Radiation Doses and Risks in Breast Screening

Abstract

This article describes radiation doses and cancer risks of digital breast imaging technologies used for breast cancer detection. These include digital mammography (DM), digital breast tomosynthesis (DBT), and newer technologies such as contrast-enhanced digital or spectral mammography (CEM), whole-breast computed tomography, breast-specific gamma imaging (BSGI), molecular breast imaging (MBI), and positron emission mammography (PEM). This article describes the basis for radiation risk estimates, compares radiation doses and risks, and provides benefit-to-radiation-risk ratios for different breast imaging modalities that use ionizing radiation.

Current x-ray–based screening modalities such as DM and DBT have small to negligible risks of causing radiation-induced cancers in women of normal screening age. Possible new screening modalities such as CEM have similar small cancer risks. Potential screening modalities that involve radionuclide injection such as BSGI, MBI, and PEM have significantly higher cancer risks unless efficient detection systems and reduced administered doses are used. Benefit-to-radiation-risk estimates are highly favorable for screening with DM and other modalities having comparable (or higher) cancer detection rates and comparably low radiation doses.

Introduction

Tissue ionization can occur due to exposure to either particulate radiation or electromagnetic radiation of adequate energies. Particulate radiation (charged particles such as alpha particles, beta particles [electrons], protons, and neutrons) are components of cosmic radiation, most of which are filtered out by the earth’s atmosphere. Particulate radiation also occurs due to terrestrial sources: uranium and its decay products such as thorium, radium, and radon. The average person in the United States receives about 3.1 milliSieverts (mSv) of natural background radiation per year, about 70% of this from inhaled radon gas in the air of homes, and about 10% each from cosmic radiation, ingested food and water, and terrestrial ground sources (1).

Tissue ionization can occur due to electromagnetic radiation (photons) with energies above a few electron volts (eV). This energy threshold occurs in the ultraviolet region of the electromagnetic spectrum (Figure 1). X-rays are defined as the region of the electromagnetic spectrum above the ultraviolet region, with photon energies between 100 eV and 100 000 eV (100 keV), so all x-rays can cause tissue ionization (2). Radio waves used in MRI are part of the electromagnetic spectrum but fall well below the threshold of photon energy causing tissue ionization. High-frequency sound waves used for ultrasound also cause no tissue ionization.

Averaged over the entire U.S. population, diagnostic medical procedures contribute an average whole-body dose of approximately 3 mSv per year, comparable to the average natural background radiation dose received each year. Most of this diagnostic population dose comes from computed tomography (CT) and nuclear medicine procedures (1).

Breast screening procedures that use ionizing radiation include digital mammography (DM) and digital breast tomosynthesis (DBT) and potential new breast screening techniques such as contrast-enhanced mammography (CEM), whole-breast CT (WBCT), breast-specific gamma imaging (BSGI), molecular breast imaging (MBI), and positron emission mammography (PEM). Most of these technologies limit radiation exposure primarily to breast tissue, but procedures that involve the injection of radionuclides, such as BSGI, MBI, and PEM, expose all body organs to ionizing radiation, and those additional exposed tissues must be taken into account when assessing radiation risks (3). Both public media and peer-reviewed articles have raised the issue of other organs being exposed to ionizing radiation due to DM and DBT (4–6), but both measurements and Monte Carlo estimates have shown that the exposure and risk to organs beyond the breast from DM and DBT are negligible (7, 8).

This article begins with a summary of radiation doses from various breast screening procedures, then describes the scientific basis for risk estimates and goes on to estimate risks from various breast screening procedures based on the linear no-threshold model of radiation risk.

Units of radiation dose

Radiation doses in x-ray imaging of the breast are typically expressed in terms of mean glandular dose (MGD), the absorbed dose to fibroglandular breast tissues averaged over the full extent of fibroglandular tissue (9). Averaging across the breast is important because at mammography x-ray energies, each 1–2 cm of breast tissue decreases the quantity of x-rays by 50%, so the dose to fibroglandular tissues falls by a factor of 10x–100x from the entrance surface to the exit surface of the breast. Units of MGD are typically milliGray (mGy), a unit of absorbed dose (9, 10).

