Abstract
Objective:
The study aims to evaluate the diagnostic efficacy of radiologists and radiology trainees in digital breast tomosynthesis (DBT) alone vs DBT plus synthesized view (SV) for an understanding of the adequacy of DBT images to identify cancer lesions.
Methods:
Fifty-five observers (30 radiologists and 25 radiology trainees) participated in reading a set of 35 cases (15 cancer) with 28 readers reading DBT and 27 readers reading DBT plus SV. Two groups of readers had similar experience in interpreting mammograms. The performances of participants in each reading mode were compared with the ground truth and calculated in term of specificity, sensitivity, and ROC AUC. The cancer detection rate in various levels of breast density, lesion types and lesion sizes between ‘DBT’ and ‘DBT + SV’ were also analyzed. The difference in diagnostic accuracy of readers between two reading modes was assessed using Man–Whitney U test. p < 0.05 indicated a significant result.
Results:
There was no significant difference in specificity (0.67-vs-0.65; p = 0.69), sensitivity (0.77-vs-0.71; p = 0.09), ROC AUC (0.77-vs-0.73; p = 0.19) of radiologists reading DBT plus SV compared with radiologists reading DBT. Similar result was found in radiology trainees with no significant difference in specificity (0.70-vs-0.63; p = 0.29), sensitivity (0.44-vs-0.55; p = 0.19), ROC AUC (0.59-vs-0.62; p = 0.60) between two reading modes. Radiologists and trainees obtained similar results in two reading modes for cancer detection rate with different levels of breast density, cancer types and sizes of lesions (p > 0.05).
Conclusion:
Findings show that the diagnostic performances of radiologists and radiology trainees in DBT alone and DBT plus SV were equivalent in identifying cancer and normal cases.
Advances in knowledge:
DBT alone had equivalent diagnostic accuracy as DBT plus SV which could imply the consideration of using DBT as a sole modality without SV.
Introduction
Full-field digital mammography (FFDM) is widely used as the standard examination for breast cancer screening, with a reduced mortality by up to 20% due to early detection. 1 However, FFDM is a two-dimensional (2D) modality and is not as effective in dense breasts in which superimposed glandular tissue can obscure cancers, resulting in missing important findings in 15–30% of cases. 2 Digital breast tomosynthesis (DBT) is a relatively new three-dimensional (3D) technique in mammography which uses a reconstruction algorithm of low-dose 2D breast data acquired by movement of the X-ray tube in an arc ranging from 15° to 60°. This tomographic technique allows the separation of overlapping tissue, therefore improving the limiting capabilities of FFDM. 3
Since being accepted by United States Food and Drug Administration (USFDA )for clinical practice in 2011, numerous studies have investigated the efficacy of DBT. 4–12 This research includes a Japanese study that found DBTwith images taken in the mediolateral oblique (MLO) and craniocaudal (CC) planes, combined with FFDM improves breast screening performance by over 20%. 4 In comparing the cancer detection rate and recall rate across international screening programs, a meta-analysis has placed the increase in incremental cancer detection rate at 1.6 cancers per 1000 screens for European/Scandinavian screening programs with a reduction of the recall rate by 2.2% in United States screening contexts. 5 Nevertheless, the increase in reporting time by radiologists and radiation dose associated with combined FFDM plus DBT is of concern. It has been shown that the reading time for FFDM + DBT (mapped on average to be 91 s) is double than for FFDM (45 s) in a Norwegian study undertaken by Skanne et al, 6 which illustrates the significant strain on screening service resources. In addition, DBT in combination with FFDM increases radiation dose to above that of FFDM alone: one study has calculated the average glandular dose values in FFDM per exposure for the standard breast at 1.57 mGy (CC) and 1.66 mGy (MLO), while for 3D imaging mode alone was 2.19 mGy and 2.29 mGy. 7 When combined, the average glandular dose per breast could be as high as 4 mGy.
