Distribution of pressure on the breast in mammography using flexible and rigid compression plates: implications on patient handling

Abstract

Background

Breast compression in mammography is important but is a source of discomfort and has been linked to screening non-attendance. Reducing compression has little effect on breast thickness, and likely little effect on image quality, due to force being absorbed in the stiff juxta thoracic area instead of in the central breast.

Purpose

To investigate whether a flexible compression plate can redistribute force to the central breast and whether this affects perceived pain.

Material and Methods

Twenty-eight women recalled from mammography screening were compressed with flexible and rigid plates while retaining force and positioning, 15 in the craniocaudal (CC) view and 13 in the mediolateral oblique (MLO) view. Pressure distribution was continuously measured using pressure sensors.

Results

The flexible plate showed greater mean breast pressure in both views: 2.8 versus 2.3 kPa for CC (confidence interval [CI] = 0.2–0.8) and 1.0 versus 0.5 kPa for MLO (CI = 0.2–0.6). The percentage of applied force distributed to the breast was significantly higher with the flexible plate, both on CC (36% vs. 22%, CI = 1–11) and MLO (30% vs. 14%, CI = 4–13).

Conclusion

The flexible plate redistributes pressure to the central breast, achieving a better compression, particularly in the MLO view, though much applied force is still applied to the juxta thoracic region.

Keywords

Mammography breast breast compression pain

Introduction

Mammography screening facilitates early detection of breast cancer (1 –3). In any screening, high attendance maximizes the public health benefits. Breast compression is intended to immobilize the breast, reduce scattered and absorbed radiation, and separate overlapping tissue, but is a primary reason for non-attendance due to pain or fear of pain (4 –8). Any discomfort should be accompanied by an improvement of outcomes, i.e. pain should be justified by improved image quality.

In Europe, recommended practice is to compress the breast no more than can be tolerated, with the force not to exceed a certain limit (9). Evidence exists that, in practice, a standard force level is used for most breasts with little regard for size and compressibility, typically 100–130 N, but sometimes higher (10 –13). An approach based on pressure, rather than force, could standardize the practice (11). This requires adapting compression force to breast size so that a desired average pressure can be achieved.

A 50% reduction of compression force changes average thickness by less than 5 mm (14,15). Monte Carlo modelling by Saunders and Samei suggests that for changes in thickness of less than 12%—about 6 mm for a 55-mm average breast—extra scattered radiation will have no discernible effect on image quality (16). Both relatively softer fatty and relatively stiffer fibroglandular breast tissue become proportionally stiffer under compression, meaning that the ratio between applied force and thickness decrease drops as force is increased. Breasts containing more fibroglandular tissue will be less flattened by the same compression force. A substantial portion of the applied force is absorbed in the stiff and thick juxta thoracic tissue, including the pectoral muscle, especially in the mediolateral oblique (MLO) view (12,17). Because of their stiffness, even high pressure in these areas causes little overall decrease in thickness. Consequently, only a small fraction of the applied force is distributed to the clinically important central breast. It has been shown that the distribution of pressure on the central breast varies, but that neither the distribution nor magnitude is substantially affected by the applied force (12).

Flexible compression plates that tilt in relation to the compressed breast can be used to increase patient comfort. Theoretically, a flexible compression plate should redistribute force from the retro-glandular area, proportionally increasing pressure on and compression of the central breast, though it has been suggested that it is detrimental to retro-glandular image quality (18).

The aim of the present study was to investigate the difference in pressure distribution on breasts compressed with rigid or flexible compression plates in the craniocaudal (CC) and MLO views, and to quantify the effect on experienced pain. The main hypothesis was that the flexible plate will redistribute force from the juxta thoracic area to the central breast.

Material and Methods

Study population

A total of 28 women participated in this prospective study, approved by the regional ethical review board at Lund University, Sweden (addendum to dnr 2013/620). Participants were recruited at Skåne University Hospital, Lund, Sweden among women who had been recalled from breast screening between 17 March and 15 April 2016. Women were asked to sign informed consent upon arrival. Those who required an interpreter (or otherwise were not capable of understanding the written and oral information) or had breast implants were not included. The mean age was 56.4 ± 11.3 years (age range = 40–70 years). Images were acquired on a Senographe Essential (GE Healthcare, Buc, France) mammography unit with two different designs of compression plates, rigid (5173493-2 Compression Paddle) or flexible (Flex 5172164).