One Gray is defined as the absorption of one joule of radiation energy per kilogram of tissue, so a milliGray is an absorbed dose of 1/1000th joule of radiation energy per kg (10). Older units of absorbed dose are the rad and millirad (mrad, 1 one-thousandth of a rad), with 100 rad = 1 Gy (or 100 mrad = 1 mGy). Since radiation doses are measured in units of energy deposited per unit mass, it is not appropriate to combine radiation doses to different breast tissues, such as adding the doses to left and right breasts, as some authors have done. It is appropriate to combine radiation doses from different exposures of the same breast tissue, such as radiation doses from mediolateral oblique (MLO) and craniocaudal (CC) views to get a two-view breast dose.

Two other units of radiation dose are sometimes used (10). One is equivalent dose, which takes into account the biological effect of the particular type of radiation by multiplying absorbed dose by a radiation weighting factor. For example, alpha particles and protons have radiation weighting factors of 20 and 2, respectively, while x-rays and electrons have radiation weighting factors of 1. Equivalent dose is measured in units of Sieverts (Sv), so for x-rays and electrons, the equivalent dose in Sv is equal to the absorbed dose in Gy.

Another useful dose quantity is effective dose, also measured in Sv, which takes into account the radiosensitivity of various tissues by multiplying the equivalent dose by a tissue weighting factor. Table 1 lists the tissue weighting factors for various tissues, as established by the International Commission on Radiological Protection (ICRP) in Report 103 (11). These weighting factors sum to 1 across all body organs, allowing a rough comparison of detrimental effects (primarily in terms of cancer induction) across various organs and various radiologic procedures. Effective dose is defined as the whole-body dose having equivalent risk of detriment as a more limited exposure to one or more organs, and it is the dose quantity that is used to describe natural background radiation. To determine the effective dose from a mammography exam, one multiplies the mean glandular dose (in mGy) by the radiation weighting factor (which is 1 for x-rays) to get equivalent dose (in mSv), then multiplies that equivalent dose by the tissue weighting factor for breast tissue (which is 0.12) to get the effective whole body dose (in mSv).

Radiation doses from breast imaging procedures

As of December 1, 2019, of the 21 305 mammography units at 8668 certified facilities in the U.S., 61% are DM units, 39% are digital units with DBT capability, and only 26 units (0.12%) are screen-film mammography (SFM) units (12). The adoption of digital in U.S. breast imaging is essentially complete, and the dissemination of DBT is widespread and increasing.

Measured radiation doses to the breasts from DM or DBT are best measured on actual breast tissue rather than on phantoms and are best measured averaged over a representative population of women. That was done in approximately 10% of the 49 528 women participating in the American College of Radiology Imaging Network’s (ACRIN’s) Digital Mammography Imaging Screening Trial (DMIST), which compared early (2001–2003) DM clinical performance to SFM in a paired study, in which each woman received both exams (13, 14).

DM Doses

Digital mammography doses depend on five factors: compressed breast thickness, breast density, x-ray beam quality (half-value layer, HVL), x-ray tube output (milliamperes), and exposure time, with the last two determining milliampere-second (mAs). Compressed breast thickness depends on the patient, patient positioning, and compression force applied by the technologist. Breast density is an inherent patient characteristic, and HVL and mAs are determined by the equipment (target, filter, and beam energy) and automatic exposure control design.

The ACRIN’s DMIST dose results, averaged over five different types of DM systems, based on 19 923 images on 5021 women, resulted in a two-view mean glandular dose of 3.72 mGy dose for DM compared with a 4.74 mGy dose for SFM (14). This comparison used first-generation DM units, including 8% of prototype charge-coupled device units that never became available commercially, 21% of units by a manufacturer no longer selling DM units (Fischer, Wheat Ridge, CO), and 18% of computed radiography (CR) units. Based on Food and Drug Administration (FDA) data collected from 2007–2014 on standardized phantoms, CR units consistently delivered higher MGD than either direct digital units or SFM, while CR image quality based on phantom scores was assessed to be below that of direct digital units (Figure 2).

Collective DM doses from the ACRIN’s DMIST study are no longer representative of DM doses now in the field, as current units are second- and third-generation DM systems operating at lower doses per exposure than those used in the DMIST. For example, in clinical screening data collected in the UK between 2010 and 2012 on 25 408 women, Young and Oduko found that the average MGD per two-view examination was 3.03 mGy for direct digital radiography (DR) and 4.69 mGy for CR compared with 4.01 mGy for SFM (15). This work also provided dose data by manufacturer and model, confirming that, in general, more recently released DM systems operate at lower dose levels than first-generation systems. Like the DMIST study, which showed a 22% lower dose for digital systems (including CR) than screen-film, the UK screening data showed a 25% lower dose for DM than for SFM.