In an attempt to reduce overall radiation exposure, DBT manufacturers have developed a method to create a synthesized 2D image from the DBT data set to replace FFDM. 4,8 However, the use of synthesized view (SV) as a replacement for the FFDM view is controversial. Two studies in 2014 have reported that SV showed equivalent diagnostic performance compared to FFDM, 4,8 however, other authors reported lower sensitivity in DBT plus SV (0.77) vs DBT plus FFDM(0.83). 9 The previous analysis demonstrated that compared to FFDM, the SV image offered a better depiction of objects of certain size and contrast but provides poorer overall resolution and noise properties which could impact on the evaluation of breast lesions. 10 Although there were a number of studies 5–12 investigating the performances of radiologists in DBT vs FFDM, or FFDM vs SV, or DBT plus SV vs DBT plus FFDM, there are limited publications exploring the value of SV in addition to DBT only. Hence, the practicalities are that the SV is an additional component of DBT units and has a cost involved to integrate this module to the DBT system, and radiologists will also need to spend extra time to review/access the SV images alongside the DBT. The aim of this study was therefore to investigate the diagnostic accuracy of radiologists in reading DBT alone vs DBT with the addition of the SV, hence providing information on the adequacy of DBT images alone to detect cancer lesions.
Methods and materials
This study was conducted with data retrospectively collected via DBT self-paced evaluation module provided to Australian radiologists between March 2019 and June 2021 from the BREAST (Breastscreen REader Assessment STrategy) program. BREAST is an educational and training platform for continuous professional development of BreastScreen readers in Australia. 13,14 Human ethics institutional approval (2019/013 from the University of Sydney) was granted for the data collection of the study. Consent was sought from each participant to record their performance.
Case selection
Thirty-five cases consisting of bilateral MLO and CC DBT examinations were gathered from BreastScreen New South Wales. The cases were derived from a national screening population library with FFDM, and where DBT images were acquired after the females had been recalled to an assessment clinic. At BreastScreen Australia, the process for FFDM screening is for double blinded reading with a third reader arbitration. Background information about the cases included in this study in terms of reasons for recall, such as agreement between both or an arbitrator was provided to the senior leading radiologists in Australia (with more than 25 years in reading screening mammograms) who collected the cases for the test set, and these cases were generally considered more challenging than regular screening cases. The DBT and SV were performed prior to any core biopsy and obtained using automatic exposure control (AEC) with a Selenia Dimension machine (Hologic, Bedford, MA).
Synthesized 2D images of each patient were generated from the acquired DBT slices through the image processing software called C-View provided by Hologic. This software is an image processing application for post-processing the pixel data of DBT and does not need additional radiation exposure. Among 35 cases, 23 cases had scattered fibroglandular density and 12 cases were heterogeneously dense. There were 15 cancer cases selected (biopsy confirmation) with each having only one cancer present in the same breast. These included four discrete masses, six asymmetric density and five with other types (one calcification, one stellate and three architectural distortion). The mean size of cancer lesions was 12.5 mm. Twenty normal cases were confirmed with a 1-year negative follow-up results.
Reading procedure
Fifty-four observers (30 radiologists and 24 radiology trainees) participated in this study between March 2019 and June 2021. 28 readers (12 radiologists and 16 radiology trainees) read the DBT images alone, and 26 readers (18 radiologists and 9 radiology trainees) read DBT plus SV. Readers were shown the images of both breasts in both views (MLO and CC) and asked to find for the presence of suspicious findings linked to breast cancer. The readers did this task as an independent assessor without peer review, as this is the current process in BreastScreen Australia assessment clinics with single reading. The assessment criteria was conducted based on the RANZCR classification, 15 with scores ranging from 1 to 5: Category 1 indicates no suspicion of malignancy, Category 2 for benign findings, Category 3 for equivocal which required closer follow-up, Category 4 for suspicious lesions warranting histological confirmation and Category 5 for highly suggestive of malignant lesions with an appearance of cancer with recommendation of biopsy to confirm diagnosis. Readers were asked to give rating for each of the cases and mark any lesions on the DBT slice in which the lesion was shown most obviously (Figure 1). Radiologists and radiology trainees were also asked to provide the information about the number of years they had experience in reading DBT images. Participants were blinded to the clinical history of each case and read the cases all independently.
Figure 1.
Display of DBT slices (First row) plus SV (second row). DBT, digital breast tomosynthesis; SV, synthesized view.
Readings were conducted either at a BREAST workshop appended to a scientific meeting or the readers’ clinical screening and assessment workplaces, with DICOM images displayed on dual 5-megapixel monitors. Readers were able to change the contrast and brightness level of the images by using the window and level function available in the PACS system. The BREAST platform 13 was used to record the decision of each reader for each case. Participants were provided the instruction on viewing the images and using the software before starting the test set. Readers were allowed to backtrack and adjust their decisions. When the test set was completed, the participants could review their answers compared with the ground truth (Figure 2).
