Deformation of breast masses between supine ultrasound and prone magnetic resonance imaging

Abstract

Background

Breast masses will deform between different postures and modalities in the background mammary gland. However, this difference in deformation between benign and malignant masses has not been studied.

Purpose

To investigate the feasibility of using the change of the longitudinal-transverse ratio (L/T) of a breast mass from supine ultrasonography (US) and prone magnetic resonance imaging (MRI) as representing deformation in differentiating between benign and malignant masses.

Material and Methods

The participants were 197 consecutive patients with 201 breast masses (55 benign, 146 malignant) who had undergone both US and MR examinations with histological diagnoses at our hospital from January 2012 to December 2014. On US, the largest transverse diameter of the mass (T_US) parallel to the pectoral muscle and the longitudinal diameter (L_US) perpendicular to the muscle were measured. On prone MRI, the largest transverse diameter of the mass (T_MR) parallel to the pectoral muscle and the orthogonally oriented diameter as the longitudinal direction (L_MR) were measured. The change of the L/T between these modalities was compared between benign and malignant masses.

Results

Malignant breast masses showed significantly smaller change of the L/T (median [interquartile range (IQR)] = 13.5% [4.29–24.4]) than benign ones (36.7% [24.3–52.4]) (P < 0.001). The optimum cut-off value of the change of the L/T for differentiating malignant from benign was 19.1%, with sensitivity, specificity, and area under the curve of 63.7%, 90.9%, and 0.834, respectively.

Conclusion

The change of the L/T may be a feasible and useful quantitative index for differentiating breast masses.

Keywords

Breast mass deformation ultrasonography magnetic resonance imaging supine position prone position

Introduction

Breast cancer is one of the most globally common cancers and the leading cause of cancer deaths among women (1). Currently, mammography, ultrasonography (US), and dynamic contrast-enhanced (DCE) magnetic resonance imaging (MRI) are the most widespread imaging modalities for both breast cancer screening and clinical breast examination. US is routinely used in conjunction with mammography and it is widely accepted as a non-invasive and useful modality for the detection and characterization of breast masses. Since DCE-MRI is the most sensitive modality for breast cancer detection (2), it is used to screen women at high risk. Additionally, MRI enables accurate evaluation of the extent of cancers and it is used to further evaluate the abnormalities seen on mammography and/or US.

Breast US is usually performed in the supine position, whereas breast MRI is done in the prone position. During breast US, all tissue layers of the breast are widened and flattened, especially the fatty tissue, which is relatively more highly compressible and stretchable than other breast structures. In contrast, in breast MRI performed in the prone position, the breasts are pendant and extended. Thus, the breast tissue, including breast lesions, tends to be more compact at US than at MRI (3). From our clinical experience, we found that breast masses tend to present a slightly flatter form on US, and they tend to show extension towards the direction of gravity on MRI. Furthermore, the degree of deformation seemed to be larger for benign masses than for malignant ones. However, the difference in deformation between benign and malignant masses in the comparison of supine US and prone MRI has not been studied so far.

It has been reported that, on breast US, masses that have a larger longitudinal-transverse ratio (L/T) are likely to be malignant (4 –6). The L/T is known to be the only quantitative index among several criteria used to differentiate malignant from benign breast masses (7,8). On the other hand, there has been no discussion about the L/T of breast masses on breast MRI, still less the difference compared to breast US.

In this study, the L/T of breast masses was retrospectively measured using US images, as was the ratio on the MR image in the same orientation as the US image. The rate of the change of the L/T of the two modalities was calculated, as representing deformation. The purpose of this study was to investigate the feasibility of using the change of the L/T from supine US and prone MR images in differentiating benign and malignant breast masses.

Material and Methods

Participants

This was a retrospective study approved by the institutional review board of our university and written informed consent from each patient was waived. Data were collected by reviewing the imaging records in the electronic medical records at our university hospital between January 2012 and December 2014. There was a total of 211 consecutive patients who had undergone pretreatment US and MR examinations, all of whom had histological diagnoses. Of these 211 patients, five were excluded because their large-size masses did not fit within the field of view of US, making it impossible to accurately measure the length of such lesions. Two patients were excluded because they had more than three lesions on one side of the breast in close proximity, making it difficult to do a one-to-one correlation between US and MR images. Several other patients were excluded due to improper positioning at MRI (n = 2), nipple deformation resulting from traction by subareolar masses (n = 2), and patients who had undergone breast-conserving surgery or nipple excision (n = 2), making it difficult to accurately measure the distances from the nipple to the pectoral muscle. In addition, one male patient was excluded. The final study population involved 197 patients with 201 masses.

