Pearls and Pitfalls of Interpretation of Automated Breast US
Abstract
Dense breast tissue is an independent risk factor for breast cancer and reduces the sensitivity of mammography. Patients with dense breast tissue are more likely to present with interval cancers and higher-stage disease. Successful breast cancer screening outcomes rely on detection of early-stage breast cancers; therefore, several supplemental screening modalities have been developed to improve cancer detection in dense breast tissue. US is the most widely used supplemental screening modality worldwide and has been proven to demonstrate additional mammographically occult cancers that are predominantly invasive and node negative. According to the American College of Radiology, intermediate-risk women with dense breast tissue may benefit from adjunctive screening US due to the limitations of mammography. Several studies have demonstrated handheld US (HHUS) and automated breast US (AUS) to be comparable in the screening setting. The advantages of AUS over HHUS include lack of operator dependence and a formal training requirement, image reproducibility, and ability for temporal comparison. However, AUS exhibits unique features that can result in high false-positive rates and long interpretation times for new users. Familiarity with the common appearance of benign mammographic findings and artifacts, technical challenges, and unique AUS features is essential for fast, efficient, and accurate interpretation. The goals of this article are to (a) examine the role of AUS as a supplemental screening modality and (b) review the pearls and pitfalls of AUS interpretation.
©RSNA, 2023
Quiz questions for this article are available in the supplemental material.
Introduction
Mammography continues to be considered the standard of reference for breast cancer screening due to its proven mortality benefit. However, mammography has significant limitations in patients with dense breast tissue (1–3). Breast density categories are defined by the Breast Imaging Reporting and Data System (BI-RADS) atlas of the American College of Radiology (ACR). Categories C (heterogeneously dense) and D (extremely dense) are considered dense categories (4).
Dense breast tissue is common and has two significant implications. First, dense breast tissue reduces the sensitivity of mammography by up to 30% by masking breast cancers (1–3,5,6) (Fig 1). Second, breast density is an independent risk factor for breast cancer (7), with an approximately fourfold increased risk for patients with extremely dense tissue compared with those with fatty tissue (8). The relative risk is roughly 1.5-fold greater when comparing individuals with heterogeneously dense breasts to those with scattered fibroglandular density, the two most common density categories (9,10).
Figure 1. Dense breast tissue in a 55-year-old woman. Mediolateral oblique (A) and craniocaudal (B) standard screening synthesized two-dimensional mammograms of the right breast show dense tissue (arrows) in the subareolar breast. The fifth edition of BI-RADS emphasizes the masking effect and states that a mammogram should be labeled “heterogeneously dense” if there are regions of sufficient density to obscure small masses. Although the overall amount of fibroglandular tissue in this patient is less than 50% of the breast, the tissue in the subareolar breast was considered sufficiently dense to obscure small masses. Therefore, the BI-RADS classification for this mammogram is category C or heterogeneously dense.
The importance of breast density is underscored by new federal legislation, released in early 2023, which mandates breast density notification under an update to the Mammography Quality Standards Act of 1992. Additionally, there is a trend toward personalized breast cancer screening regimens based on individual risk versus traditional population risk. Breast density is now incorporated into the Tyrer-Cuzick, Breast Cancer Surveillance Consortium, and CanRisk models (9,11,12).
Detection of early-stage cancers is essential for successful outcomes of breast cancer screening (13). Women with dense breast tissue are more likely to present with interval cancers, larger tumor size, and positive lymph nodes at diagnosis (5,14–16). Therefore, adjunctive screening modalities are used to detect mammographically occult cancers. Current options for supplemental screening include US, MRI, contrast-enhanced mammography (CEM), and molecular breast imaging (MBI).
Both handheld US (HHUS) and automated US (AUS) have been shown to improve detection rates of invasive cancer in women with dense breast tissue (17,18). US is widely available, incurs no ionizing radiation dose, and does not require intravenous access for contrast material or radiotracer administration. According to a survey of ACR lead interpreting physicians in 2017, 68.4% of radiology facilities offer supplemental screening, with US ranking as the most common modality (53%) (19). Disadvantages of US include a high false-positive rate and lower cancer detection rate (CDR) compared with those of other screening methods, such as MRI and CEM (10,20).
The updated ACR screening recommendations for women at elevated risk indicate that US should be considered in women who qualify for MRI but cannot undergo itAdditionally, the European Society of Breast Imaging guidelines recommend HHUS or AUS after a negative mammographic study in women at average or intermediate risk with dense breasts (23).
The purpose of this article is twofold. First, we provide an overview of AUS, including comparison with HHUS, discussion of AUS image acquisition, and recent technology updates. Second, we detail the major pearls and pitfalls of AUS interpretation and familiarize the radiologist with advanced three-dimensional software.
Performance of US in Supplemental Screening
US has been used for breast cancer screening for over 20 years. Early efforts to demonstrate the effectiveness of breast US screening before the mid-1990s failed, largely due to limited resolution. One of the first large-scale trials that evaluated breast US for screening, by Kolb et al (24) in 1998, demonstrated an incremental CDR of 3 per 1000.
The ACR Imaging Network (ACRIN) 6666 multicenter prospective randomized controlled trial found that HHUS in women at elevated risk allowed detection of an additional 1.1–7.2 cancers per 1000 women (17,25). The first-generation AUS device was approved by the U.S. Food and Drug Administration (FDA) in 2012. The SomoInsight multicenter study was the first to evaluate AUS for breast cancer screening in patients with dense breast tissue and reported incremental detection of 1.9 cancers per 1000 women (18).
Screening US has the potential to reduce mortality and morbidity from breast cancer by its proven ability to demonstrate mammographically occult, invasive, node-negative cancers (26). However, the major reported disadvantages of screening breast US are the undesired high recall rate and low positive predictive value (PPV) (Table). The high recall rate (13.5%) and low PPV1 (PPV based on abnormal findings at screening) (1.4%) in the SomoInsight study have been attributed to use of the first-generation transducer, radiologist unfamiliarity with the modality, and minimal training for both technologists and radiologists (18).
 |
A study of over 5500 patients in 2014 reported a lower recall rate for AUS (2.57%) than for HHUS (3.57%) while maintaining high diagnostic accuracy and a CDR of 3.8 per 1000 for AUS (27). In our single-institution review of 8891 AUS examinations performed between 2013 and 2020, we found that the recall rate and CDR can be maintained at acceptable performance levels (recall rate, 5%; CDR, 2.2 per 1000) even in the setting of a screening program fully transitioned to digital breast tomosynthesis (DBT) (28). A recent multicenter prospective trial that compared the addition of HHUS to DBT reported a modest incremental CDR for HHUS of 1.1 per 1000, with 23 of 126 cancers visible at US although missed at screening by the technologist (29). The authors of this trial suggested that AUS may overcome the low cancer yield of HHUS due to the standardized automated technique and interpretation by physicians. While AUS has limitations, many can be circumvented with appropriate training and knowledge of the software and imaging features.
Comparison of HHUS and AUS
Automated techniques for breast US screening have gained popularity due to several advantages.
AUS overcomes major limitations of HHUS, including operator dependence, image reproducibility, lack of objective temporal comparison, and a potentially more challenging audit, depending on HHUS workflowsThere is no mandated operator training for AUS in the United States, and requirements vary worldwide. Therefore, sonographers, mammography technologists, and medical assistants can perform AUS examinations, which improves workflow efficiency and resource utilization (10). Barr et al (31) reported no difference in lesion characterization when AUS images were acquired by a sonographer or mammography technologist. ACR-accredited Centers of Excellence do require that a certified technologist perform AUS, similar to HHUS. While AUS is more standardized and reproducible than HHUS, variations in operator skill and technique do exist, and interpreting radiologists should closely assess image quality and provide feedback to improve performance, as with other imaging modalities.