Similarly, a 2018 paper by Gennaro et al measured the breast doses by view on 4780 digital images in 1208 women acquired on a single manufacturer’s system (Selenia Dimensions, Hologic, Bedford, MA), finding the average CC view MGD to be 1.366 mGy and the average MLO view MGD to be 1.374 mGy for a two-view dose 2.74 mGy (16). Working with the same system, Skaane et al, in a 2013 comparison of DM and DBT, reported a two-view MDG of 3.16 mGy for DM (17).

DBT Doses

Digital breast tomosynthesis doses depend on the number of different angle views acquired and the dose per angle view, but the primary determinant of DBT doses is the protocol used for acquisition. Initial FDA approval of Hologic’s and Siemens’ (Erlangen, Germany) DBT systems involved collecting both CC and MLO two-dimensional (2D) DM images along with CC and MLO multiangle data sets that enabled DBT reconstruction of dozens of planar images through the entire breast in both CC and MLO projections (two-view DM + two-view DBT). Subsequently, manufacturers added software (and received FDA approval) to collect only the multiangle DBT data sets in each view projection (CC and MLO) and reconstruct synthetic (SDM) 2D CC and MLO images from each DBT data set (two-view DBT + SDM). GE Healthcare (Chicago, IL) was initially approved for acquisition of DBT in the MLO view, with the CC view acquired as a 2D image. Subsequently, they received approval to acquire both CC and MLO DBT views, with simulated 2D CC and MLO views reconstructed from each 2D data set.

Typically, the multiview DBT data set is acquired at doses ranging from those equal to 2D DM doses up to 1.5x higher than 2D DM doses (18). For example, in the large study mentioned above using the Hologic Dimension DBT system, Gennaro et al found the mean two-view DBT dose to be 3.74 mGy, compared to 2.74 mGy for DM, a 36% higher dose for DBT (16). In the Oslo Screening Trial, using the same DBT system, Skaane et al reported a mean DBT dose 23% higher than the mean DM doses (17). A 2015 literature review by Svahn et al found that protocols adding two-view DBT to DM had MGDs 2.0 to 2.23 times higher than DM alone (18). When synthetic 2D views replaced 2D acquisitions, breast dose was reduced by 45%.

Manufacturers such as GE Healthcare acquire DBT images at approximately the same MGD as DM images (19). A study performed on a Siemens Inspiration DBT system using patient acquisitions with an average breast thickness of 53 mm reported MGDs of 1.06 mGy for DM compared with 2.39 mGy for DBT (20). Figure 3 shows a comparison of DBT and DM doses as a function of simulated breast thickness using acrylic phantoms for GE Healthcare, Hologic, and Siemens DBT-capable units (19).

CEM Doses

Contrast-enhanced mammography acquires a pair of 2D images within 1–2 seconds during the same breast compression, with one image acquired with x-ray energies below the k-edge energy of iodine (33.2 keV) and the other with x-ray energies mostly above the k-edge. These dual-energy images are acquired several minutes after injection of a nonionic iodinated contrast agent. The low-energy image is typically acquired with the same x-ray target, filtration, and beam energy (and at the same radiation dose) as a conventional 2D image. The high-energy images are typically acquired at 45–49 kVp, with additional filtration to harden the x-ray beam, at a breast radiation dose 20%–45% that of the low-energy image. The two acquired images are combined with appropriate weighting factors to produce an image maximizing the conspicuity of lesions taking up iodinated contrast agent and minimizing the structured noise of fibroglandular tissues. The resulting CEM dose is therefore 20%–45% higher than the DM dose. Figure 4 shows MGDs for the GE Healthcare Senobright CEM system compared to DM doses on the same system (21). Using a different system (Hologic Selenia Dimensions) capable of performing DM, DBT, and CEM, James et al determined doses for 6214 patients who underwent DM, 3662 patients who underwent DBT, and 173 patients who underwent CEM (22). At the same compressed breast thickness (average for CEM: 63 mm), MGD for DM was 2.1 mGy, for DBT was 2.5 mGy (19% higher than DM), and for CEM was 3.0 mGy (43% higher than for DM).