Sensors

We employed a BPMS 5350 pressure sensor (Tekscan Inc., South Boston, MA, USA) (Fig. 1), a flexible matrix of 38 × 41 sensor elements, each a square of 10.16 × 10.16 mm. It is designed for mapping pressure on seating, with a recommended pressure range of 0–41 kPa. The sensor was calibrated using a manufacturer-supplied vacuum equilibration and calibration system (VB5A) allowing pressures up to 24 kPa, and continuously conditioned before use. The system allows a non-linear calibration with up to 10 pressure calibration points, which were clustered in the lower range to optimize low-pressure accuracy. Pressure resolution is equal to 1/256th of the full digital output or 0.15 kPa, with an extrapolated saturation pressure of 39.3 kPa.

Fig. 1.

The Tekscan CONFORMAT 5350 pressure sensor used in the study. The sensor has 1558 sensor elements, arranged in 38 rows of 41 elements each. The spatial resolution is 10.16 mm. The sensor is very flexible and though larger than the breast support could still easily be placed on it (folding as needed) to cover the mammography unit’s field of view.

Data collection

The sensor was placed with one corner aligned with the support table corner. It was attached to the support table by rubbing the surface with disinfectant alcohol before application, creating a seal by suction. The entire field of view was covered with sensor elements, except for a small strip along the edges of detector, amounting to a couple of millimeters close to the chest wall (on MLO) and the axillary area.

Pressure was recorded from immediately before compression until the breast was released after mammogram acquisition. A representative measurement was selected after a stable compression force had been achieved, defined as the midpoint of the plateau (Fig. 2). Values are thus representative of the “clamping” phase (19). Fifteen women were positioned and imaged in the CC view and 13 in the MLO view. Each woman was compressed with both plates in sequence, with care being taken to maintain the same positioning and to use the same force. To avoid bias, half of the women were first compressed with the rigid plate (even-numbered patients) and the other half with the flexible plate (odd-numbered patients).

Fig. 2.

Sample chart of the evolution of mean pressure (normalized) during breast compression from initial application until release. Data taken from case 5 using the rigid compression plate. This is representative of the cases included in the study, exhibiting four distinct regions: 1 = initial positioning and application of pressure, characterized by low, variable pressure; 2 = full compression, achieved immediately before the end of positioning and application of force, usually characterized by a (or sometimes a series of) pressure peak; 3 = stable compression, which lasts until imaging is finished, is usually around 80%–90% of the mean pressure at maximum compression; and 4 = a rapid decrease of pressure when the breast is released. A measurement in the stable area was selected for each case and used for further analysis.

Minimal-dose mammograms (5 mAs) were acquired to match pressure readings to breast anatomy, one image each for the rigid and flexible plate. Breast contours were extracted from these images through segmentation and all pressure readings fully within the contours were defined as being on the breast. Other clinical mammograms were acquired as needed but were not part of the study. Thickness (mm) and force readouts (decaNewton [daN]) from the mammography device were calibrated by a service engineer immediately before the start of the study.

The participants rated their experience of pain during the two compressions on two 100-mm visual analog scales (VAS) subsequent to the end of the examination (20). They were not informed of the difference between the plates.

Data analysis

Pressure readings were matched with breast anatomy using the low-dose mammograms. The breast area was determined from the mammogram as the projected area. The measured force was determined by adding the force on each sensor element located on the breast. This was compared with compression force to find the portion of that force applied to the breast and thus not to the juxta thoracic region outside of the sensors active area. Mean pressure was defined as the force on the breast divided by breast area as measured on the mammogram. The ratio between mean breast pressure during stable compression and highest achieved mean breast pressure was determined in order to quantify changes after the end of force application.

Wilcoxon’s signed-rank test was used to evaluate differences between groups, and descriptive statistics report median values with interquartile range and range of values, unless otherwise noted.

