ABSTRACT
Introduction
Imaging surveillance of patients post breast surgery for breast cancer is routinely performed with mammography. In practice, a large number of patients receive a concomitant ultrasound. The latter has not been established as routine practice and varies across imaging departments and referrers. This study aims to determine if the use of digital breast tomosynthesis (DBT) has decreased the need for concomitant ultrasound and extra mammographic views compared to digital mammography (DM).
Methods
Data was retrospectively reviewed for patients undergoing surveillance mammography for two periods: 2015–2016 (DM) and 2018–2019 (DBT). Patients identified as first‐year post breast‐conserving surgery (BCS) were included in the study. Comparison of proportions of concomitant ultrasound use between the two periods, with subgroup analysis for breast density categories, was performed using Chi‐squared tests.
Results
The study included 499 patients, of which 245 and 254 represented the DM and DBT periods, respectively, with a mean age of 60.1 years (SD 13.1) for the entire cohort.
During DBT period, the percentage of patients having concomitant ultrasounds decreased by 15 percentage points (95% CI 8.9%–21.1%) (p < 0.0001) when compared to the DM period. During DBT period, the percentage of patients having extra views decreased by 44 percentage points (95% CI 36.1%–51.1%) (p < 0.0001).
Conclusion
The introduction of DBT resulted in fewer concomitant ultrasound examinations and extra mammographic views performed in the first‐year post breast cancer treatment. The role of ultrasound in cancer surveillance post‐treatment is debatable and large prospective studies investigating its value are warranted.
Keywords: breast, mammography, tomosynthesis, ultrasound
The introduction of digital breast tomosynthesis (DBT) has reduced the number of concomitant ultrasound examinations performed in the first year post breast cancer treatment. DBT has also reduced the number of extra mammographic views in this cohort. The role of ultrasound in cancer surveillance post‐treatment is debatable, and large prospective studies investigating its value are warranted.
1. Introduction
Breast cancer is the most frequently diagnosed cancer for women, with 2.3 million new detections globally each year [1]. In the last few decades, surgical treatment has evolved, resulting in minimising the necessity of mastectomies for small breast cancers. Mammography is the screening and diagnostic test of choice for detecting breast cancer in women aged 40 years or older [2]. BreastScreen Australia invites women aged 50–74 for biennial mammography. Women aged 40–49 and 75+ are also eligible for screening through the programme. However, the cost‐effectiveness of routine screening in these age groups depends on several factors, including background risk of breast cancer in the 40–49 year old age group and competing risk factors for mortality in the 75+ year old age group [3]. BreastScreen Australia continues to use 2D digital mammography (DM) for screening, and whereas the use of digital breast tomosynthesis (DBT) for screening has been trialled with successful results, its widespread use has not been established [4].
Digital mammography produces a flat, two‐dimensional image of the breast. Routinely, two images are obtained for each breast: cranio‐caudal (CC) and medio‐lateral oblique (MLO) views. Advances in mammographic technology have seen the implementation of DBT, which creates a stack of images of the breast by taking multiple low dose projections along an arc, varying between 15° and 60° [5]. The resultant series of stacked images reduces superimposition of breast tissue and can improve lesion detection and characterisation, along with better lesion localisation [6]. Recent studies have shown that DBT has an increased sensitivity and specificity in comparison to DM, resulting in increased cancer detection and a reduction in false positives [5, 7]. The requirement for additional mammographic views such as spot compression and spot magnification views has also decreased for non‐calcified findings in DBT [8]. However, evidence for decreased use of additional views comes from screening and diagnostic work‐up cohorts rather than post‐treatment cohorts [8].
It is recommended that women undergoing breast‐conserving surgery (BCS), and mastectomy undertake routine annual mammograms for at least 5 years after initial diagnosis as the risk of recurrence is highest in this time frame post diagnosis [5]. Some women will also receive a concomitant ultrasound with mammography to improve the detection of cancer [9]. In current practice, the role of concomitant ultrasound in post‐operative patients has not been established and practice varies amongst imaging departments and referrers.
The primary aim of this study is to determine if the introduction of DBT has reduced the need for concomitant ultrasound for patients undergoing surveillance imaging post breast cancer surgery. A secondary aim is to assess the reduction in extra mammographic views before and after DBT introduction.
2. Materials and Methods
This retrospective audit was performed at Austin Health, Repatriation Campus, Melbourne, Australia, and was given approval by the Austin Health Human Research and Ethics Committee (HREC/23604/Austin‐2024).