The patients were all women (age range = 19–88 years; mean = 57 years). The masses were benign in 55 masses (fibroadenoma [n = 35], intraductal papilloma [n = 7], benign phyllodes tumor [n = 5], hamartoma [n = 3], fibrocystic disease [n = 2], mastitis [n = 1], adenosis [n = 1], and adenomyoepithelioma [n = 1]) and malignant in 146 masses (ductal carcinoma in situ [DCIS; n = 8], invasive ductal carcinoma [IDC; n = 124], mucinous carcinoma [n = 9], invasive lobular carcinoma [n = 3], metaplastic carcinoma [n = 1], and metastasis [n = 1]).

For 189 of the 201 masses, MRI was performed for those suspected of being malignant on US. For the remaining 12 masses, MRI was initially performed and US was done later for the lesions detected on MRI. The median time interval between US and MRI was seven days (interquartile range [IQR] = 5–14 days).

US and MRI protocols

US examinations were performed by breast surgeons with > 3 years of experience using either the US systems of HI VISION Avius or HI VISION Preirus (both Hitachi-Aloka Medical, Tokyo, Japan), equipped with a 10-MHz linear transducer, in accordance with the guidelines of the Japan Association of Breast and Thyroid Sonology (9). In the guidelines, it is stated that mass lesions should be recorded with the largest cross-section and the cross-section orthogonal to the section. Recoding and measurement of lesions on transverse and sagittal sections were done in all masses. Patients were examined in the supine position with their ipsilateral arm over the head.

MR examinations were performed using either a 1.5-T MR system (Intera Nova Dual; Philips Medical Systems, Best, Netherlands) with a dedicated four-channel phased array breast coil between January 2012 and February 2013, or a 3-T MR system (MAGNETOM Skyra; Siemens Healthcare, Erlangen, Germany) with a dedicated 16-channel phased array breast coil between March 2013 and December 2014. Patients were set in the prone position with both arms over the head and both breasts were examined in the transverse plane.

The DCE-MRI protocols and parameters were as follows: three-dimensional (3D) fat-suppressed T1-weighted (T1W) high-resolution isotropic volume excitation sequence on the 1.5-T system; TR/TE = 4.4/2.2 ms; flip angle (FA) = 15°; field of view (FOV) = 320 mm; matrix = 304 × 304; slices = 150; slice thickness = 1 mm; sensitivity encoding factor = 1.8; fat suppression = spectral attenuated inversion recovery; number of excitations = 1; and acquisition time = 56 s. A 3D fat-suppressed volume-interpolated breath-hold examination sequence was done on the 3-T system, with: TR/TE = 3.3/1.4 ms; FA = 15°; FOV = 320 mm; matrix = 352 × 352; slices = 144; slice thickness = 1 mm; generalized autocalibrating partial parallel acquisition acceleration factor = 3; fat suppression = spectral attenuated inversion recovery; number of excitations = 1; and acquisition time = 60 s.

Gadopentetate dimeglumine (Magnevist; Bayer Yakuhin, Limited, Osaka, Japan) or gadodiamide hydrate (Omniscan; Daiichi Sankyo Company, Limited, Tokyo, Japan) were power-injected and the early phases of dynamic MRI were obtained at the centers of k-space at 90 s on the 1.5-T system and 115 s on the 3-T system after bolus injection. Early-phase DCE-MR images were used for mass measurement. Pre-contrast images were used for assessment of fibroglandular tissue (FGT).

Image analysis

Two of the authors with seven and 17 years of experiences as radiologists, respectively, measured breast masses using recorded transaxial images of B-mode US and early-phase DCE-MRI. The method used to measure the lesions is illustrated in Fig. 1.

Fig. 1.

Measurements of a breast mass on US and MRI. Schemas show a transaxial US image in the supine position (a) and an MR image in the prone position (b). (a) T_US, the largest transverse diameter of the mass parallel to the superficial surface of the pectoral muscle. L_US, the longitudinal diameter perpendicular to the muscle; D_US, the distance between the skin and the anterior border of the mass. (b) T_MR, the largest transverse diameter of the mass parallel to the pectoral muscle. L_MR, the orthogonally oriented diameter as the longitudinal direction; D_MR, the distance between the skin and the anterior border of the mass. Breast projection, the distance from the nipple to the pectoral muscle. The change of the L/T = (L_MR/T_MR – L_US/T_US)/(L_MR/T_MR)×100 (%).