Results

Results are presented in Tables 1 and 2 for the rigid and flexible plates, respectively. The pressure readings with the rigid paddle for patients 9 and 20 were excluded due to technical issues with the sensor. The corresponding measurements with the other paddle were not affected and are included in the mean pressure values for the flexible paddle measurements, though they are excluded for hypothesis testing. All other values for the patients (thickness, force, area) are included.

Table 1.

Summary of data for all included women using the rigid plate.

Case no.	Projection	Mean pressure (kPa)	Breast area (cm²)	Thickness (mm)	VAS pain (mm)	Applied force (daN)	Force on breast (N)	Stable to peak mean pressure ratio
1	CC	2.0	60	48	27	10	12	0.96
2	CC	2.4	111	65	7	10	27	0.97
3	CC	2.9	112	50	52	9	32	0.93
4	CC	2.0	108	55	11	10	22	0.95
5	CC	1.9	167	59	11	10	31	0.89
6	CC	2.3	121	61	64	10	28	0.66
7	CC	1.3	127	76	11	9	16	0.98
8	CC	2.7	83	56	64	10	23	0.89
9*	CC	N/A	71	40	8	11	N/A	N/A
10	CC	3.5	107	50	22	11	37	0.96
11	CC	1.6	166	45	92	11	27	0.94
12	CC	5.0	68	58	1	10	34	0.89
13	CC	2.5	86	61	11	9	22	0.86
14	CC	0.8	168	58	10	9	13	0.49
15	CC	4.3	130	44	29	9	56	0.91
16	MLO	0.8	407	109	24	14	33	0.94
17	MLO	0.4	185	74	49	6	7	0.27
18	MLO	1.0	343	65	8	8	34	0.73
19	MLO	0.4	160	68	35	10	6	0.38
20*	MLO	N/A	241	73	89	11	N/A	N/A
21	MLO	1.0	240	60	28	10	25	0.89
22	MLO	0.5	240	74	7	9	13	0.46
23	MLO	0.5	263	77	14	10	14	0.40
24	MLO	0.2	136	48	46	8	3	0.20
25	MLO	1.0	165	63	21	10	17	0.96
26	MLO	0.4	85	45	68	9	3	0.32
27	MLO	0.4	344	77	49	11	15	0.46
28	MLO	1.5	211	62	15	9	31	0.79

*Case 9 and 20 excluded due to technical errors resulting in missing data (using the rigid plate).

CC, craniocaudal; MLO, mediolateral oblique; N/A, not applicable; VAS, visual analog scale.

Table 2.

Summary of data for all included women using the flexible plate.

Case no.	Projection	Mean pressure (kPa)	Breast area (cm²)	Thickness (mm)	VAS pain (mm)	Applied force (daN)	Force on breast (N)	Stable to peak mean pressure ratio
1	CC	2.7	60	37	10	10	16	0.89
2	CC	2.3	112	52	7	10	26	0.97
3	CC	4.1	113	42	49	10	47	0.98
4	CC	2.5	109	46	8	10	27	0.95
5	CC	2.4	159	44	7	10	37	0.93
6	CC	3.5	117	49	51	9	41	0.78
7	CC	2.5	131	60	11	9	33	0.72
8	CC	3.0	82	45	80	10	25	0.79
9*	CC	7.0	74	30	58	11	52	0.94
10	CC	3.5	107	42	45	11	38	0.76
11	CC	2.8	166	32	4	10	46	0.90
12	CC	5.4	70	47	1	10	38	0.87
13	CC	2.4	81	45	0	10	19	0.96
14	CC	1.0	167	50	18	10	17	0.60
15	CC	4.4	128	43	48	9	56	0.87
16	MLO	1.0	402	95	45	13	42	0.84
17	MLO	0.5	199	60	23	8	10	0.49
18	MLO	1.7	333	62	8	10	55	0.84
19	MLO	0.5	159	59	32	8	8	0.50
20*	MLO	1.8	241	63	49	12	44	0.90
21	MLO	1.6	239	51	39	9	39	0.82
22	MLO	1.0	242	61	6	10	24	0.76
23	MLO	0.6	264	64	11	10	16	0.72
24	MLO	0.6	139	39	39	9	9	0.86
25	MLO	1.7	164	55	29	9	29	0.98
26	MLO	1.1	86	32	68	9	10	0.95
27	MLO	0.9	348	68	51	10	30	0.98
28	MLO	1.9	210	50	17	10	39	0.94

*Case 9 and 20 excluded due to technical errors resulting in missing data (using the rigid plate).