The ability of AUS to provide visibility of all breast tissue, rather than use of a single HHUS image, allows the reader to assess the symmetry, bilaterality, and multiplicity of findings, which can reduce false positives (30). AUS also allows easier decoupling of image acquisition and examination interpretation, and examinations are typically assigned a traditional screening BI-RADS category 0, 1, or 2 assessment. Recalls are then performed with a separate HHUS study, allowing standardized auditing comparable to that of mammography.
In contrast, HHUS commonly serves as both a screening and diagnostic examination in the same setting, with a single overall BI-RADS assessment. Therefore, auditing performance measures such as recall rate and PPV can be more difficult to calculate. The patient convenience of a single US study is a significant advantage of HHUS, as is its capability for Doppler and elastography evaluation. Real-time evaluation of more difficult to assess locations—such as the axilla, subareolar region, and far medial or lateral tissue—is another potential advantage of HHUS, although routine axillary screening has not been shown to be beneficial (32). While HHUS can also be interpreted off-line, allowing a standard screening audit, this practice negates the HHUS benefit of the single US examination.
Technologist scanning time for AUS is approximately 12 minutes for a bilateral examination (33,34). This is comparable to reported times for HHUS, which are variable, ranging from less than 5 minutes in early studies (2,24) to 13–19 minutes in the ACR Imaging Network 6666 trial for physicians (25) and 10 minutes for sonographers (33,35). Experienced practices likely have reduced variability in performance times for both HHUS and AUS. Reported acquisition times for HHUS rarely account for routine physician overscanning of a positive study, which can significantly lengthen the examination and interpretation times.
The early literature reported lengthy interpretation times for AUS; however, these studies often included a significant number of diagnostic examinations. Recently published studies of screening populations have demonstrated that AUS interpretation times decrease significantly with experience and that the study can be consistently interpreted in less than 4 minutes (36–38). Computer-aided detection (CAD) software has the potential to further improve interpretation times, with one study demonstrating a decrease in reader interpretation time by 33%, from 3.33 to 2.24 minutes per case, without impacting diagnostic performance (38).
Many studies have confirmed the overall comparable performance of HHUS and AUS for lesion identification, lesion characterization, reader agreement, and diagnostic performance. In a study of 1293 women, Zhang et al (39) found similar CDRs for HHUS and AUS as well as a strong agreement rate of 94% (κ = 0.860). Even higher agreement was demonstrated by Vourtsis and Kachulis (40), who found HHUS and AUS agreement of 99.8% (κ = 0.994) and nearly 100% interobserver agreement for double-blinded AUS interpretation.
In a prospective study of 411 lesions by An et al (41), the image quality of AUS was identical or superior to that of HHUS in 97% of cases regarding lesion coverage, lesion conspicuity, and artifacts. Similarly, Kuzmiak et al (42) demonstrated no significant difference in visibility of suspicious lesions when HHUS and AUS were compared. Interestingly, in that study, there was a statistically significant (P < .001) increase in reader confidence for assessing lesion shape and margin at AUS, thought to be secondary to the three-dimensional nature of AUS and the availability of the coronal plane, a unique advantage of AUS over HHUS.
A meta-analysis published in 2019 by Wang et al (43) found overall similar performance of HHUS and AUS, with all nine included studies using a first-generation AUS device. Lastly, multiple prior studies outlined in a recent review article by Berg and Vourtsis (10) demonstrated similar CDR, recall rate, and PPV3 (PPV based on results of biopsy) between HHUS and AUS.
AUS Image Acquisition and Technology Updates
As with mammography, consistent standardized positioning during AUS is essential. The patient is scanned in the supine or supine oblique position, with the breast equally sloped on all sides and a positioning wedge preferred when imaging the outer breast (44). The arm is lifted overhead, and the elbow and shoulder can be supported with a towel or pillow to maximize patient comfort and reduce motion (44). The patient's head should be turned to the opposite breast, and talking should be kept to a minimum.
A standard AUS examination includes a minimum of three data sets per breast, labeled based on probe position during acquisition: anteroposterior (AP), lateral (LAT), and medial (MED) (45). The AP view should be centered over the nipple and include the 6-o'clock axis and inframammary fold. The LAT view should focus on the lateral tissue, including both the axillary tail and nipple. The MED view should include medial and inferior tissue, as well as the inframammary fold. Additional views may be necessary to include all glandular tissue, depending on individual breast size and composition (10,44). The location of the nipple and optimal scan depth are manually selected by the operator.
The U.S. Food and Drug Administration (FDA) mandates that the AUS vendor provide an initial 8-hour training course for individuals operating the device at the time of installation. Often, additional technologists are trained at the institution after initial implementation, and it can be advantageous to have a standardized review process documenting mastery of skills. At our institution, new technologists performing AUS are evaluated on 10 patients with an examination competency checklist (Fig S1).
The FDA does not mandate AUS-specific training or continuing medical education (CME) for interpreting radiologists, although the vendor is required to provide 8 hours of training at installation. Training typically includes review of workstation knobology and didactic lectures as well as instructor-directed and self-guided case reviews (46,47). A dedicated workstation or software upgrade to view AUS images on a standard picture archiving and communication system (PACS) can be used for interpretation.
Adequate knowledge of the unique workstation tools is essential for rapid accurate interpretation of images, as recall rate and interpretation time have been shown to improve with experience and proper trainingMost of the literature evaluating the outcomes data and clinical use of AUS uses earlier-generation software and probes, including a flat-head transducer known to produce more artifacts at the edges of acquisition due to poor contact. The second-generation curvilinear transducer follows the natural contour of the breast and eliminates many of the previously experienced artifacts by improving compression and patient comfort (Fig 2). The images in this article were obtained using the Invenia ABUS 2.0 system (GE HealthCare), which includes the second-generation curved transducer.
Figure 2. Comparison of transverse AUS images from the first-generation flat-head transducer (A) and the second-generation curvilinear transducer (B) in the same patient. Note the improved image resolution and inclusion of peripheral tissues at the edges of the image due to better contact when the curved transducer is used. Artifactual shadowing is also reduced due to improved compression at the periphery of the images.
Pearls of AUS Interpretation
Correlate with Current and Prior Imaging
Whether performed concurrently or separately, correlation with mammography is essential for characterization of AUS findings and may reduce recall rates. There are many characteristically benign mammographic findings that can demonstrate a suspicious appearance at US, including popcornlike calcifications associated with fibroadenomas (Fig 3), dystrophic calcifications, and calcifications associated with fat necrosis. Surgical scars and noncalcified fat necrosis also require correlation with mammography (Fig 4). Additionally, 2-year mammographic stability of solid, oval, circumscribed masses obviates the need for targeted US or short-interval imaging follow-up (Fig 5) (4).
Figure 3. Benign mass with calcifications in a 48-year-old woman who presented for baseline screening AUS. (A) Coronal (top) and transverse (bottom) images from the AP (left) and LAT (right) data sets show an irregular hypoechoic mass (arrow) with posterior acoustic shadowing at the 11:30 position, 7 cm from the nipple. These features are typically associated with a suspicious finding, which would necessitate recall. (B) However, comparison with a mediolateral oblique synthesized two-dimensional screening mammogram from earlier in the year shows characteristically benign coarse calcifications (arrow) in the same location, obviating the need for further evaluation. It is common for larger calcifications to cause posterior acoustic shadowing at US.