On US, the largest transverse diameter of the mass (T_US) parallel to the superficial surface of the pectoral muscle was measured. The longitudinal diameter (L_US) was set perpendicular to the muscle. The distance from the skin surface to the anterior border of the mass (D_US, for depth) was also measured. When the mass showed a halo, the distances were measured in the low-echo area.

On MRI, the largest transverse diameter of the mass (T_MR) parallel to the pectoral muscle was measured and the orthogonally oriented diameter was set as the longitudinal diameter (L_MR). The distance between the skin surface and the anterior border of the mass (D_MR) was also measured. In addition, the distance from the nipple to the pectoral muscle, which was termed breast projection by Satake et al. (10), was measured. The amount of FGT was evaluated on pre-contrast images by consensus of two radiologists (with seven and 10 years of experience as radiologists, respectively) according to the BI‒RADS MR lexicon (11).

The L/T of the breast mass on US is calculated as L_US/T_US. The L/T of the mass on MRI measured in the same orientation as the US image is calculated as L_MR/T_MR. The rate of the change of the L/T is calculated as follows:

(L_{MR} / T_{MR} - L_{US} / T_{US}) / (L_{MR} / T_{MR}) \times 100 (%)

where L_US/T_US, L_MR/T_MR, and the change of the L/T were compared between benign and malignant masses.

Statistical analysis

The Mann–Whitney U test was used to compare L_US/T_US, L_MR/T_MR, and the change of the L/T between benign and malignant masses. The receiver operating characteristic (ROC) curve and the Youden index were used to identify the optimum cut-off value for the change of the L/T to differentiate malignant from benign masses. The Mann–Whitney U test was also used to assess differences in the change of the L/T among different histological types of masses and to compare the patients’ age, mass size (the larger value of the transverse diameter or the longitudinal diameter on both US and MRI), depth, and breast projection between benign and malignant masses. Fisher’s exact test was used to confirm differences in patients’ menopausal status, mass location into quadrants (mainly on palpation), and the amount of FGT between the two groups.

Logistic regression was used to investigate the change of the L/T and clinical characteristics (age, menopausal status, size, depth, location, breast projection, and the amount of FGT) related to benign or malignant masses on univariate and multivariate analyses.

Inter-observer agreement between the measurers was examined using intraclass correlation coefficients for L_US/T_US and L_MR/T_MR.

All statistical analyses were conducted using R software (version 3.3.0, available as a free download from http://www.R-project.org). A P value < 0.05 was considered significant.

Results

The clinical and imaging characteristics of the 197 patients with 201 breast masses are summarized in Table 1. There were significant differences in age and menopausal status between the benign and malignant mass groups (P < 0.001, respectively), whereas there were no significant differences in mass size, depth, location, breast projection, and the amount of FGT between the two groups.

Table 1.

Characteristics of the 197 patients with 201 breast masses.

Characteristics	Benign (n = 55)		Malignant (n = 146)
Age (years), median (range)	43 (19–76)	56 (27–88)	<0.001*
Menopausal state
Premenopausal, n (%)	44 (80)	60 (41)	<0.001^†
Postmenopausal, n (%)	11 (20)	86 (59)	<0.001^†
Size (mm), median (IQR)
US (the larger value of T_US or L_US)	18.48 (10.89–28.07)	15.67 (10.44–20.50)	0.14*
MR (the larger value of T_MR or L_MR)	15.7 (11.7–27.2)	17.5 (12.6–22.9)	0.78*
Depth (mm), median (IQR)
US (D_US)	5.15 (3.08–8.10)	5.91 (3.80–8.48)	0.28*
MR (D_MR)	14.2 (7.2–24.6)	16.0 (9.5–25.7)	0.40*
Location, n (%)
Inner-upper	13 (24)	33 (23)	0.42^†
Inner-lower	4 (7)	14 (10)
Outer-upper	24 (44)	63 (43)
Outer-lower	6 (11)	26 (18)
Subareolar	8 (15)	10 (7)
Breast projection on MR (mm), median (IQR)	56.2 (41.8–67.4)	59.8 (44.8–74.2)	0.37*
Amount of FGT on MR, n (%)
Almost entirely fat	2 (4)	6 (4)	0.12^†
Scattered FGT	7 (13)	41 (28)
Heterogeneous FGT	37 (67)	81 (55)
Extreme FGT	9 (16)	18 (12)

IQR, interquartile range; FGT, fibroglandular tissue.

