-related pain.
In our investigations, escalating compression force was increasingly less tolerated
during the course of each individual mammographic examination. The proportion of mammograms
during which patients tolerated greater force declined by half from 47.3 % in the
first projection (RCC) to 23.7 % in the final LMLO projection. Therefore, not only
past painful mammograms, but also a position causing pain in the course of the current
examination can adversely affect the examination procedure. Therefore mammography
should not begin with that breast which due to prior surgery, radiation or unilateral
mammalgia is particularly sensitive. Several authors discuss the positive effect of
psychological guidance during the examination, with explanations of the course of
the examination as well as closer observation of the patient’s sensations by the examiner
[1 ]
[11 ]
[12 ]
[13 ]
[15 ]
[17 ].
Some women with lumpectomies and radiation complained of greater pain during compression
of the operated side as compared to the non-operated side. This was also observed
by de Groot et al. [7 ]. However, the majority of our patients did not report any difference with respect
to pain.
Breast compression during mammography and topographical distribution of pain
For more than half of the women we queried, the breast was not the site that was the
most painful during compression. This particularly related to oblique projections
during which 60 % of patients experienced the greatest pain in the axillary region.
Consequently, pain directly in the breast is not solely responsible for discomfort
during the mammogram. This should be taken into account during the performance of
the mammogram as well as by the manufacturers of mammography units when designing
these devices. Several authors report a reduction of compression-related pain as a
result of technical modifications to the compression plate [20 ]
[29 ]
[30 ]
[31 ]
[32 ].
Limitations of the study
Our investigations were based on mammograms of symptomatic patients. The results therefor
have limited applicability to early detection examinations of asymptomatic women (screening
mammography). A further limitation of the study is the absence of a specified minimum
value for compression force to guide the examiners when evaluating patients’ pain
indications. In some cases, the desired compression force of at least 10 daN could
not be realized due to patient pain. This can result in inter-individual differences
in the examiners’ procedure. In contrast to mammographic screening of asymptomatic
women, mammography of symptomatic patients requires closer attention to the patient’s
individual situation, taking into account pre-existing conditions and previous breast
surgery.
Direct measurement of intramammary pressure during compression would not be possible
without technical modifications of the mammography equipment and consequent loss of
operating authorization. We had to limit ourselves to detection of compression force
and breast surface which allowed only an indirect statement regarding intramammary
pressure. A further limitation was the varying numbers of cases in the analyses, which
possibly influenced statistical evaluation. Therefore further studies of individualized
pressure-related compression during mammography are required. Such studies would employ
pressure-sensitive compression plates [20 ] and include mammographic screening of asymptomatic women.
Forced compression using 15 daN, compared to application of 10daN, resulted in an
additional average reduction of average glandular dose of 17 %.
Pain during a mammogram is not exclusively due to the physical extent of compression,
but also related to individual differences in sensitivity to pain.
Compression force should be made dependent on breast size since women with larger
breasts frequently tolerate greater compression force.
During breast compression, intramammary pressure as a quotient of compressive force
and breast surface area is a better measure of compression tolerance than the patients’
pain indication.
Mammography should not begin with that breast which is particularly sensitive, since
a painfully experienced position can have a negative influence on the further course
of the examination.
The presence of mammography-associated pain outside of the breasts should be taken
into account both in positioning the images as well as in the design of mammography
units.