Figure 4. Benign fat necrosis in a 55-year-old woman with a history of right breast conservation therapy who presented for screening AUS. (A) Coronal (top) and transverse (bottom) images from the AP (left) and MED (right) data sets show a suspicious, irregular, hypoechoic mass (arrow) with indistinct margins and not parallel orientation in the medial right breast. (B) A single 1-mm-thick tomosynthesis section from mammography performed the same day shows a rim-calcified mass with internal fat density (yellow arrow) in the same location, consistent with benign fat necrosis. Additional areas of benign fat necrosis within the lumpectomy bed were also noted (blue arrow), consistent with the patient's history of surgery and radiation therapy in this breast. Skin thickening is also noted on both the AUS images and mammogram. No further imaging was needed, and the patient was given a BI-RADS category 2 assessment with a recommendation to return to routine screening.
Figure 5. Fibroadenoma in a 47-year-old woman who presented for screening AUS. (A) Coronal (top) and transverse (bottom) images from the LAT data set show a heterogeneous, oval, circumscribed mass (arrow) with parallel orientation and posterior enhancement. (B) Comparison with corresponding images from a prior AUS examination shows stability of the mass (arrow). (C) Correlation with a single mediolateral oblique tomosynthesis image shows a stable, oval, circumscribed mass (arrow). The mass contains a tissue marker clip, which was not visible at US. The medical chart confirmed a prior biopsy at this location, with pathologic analysis yielding fibroadenoma.
There is potential for reduction of recall rates at screening mammography when AUS is interpreted as a combination examination with screening mammography performed in the same setting (43). This has been demonstrated for asymmetries caused by superimposition of normal fibroglandular tissue and masses representing benign cysts. Figure 6 highlights a common scenario of same-day AUS preventing a screening mammography recall.
Figure 6. Cyst in a 51-year-old woman who presented for screening AUS. (A) Single mediolateral oblique 1-mm-thick tomosynthesis section from screening mammography shows a new, oval, circumscribed, equal-density mass (arrow) in the upper right breast. Same-day AUS was performed and the results were interpreted concurrently, allowing the reader to correlate findings. (B) Coronal (top) and transverse (bottom) images from the AP (left) and MED (right) data sets of the right breast show an oval, circumscribed, anechoic mass (arrow) with parallel orientation at the 12-o'clock position, 3 cm from the nipple. This mass corresponds in size, location, and shape to the mass seen on the mammogram. (C) Additional coronal image from the LAT data set captured posterior to the mass shows posterior enhancement (arrows), confirming that the finding is a cyst. Given the benign characteristics at AUS, the reader was able to avoid recall for the mammographic finding.
Standardized AUS image acquisition is highly reproducible and allows direct temporal comparison when prior AUS studies are available. Prior comparisons can be used not only to prove stability and decrease recall rates (Fig S2) but also to confirm new subtle findings, which may otherwise be overlooked (Fig 7) or thought to be benign (Fig 8).
![Invasive ductal carcinoma (IDC) in a 63-year-old woman with a history of breast conservation therapy (ipsilateral lumpectomy and radiation therapy) 3 years earlier who presented for screening AUS. (A) Coronal (top) and transverse (bottom) AUS images from the LAT data set of the right breast show a 0.3-cm, irregular, hypoechoic mass (arrow) near the surgical scar (seen inferior to the mass). Given the postsurgical appearance of the breast and the small size of the mass, this finding could easily have been overlooked. (B) However, corresponding images from a prior AUS examination show that this was a new finding (arrow = normal tissue in this area), thus prompting additional evaluation with HHUS and ultimately US-guided biopsy, yielding a 0.4-cm grade 2 IDC (estrogen receptor [ER] positive, progesterone receptor [PR] positive, human epidermal growth factor receptor 2 [HER2] negative). Results of final surgical pathologic analysis were negative for nodal involvement.](/cms/10.1148/rg.230023/asset/images/medium/rg.230023.fig7.gif "Invasive ductal carcinoma (IDC) in a 63-year-old woman with a history of breast conservation therapy (ipsilateral lumpectomy and radiation therapy) 3 years earlier who presented for screening AUS. (A) Coronal (top) and transverse (bottom) AUS images from the LAT data set of the right breast show a 0.3-cm, irregular, hypoechoic mass (arrow) near the surgical scar (seen inferior to the mass). Given the postsurgical appearance of the breast and the small size of the mass, this finding could easily have been overlooked. (B) However, corresponding images from a prior AUS examination show that this was a new finding (arrow = normal tissue in this area), thus prompting additional evaluation with HHUS and ultimately US-guided biopsy, yielding a 0.4-cm grade 2 IDC (estrogen receptor [ER] positive, progesterone receptor [PR] positive, human epidermal growth factor receptor 2 [HER2] negative). Results of final surgical pathologic analysis were negative for nodal involvement.")
Figure 7. Invasive ductal carcinoma (IDC) in a 63-year-old woman with a history of breast conservation therapy (ipsilateral lumpectomy and radiation therapy) 3 years earlier who presented for screening AUS. (A) Coronal (top) and transverse (bottom) AUS images from the LAT data set of the right breast show a 0.3-cm, irregular, hypoechoic mass (arrow) near the surgical scar (seen inferior to the mass). Given the postsurgical appearance of the breast and the small size of the mass, this finding could easily have been overlooked. (B) However, corresponding images from a prior AUS examination show that this was a new finding (arrow = normal tissue in this area), thus prompting additional evaluation with HHUS and ultimately US-guided biopsy, yielding a 0.4-cm grade 2 IDC (estrogen receptor [ER] positive, progesterone receptor [PR] positive, human epidermal growth factor receptor 2 [HER2] negative). Results of final surgical pathologic analysis were negative for nodal involvement.
Figure 8. IDC in a 68-year-old woman with a history of left breast conservation therapy 2 years earlier who presented for screening AUS. Coronal (top) and transverse (bottom) AUS images from the AP (left) and LAT (right) data sets of the left breast show a new, nearly anechoic, irregular mass with posterior enhancement (solid arrow) inferior and medial to the lumpectomy bed (open arrow). This appeared similar to a seroma at the lumpectomy site (open arrow), which had been visualized previously and confirmed at both mammography and HHUS as a benign finding. The mass inferomedial to the seroma at AUS was initially suspected to be a postsurgical seroma, given the proximity to the lumpectomy site. However, correlation with the prior AUS study proved this to be a new finding. Further evaluation with HHUS and US-guided biopsy yielded a recurrent, 1.2-cm, grade 3 IDC with metaplastic features (ER negative, PR negative, HER2 negative), which was negative for nodal involvement at final surgical pathologic analysis.
Know the Patient's History
Knowledge of the patient's clinical, surgical, and breast-specific history is invaluable for interpretation. In our experience, it is common for a baseline AUS examination to depict a suspicious finding that has previously been evaluated and/or biopsied. Correlation with the patient's history and comparison imaging can prove that a finding is stable and warrants no further evaluation (Fig S3). While many patients undergoing supplemental AUS have a history of excisional biopsy or breast conservation therapy, occasionally non–breast-related surgery can result in findings at AUS (Fig 9).
Figure 9. Benign findings in a 41-year-old woman who presented for screening AUS. Coronal (top) and transverse (bottom) images from the AP (left) and LAT (right) data sets of the right breast are shown. Both coronal images show a suspicious hypoechoic mass (arrow) at the 6-o'clock position, which was new from prior AUS studies. However, use of the software rotation tool in the transverse image from the LAT data set (bottom right) shows that this finding is linear, extending from the skin surface to the chest wall in a diagonal fashion, inconsistent with most malignant processes arising in the breast. Careful review of the medical chart and correlation with the surgical history revealed that this finding corresponded to a thoracoscopic port site from recent mitral valve repair.
The clinical history is critical when evaluating the axilla, especially in the current era of COVID-19 vaccination. The axilla is often partially included in the field of view of the LAT data set, and knowledge of the patient's vaccination status (via either the medical record or an intake form) can help distinguish pathologic (Fig 10) from reactive (Fig 11) lymphadenopathy. Knowledge of systemic diseases or medications that may result in bilateral lymphadenopathy—such as granulomatous disease, autoimmune disease, or lymphoma—can also help reduce recall rates, as these should be categorized as BI-RADS category 2 (Fig S4).