*According to Mann–Whitney U test.

†

According to Fisher’s exact test.

The L/T of US and MRI

The L_US/T_US was significantly smaller in benign masses (median [IQR] = 0.621 [0.484–0.712]) than in malignant ones (median [IQR] = 0.778 [0.640–0.872]) (P < 0.001). Conversely, the L_MR/T_MR was significantly larger in benign masses (median [IQR] = 1.00 [0.887–1.26]) than in malignant ones (median [IQR] = 0.900 [0.804–1.00]) (P < 0.001). There was substantial agreement between observers for the L_US/T_US (intraclass correlation coefficient [ICC] = 0.755) and almost perfect agreement for the L_MR/T_MR (ICC = 0.880).

The change of the L/T between US and MRI

Representative US and MR images of benign and malignant masses are shown in Fig. 2. The median change of the L/T was 36.7% (IQR = 24.3–52.4) for benign masses and 13.5% (IQR = 4.29–24.4) for malignant masses; the difference was significant (P < 0.001) (Fig. 3). Although there were some overlaps, the change of the L/T between US and MR images was significantly smaller for malignant masses than for benign ones.

Fig. 2.

Typical transaxial B-mode US images and early-phase DCE-MR images. A 46-year-old woman with fibroadenoma (a) and a 45-year-old woman with invasive ductal carcinoma (b). Deformation of the mass (flattening) is obvious in fibroadenoma (a), whereas in invasive ductal carcinoma (b), deformation is not so apparent. The change of the L/T of the masses in (a) and (b) are 59.2% and –0.618%, respectively.

Fig. 3.

Difference in the change of the L/T: benign vs. malignant breast masses.

The ROC analysis showed that the optimal cut-off value of the change of the L/T for differentiating malignant from benign masses was 19.1%, with an area under the curve (AUC) of 0.834 (Fig. 4). Assuming that masses with a change of the L/T < 19.1% are malignant, the sensitivity and specificity for differentiating malignant from benign using the change of the L/T alone were 63.7% and 90.9%, respectively.

Fig. 4.

ROC analysis using the change of the L/T to differentiate malignant from benign breast masses.

The small change of the L/T for malignant masses was significant on univariate analysis, with an odds ratio (OR) of 0.933 and a 95% confidence interval (CI) of 0.913–0.954 (P < 0.001). In addition, even when the change of the L/T was adjusted by clinical characteristics on multivariate analysis, the result was still significant with an OR of 0.900 and a 95% CI of 0.868–0.934 (P < 0.001).

Figure 5 shows the change of the L/T for different tissue characteristics. A significant difference in the change of the L/T was detected between hamartoma and all other benign masses (median [IQR] = 77.8% [77.6–81.5] and 35.8% [23.8–51.2], respectively; P = 0.0065). No significant differences in the change of the L/T were detected among malignant masses with different tissue types.

Fig. 5.

The change of the L/T for various histological types of breast masses. When there are three or more cases, the cases are shown by a boxplot. Regarding benign masses, the median change of the L/T is 35.0% (IQR = 22.6–51.3%) for fibroadenoma (35 patients) (a), 37.0% (IQR = 29.8–48.5) for intraductal papilloma (7 patients) (b), 36.5% (IQR = 33.1–41.4) for benign phyllodes tumor (5 patients) (c), and 77.8% (IQR = 77.6–81.5%) for hamartoma (3 patients) (d). There is a significant difference in the change of the L/T between hamartoma and all other benign tumors (median [IQR] = 35.8% [23.8–51.2]; P = 0.0065). For malignancy, the median change of the L/T is 12.3% (IQR = 1.92–18.9) for ductal carcinoma in situ (8 patients) (e), 13.7% (IQR = 4.12–24.8) for invasive ductal carcinoma (124 patients) (f), 10.3% (IQR = 8.37–23.2) for mucinous carcinoma (9 patients) (g), and 20.1% (IQR = 15.2–22.2) for invasive lobular carcinoma (3 patients) (h). No significant differences in the change of the L/T are noted among malignant tumors.