CC, craniocaudal; MLO, mediolateral oblique; N/A, not applicable; VAS, visual analog scale.

The median breast thickness was 60.5 mm (interquartile range [IQR] = 20.5, range = 40–109 mm) with the rigid plate and 49.5 mm (IQR = 17.5, range = 30–95 mm) with the flexible plate (P < 0.0001, confidence interval [CI] = 9.5–12). Applied force was equivalent (P = 0.56, CI = –1 to 0.5); 10 daN (IQR = 1, range = 6–14 daN) for the rigid plate, 100 N (IQR = 1, range = 8–13 N) for the flexible plate. Figs. 3 –7 show sample pressure distributions.

Fig. 3.

Case 11. Pressure distribution (left = rigid, 11 daN, right = flexible, 10 daN). This case illustrates the generally increased pressure over the breast using a flexible plate. Pressure scale: low = dark red, high = bright yellow, 0.0–8.0 kPa.

Fig. 4.

Case 5. Pressure distribution (left = rigid, 10 daN, right = flexible, 10 daN). This case illustrates the smoothing effect of the pressure over the breast using a flexible plate. Note the higher-pressure column close to the chest wall on the rigid pressure image to the left. Pressure scale: low = dark red, high = bright yellow, 0.0–8.3 kPa.

Fig. 5.

Case 18. Pressure distribution (left = rigid, 8 daN, right = flexible, 10 daN). This case illustrates the generally increased pressure over the breast using a flexible plate. Pressure scale: low = dark red, high = bright yellow, 0.0–11.2 kPa.

Fig. 6.

Case 21. Pressure distribution (left = rigid, 10 daN, right = flexible, 9 daN). This case illustrates how the pressure is removed from the axillary area to the centrally breast using a flexible plate. Note the higher-pressure area in the axillary on the rigid pressure image to the left. Pressure scale: low = dark red, high = bright yellow, 0.0–11.9 kPa.

Fig. 7.

Case 23. Pressure distribution (left = rigid, 10 daN, right = flexible, 10 daN). This case illustrates, to a lesser degree, how the pressure is removed from the pectoral area to the central breast using a flexible plate. However, we believe that the compression is suboptimal for both cases. Pressure scale: low = dark red, high = bright yellow, 0.0–4.5 kPa.

On the CC view, the median mean pressure on the breast with the rigid plate was 2.3 kPa (IQR = 1.0, range = 0.8–5.0 kPa), which was significantly lower (P = 0.001, CI = 0.2–0.8) than 2.8 kPa (IQR = 1.6, range = 1.0–7.0) measured on the flexible plate. The difference was likewise significant on the MLO view (P = 0.0004, CI = 0.2–0.6), with the rigid plate showing 0.5 kPa (IQR = 0.6, range = 0.2–1.5 kPa) and the flexible plate showing 1.0 kPa (IQR = 1.1, range = 0.5–1.9 kPa).

Concerning breast area, there was a negligible difference between the two plates (P = 0.47, CI = –1 to 2)—rigid = 148 cm² (IQR = 118, range = 60–407 cm²) and flexible = 149 cm² (IQR = 117, range = 60–402 cm²)—but a non-paired t-test showed a significantly greater area on MLO view than CC view (P < 0.0001, CI = 70–172) with a mean of 232 ± 88 cm² (range = 85–407 cm²) and 112 ± 34 cm² (range = 60–168 cm²), respectively.

The ratio of force applied to force measured on the breast was significantly lower with the rigid plate, both on CC view (25%, IQR = 12%, range = 12%–62% vs. 36%, IQR = 21%, range = 16%–62%; P = 0.007, CI = 1–11) and MLO (14%, IQR = 15%, range = 3%–43% vs. 30%, IQR = 25%, range = 10%–55%; P < 0.0004, CI = 4–13).