Figure 10. Bilateral breast malignancy and right axillary nodal metastasis in a 55-year-old woman who presented for screening AUS. (A) Transverse image from the right LAT data set initially shows an enlarged lymph node (arrow) with cortical thickening and hilar effacement in the axilla. The patient endorsed no recent history of vaccination or systemic causes of lymphadenopathy. (B) Additionally, coronal (top) and transverse (bottom) images from the AP (left) and MED (right) data sets show an irregular hypoechoic mass (arrow) in the 12:30 position of the right breast. (C) Coronal (top) and transverse (bottom) images from the LAT data set of the left breast show an irregular hypoechoic mass (arrow). All findings were recalled for additional HHUS. Ultimately, the patient was diagnosed with a 5-cm right breast grade 3 IDC (ER negative, PR negative, HER2 positive) with metastatic lymphadenopathy (one of six positive nodes at final surgical pathologic analysis), as well as a 0.8-cm left breast grade 1 IDC (ER positive, PR positive, HER2 negative).
Figure 11. Benign lymph nodes in a 58-year-old woman who presented for screening AUS. Transverse image from the LAT data set shows multiple left axillary lymph nodes (arrows) with uniformly increased cortical thickness (>0.3 cm). While this might have prompted recall, review of the medical chart indicated recent left arm COVID-19 vaccination. This explains the imaging findings; the examination was assessed as BI-RADS category 2, benign.
Build Confidence Quickly
When AUS is introduced as a new screening tool in a breast imaging program, it is helpful for users to gain experience as quickly as possible. One strategy is to perform a limited AUS examination (including one or two data sets in the breast of interest) on patients undergoing US-guided biopsy at no cost to the patient. Radiologists new to interpretation of AUS will then have comparison HHUS images as well as pathologic findings to rapidly build confidence in interpretation. When AUS was introduced at our facility in 2009, we scanned approximately 50 patients in this fashion.
Visualize Lesions in at Least Two Data Sets
In a properly positioned AUS examination, the majority of nonmarginal breast tissue should be visualized in at least two data sets. This allows improved discrimination between real findings and artifact, as well as multiple-view assessment of morphologic features. In one study of nearly 8900 patients, 85% of the mammographically occult AUS-detected cancers were visualized in two or more data sets (28). When a finding is seen in multiple data sets, this eliminates artifact due to poor compression or positioning as the cause. Mild variations in positioning between data sets may manifest as potential lesions not appearing in the same location, although true findings should be within an approximately 1-cm radius of one another.
Interpreting physicians can use the “volume sync” feature to unlink data sets and reposition transverse images to display similar tissue planes and confirm positive findings or dismiss artifacts. Tissue landmarks, such as fat lobules or cysts, can also be used to confirm location. Visualization of a finding in multiple data sets also allows the reader to better assess morphologic features, as benign masses will sometimes appear more suspicious in a data set with decreased compression (Fig 12). One caveat is that findings in the peripheral portions of the breast will often not be in the field of view of more than one data set (Fig 13).
Figure 12. Cyst in a 43-year-old woman who presented for screening AUS. Coronal (top) and transverse (bottom) images from the AP data set (left) show an irregular hypoechoic mass with indistinct margins (yellow arrow) in the superior lateral right breast. Note the far lateral location of the mass on the AP view, where there is decreased compression. More central positioning and improved compression of the lateral breast on the coronal (top) and transverse (bottom) images from the LAT data set (right) show the mass to be an oval, circumscribed, anechoic mass with parallel orientation (blue arrow), findings consistent with a cyst. The patient had multiple additional cysts bilaterally. This examination was given a BI-RADS category 2 assessment.
Figure 13. IDC in a 65-year-old woman who presented for screening AUS. Coronal (top) and transverse (bottom) images from the LAT data set (right) show a suspicious, irregular, hypoechoic mass with angular margins (yellow arrow) in the 2:30 position of the left breast, 7.5 cm from the nipple. When the cursor is placed over the finding in the LAT data set, the positioning is noted to be outside the field of view (blue arrow) on coronal (top) and transverse (bottom) images from the AP data set (left). Despite being visualized in only one data set, the mass was still recalled for additional imaging, as the mass was not expected to be seen in additional data sets and remained suspicious. Ultimately, US-guided biopsy was performed. Final surgical pathologic analysis revealed a 1.2-cm grade 2 IDC (ER positive, PR positive, HER2 negative) with intermediate-grade ductal carcinoma in situ (DCIS), which was negative for nodal involvement.
Use the Unique Coronal Plane
Interpretation of standard breast US in the transverse plane is familiar to most readers. However, the coronal plane is a unique feature of AUS, which offers distinct imaging characteristics that can aid readers in distinguishing benign versus malignant findings. In addition to the three transverse data sets acquired during an AUS examination, the data are reconstructed to be viewed in two additional orthogonal planes simultaneously: sagittal and coronal (Fig 14).
Figure 14. IDC in a 69-year-old woman who presented for screening AUS. The AP data set is shown in three orthogonal planes. A suspicious, irregular, hypoechoic mass with spiculated and angular margins (arrow) is seen in the upper medial left breast on coronal (top left), transverse (bottom left), and sagittal (top right) images. This finding was recalled for HHUS and underwent US-guided biopsy. Final surgical pathologic analysis revealed a 1.0-cm grade 1 IDC (ER positive, PR positive, HER2 negative), which was negative for nodal involvement.
The retraction phenomenon described in the literature is simply a manifestation of architectural distortion seen at mammography and refers to a stellate pattern surrounding a mass on the coronal view. This is thought to be secondary to the desmoplastic reaction created by malignancy on the surrounding tissues (Fig 15) (33,49). Architectural distortion is characterized as an associated feature in the BI-RADS US lexicon (4), and this finding has been cited as an advantage over HHUS (40).
Figure 15. IDC in a 57-year-old woman who presented for screening AUS. Coronal (top) and transverse (bottom) images from the AP (left) and LAT (right) data sets show an irregular hypoechoic mass (arrow), which stands out in the coronal plane due to the associated retraction phenomenon. Note that the mass is less conspicuous in the transverse plane, with parallel orientation and far posterior depth location. This finding was recalled for HHUS, and US-guided biopsy was performed. Final surgical pathologic analysis revealed a 1.3-cm grade 1 IDC (ER positive, PR positive, HER2 positive), with micrometastatic carcinoma measuring 0.2 cm in one of 31 lymph nodes.
The retraction phenomenon has been well described in the literature (33,50–52) and has been shown to be the strongest independent feature of malignancy at AUS, with one study by Chen et al (51) demonstrating a PPV of nearly 100%. However, this tissue fibrotic response is generally associated with lower-grade malignancies, so the retraction phenomenon is not typical for high-grade cancers, such as the triple-negative subtype.
Use of the AUS coronal plane for initial evaluation of the breasts allows fast and efficient diagnosis. Several studies have demonstrated that use of the coronal plane not only improves performance but also shortens interpretation timeUse of the coronal plane as a stand-alone tool for AUS evaluation is not currently accepted as a standard method of interpretation, and complete review of transverse images is recommended for optimal interpretation (54).
Use Unique Software for Lesion Characterization
Acquisition of images in a volumetric three-dimensional data set allows interrogation of features exclusive to the automated technique. Each data set can be reconstructed into various orthogonal radial and antiradial planes and compared in a variety of formats. Specific findings can be rotated 360° using a software tool, allowing more thorough evaluation. This technique proves useful when trying to determine whether a finding is a true space-occupying lesion, particularly when it is visualized in only one data set (Fig 16). Peripheral portions of fat lobules often appear masslike and edges of the Cooper ligaments commonly cause shadowing, both of which can often be resolved with this AUS software feature (Fig 17).