Patients who underwent a mammogram with or without an ultrasound from the time periods of January 2015 to December 2016 and January 2018 to December 2019 were reviewed. These time periods were selected to enable comparison between two types of mammographic technologies; DM (Siemens Mammomat Inspirations, Forchheim, Germany) (2015–2016), to DBT (Hologic Selenia Dimensions, Danbury, USA) (2018–2019). The year of 2017 was excluded due to a changeover of technology from DM to DBT. All examinations included CC and MLO views of each breast, (or one breast in the case of a previous mastectomy), consisting of a synthesised 2‐dimensional view (C‐view) and a DBT series. Both the C‐view and DBT images were assessed by a single radiologist at the time of reporting, with a total of nine radiologists reporting mammographic and ultrasound images throughout the timeframe of the study (with breast reporting experience ranging from 2 to 20 years). Radiologists also had access to previous imaging. When previous imaging was not performed at Austin Health, images were transferred electronically. If extra views were required, they were performed immediately and reviewed by the same radiologist. If ultrasound was performed, this occurred straight after the mammogram or any extra views requested.
Inclusion criteria were (1) one year post‐surgical treatment; (2) imaging performed at Austin Health; (3) with or without concomitant ultrasound on the same day. Further data collected included breast density (categories A, B, C, D), mammographic findings, supplementary mammographic views, ultrasound findings and biopsy findings if applicable [10]. Density categories were assigned by Consultant Radiologists at the time of reporting, with the cohort of reporting radiologists remaining the same for the duration of the study time frame. Breast density categories were defined as follows: category A—almost entirely fat, category B—scattered fibroglandular densities, category C—heterogeneously dense and category D—extremely dense [11]. A summary of the selection process can be seen in Figure 1.
FIGURE 1.
Patient inclusion process. Comparison of diagnostic clinical outcomes using digital mammography (DM) and digital breast tomosynthesis (DBT) technologies. Patients included in the study were first‐year post breast‐conserving surgery.
Descriptive results are presented in frequencies and percentages. Continuous variables are presented as mean with standard deviation (SD). We compared proportions of concomitant ultrasound use between the two periods for the two cohorts (DM and DBT). Comparisons across the two cohorts, including subgroup analyses, were performed using the “N‐1” Chi‐squared tests. Comparison of age between the DM and DBT groups was performed using the two‐sample t‐test. Statistical analysis was performed using MedCalc for Windows, v19.4 (MedCalc Software Ltd., Ostend, Belgium).
3. Results
A total of 5554 patient examinations were reviewed, which included 2547 with DM and 3007 with DBT. A total of 499 met the specified inclusion criteria, consisting of 245 and 254 patients with DM and DBT technologies, respectively. The average age was 60.1 years (13.1 SD).
The percentage of patients having concomitant ultrasounds changed from 93% (229 out of 245) using DM to 78% (199 out of 254) using DBT, with an absolute reduction of 15% (95% CI 8.9%–21.1%) (p < 0.0001) (Table 1) (Figure 2).
TABLE 1.
Proportions of digital mammography (DM) and digital breast tomosynthesis (DBT) patients with different breast densities and the number that went on to have extra mammographic views and ultrasound.
| Year | 2015–2016 | 2018–2019 |
|---|---|---|
| Technology | DM (n = 245) | DBT (n = 254) |
| Number of patients undergoing ultrasound | 229 (93%) | 199 (78%) |
| Breast Density A | 42/43 (98%) | 24/35 (69%) |
| Breast Density B | 83/91 (91%) | 81/105 (77%) |
| Breast Density C | 77/82 (94%) | 67/80 (84%) |
| Breast Density D | 27/29 (93%) | 27/34 (79%) |
| Patients requiring extra views (Total) | 143 (58%) | 36 (14%) |
| Due to density | 119 (48%) | 22 (9%) |
| Due to microcalcifications | 24 (10%) | 14 (6%) |
| Ultrasound results led to biopsy | 10 | 4 |
| Positive ultrasound biopsy | 1 | 1 |
Note: Breast Density A—Almost entirely fatty (less than 25% fibroglandular tissue), Breast Density B—Scattered areas of fibroglandular density (25%–49% fibroglandular tissue), Breast Density C—Heterogeneously dense (50%–75% fibroglandular tissue) and Breast Density D—Extremely dense (> 75% fibroglandular tissue) [12].