Figure 2.
A cancer is shown on DBT slice and SV. DBT, digital breast tomosynthesis; SV, synthesized view.
Statistical method
The performance datain DBT plus SV and DBT alone completed by readers in each group (radiologists and trainees) were compared with the truth and calculated in term of specificity, sensitivity and ROC AUC (receiver operating characteristic area under the curve). The rating 1 and 2 by readers were considered as normal cases while rating 3, 4 and 5 were classified as abnormal. A lesion was considered as correctly localized when the distance from the mathematical center of the lesion that a reader marked to the center of true cancer location was equal or less than radius of the true cancer lesion, which were biopsy-proven. Sensitivity is the proportion of cancer cases correctly identified as positive; specificity is the proportion of normal/benign cases correctly identified as negative. ROC AUC is acquired by combining sensitivity, specificity and confidence ratings of each reader for all cases in the test set. If readers reported multiple lesion locations in a case, only the highest rating was used for data analysis. ROC AUC values were calculated usingDorfman, Berbaum, and Metz multireader–multicase method (DBM MRMC). 16 The difference indiagnostic accuracy between two group of readers was assessed using Mann–Whitney U test. A subanalysis exploring diagnostic performances of radiologists and trainees in cancer cases with different levels of breast density, lesion types and lesion sizes between ‘DBT only’ and ‘DBT + SV’ were also analyzed using the same method. A two-tailed test with a significance level of p < 0.05 was used. The data analysis was performed with SPSS software (v. 23, Chicago, IL).
Results
Performance of radiologists in DBT+SV and DBT alone
There were 18 radiologists reading DBT plus SV and 12 radiologists reading DBT only. The average DBT experience of radiologists in DBT plus SV group was 3.2 years and 2.7 years in DBT group with no significant difference in two groups (p = 0.38). No significant difference was also found in specificity (0.67 vs 0.65; p = 0.69), sensitivity (0.77 vs 0.71; p = 0.09), ROC AUC (0.77 vs 0.73; p = 0.19) between radiologists reading DBT plus SV and the ones reading DBT only (Table 1).
Table 1.
Comparison of diagnostic performances of radiologists and radiology trainees in DBT plus SV versus DBT only
| Radiologists | DBT + SV a | DBT | P value |
|---|---|---|---|
| (18) a | (12) a | (Male Whitney-U) | |
| (Mean ± SDEV) | (Mean ± SDEV) | ||
| Specificity | 0.672 ± 0.20 | 0.646 ± 0.174 | 0.69 |
| Sensitivity | 0.769 ± 0.20 | 0.705 ± 0.117 | 0.09 |
| ROC AUC | 0.771 ± 0.11 | 0.728 ± 0.077 | 0.19 |
| Radiology trainees | DBT + SV a | DBT | P value |
|---|---|---|---|
| (9) a | (16) a | (Male Whitney-U) | |
| (Mean ± SDEV) | (Mean ± SDEV) | ||
| Specificity | 0.700 ± 0.142 | 0.631 ± 0.154 | 0.29 |
| Sensitivity | 0.441 ± 0.186 | 0.554 ± 0.186 | 0.19 |
| ROC AUC | 0.591 ± 0.082 | 0.618 ± 0.098 | 0.6 |
DBT, digital breast tomosynthesis; ROC AUC, receiver operating characteristic area under the curve; SV, synthesized view.
Number of readers
Performances of radiology trainees in DBT+SV and DBT alone
The DBT plus SV set was completed by nine trainees whilst DBT set was assessed by 16 trainees. Trainees in the first group had average experience of 9 months in reading DBT while trainees in the second group had an average of 5-month experience (p = 0.75). There was no significant difference in specificity (0.70 vs 0.63; p = 0.29), sensitivity (0.44 vs 0.55; p = 0.19), ROC AUC (0.59 vs 0.62; p = 0.60) between two groups of readers in DBT plus SV and DBT only ().