Discussion

This is the first study that evaluated the L/T on prone MRI and the change of the L/T by comparing supine US and prone MRI. The L/T on MRI was significantly larger in benign masses than in malignant ones, and the change of the L/T between US and MRI was also significantly larger for benign masses than for malignant ones.

The breast itself can easily deform due to gravity because it has a wide range of motion and it is a soft structure that consists mainly of fat and glandular tissue. It can be speculated that breast masses located within the deformable breast will also deform due to posture under the influence of gravity. In fact, there is a previous report with a limited number of cases (n = 6) that showed that there were changes in volume, surface area, and the maximum diameter of breast cancers between prone and supine MRI (12). They concluded that breast cancers could show some degree of deformation depending on posture. The present study is in agreement with that paper, because even malignant masses showed slight deformation between supine US and prone MRI.

In the present study, benign masses showed a relatively larger L/T on MRI and a larger change of the L/T between US and MRI than malignant ones. It is reasonable to assume that the difference in these can be attributed to the difference in stiffness. It is known from ex vivo surgical pathological studies that benign breast lesions are often softer than breast cancers (13,14). Histopathologically, breast cancers often show a desmoplastic reaction with reactive proliferation of connective tissue and fibroblasts around malignant epithelial cells; this desmoplastic reaction would be the source for the stiffness of cancers (15).

The influence of mass deformation by US probe compression is one of the factors that could affect the difference in the change of the L/T between benign and malignant masses. In the previous US report, it was shown that benign lesions are more easily deformed and flattened by compression of US probe than malignant lesions (that is a smaller L/T), even with slight compression (16). It is considered that, in benign masses, the change of the L/T better expresses the degree of deformation due to the magnification effect by subtracting the relatively smaller L/T on US from the larger L/T on MRI.

Mass rotation may be another factor that could affect the difference in the change of the L/T between benign and malignant masses. The desmoplastic reaction takes place within the immediate periphery of cancers (15) and therefore malignant masses would have less chance to rotate within the breast tissue, whereas benign masses are less fixed to the surrounding tissue and would, therefore, be freer to rotate. Since there are individual differences in the degree of mammary gland development and the size of breasts, mass size, depth, location, breast projection, and the amount of FGT might affect the change of the L/T. However, even after adjustment for these factors, the difference in the change of the L/T between begin and malignant was present. Thus, the influence of mass rotation may not be so large.

In the present study, the change of the L/T between US and MRI showed a high AUC (0.834) and high specificity (90.9%) for discriminating malignant from benign masses with high inter-rater reliability, even though the US examinations were performed by many different operators. It is known from a meta-analysis that MRI has high sensitivity (90%), but relatively lower specificity (72%) for diagnosing breast cancers (17). From the present result, retrospective measurement of the change of the L/T between these modalities may be able to improve the specificity of breast MRI and it can be easily applied to our clinical practice without any additional examinations.

There were no significant differences in the change of the L/T among different histological types of malignancy. This may contradict the fact that IDC tends to be stiffer than DCIS (13,14,18 –20). We speculated that this is because, in the present study, the number of DCIS cases recognized as masses was small and typical DCIS that appears as non-mass enhancement on MRI (21) was not included. Among the various histological types of benign masses, a significant difference was detected between the change of the L/T of hamartoma and that of all other benign masses. This coincides with a report stating that hamartomas were mostly soft and never had a consistency as hard as that of fibroadenomas, even when they were composed mostly of glandular tissue (22).

The present study has some limitations. First, two different imaging modalities had to be used to calculate L/T. US strongly reflects the morphological aspects of lesions, whereas DCE-MRI contains functional aspects, i.e. tumor vascularity (3). Therefore, what is observed by these two modalities may be slightly different in nature. Second, there may be operator-dependent technical issues in the visualization of lesions on US. A prospective study using the same modality (e.g. prone and supine MRI) and acquisition of 3D data may aid in validation of the present results.

In conclusion, breast masses show deformation expressed by the change in the L/T comparing supine US and prone MRI. Malignant masses showed a smaller change of the L/T than benign ones. The change of the L/T may become one of the quantitative indices for differentiating breast masses.

Footnotes

Acknowledgements

The authors are grateful to the statistician at their institute, Satoshi Teramukai, for his help with statistics.

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) received no financial support for the research, authorship, and/or publication of this article.