Abbreviations: DBT, Digital breast tomosynthesis; DM, Digital mammography.
FIGURE 2.
Percentage of patients requiring ultrasound with mammographic imaging categorised into breast density groups. Lower numbers of concomitant ultrasound were noted in the DBT cohort compared to the DM cohort across all densities (p = 0.0117). DBT, digital breast tomosynthesis; DM, digital mammography.
Of the 428 ultrasound examinations conducted, 414 (96.2%) cases had benign or no findings. Fourteen cases led to a biopsy, which consisted of 10 during the DM period and four during DBT. Only one biopsy in each period yielded a positive (malignant) result. The positive biopsy result from the DM period was a recurrence in the scar tissue which was clinically palpable. The positive case from the DBT period was a recurrence in a clinically suspected axillary lymph node in a patient who had axillary clearance with nine malignant nodes out of 12 nodes removed at the time of surgery.
Over the total 4 year time frame, 179 (35.9%) patient examinations required extra or supplementary mammographic views. The percentage of patients have extra views changed from 58% (143 out of 245) using DM to 14% (36 out of 254) using DBT, with an absolute reduction of 44% (95% CI 36.1%–51.1%) (p < 0.0001) (Table 1). Overall, 97% of extra views were spot compression and magnification views and 3% were due to implants and extended CC views. Extra views due to the presence of microcalcifications were higher during DM period (24 out of 245 [9.7%]) compared to DBT period (14 out of 254 [5.5%]) but this was not statistically significant (p = 0.076), (Figure 3).
FIGURE 3.
Number of extra views performed per mammographic technology. A decrease in extra views from digital mammography to digital breast tomosynthesis can be appreciated. Extra views due to microcalcifications make up the greater portion of total extra views in digital breast tomosynthesis compared to digital mammography. Microcalcifications must always be assessed further regardless of the technology used.
4. Discussion
4.1. Ultrasound Findings
The study demonstrated that ultrasound is used in the majority of women as a concomitant modality to mammography in the surveillance of patients post BCS. It also demonstrated that replacing DM with DBT led to a reduction in the number of concomitant ultrasound studies performed with surveillance mammography. Patient breast density was potentially a contributing factor to reduced concomitant ultrasound use, given the lowest numbers of concomitant ultrasound were observed in patients with breast density category A (fatty breasts). However, these findings merit further investigation in larger cohorts.
Less ultrasound examinations were performed comparing DM and DBT, with the largest difference amongst breast density A patients. In both DM and DBT, the number of ultrasound examinations performed for those having breast density C and D remained highest, likely due to the increased superimposition of dense fibroglandular tissue. Other studies have demonstrated breast cancer may be obscured by dense fibroglandular tissue on mammography and the potential clinical significance of ultrasound in detecting breast cancers in these cases [13].
According to the findings of our study, ultrasound examinations are ideally performed on a case‐by‐case basis, with particular emphasis being placed on breast density. It would be reasonable to recommend that patients with breast density A and B, which are breasts that are almost entirely fatty or composed of scattered areas of fibroglandular tissue, receive a DBT only. Category C and D patients with heterogeneous or extremely dense breast tissue probably require concomitant ultrasound to overcome the limitations of the resultant reduced sensitivity in mammography due to this increase in dense fibroglandular tissue. Indeed, Berg et al. identified that additional screening of ultrasound in patients with dense breasts was beneficial in reducing the rate of interval cancers in a screening setting [14]. An exception to this arises in cases where patient‐related limitations are present, such as body habitus or movement limitations, as these can impact the amount of breast tissue mammographically visualised.
The role of ultrasound in surveillance imaging post‐surgical treatment is debatable. The European Society for Medical Oncology is the only body which recommends annual ultrasound (when needed) in addition to mammography in this setting. Most guidelines do not routinely recommend concomitant ultrasound with mammography, and its role in post‐operative patients varies amongst imaging departments and referrers [15, 16].
4.2. Mammographic Findings
The present study demonstrated that the introduction of DBT has resulted in a significant reduction in the number of extra mammographic views in the DBT era compared to the DM era. This finding aligns with the literature as DBT has the ability to resolve apparent densities resulting from tissue summation, along with better mass visualisation, an increased ability to effectively portray the margins of lesions, and detecting suspicious lesions [17, 18]. This does not however rule out the need for extra views as some lesions require further investigation (Figure 4).
FIGURE 4.