Figure 16. Normal results of screening AUS in a 73-year-old woman. (A) Coronal (top) and transverse (bottom) images from the AP data set (left) show a possible irregular hypoechoic mass (yellow arrow) in the lower outer right breast. However, coronal (top) and transverse (bottom) images from the LAT data set (right) do not show a mass (blue arrow). (B) Further interrogation with the software rotation tool on coronal (top) and transverse (bottom) images from the AP data set shows that the finding (arrow) represents normal fibroglandular tissue.
Figure 17. Normal results of screening AUS in a 45-year-old woman. (A) On images from the AP data set, a possible spiculated mass (arrow) is seen on the coronal image (top), although it is difficult to confirm whether this is a true mass or artifactual shadowing on the transverse view (bottom). (B) Use of the software rotation tool on corresponding images shows that both shadowing from an overlying Cooper ligament on the edge of a fat lobule (blue arrow) and an underlying benign entrapped fat lobule (yellow arrow) contribute to the original finding, with no true suspicious mass to warrant recall. Results of 1-year follow-up AUS were negative.
As a result of the volumetric acquisition of data, users are able to customize the thickness and spacing of displayed images. Knowledge of these settings can optimize interpretation. Images can be adjusted from 0.5 mm to 3 mm in thickness and from 0.5 mm to 2 mm in spacing. Coronal images are most commonly set at 2-mm thickness with spacing of 0–1 mm, which optimizes display of the retraction phenomenon and disease arising from the terminal ductal–lobular unit (44,56). Many experienced readers use a size threshold for recall of approximately 5–6 mm. Therefore, when reviewing the coronal plane and visualizing a lesion on three consecutive coronal images (assuming 2-mm thickness with no overlap), the threshold for callback is met.
Trust Traditional BI-RADS Lesion Characterization
BI-RADS descriptors were originally developed for HHUS, although they are applicable and accurate for use with AUS. BI-RADS descriptors including irregular shape, not parallel orientation, not circumscribed margins (angular, spiculated, or microlobulated), and associated calcifications have all been shown to have a statistically significant association with malignant masses (50). We confirmed these findings in our own practice when evaluating 20 mammographically occult AUS-detected cancers over a 7-year period. Our results demonstrated that 100% of the AUS-detected cancers exhibited two or more of these suspicious BI-RADS features (28). In our review, no biopsy-proven malignancies had a combination of oval shape, circumscribed margins, and parallel orientation, findings typically appropriate for a BI-RADS category 3 assessment (28).
Use Signs and Artifacts to Your Advantage
The unique nature of the wide-format US probe and acquisition of each data set in a single recorded sweep may result in unique reconstruction artifacts and features related to AUS, of which readers should be aware.
Skip artifact is identified on coronal and sagittal views. This occurs when there is nonlinear transducer movement during image acquisition, manifesting as a transverse hypoechoic or hyperechoic line on the coronal images (48). Skip artifact should prompt the reader to investigate the underlying tissue in more detail, particularly on the originally scanned transverse images, which are unaffected by the artifact, and correlate with available images from mammography. Skip artifact can be due to benign (Fig 18) or malignant (Fig 19) findings, including a rib in a thin patient or sometimes focally dense breast tissue. Recently, a retrospective study of 457 pathologically confirmed lesions demonstrated skip artifact to be a predictor of malignant nonmass lesions (57).
Figure 18. Skip artifact caused by a cyst in a 49-year-old woman who presented for screening AUS. Coronal image (top) from the AP data set shows an echogenic transverse line (yellow arrows) in the upper breast. This is known as skip artifact and should alert the reader to evaluate the underlying tissue in more detail. A hypoechoic mass (blue arrow) is seen at the 10:30 position, 2 cm from the nipple, which is better visualized on the transverse image (bottom) and was shown to represent a cyst. Skip artifact can be caused by both benign and malignant masses.
Figure 19. Skip artifact caused by malignancy in a 56-year-old woman who presented for screening AUS. Coronal image (top) from the AP data set shows a thin transverse hypoechoic line (yellow arrows) in the upper right breast, consistent with skip artifact. Careful interrogation of the underlying tissue revealed a suspicious, irregular, hypoechoic mass (blue arrow) at the 12-o'clock position on the transverse image (bottom), likely causing the artifact. Subsequent HHUS confirmed the mass, and US-guided biopsy was performed. Final surgical pathologic analysis yielded a 2.2-cm right breast grade 3 IDC (ER negative, PR negative, HER2 positive) with metastatic lymphadenopathy (one of five positive nodes).
Posterior enhancement is commonly seen and is often associated with benign lesions such as cysts, as with HHUS (57) (Fig 20). The radiologist should carefully evaluate masses thought to represent benign cysts, particularly if solitary or new, as high-grade malignancies are also known to demonstrate posterior enhancement at US due to their high cellularity (4).
Figure 20. Multiple cysts in a 50-year-old woman who presented for screening AUS. Two AP data sets show innumerable cysts, characterized by oval, anechoic, circumscribed masses (yellow arrows) on the transverse images (bottom). Scrolling posteriorly on the coronal images (top) shows the “whitewall sign” (blue arrows), which is the manifestation of posterior enhancement at AUS, helping confirm the benign origin.
Use Custom Hanging Protocols and Computer-aided Detection
Similar to interpretation of screening mammography, a customized AUS hanging protocol can be created to facilitate rapid standardized review of all data sets. This will ensure visualization of all images, create a reliable and consistent method for interpretation, increase efficiency, and decrease user error (58). Batch reading of screening examinations has been shown to decrease recall rates in mammography while maintaining CDR and PPV (59), although this has not been studied in the screening breast US setting.
As mentioned previously, computer-aided detection (CAD) (QVCAD; Qview Medical) has been approved for use in AUS interpretation and has been shown to reduce interpretation time without loss of diagnostic accuracy, using both first- and second-generation transducers (38,60). With CAD, a maximum intensity projection (MIP) image is provided, which highlights areas of concern detected by the program (38). Upon clicking on a highlighted area, the CAD program displays the potential lesion within the volume, allowing the reader to quickly identify potentially suspicious findings and efficiently evaluate the area on the conventional multiplanar images (Fig S5).
Pitfalls of AUS Interpretation
Selection of Depth, Tissue Coverage, and Nipple Marker
Proper operator training is paramount to avoid technical and acquisition errors related to positioning. Obtaining optimal AUS images starts with appropriate selection of tissue depth and identification of the number of data sets needed to cover all breast tissue (Fig 21). Variable depth settings are available, with the goal of having the breast tissue cover approximately three-fourths of the image view. It is important to note differences in depth selection when comparing with prior AUS examinations, as lesions may appear larger visually due to depth selection, not from actual change in size (Fig S6).
Figure 21. Fibroadenoma in a 44-year-old woman who presented for screening AUS. (A) On coronal (top) and transverse (bottom) images from the AP data set, the technologist noted that the most superior portion of the image still includes a significant amount of fibroglandular tissue (arrows). (B) Therefore, corresponding images from an additional data set were obtained for inclusion of superior tissue; the data set was labeled “SUP.” Note that the most superior portion of the coronal image (arrows) corresponds to fatty tissue in the transverse plane, ensuring inclusion of all fibroglandular tissue. (C) In this case, this was important, as a suspicious, irregular, hypoechoic mass (arrow) was visualized only on corresponding images from the SUP data set at the 12:30 position, 9 cm from the nipple. (D) On a targeted HHUS image, the mass is oval and hypoechoic with parallel orientation and predominantly circumscribed margins. A single margin was called angular (arrow); thus, US-guided biopsy was recommended. Pathologic analysis revealed a fibroadenoma, considered concordant.