References

Fitzmaurice

Allen

Barber

, et al. Global, regional, and national cancer incidence, mortality, years of life lost, years lived with disability, and disability-adjusted life-years for 32 cancer groups, 1990 to 2015: a systematic analysis for the global burden of disease study. JAMA Oncol 2017; 3:524–548.

Berg

Gutierrez

NessAiver

, et al. Diagnostic accuracy of mammography, clinical examination, US, and MR imaging in preoperative assessment of breast cancer. Radiology 2004; 233:830–849.

Park

Kim

, et al. Second-look US: how to find breast lesions with a suspicious MR imaging appearance. Radiographics 2013; 33:1361–1375.

Ueno

Tohno

Soeda

, et al. Dynamic tests in real-time breast echography. Ultrasound Med Biol 1988; 14(Suppl. 1):53–57.

Fornage

Lorigan

Andry

Fibroadenoma of the breast: Sonographic appearance.

Radiology 1989; 172:671–675.

Stavros

Thickman

Rapp

, et al. Solid breast nodules: Use of sonography to distinguish between benign and malignant lesions. Radiology 1995; 196:123–134.

Ishii

Ultrasonographic diagnosis of breast diseases: A review of diagnostic criteria of sonomammography on a real-time scanner. Nihon Igaku Hoshasen Gakkai Zasshi 1993; 53:1141–1159. [in Japanese]

Kato

Ueno

Kawauchi

, et al. Threshold value of depth width ratio (D/W) in the breast ultrasonogram. J Med Ultrasonics 2002; 29:450. [in Japanese]

Japan Association of Breast and Thyroid Sonology. Guideline for Breast Ultrasound: management and diagnosis. 3rd ed. Tokyo, Japan: Nankodo Co., 2014. [in Japanese]

10.

Satake

Ishigaki

Kitano

, et al. Prediction of prone-to-supine tumor displacement in the breast using patient position change: investigation with prone MRI and supine CT. Breast Cancer 2016; 23:149–158.

11.

American College of Radiology. Breast Imaging Reporting and Data System (BI‒RADS) atlas. 5th ed. Reston, VA: American College of Radiology, 2013.

12.

Gombos

Jayender

Richman

, et al. Intraoperative supine breast MR imaging to quantify tumor deformation and detection of residual breast cancer: preliminary results. Radiology 2016; 281:720–729.

13.

Samani

Zubovits

Plewes

Elastic moduli of normal and pathological human breast tissues: an inversion-technique-based investigation of 169 samples.

Phys Med Biol 2007; 52:1565–1576.

14.

Ramiao

Martins

Rynkevic

, et al. Biomechanical properties of breast tissue, a state-of-the-art review. Biomech Model Mechanobiol 2016; 15:1307–1323.

15.

Meng

Zhou

Sasano

, et al. Tumor necrosis factor alpha and interleukin 11 secreted by malignant breast epithelial cells inhibit adipocyte differentiation by selectively down-regulating CCAAT/enhancer binding protein alpha and peroxisome proliferator-activated receptor gamma: mechanism of desmoplastic reaction. Cancer Res 2001; 61:2250–2255.

16.

Nonomura

Shimamoto

Hatasa

, et al. Differential diagnosis of breast mass image-forming lesions based on changes in depth-width ratio and internal echo intensity by hand-held probe compression. J Med Ultrason (2001) 2008; 35:63–69.

17.

Peters

Borel Rinkes

Zuithoff

, et al. Meta-analysis of MR imaging in the diagnosis of breast lesions. Radiology 2008; 246:116–124.

18.

Krouskop

Wheeler

Kallel

, et al. Elastic moduli of breast and prostate tissues under compression. Ultrason Imaging 1998; 20:260–274.

19.

Matsumura

Umemoto

Fujihara

, et al. Measurement of elastic property of breast tissue for elasticity imaging. In: Proceedings of IEEE Ultrasonic Symposium. Rome: IEEE, 2009:1451–1454.

20.

Umemoto

Ueno

Matsumura

, et al. Ex vivo and in vivo assessment of the non-linearity of elasticity properties of breast tissues for quantitative strain elastography. Ultrasound Med Biol 2014; 40:1755–1768.

21.

Liberman

Morris

Lee

, et al. Breast lesions detected on MR imaging: features and positive predictive value. Am J Roentgenol 2002; 179:171–178.

22.

Hessler

Schnyder

Ozzello

Hamartoma of the breast: diagnostic observation of 16 cases.

Radiology 1978; 126:95–98.