Importance of DBT and the role of extra views. (A) Two‐dimensional image of a breast post BCS. There appears to only be distortion at the scar site (arrow). (B) DBT slice showing an area of spiculation in the lateral aspect of the breast (dotted arrow). (C) Spot compression view allows much better delineation of the lesion (dotted arrow) which was cancerous. Scar tissue (arrow) appears less spiculated with extra compression.
The increased accuracy of cancer detection and reduction in the need for supplementary mammographic imaging using DBT can be attributed to the way the images are acquired in an arc. This image acquisition technique gives the viewer the ability to assess breast parenchyma cross‐sectionally and better separates superimposed tissue (Figure 5) [19]. This increases the sensitivity of DBT to 84%–90% compared to 69%–86% for DM [20]. A study by Kulkarni et al. showed that DBT allowed for confident diagnosis and assessment of scar tissue, calcifications, more accurate definition of calcified areas, along with better representation of the contents of lesions (e.g., adipose tissue) [21].
FIGURE 5.
Mammographic images of dense breasts. (A) Two‐dimensional CC (cranio‐caudal) image of a breast with density category D. The spiculated lesion (arrow) is difficult to see due to overlapping breast parenchyma. (B) Corresponding tomosynthesis slice of A at the level of the lesion (arrow) making it much more easily visible. The margins of the spiculated mass also appear larger than the 2D view. (C) 2D right MLO (medio‐lateral oblique) with breast density D appearing unremarkable. (D) Corresponding tomosynthesis slice of C demonstrating a large spiculated and obvious mass (arrow), clearly demonstrating the advantages of DBT.
Individuals with dense breasts not only have a four to six times greater risk of breast cancer, but are also at a higher risk of interval cancers [22, 23, 24]. This is due to dense fibroglandular tissue masking small lesions, resulting in them going undetected until the patient presents for their next mammogram, leading to late detection and a reduced survival rate. Mammographic sensitivity has been shown to be 75%–94% in non‐dense breasts (BI‐RADS A and B), but decreases to 50%–68% for those with dense breasts [24]. Therefore, patients with dense breasts (BI‐RADS category C and D) may benefit from receiving concomitant ultrasound to ensure that potential small cancers do not go undetected.
At our centre, diagnostic mammography is reviewed by a breast radiologist to decide on whether extra views are required prior to ultrasound imaging. Concomitant ultrasound is requested, in addition to surveillance mammography, by a variety of referring clinicians (surgeons, oncologists, radiation oncologists and general practitioners). Although many clinicians request ultrasound routinely in this setting, some referring clinicians request ultrasound to be performed at the discretion of the reporting breast radiologist, allowing flexibility in the workflow and better resource utilisation. Several factors play a role in clinicians' decision to request an ultrasound in addition to mammography, including breast density, modality of detection of the initial malignancy, type of malignancy, findings on mammography, presence of symptoms and patient's preference. The implications of our findings, from a practical perspective, depend on individual radiology practices and departments workflow patterns. For practices where concomitant ultrasound is not routinely used, the implications are potentially minimal or none. For practices where concomitant ultrasound is performed routinely without an individualised approach, it is possible that our findings will prompt healthcare providers to reassess their workflow to assess whether routine concomitant ultrasound is required for all patients in this setting.
4.3. Limitations
The present study has some limitations. First, this was a retrospective single centre study with limitations inherent in data collection. These include different radiologists reporting the different studies with possible differences in preference for use of additional imaging (extra mammographic view and concomitant ultrasound) and differences in reporting breast density categories [21]. Second, surveillance imaging was only evaluated in patients who were in their first‐year post cancer treatment and not assessed in subsequent years. Third, referring clinicians' preferences for concomitant ultrasound along with the skill set and experience of different radiologists may have influenced the use of concomitant ultrasound and extra mammographic views. However, the study included patients over 2 years in each period, which likely captured a range of preferences and skill sets. Fourth, the study was conducted using machines from different manufacturers: a Siemens system for DM and a Hologic system for DBT, which may have influenced the interpretation of the imaging outcomes. Finally, the study did not assess interval cancers or subsequent development of cancer beyond the first‐year post cancer treatment. Instead, we focused on the use of concomitant ultrasound and extra mammographic views and their rate of positive findings. Further follow‐up of such patients is required to assess such outcomes.