While a minimum of three data sets is recommended for each breast, additional data acquisitions may be required for larger breasts. Overlap is desirable to allow cross-referencing of findings and confirmation of true lesions. Operators should be trained to recognize the typical borders of fibroglandular tissue and obtain additional data sets if properly positioned initial images do not include all fibroglandular tissue. The operator manually selects placement of the nipple marker after image acquisition, which can be a potential source of error (Fig S7). The transducer must also be placed in the proper orientation during image acquisition to ensure that data are displayed in the correct orientation (Fig 22).
Figure 22. IDC in a 58-year-old woman who presented for screening AUS. Coronal (top) and transverse (bottom) images from the first AP data set (left) show a suspicious, irregular, hypoechoic mass (yellow arrow) with the retraction phenomenon on the coronal view at the 6-o'clock position. A second AP view was obtained to include additional inferior tissue. Corresponding images from the second AP data set (right) show a similar-appearing mass (blue arrow) at the 12-o'clock position. Given that the findings were nearly identical and that each AP view showed only a single mass, the case was reviewed in detail with the operator. It was discovered that the operator had inadvertently flipped the transducer when obtaining the second image; thus, the images were transmitted in the incorrect orientation. Careful attention should be paid to transducer orientation to ensure correct image acquisition. A single mass was visualized at HHUS, and subsequent US-guided biopsy was performed. Final surgical pathologic analysis yielded a 1.4-cm grade 2 IDC (ER positive, PR negative, HER2 negative), which was negative for nodal involvement.
Ensure Adequate Compression
While previously described artifacts are due to the nature of breast tissue and reconstruction format, additional artifacts are secondary to scanning technique and patient factors and should be recognized to ensure accurate interpretation.
Low scanning pressure, typically seen at the periphery of images, can result in artifactual shadowing and loss of contact of the transducer and underlying tissues, potentially resulting in false-positive recall or limiting visualization of malignanciesAs with HHUS, adequate compression is essential for proper AUS acquisition. Adequate compression can be confirmed by the presence of tissue toward the edge of the image on both coronal and transverse images (Fig 23). Poor compression can cause otherwise benign-appearing masses or normal tissue to demonstrate suspicious features, such as irregular or indistinct margins (Fig 24). This can be a source of false positives. Therefore, readers of AUS studies should give credence to findings on the view on which they are best compressed (Fig 25). Second-generation AUS with a curved transducer offers improved side-to-side compression, as well as user-selectable transducer compression levels, which allow optimal imaging of patients with various levels of pressure tolerance.
Figure 23. AUS examination highlighting how differences in scanning pressure can affect image quality. Left: The AP data set has low scan pressure, denoted by an “L” (purple circle), with loss of contact at the periphery of the image (yellow arrows) on both the coronal (top) and transverse (bottom) images. Right: A second AP data set was obtained with high scan pressure, denoted by an “H” (green circle). This second data set demonstrates the presence of tissue at the edges of the image, with improved visualization of the tissue both medially and laterally (blue arrows) on the coronal (top) and transverse (bottom) images. Additionally, there is less artifactual shadowing of the normal fibroglandular tissue throughout the transverse image, resulting in improved visualization. Technologists should strive to use the highest level of pressure tolerated by the patient.
Figure 24. Normal results of screening AUS in a 61-year-old woman. On both the coronal (top) and transverse (bottom) images from the AP data set (left), there is a possible large, irregular, hypoechoic mass (yellow arrow) with posterior shadowing. However, improved compression in this portion of the breast on corresponding images from the LAT data set (right) shows that the area represents normal fibroglandular tissue with no underlying mass (blue arrow). Decreased compression is often experienced at the periphery of the image, and credence should be given to the view on which the finding is best compressed. Additionally, a finding this large would be expected to be confirmed in multiple data sets.
Figure 25. Normal results of screening AUS in a 75-year-old woman. Both AP (top) and SUP (bottom) data sets are provided. In the AP data set, there is a possible irregular hypoechoic mass (yellow arrow) at the 12-o'clock position seen on coronal (center) and transverse (right) images. However, the SUP data set shows that this area represents focally dense breast tissue (blue arrows), with no persistent suspicious underlying mass on coronal (center) and transverse (right) images. Additional reformatted sagittal images are provided (left), with the orientation highlighted by the blue circle. The area of interest is shown to be at the edge of the image in the AP data set (top purple arrow), where there is poor compression. However, in the SUP data set, the area of interest (bottom purple arrow) is more centered from superior to inferior, and the reader can be confident there is improved compression.
Recognize Limiting Patient Factors
Patient factors, most commonly motion and body habitus, can lead to artifacts that limit interpretation. Respiratory motion, cardiac motion, and patient movement can result in zigzag artifact, which can degrade images (Fig 26). Unique patient anatomy, including musculoskeletal conditions such as pectus excavatum or the presence of breast implants, can create challenges for acquisition of optimal AUS images. While the presence of breast implants is not an absolute contraindication to performance of AUS, some institutions elect to exclude patients with implants from AUS due to difficulty in maintaining uniform contact. In these cases, HHUS is an appropriate alternative to ensure optimal image acquisition.
Figure 26. Motion artifact in a 60-year-old woman who presented for screening AUS. Significant motion artifact is visualized due to coughing during the study. Motion can manifest as zigzag artifact (yellow arrows) or skip artifact (blue arrow) on coronal images and as blur on transverse images.
Improper Characterization of Benign Findings
In keeping with the BI-RADS lexicon, multiple bilateral similar-appearing circumscribed masses should be assessed as benign, BI-RADS category 2 (61). The global bilateral nature of AUS improves detection of incidental synchronous lesions, allowing increased confidence over HHUS in this scenario. This can avoid unnecessary recall and subsequent BI-RADS category 3 follow-ups (Fig 27). Berg (62) reported that lesions that were circumscribed, oval, and hypoechoic or isoechoic masses with minimal posterior enhancement or no posterior features could be assessed as BI-RADS category 3 and reevaluated at annual screening.
Figure 27. Multiple bilateral benign masses in a 44-year-old woman who presented for baseline screening AUS. Both coronal (top) and transverse (bottom) images from the LAT data set of the right breast (left) and the AP data set of the left breast (right) are shown. Oval, circumscribed, hypoechoic masses with parallel orientation are visualized in the upper outer right breast (yellow arrow) and the lower left breast (blue arrow). The patient had additional similar-appearing masses bilaterally, which fulfills criteria for a benign finding, or BI-RADS category 2.
However, a prospective clinical trial recently recommended return to routine screening for lesions that would typically qualify for BI-RADS category 3 found at AUS, as this would result in a substantial decrease of recall rate and be unlikely to result in adverse outcomes (63) (Fig S8). In this study, the recall rate was reduced from 21.3% to 3.8%, with no cancers diagnosed in the same quadrant as the BI-RADS category 3 lesion in a 2-year follow-up period. Features of BI-RADS category 3 lesions in this study included oval masses parallel to the skin with circumscribed margins and no posterior features or minimal posterior enhancement, among others. Cancers can coexist with multiple circumscribed masses (61,64); therefore, each mass should be carefully scrutinized, and irregular masses or those with noncircumscribed margins merit additional imaging.
Learn from False Positives
As with any imaging modality, review of false positives allows radiologist practice groups to improve interpretive skills and prevents similar recalls in the future. The operator performing the targeted US of an AUS recall should first review the AUS images. Because AUS images are not viewed in traditional radial and antiradial planes, it may be helpful for the operator to begin by imaging with the probe positioned in the transverse plane to reproduce the AUS finding (Fig 28). Images should then be obtained in the conventional HHUS format. The reader should also review the original AUS images to determine the level of suspicion of the original finding, as well as to ensure that the correct area was interrogated.