Mean breast pressure during stable compression was lower than peak pressure, but though the difference between the rigid and flexible plate was non-significant on CC (rigid = 92% [IQR = 7%, range = 49%–98%]) and flexible = 89% [IQR = 17%, range = 60%–98%]; P = 0.79, CI –5 to 12) it was substantial on MLO (rigid = 46% [IQR = 49%, range = 20%–96%] and flexible = 84% [IQR = 19%, range = 49%–98%]; P = 0.005, CI = 6–43).

Concerning which type of plate was most painful, eight women rated the rigid plate as more painful, eight rated the flexible plate as more painful, and the remaining 12 rated them as equally painful (within 3% of the full VAS readout). Six women rated the rigid plate, and five the flexible plate, as ≥ 50% of the maximum.

Concerning the order of compression, 10 women rated the first compression as most painful, 12 rated the second one as most painful, and six rated them as equally painful. Six women rated the first compression and five rated the second as ≥ 50% of the maximum.

Discussion

Due to its inclined thickness gradient, thickness measurements with the flexible plate are not directly comparable to measurements with the rigid plate (21). Thus, even though the measured breast thickness was lower with the flexible plate, this should not on its own be construed as evidence for an improved compression.

Other publications report breast pressures much higher than those obtained here, with de Groot et al. reporting 21.3 kPa for the CC view and 14.2 kPa for the MLO view and recommending a target level of 10 kPa so as to not exceed diastolic arterial blood pressure (11,22). However, the methodologies were different. De Groot et al. determined a contact area between the breast and compression plate (which correlates with breast size) and assumed an even distribution of force, whereas we measured the actual pressure distribution under the compressed breast. Recommendations based on an evenly distributed pressure, while supporting that compression force should be based on the size of the breast, do not reflect actual pressure on the breast. A large portion of the applied force is absorbed in the juxta thoracic, axillary, and pectoral regions that are outside the field of view. As an example, consider this attempt at modifying de Groot et al.’s value of 21.3 kPa for the CC view to be comparable to our mean pressure value from this study, 2.5 kPa (with the rigid plate). First, compensate for the higher average compression force, 18 daN compared to 10 daN by multiplying the pressure by 55%. Second, for the rigid plate, only 25% of the applied force is distributed to the breast (with the remainder in the juxta thoracic area, outside the sensor’s sensitive area). Third, to correct for the different definitions of breast area (contact area in the study by de Groot et al. and total area in the present study) the pressure should be further reduced, but as a conversion factor would be subject to large individual variations the step is skipped. This gives a rough estimated breast pressure of less than 2.9 kPa, similar to the 2.3 kPa value from this study and considerably lower than diastolic pressure. This does not, however, guarantee that blood flow is not obstructed. As the distribution of pressure is highly heterogeneous (Figs. 3 –7), there are often parts of the breast that are subject to pressures exceeding 40 kPa (12,17).

We did not investigate pressure in different regions on the breast (beyond the division between breast and juxta thoracic region), but images (Figs. 3 –7) suggest that the flexible plate gave higher pressure over a larger pressurized breast area (area with non-zero pressure). The pressure gradient was “smoothed” towards the nipple with the flexible plate, while conversely pressure was higher in the juxta thoracic region with the rigid plate. This effect, including thickness reduction, is similar to that obtained by repositioning the breast to exclude part of the juxta thoracic area (17).

There was no change in breast area between the two plates, contrary to the results obtained by Broeders et al. (18) who used a similar set-up with the Hologic Selenia Dimensions system. This could imply either that there is a difference between the flexible plates of the two manufacturers or that this is more dependent on positioning than the plate itself, as we were careful to maintain the same positioning by instructing women to stay absolutely still, with one radiographer providing support while another changed the plate. Another possibility is that while the juxta thoracic tissue is moved out of the field of view, the overall area is kept constant as the remaining breast undergoes greater deformation due to increased compression. If this hypothesis is correct, one must weigh the benefit of greater juxta thoracic tissue inclusion against better overall compression and thus presumably higher image quality of the breast itself. Broeders et al. also suggests that image quality in the juxta thoracic area suffers. This implies that although the idea behind the flexible plate is sound, it is limited by the fact that its inclination passively adapts to the breast, increasing juxta thoracic thickness. An adjustable and lockable inclination may allow additional compression of the clinically relevant parts without affecting the juxta thoracic area.