5. Conclusion
The introduction of DBT has resulted in fewer numbers of concomitant ultrasound examinations and extra mammographic views performed in the first‐year post breast cancer treatment. The role of ultrasound in breast cancer surveillance post‐treatment is debatable. Large prospective studies investigating the utility of concomitant ultrasound in surveillance post surgery may be required to justify its use in the era of DBT.
Ethics Statement
This study was given approval by ethics before commencement.
Conflicts of Interest
The authors declare no conflicts of interest.
Funding: The authors received no specific funding for this work.
Data Availability Statement
The authors have nothing to report.
References
- 1. Łukasiewicz S., Czeczelewski M., Forma A., Baj J., Sitarz R., and Stanisławek A., “Breast Cancer—Epidemiology, Risk Factors, Classification, Prognostic Markers, and Current Treatment Strategies—An Updated Review,” Cancers 13, no. 17 (2021): 4287, https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8428369/. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Joshi P., Singh N., Raj G., Singh R., Malhotra K. P., and Awasthi N. P., “Performance Evaluation of Digital Mammography, Digital Breast Tomosynthesis and Ultrasound in the Detection of Breast Cancer Using Pathology as Gold Standard: An Institutional Experience,” Egyptian Journal of Radiology and Nuclear Medicine 53, no. 1 (2022): 1, 10.1186/s43055-021-00675-y. [DOI] [Google Scholar]
- 3. Carle C., Velentzis L. S., and Nickson C., “BreastScreen Australia National Data by Factors of Interest for Risk‐Based Screening: Routinely Reported Data and Opportunities for Enhancement,” Australian and New Zealand Journal of Public Health 46, no. 2 (2022): 230–236, https://pubmed.ncbi.nlm.nih.gov/35112749/. [DOI] [PubMed] [Google Scholar]
- 4. Houssami N., Lockie D., Giles M., Noguchi N., Marr G., and Marinovich L., “Two‐Year Follow‐Up of Participants in the BreastScreen Victoria Pilot Trial of Tomosynthesis Versus Mammography: Breast Density‐Stratified Screening Outcomes,” British Journal of Radiology 96, no. 1148 (2023): 20230081, https://pubmed.ncbi.nlm.nih.gov/37191331/. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Asbeutah A. M., Karmani N., Asbeutah A. A., Echreshzadeh Y. A., AlMajran A. A., and Al‐Khalifah K. H., “Comparison of Digital Breast Tomosynthesis and Digital Mammography for Detection of Breast Cancer in Kuwaiti Women,” Medical Principles and Practice 28, no. 1 (2019): 10–15, https://www.karger.com/Article/FullText/495753. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Chong A., Weinstein S. P., McDonald E. S., and Conant E. F., “Digital Breast Tomosynthesis: Concepts and Clinical Practice,” Radiology 292, no. 1 (2019): 1–14, 10.1148/radiol.2019180760. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Butler R., Conant E. F., and Philpotts L., “Digital Breast Tomosynthesis: What Have We Learned?,” Journal of Breast Imaging 1, no. 1 (2019): 9–22, https://academic.oup.com/jbi/article/1/1/9/5379166. [DOI] [PubMed] [Google Scholar]
- 8. Noroozian M., Hadjiiski L., Rahnama‐Moghadam S., et al., “Digital Breast Tomosynthesis Is Comparable to Mammographic Spot Views for Mass Characterization,” Radiology 262, no. 1 (2012): 61–68, https://pmc.ncbi.nlm.nih.gov/articles/PMC3244671/. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Gemici A. A., Bayram E., Hocaoglu E., and Inci E., “Comparison of Breast Density Assessments According to BI‐RADS 4th and 5th Editions and Experience Level,” Acta Radiologica Open 9, no. 7 (2020): 205846012093738, 10.1177/2058460120937381. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Sickles E. A., D'Orsi C. J., Bassett L. W., et al., “ACR BI‐RADS Mammography,” in ACR BI‐RADS Atlas, Breast Imaging Reporting and Data System, 5th ed., ed. D'Orsi C. J., Sickles E. A., Mendelson E. B., et al. (Reston, VA, 2013). [Google Scholar]