Figure 28. Normal results of baseline screening AUS in a 58-year-old woman. (A) Coronal (top) and transverse (bottom) images show a possible irregular hypoechoic mass (arrow) in the lower right breast. The patient was recalled for HHUS, although no correlate was initially identified by the sonographer when the traditional radial and antiradial scanning technique was used. (B) The patient was then imaged in the transverse plane, and a correlate for the AUS finding was visualized (arrow). (C) However, in the longitudinal plane, this finding corresponded to an elongated fat lobule (arrows). (D) Retrospective interrogation of the AUS finding on the dedicated workstation using the software rotation tool revealed that the finding initially seen on the coronal image (yellow arrow) represented an elongated fat lobule when rotated into the transverse plane (blue arrows). The interpreting radiologist should always review the original AUS images when performing a recall examination with HHUS; doing so will both confirm successful identification of the finding of concern and allow the reader to learn from false positives. Additionally, multiplanar review using the software rotation tool at the time of the original screening examination may have prevented this recall.
Understand Differences in Transducer Size
AUS data are acquired with a unique large-format automated transducer (frequency range, 6–15 MHz) scanning the breast from bottom to top, ultimately resulting in an image that is 15.4 cm in width and 17 cm in length. In comparison with an image obtained with a handheld probe, disease in the coronal plane will appear minified in size, with the potential for less-experienced users to dismiss smaller lesions that may represent true disease (Fig 29). Users are encouraged to review coronal images in one of the larger display formats.
Figure 29. IDC in a 61-year-old woman who presented for screening AUS. (A) Photograph shows the significant size difference between a traditional handheld transducer and the large-format AUS probe. The large amount of tissue captured in a single coronal view from AUS can result in findings appearing smaller than readers are accustomed to with single HHUS images. (B) Coronal (top) and transverse (bottom) images from an AP data set of an AUS examination show a 0.8-cm, irregular, hypoechoic mass (arrow) with not parallel orientation, which is suspicious. (C) This finding was confirmed with HHUS (arrow); subsequent biopsy yielded grade 2 IDC. Visualization of this 0.8-cm mass at HHUS is not difficult, but perception is harder on the coronal AUS image given the large field of view of 15.4 cm. When interpreting AUS images, readers must become familiar with these differences to identify small malignancies.
Conclusion
Dense breast tissue is common and elevates breast cancer risk. Multiple studies confirm the benefits of supplemental screening in this intermediate-risk patient population. Owing to patient breast density notification laws and inclusion of breast density in some risk models, use of supplemental screening is expected to increase. US is an attractive adjunctive screening option because it lacks patient exposure to ionizing radiation, does not require contrast material administration or intravenous access, and is widely accessible. Screening US is proven to demonstrate mammographically occult cancers in dense breast tissue, the majority of which are small and node negative.
The advantages of AUS over HHUS include standardized image acquisition, reproducible comparison with prior studies, potentially improved auditing, decreased operator dependence, and no formal operator training requirement. Readers of AUS studies should use available comparison images and correlate with mammography to avoid unnecessary recalls. Understanding the unique software features of AUS, as well as the technical challenges and artifacts specific to it, can aid in accurate characterization of findings and facilitate constructive feedback to operators. Knowledge of the pearls and pitfalls of the AUS technique is essential for image interpretation and can decrease false-positive rates and interpretation times.
Acknowledgment
We would like to thank Susan Roux, MD, for her contribution to the manuscript due to her extensive experience with and knowledge of AUS systems.
* A.I.H. and M.F.I. contributed equally to this work.
Recipient of a Certificate of Merit award for an education exhibit at the 2022 RSNA Annual Meeting.
References
- 1. . Breast cancer detected with screening US: reasons for nondetection at mammography. Radiology 2014;270(2):369–377.
- 2. . Comparison of the performance of screening mammography, physical examination, and breast US and evaluation of factors that influence them: an analysis of 27,825 patient evaluations. Radiology 2002;225(1):165–175.
- 3. . Screening US in patients with mammographically dense breasts: initial experience with Connecticut Public Act 09-41. Radiology 2012;265(1):59–69.
- 4. . ACR BI-RADS Manual. Reston, Va: American College of Radiology, 2013.
- 5. . Mammographic density and the risk and detection of breast cancer. N Engl J Med 2007;356(3):227–236.
- 6. . Volumetric breast density affects performance of digital screening mammography. Breast Cancer Res Treat 2017;162(1):95–103.
- 7. . Quantitative assessment of mammographic breast density: relationship with breast cancer risk. Radiology 2004;230(1):29–41.
- 8. . Breast density and parenchymal patterns as markers of breast cancer risk: a meta-analysis. Cancer Epidemiol Biomarkers Prev 2006;15(6):1159–1169.
- 9. . Combining quantitative and qualitative breast density measures to assess breast cancer risk. Breast Cancer Res 2017;19(1):97.
- 10. . Screening Breast Ultrasound Using Handheld or Automated Technique in Women with Dense Breasts. J Breast Imaging 2019;1(4):283–296.
- 11. . Mammographic breast density refines Tyrer-Cuzick estimates of breast cancer risk in high-risk women: findings from the placebo arm of the International Breast Cancer Intervention Study I. Breast Cancer Res 2014;16(5):451.
- 12. . Performance of Breast Cancer Risk-Assessment Models in a Large Mammography Cohort. J Natl Cancer Inst 2020;112(5):489–497.
- 13. . Influence of tumour stage at breast cancer detection on survival in modern times: population based study in 173,797 patients. BMJ 2015;351:h4901.
- 14. . Association between mammographic breast density and breast cancer tumor characteristics. Cancer Epidemiol Biomarkers Prev 2005;14(3):662–668.
- 15. . Invasive cancers detected after breast cancer screening yielded a negative result: relationship of mammographic density to tumor prognostic factors. Radiology 2004;230(1):42–48.
- 16. . Localized mammographic density is associated with interval cancer and large breast cancer: a nested case-control study. Breast Cancer Res 2019;21(1):8.
- 17. . Ultrasound as the Primary Screening Test for Breast Cancer: Analysis from ACRIN 6666. J Natl Cancer Inst 2015;108(4):djv367.
- 18. . Assessing improvement in detection of breast cancer with three-dimensional automated breast US in women with dense breast tissue: the SomoInsight Study. Radiology 2015;274(3):663–673.
- 19. . Trends of Supplemental Screening in Women with Dense Breasts. J Am Coll Radiol 2020;17(8):990–998.
- 20. . A Review of Supplemental Screening Ultrasound for Breast Cancer: Certain Populations of Women with Dense Breast Tissue May Benefit. Acad Radiol 2016;23(12):1604–1609.
- 21. . Breast Cancer Screening in Women at Higher-than-Average Risk: Recommendations from the ACR. J Am Coll Radiol 2018;15(3 3 Pt A):408–414.
- 22. . Breast Cancer Screening for Women at Higher-than-Average Risk: Updated Recommendations from the ACR. J Am Coll Radiol 2023. https://doi.org/10.1016/j.jacr.2023.04.002. Published online May 5, 2023. https://doi.org/10.1016/j.jacr.2023.04.002.
- 23. . Breast ultrasound: recommendations for information to women and referring physicians by the European Society of Breast Imaging. Insights Imaging 2018;9(4):449–461.
- 24. . Occult cancer in women with dense breasts: detection with screening US—diagnostic yield and tumor characteristics. Radiology 1998;207(1):191–199.
- 25. . Combined screening with ultrasound and mammography vs mammography alone in women at elevated risk of breast cancer. JAMA 2008;299(18):2151–2163.
- 26. . Detection of breast cancer with addition of annual screening ultrasound or a single screening MRI to mammography in women with elevated breast cancer risk. JAMA 2012;307(13):1394–1404.