WBCT Doses

Unlike the limited-angle acquisitions of DBT, WBCT is designed to acquire a 360^o data set of x-ray beams through the breast with the patient lying prone, reconstructing true 3D images of the breast. Several prototype WBCT units have been studied in clinical trials for diagnostic use. Lindfors et al reported a mean MGD of 6.0 mGy (range: 2.5–10.3 mGy) for an early prototype WBCT (23). Using a different prototype WBCT system, O’Connell et al reported a mean MGD of 8.2 mGy (range: 4–12.8 mGy) for WBCT, compared to a mean diagnostic mammography MGD of 6.5 mGy (range: 2.2 to 15 mGy), 26% higher for WBCT (24). One manufacturer’s system, the Konig cone-beam breast CT unit, has been approved for diagnostic applications as a supplement to two-view mammography. In its clinical trial data, the mean MGD for the Konig WBCT was 10.6 mGy (10% higher) compared with 9.6 mGy for diagnostic mammography doses on the same subjects (25).

BSGI and MBI Doses

Breast-specific gamma imaging and MBI rely on the injection of a radionuclide: ^99mTc-sestamibi. Label-recommended doses for breast imaging are 740–1110 MBq (20–30 mCi), the same dose recommended for single-day cardiac stress tests (26). Unlike x-ray imaging of the breasts, radionuclides expose all organs of the body, including breasts, to ionizing radiation. Because of the way ^99mTc-sestamibi is taken up and cleared by the body, highest organ doses are to the large intestine wall (40.0–55.5 mGy or mSv), small intestine wall (30 mGy), and kidneys, urinary bladder wall, and gallbladder wall (20 mGy each) for a 1110 MBq administered dose (26).

Within the last decade, new devices have been developed with more efficient dual-headed detectors built into compression paddles to enable using lower-dose ^99mTc-sestamibi administrations (27). These more efficient devices go by the name molecular breast imaging (MBI) and have demonstrated good clinical results at administered doses of 300 MBq (8 mCi) (28). A 2013 study found that for drawn doses of 150–300 MBq of ^99mTc-sestamibi, approximately 20% of the radionuclide was retained by the syringe and tubing (29). For BSGI drawn doses of 300 mBq, this means the received dose was 80% of the drawn dose, or 240 MBq (6.4 mCi) of ^99mTc-sestamibi. Table 2 shows estimated organ doses for both the label-recommended dose of 1110 MBq (30 mCi) (26) and also for a drawn dose of 300 MBq (with a received dose of 240 MBq [6.4 mCi]), assuming a 2-hour void.

PEM Doses

PEM uses a dedicated breast imaging device to place a woman’s breast under mild compression between two parallel paddles, each consisting of gamma ray detectors that register coincident back-to-back 512 keV gamma rays resulting from electron-positron annihilation within the breast (30). PEM requires the administration of fluorine-18 fluorodeoxyglucose (¹⁸F-FDG), the positron-emitting radionuclide used in whole-body positron emission tomography studies for the detection of metastatic cancer. Fluorodeoxyglucose, like glucose, is selectively taken up in solid tumors as well as lymph nodes due to its increased uptake and retention in metabolically active tissue. The label-recommended adult dose of ¹⁸F-FDG is 185–370 MBq (5–10 mCi) (31). Because FDG is taken up by the heart and excreted primarily through the urinary tract, the highest doses are to the bladder wall, heart wall, spleen, pancreas, and kidneys. Organ doses for 370 MBq (10 mCi) of ¹⁸F-FDG are given in Table 3.