There was no pain difference between the two plates. The flexible plate reduces high retro-glandular pressures—presumably relieving pain—but increases pressure over the likely more tender breast itself. Since experienced pain is equal, it seems prudent to use the flexible plate, as it provides better compression (higher pressure over a larger area) of the clinically relevant parts of the breast. This might be interpreted in different ways: (i) with equal compression force, the flexible plate provides better compression of the relevant parts of the breast and the perceived patient discomfort is thus more justified; and (ii) with equal desired compression of the relevant parts of the breast, compression force may be reduced with the flexible plate relative to the rigid plate, likely resulting in increased patient comfort.

Both plates are inefficient in the MLO view. The MLO view has a greater compressed area than the CC view, but nevertheless receives less force on the breast. This heavily implies that breast tissue is less compressed due to increased distribution of force to the juxta thoracic and axillary regions. The flexible plate redistributes force from the juxta thoracic area, which is seen as relatively higher mean pressure in both projections but especially on the MLO view. A potential solution could be to implement a compression plate that is flexible in an additional degree of freedom making it able to redistribute force from the axillary region as well. Fig. 7 presents a case in which neither plate provides much pressure on the breast itself in the MLO view. The shortcomings of the rigid plate are highlighted by considering the ratio between mean pressure during stable compression, i.e. during the imaging phase, and the peak achieved mean breast pressure, immediately before the end of force application (Fig. 2). For the CC view, stable pressure was around 90% of the peak value with either plate. This indicates that compression relaxes over time, which was reported by radiographers participating in the study. There is room for improvement, but compression loss is reasonably low. However, for the MLO view, the effect is pronounced, as although the flexible plate maintained 84% of peak pressure, the rigid plate managed only 46%. We speculate that in the MLO view, the radiographer supports the breast during compression so it does not sag, ensuring that the flexible plate, by partly avoiding the juxta thoracic area, can make contact with and compress the breast, making it stay compressed when the radiographer stops supporting it. However, with the rigid plate, while this allows the plate to make contact with the breast, when the supporting hand is removed, the breast will sag as tissue is now displaced rather than compressed, reducing pressure. This result has implications for patient handling during positioning, but we believe that with further investigation this could be compensated for with improved radiographer training.

The present study has several limitations. It had a relatively small study population. The presence of malignant or benign lesions might have affected pressure distribution. Bilateral pressure readings with both plates would have been valuable, as this could have allowed separation of positioning effects and inherent properties of the breast and compression system. An investigation of the relative mammographic image quality of women imaged with the two different plates would have been valuable but would require a much larger study population that would have to be exposed to substantial additional doses of ionizing radiation.

The sensor was calibrated using manufacturer-supplied equipment and recommended procedures. From previous experience, we have found that if known loads are applied to a test object, the sensor is accurate to within a few percent. As the spatial resolution is limited, partial-area effects may have influenced readings, especially at breast edges, and would have a greater relative effect for smaller breasts. The use of paired data should partially compensate for this. Much care was taken to maintain exact positioning using both plates, but there were likely some differences. As the plate order was alternated, the effect of systematic errors of this kind is limited.

In conclusion, the flexible plate improves pressure distribution by distributing more pressure to the breast and less to areas outside of the clinically relevant parts, such as thick tissue at the chest wall and axillary area. We believe that this results in a better breast compression, particularly in the MLO view. Radiographers should be aware that when using the rigid plate for the MLO view, compression will drop considerably during the imaging phase. The flexible plate can improve compression relative to the rigid plate without increasing patient discomfort, or alternatively achieve equal compression with improved patient comfort. There is, however, potential for improvement, as the greater part of the force is still distributed to the juxta thoracic area and there are possible concerns about tissue inclusion and image quality in the juxta thoracic region of the breast.

Footnotes

Acknowledgements

The authors acknowledge the staff at the Lund mammography clinic, Unilabs AB, for their contributions to data collection and study set-up.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The study was supported by GE Healthcare, Women?s Health.