- 27. . Comparison of automated breast volume scanning and hand-held ultrasound in the detection of breast cancer: an analysis of 5,566 patient evaluations. Asian Pac J Cancer Prev 2014;15(21):9101–9105.
- 28. . Mammographically Occult Automated Ultrasound Screening-Detected Cancers: Analysis of Common and Unique Descriptors.
Presented at the 2022 SBI/ACR Breast Imaging Symposium ,Savannah, Ga ,May 16–19, 2022 . - 29. . Prospective Multicenter Diagnostic Performance of Technologist-Performed Screening Breast Ultrasound after Tomosynthesis in Women with Dense Breasts (the DBTUST). J Clin Oncol 2023;41(13):2403–2415.
- 30. . Automated breast ultrasound: supplemental screening for average-risk women with dense breasts. Clin Imaging 2021;76:15–25.
- 31. . Agreement between an Automated Volume Breast Scanner and Handheld Ultrasound for Diagnostic Breast Examinations. J Ultrasound Med 2017;36(10):2087–2092.
- 32. . Supplemental Screening Breast US in Women with Negative Mammographic Findings: Effect of Routine Axillary Scanning. Radiology 2018;286(3):830–837.
- 33. . Analysis of eighty-one cases with breast lesions using automated breast volume scanner and comparison with handheld ultrasound. Eur J Radiol 2012;81(5):873–878.
- 34. . Dependability of Automated Breast Ultrasound (ABUS) in Assessing Breast Imaging Reporting and Data System (BI-RADS) Category and Size of Malignant Breast Lesions Compared with Handheld Ultrasound (HHUS) and Mammography (MG). Int J Gen Med 2021;14:9193–9202.
- 35. . Clinical utility of bilateral whole-breast US in the evaluation of women with dense breast tissue. Radiology 2001;221(3):641–649.
- 36. . Automated Breast Ultrasound Interpretation Times: a Reader Performance Study. Acad Radiol 2018;25(12):1577–1581.
- 37. . 3D Automated Breast Ultrasound System: Comparison of Interpretation Time of Senior versus Junior Radiologist. Eur J Breast Health 2019;15(3):153–157.
- 38. . Interpretation Time Using a Concurrent-Read Computer-aided Detection System for Automated Breast Ultrasound in Breast Cancer Screening of Women with Dense Breast Tissue. AJR Am J Roentgenol 2018;211(2):452–461.
- 39. . A multicenter hospital-based diagnosis study of automated breast ultrasound system in detecting breast cancer among Chinese women. Chin J Cancer Res 2018;30(2):231–239.
- 40. . The performance of 3D ABUS versus HHUS in the visualisation and BI-RADS characterisation of breast lesions in a large cohort of 1,886 women. Eur Radiol 2018;28(2):592–601.
- 41. . The image quality and lesion characterization of breast using automated whole-breast ultrasound: a comparison with handheld ultrasound. Eur J Radiol 2015;84(7):1232–1235.
- 42. . Whole Breast Ultrasound: Comparison of the Visibility of Suspicious Lesions with Automated Breast Volumetric Scanning versus Hand-held Breast Ultrasound. Acad Radiol 2015;22(7):870–879.
- 43. . Automatic Breast Volume Scanner versus Handheld Ultrasound in Differentiation of Benign and Malignant Breast Lesions: a Systematic Review and Meta-analysis. Ultrasound Med Biol 2019;45(8):1874–1881.
- 44. . Three-dimensional automated breast ultrasound: technical aspects and first results. Diagn Interv Imaging 2019;100(10):579–592.
- 45. . Automated Three-dimensional Breast US for Screening: Technique, Artifacts, and Lesion Characterization. RadioGraphics 2018;38(3):663–683.
- 46. . Automated Breast Ultrasound: Technical Aspects, Impact on Breast Screening, and Future Perspectives. Curr Breast Cancer Rep 2021;13(3):141–150.
- 47. . Recall rate of screening ultrasound with automated breast volumetric scanning (ABVS) in women with dense breasts: a first quarter experience. Clin Imaging 2014;38(4):439–444.
- 48. . Three-dimensional Automated Breast US: Facts and Artifacts. RadioGraphics 2019;39(4):913–931.
- 49. . The contribution of three-dimensional power Doppler imaging in the preoperative assessment of breast tumors: a preliminary report. Obstet Gynecol Int 2009;2009:530579.
- 50. . Comparison of retraction phenomenon and BI-RADS-US descriptors in differentiating benign and malignant breast masses using an automated breast volume scanner. Eur J Radiol 2015;84(11):2123–2129.
- 51. . Comparative study of automated breast 3-D ultrasound and handheld B-mode ultrasound for differentiation of benign and malignant breast masses. Ultrasound Med Biol 2013;39(10):1735–1742.
- 52. . Comparison of automated breast volume scanning to hand-held ultrasound and mammography. Radiol Med (Torino) 2012;117(8):1287–1293. [Published correction appears in Radiol Med 2012;117(8):1443.]
- 53. . Diagnostic performance of coronal view in comparison with transverse view of three-dimensional automated breast ultrasound. Acta Radiol 2021;62(1):27–33.
- 54. . The value of coronal view as a stand-alone assessment in women undergoing automated breast ultrasound. Radiol Med (Torino) 2021;126(2):206–213.
- 55. . Comparison of lesion detection in the transverse and coronal views on automated breast sonography. J Ultrasound Med 2015;34(1):125–135.
- 56. . Automated 3-D breast ultrasound as a promising adjunctive screening tool for examining dense breast tissue. Semin Roentgenol 2011;46(4):302–308.
- 57. . Automated Breast Ultrasound: Interobserver Agreement, Diagnostic Value, and Associated Clinical Factors of Coronal-Plane Image Features. Korean J Radiol 2020;21(5):550–560.
- 58. . Optimizing radiologist productivity and efficiency: work smarter, not harder. Eur J Radiol 2022;155:110131.
- 59. . Batch Reading and Interrupted Interpretation of Digital Screening Mammograms without and with Tomosynthesis. J Am Coll Radiol 2021;18(2):280–293.
- 60. . Performance and Reading Time of Automated Breast US with or without Computer-aided Detection. Radiology 2019;292(3):540–549.
- 61. . Multiple bilateral circumscribed masses at screening breast US: consider annual follow-up. Radiology 2013;268(3):673–683.
- 62. . BI-RADS 3 on Screening Breast Ultrasound: What Is It and What Is the Appropriate Management? J Breast Imaging 2021;3(5):527–538.
- 63. . Outcomes of Return to Routine Screening for BI-RADS 3 Lesions Detected at Supplemental Automated Whole-Breast Ultrasound in Women with Dense Breasts: a Prospective Study. AJR Am J Roentgenol 2021;217(6):1313–1321.
- 64. . Multiple Bilateral Circumscribed Masses at Screening Breast Ultrasound: Outcomes of New or Enlarging Masses at Follow-up. AJR Am J Roentgenol 2022;218(3):435–443.
- 65. . Breast cancer detection using automated whole breast ultrasound and mammography in radiographically dense breasts. Eur Radiol 2010;20(3):734–742.
- 66. . Adding 3D automated breast ultrasound to mammography screening in women with heterogeneously and extremely dense breasts: report from a hospital-based, high-volume, single-center breast cancer screening program. Eur J Radiol 2016;85(9):1554–1563.
- 67. . Can automated breast ultrasound be the key to increased cancer detection and a low callback rate? Washington, DC: American Roentgen Ray Society, 2017.
- 68. . Screening Outcomes of Supplemental Automated Breast US in Asian Women with Dense and Nondense Breasts. Radiology 2023;307(4):e222435.
Article History
Received: Feb 16 2023Revision requested: May 24 2023
Revision received: May 25 2023
Accepted: June 5 2023
Published online: Oct 04 2023