Scientific basis for risk estimates

Risks of radiation-induced cancer incidence and deaths are based on long-term studies following subjects receiving sizable radiation doses. Investigators have monitored cancer incidence and mortality rates within each exposed study cohort and compared those rates to those in comparable unexposed or low-exposure cohorts. An important cohort is the group of 120 000 Japanese atomic bomb survivors from Hiroshima and Nagasaki who have been followed for over 60 years (the Life Span Study [LSS]) (32–36). Others involve patients exposed to diagnostic radiation for benign conditions (repeated fluoroscopy of tuberculosis patients and radiography of scoliosis patients) (37, 38) and therapeutic radiation for cancer (treatment of Hodgkin’s disease) (39, 40). On the basis of excess cancer incidence among these high-dose cohorts, estimates of radiation-induced cancer risks have been made by two major groups: the United States National Academy of Sciences’ Biologic Effects of Ionizing Radiation (BEIR) VII Group (35), which estimated radiation risks to the U.S. population, and the ICRP (11). These studies found a linear relationship between radiation dose and risk of radiation-induced solid cancers, including breast cancer, for estimated organ doses above 0.1 Gy (100 mGy). There is controversy over extrapolating the linear, no-threshold model to doses below 100 mGy (41, 42). Since all screening and diagnostic breast imaging tests involve organ doses well below this threshold, breast imaging tests are subject to this controversy.

A recently updated analysis of breast cancer risk from the LSS included 397 new breast cancer cases diagnosed since 1998, 75% of which were exposed before 20 years of age (43). Only 139 (9.5%) of the 1470 breast cancer cases in the full LSS analysis occurred in women over 40 years of age (43). These new data confirm a roughly linear relationship between breast cancer risk and radiation dose for doses above 250 mGy, with scattered data both above and below the linear trend line for breast radiation doses below 250 mGy (Figure 5) (43). The study confirmed that “exposure to radiation may be particularly carcinogenic when it occurs during sensitive periods in breast development, such as in utero, puberty and pregnancy, which are characterized by rapid proliferation of undifferentiated cells.” They found that “for a given age at menarche, exposure around the time of menarche results in the largest radiation effects.” In terms of estimated relative risk (relative to nonradiation-caused incidence of breast cancer), risks increased as exposure age approached menarche, then decreased as exposure age increased beyond menarche (Figure 6). Earlier age at menarche raises risks of radiation-induced breast cancer throughout a woman’s lifetime. Figure 7 from an earlier LSS publication, shows that the lifetime risk of radiation-caused breast cancer induction decreases steadily for higher ages at exposure (44). The low excess relative risk estimates and wide error bars for women ages 40 years and over reflect the limited number of radiation-caused breast cancer cases in this age group.

Risk Estimates

Risk estimates for the potential to cause cancer and cancer deaths can be made using either ICRP or BEIR VII (11, 35). Both use a linear no-threshold model of dose versus risk for both low and high doses of ionizing radiation. Using ICRP 103 weighting factors (0.12 for breast, as shown in Table 1), it is possible to compare the radiation dose of a given breast procedure to other diagnostic procedures or to natural background radiation using effective dose (defined earlier). For example, screening with modern DM delivers an average absorbed MGD of about 3.0 mGy. Converting MGD in units of absorbed dose (mGy) to equivalent dose (in mSv) for x-rays simply involves the multiplication of MGD by 1 mSv/mGy since the radiation weighting factor for x-rays is 1, so the equivalent dose to each breast is approximately 3.0 mSv. Converting equivalent dose to effective dose is done by multiplying equivalent dose by the dimensionless tissue weighting factor for breast tissue of 0.12, yielding an effective dose of 0.36 mSv. Table 4 compares effective doses of several different radiologic and nuclear medicine procedures, including two-view mammography (45).

Given that the average annual effective (whole body) dose from natural background radiation is 3.1 mSv, a digital screening mammography exam of an average-sized woman has the same risk of causing cancer as approximately 6 weeks of natural background radiation (3.1 mSv*6/52 = 0.36 mSv; see Table 4). ICRP’s Report 103 gives a risk estimate for detriment (cancer induction) of 0.041 per Sv (11), so for digital screening mammography, this would translate to a risk of cancer induction of (0.36 mSv)(1/1000 Sv/mSv)(0.041/Sv) = 1.5 per 1 000 000. The problem with using ICRP 103 estimates is that the risk of detriment is age-averaged (over the adult population) and gender-averaged (across males and females), and it is therefore not the most suitable for estimating cancer risk for breast imaging procedures.

A more acceptable method is to use BEIR VII estimates of radiation risk, which are age- and gender-specific and include estimates of both radiation-induced cancer incidence and mortality. These estimates do not take into account the recent LSS update on breast cancer risks (43). Table 5 shows BEIR VII estimates of lifetime attributable risks of cancer incidence and mortality specific to female breast exposure. The BEIR VII assumes a relatively high breast cancer mortality-to-incidence ratio of about 0.25. More recent Cancer Intervention and Surveillance Modeling Network models estimate an overall breast cancer mortality-to-incidence ratio of 0.19 for women receiving modern therapy but not screening (46). These more current estimates of cancer mortality rates in the absence of screening would lower BEIR VII mortality estimates by about 25%. Mortality estimates would be lowered even further if screening were assumed (47, 48).