- 11. Tari D. U., Santonastaso R., De Lucia D. R., Santarsiere M., and Pinto F., “Breast Density Evaluation According to BI‐RADS 5th Edition on Digital Breast Tomosynthesis: AI Automated Assessment Versus Human Visual Assessment,” Journal of Personalized Medicine 13, no. 4 (2023): 609, https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10146726/. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Sood R., Rositch A. F., Shakoor D., et al., “Ultrasound for Breast Cancer Detection Globally: A Systematic Review and Meta‐Analysis,” JGO 5 (2019): 1–17, 10.1200/JGO.19.00127. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Berg W. A., Rafferty E. A., Friedewald S. M., Hruska C. B., and Rahbar H., “Screening Algorithms in Dense Breasts: Ajr Expert Panel Narrative Review,” American Journal of Roentgenology 216, no. 2 (2021): 275–294, https://www.ajronline.org/doi/10.2214/AJR.20.24436. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Cardoso F., Kyriakides S., Ohno S., et al., “Early Breast Cancer: ESMO Clinical Practice Guidelines for Diagnosis, Treatment and Follow‐Up,” Annals of Oncology 30, no. 8 (2019): 1194–1220, https://linkinghub.elsevier.com/retrieve/pii/S0923753419312876. [DOI] [PubMed] [Google Scholar]
- 15. Swinnen J., Keupers M., Soens J., Lavens M., Postema S., and Van Ongeval C., “Breast Imaging Surveillance After Curative Treatment for Primary Non‐Metastasised Breast Cancer in Non‐High‐Risk Women: A Systematic Review,” Insights Into Imaging 9 (2018): 961–970, https://insightsimaging.springeropen.com/articles/10.1007/s13244‐018‐0667‐5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Tirada N., Li G., Dreizin D., et al., “Digital Breast Tomosynthesis: Physics, Artifacts, and Quality Control Considerations,” Radiographics 39, no. 2 (2019): 413–426, 10.1148/rg.2019180046. [DOI] [PubMed] [Google Scholar]
- 17. Houssami N. and Skaane P., “Overview of the Evidence on Digital Breast Tomosynthesis in Breast Cancer Detection,” Breast 22, no. 2 (2013): 101–108. [DOI] [PubMed] [Google Scholar]
- 18. Kleinknecht J. H., Ciurea A. I., and Ciortea C. A., “Pros and Cons for Breast Cancer Screening With Tomosynthesis—A Review of the Literature,” Medicine and Pharmacy Reports 93 (2020): 334–335, https://medpharmareports.com/index.php/mpr/article/view/1698. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Phi X. A., Tagliafico A., Houssami N., Greuter M. J. W., and De Bock G. H., “Digital Breast Tomosynthesis for Breast Cancer Screening and Diagnosis in Women With Dense Breasts—A Systematic Review and Meta‐Analysis,” BMC Cancer 18, no. 1 (2018): 380, 10.1186/s12885-018-4263-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Kulkarni S., Freitas V., and Muradali D., “Digital Breast Tomosynthesis: Potential Benefits in Routine Clinical Practice,” Canadian Association of Radiologists Journal 73, no. 1 (2022): 107–120, 10.1177/08465371211025229. [DOI] [PubMed] [Google Scholar]
- 21. Portnow L. H., Georgian‐Smith D., Haider I., et al., “Persistent Inter‐Observer Variability of Breast Density Assessment Using BI‐RADS 5th Edition Guidelines,” Clinical Imaging 83 (2022): 21–27, https://pubmed.ncbi.nlm.nih.gov/34952487/. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Khara G., Trivedi H., Newell M. S., et al., “Generalisable Deep Learning Method for Mammographic Density Prediction Across Imaging Techniques and Self‐Reported Race,” Communications Medicine 4, no. 1 (2024): 1–8, https://www.nature.com/articles/s43856‐024‐00446‐6#citeas. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Mansour S., Soliman S., Kansakar A., Marey A., Hunold C., and Hanafy M. H., “Strengths and Challenges of the Artificial Intelligence in the Assessment of Dense Breasts,” British Journal of Radiology 4, no. 1 (2022): 20220018, https://pmc.ncbi.nlm.nih.gov/articles/PMC10958665/#:~:text=AI%20in%20breast%20cancer%20screening,second%20reader%20for%20mammographic%20images.&text=The%20diagnostic%20performance%20of%20AI,aided%20performance%20in%20dense%20breasts. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Hadadi I., Rae W., Clarke J., McEntee M., and Ekpo E., “Diagnostic Performance of Adjunctive Imaging Modalities Compared to Mammography Alone in Women With Non‐Dense and Dense Breasts: A Systematic Review and Meta‐Analysis,” Clinical Breast Cancer 21, no. 4 (2021): 278–291, https://www.sciencedirect.com/science/article/pii/S1526820921000616. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The authors have nothing to report.