ORCID iD

Magnus Dustler

References

Independent UK Panel on Breast Cancer Screening. The benefits and harms of breast cancer screening: an independent review. Lancet 2012; 380:1778–1786.

Gotzsche

Jorgensen

KJ.

The benefits and harms of breast cancer screening. Lancet 2013; 381:799.

Oeffinger

Fontham

ETH

Etzioni

, et al. Breast Cancer Screening for Women at Average Risk 2015 Guideline Update From the American Cancer Society. JAMA 2015; 314:1599–1614.

Hofvind

Wang

Thoresen

The Norwegian Breast Cancer Screening Program: re-attendance related to the women's experiences, intentions and previous screening result. Cancer Causes Control 2003; 14:391–398.

Orton

Fitzpatrick

Fuller

, et al . Factors affecting women’s response to an invitation to attend for a second breast cancer screening examination. Br J Gen Pract 1991; 41:320–322.

Keemers-Gels

Groenendijk

van den Heuvel

, et al. Pain experienced by women attending breast cancer screening. Breast Cancer Res Treat 2000; 60:235–240.

Whelehan

Evans

Wells

, et al. The effect of mammography pain on repeat participation in breast cancer screening: a systematic review. Breast 2013; 22:389–394.

Miller

Livingstone

Herbison

Interventions for relieving the pain and discomfort of screening mammography. Cochrane Database Syst Rev 2008;(1):CD002942.

Perry

Broeders

de Wolf

, et al. European guidelines for quality assurance in breast cancer screening and diagnosis. Fourth edition–summary document. Ann Oncol 2008; 19:614–622.

10.

Mercer

Hogg

Lawson

, et al . Practitioner compression force variability in mammography: a preliminary study. Br J Radiol 2013; 86:20110596.

11.

de Groot

Broeders

MJM

Branderhorst

, et al. A novel approach to mammographic breast compression: Improved standardization and reduced discomfort by controlling pressure instead of force. Med Phys 2013; 40:081901.

12.

Dustler M, Andersson I, Brorson H, et al. Breast compression in mammography: pressure distribution patterns. Acta Radiol 2012;53:973--980.

13.

Helvie

Chan

Adler

, et al. Breast thickness in routine mammograms: effect on image quality and radiation dose. AJR Am J Roentgenol 1994; 163:1371–1374.

14.

Chida

Komatsu

Sai

, et al. Reduced compression mammography to reduce breast pain. Clin Imaging 2009; 33:7–10.

15.

Fornvik D, Andersson I, Svahn T, et al. The effect of reduced breast compression in breast tomosynthesis: human observer study using clinical cases. Radiat Prot Dosimetry 2010;139:118--123.

16.

Saunders

Jr Samei

The effect of breast compression on mass conspicuity in digital mammography. Med Phys 2008; 35:4464–4473.

17.

Dustler M, Andersson I, Fornvik D, et al. The Effect of Breast Positioning on Breast Compression in Mammography: a Pressure Distribution Perspective. Medical Imaging 2012: Physics of Medical Imaging 2012;8313. DOI: Artn 83134m Doi 10.1117/12.905746.

18.

Broeders

MJM

ten Voorde

Veldkamp

WJH

, et al. Comparison of a flexible versus a rigid breast compression paddle: pain experience, projected breast area, radiation dose and technical image quality. Eur Radiol 2015; 25:821–829.

19.

de Groot

Broeders

Grimbergen

, et al. Pain-preventing strategies in mammography: an observational study of simultaneously recorded pain and breast mechanics throughout the entire breast compression cycle. BMC Womens Health 2015; 15:26.

20.

Revill

Robinson

Rosen

, et al. The reliability of a linear analogue for evaluating pain. Anasthesia 1976; 31:1191–1198.

21.

Hauge

Hogg

Szczepura

, et al. The readout thickness versus the measured thickness for a range of screen film mammography and full-field digital mammography units. Med Phys 2012; 39:263–271.

22.

de Groot

Broeders

Branderhorst

, et al. Mammographic compression after breast conserving therapy: controlling pressure instead of force. Med Phys 2014; 41:023501.