The BEIR VII can be used to estimate the lifetime risk of cancer incidence and mortality for a single screening exam as a function of age at exposure (35). Table 6 presents these estimates for screening with DM, DBT, CEDM, whole-breast CT, BSGI/MBI, and PEM as a function of age at exposure for the designated doses of each procedure. The first four listings are for risks of breast cancer or breast cancer-caused deaths, since these procedures expose only breast tissue to significant amounts of radiation. The latter listings (BSGI, MBI, and PEM) include risks of cancer or cancer-caused deaths to all body organs (see Tables 2 and 3).

Benefit-to-Risk Estimates

Estimates of benefit-to-radiation-risk have been made by a number of authors (47–53). Only a few estimates are summarized here. Yaffe and Mainprize estimated the benefit-to-radiation-risk of screening mammography assuming a dose of 3.7 mGy to both breasts, a screening regimen similar to that recommended by the American Cancer Society (annual screening from ages 40–55 years followed by biennial screening from ages 55–74 years), and assuming a 24% mortality reduction from screening (47). They estimated risks in terms of breast cancer deaths due to radiation from screening (10.7 per 100 000 women) and benefits in terms of breast cancer deaths averted due to screening (497 per 100 000 women), yielding a benefit-to-radiation-risk estimate of 47:1. In terms of life-years, the risk of life-years lost due to radiation from screening was 136.4 per 100 000 and the life-years gained from screening was 10 670 per 100 000, yielding a benefit-to-radiation-risk of 78:1. The greater benefit-to-radiation-risk in terms of life-years occurs because most deaths from breast cancer occur within a decade of diagnosis, while radiation-caused breast cancer deaths are most likely to occur two to three decades after exposure. A recent modeling study by Yaffe et al estimated a 42% mortality reduction from annual screening ages 40–49 and biennial screening ages 50–74 (48). If this mortality reduction value is used, the already favorable benefit-to-radiation-risk ratio increases by 75%.

The mortality benefit from screening modalities other than mammography is not known, but it is likely that modalities with higher cancer detection rates (CDRs) in the same population will yield greater mortality reduction. Studies have shown that DBT has a 29%–89% higher CDR than DM in screening populations (54–63). In estimating benefit-to-radiation-risks for DBT or other modalities, their increase in dose relative to DM, in addition to their higher CDR (as a surrogate for mortality reduction), should be taken into account.

Benefit-to-radiation-risk is one way to compare new or proposed screening modalities to screening mammography. For example, Hendrick and Tredennick compared the benefit-to-radiation-risk of reduced dose MBI (300 MBq or 8mCi drawn dose, 240 MBq or 6.4 mCi received dose) to DM (64). While reduced dose MBI was shown to have a much higher CDR than DM (for prevalent MBI screens), the benefit-to-radiation-risk ratios were higher for DM (by a factor of 2.6 for screening ages 40–49 years up to a factor of 10 for screening ages 70–79 years) than for reduced-dose MBI. Recent work suggests that MBI may be capable of comparable breast cancer detection at approximately half this reduced dose (150 MBq or 4 mCi) with new image processing algorithms (65). Brown and Covington have recently updated benefit-to-risk estimates for DM and DBT with various acquisition strategies and reduced-dose MBI for different compressed breast thicknesses (66).

Conclusion

Most current x-ray–based screening modalities, including DM and DBT, have small to negligible risks of radiation-caused cancer incidence or deaths. Prospective screening modalities such as CEM have similar small cancer risks. Prospective screening modalities that involve radionuclide injection, such as BSGI, MBI, and PEM, have significantly higher cancer risks unless efficient detection systems and reduced administered doses are used. Benefit-to-radiation-risk estimates are highly favorable for screening mammography, with DM or other modalities having equal or greater cancer detection rates and similarly low radiation doses.

Funding

None declared.

Conflict of interest statement

Dr Hendrick is a consultant to GE Healthcare on new technologies in breast imaging.

References