Performances of radiologists and radiology trainees in DBT+SV and DBT alone in different case features
There was no significant difference in the performances of radiologists in reading DBT + SV and DBT only in detecting cancer cases with different level of breast density: low density (p = 0.27), high density (p = 0.07), cancer cases with various lesion types (Discrete masses, p = 0.20; Stellate/Architectural distortion/Calcification, p = 0.52; Asymmetric density = 0.64) and lesion sizes (≤ 10 mm, p = 0.53; >10 mm, p = 0.16) (Table 2).
Table 2.
Diagnostic accuracy of radiologists in cancer detection among different levels of breast density, lesion types and lesion sizes between the group of reading DBT + SV and the group of reading DBT only
| Categories | DBT + SV | DBT |
p-value (Mann–Whitney U ) |
|---|---|---|---|
| (Mean ± SDEV) | (Mean ± SDEV) | ||
| Low breast density (10) | 0.739 ± 0.22 | 0.692 ± 0.138 | 0.267 |
| High breast density (5) | 0.833 ± 0.22 | 0.733 ± 0.178 | 0.068 |
| Discrete masses (4) | 0.819 ± 0.254 | 0.708 ± 0.257 | 0.197 |
| Stellate/Architectural distortion/Calcification (5) | 0.767 ± 0.268 | 0.733 ± 0.197 | 0.523 |
| Asymmetric density (6) | 0.741 ± 0.183 | 0.681 ± 0.219 | 0.639 |
| Lesion ≤10 mm (7) | 0.73 ± 0.219 | 0.679 ± 0.203 | 0.529 |
| Lesion >10 mm (8) | 0.806 ± 0.228 | 0.729 ± 0.158 | 0.157 |
DBT, digital breast tomosynthesis; SV, synthesized view.
():Number of cases
Similar to radiologists, no significant difference was shown in the performance of radiology trainees between two groups of reading DBT + SV and DBT only in detecting cancer cases withlow breast density (p = 0.79), high breast density (p = 0.12), lesion types (discrete masses, p = 0.15; stellate/architectural distortion/calcification, p = 0.98; asymmetric density, p = 0.38) and lesion sizes (≤10 mm, p = 0.40; >10 mm, p = 0.52) (Table 3).
Table 3.
Diagnostic accuracy of radiology trainees in cancer detection among different levels of breast density, lesion types and lesion sizes between the group of reading DBT + SV and the group of reading DBT only
| Categories | DBT + SV | DBT |
p-value (Mann–Whitney U) |
|---|---|---|---|
| (Mean ± SDEV) | (Mean ± SDEV) | ||
| Low breast density (10) | 0.475 ± 0.243 | 0.531 ± 0.196 | 0.793 |
| High breast density (5) | 0.375 ± 0.225 | 0.600 ± 0.273 | 0.118 |
| Discrete masses (4) | 0.594 ± 0.297 | 0.656 ± 0.256 | 0.146 |
| Stellate/Architectural distortion/Calcification (5) | 0.400 ± 0.239 | 0.475 ± 0.282 | 0.976 |
| Asymmetric density (6) | 0.375 ± 0.33 | 0.552 ± 0.284 | 0.379 |
| Lesion ≤10 mm (7) | 0.411 ± 0.235 | 0.563 ± 0.293 | 0.396 |
| Lesion >10 mm (8) | 0.469 ± 0.186 | 0.547 ± 0.193 | 0.515 |
DBT, digital breast tomosynthesis; SV, synthesized view.
* (): Number of cases
Discussion
Findings from our study illustrate that the diagnostic efficiency of reading DBT images alone (without SV view) only could be equivalent to DBT plus SV reading as shown by the reading performances of radiologists and radiology trainees in this multireader, multicase study. This result complements previous studies which have shown advantages of DBT alone 3 in improving breast cancer detection and reducing recall rates, and also in line with the demand for evaluating the contributions of modalities independently. To our knowledge, this is the first study to evaluate the diagnostic accuracy of radiologists in DBT with and without SV using self-paced educational sets of cases.
It is highly likely that DBT will be increasingly used in screening programs in the near future following its application within a diagnostic context in many countries for several years, and a great number of countries are currently undertaking large trials with DBT for screening. 3,17–19 The introduction of screen reading with DBT together with FFDM has the potential to result in an increase in workload for radiology departments and screening servicesas one study has documented a near doubling of image reading time for radiologists. 6 Our study assessed the efficacy of using DBT view only compared with DBT and additional SV, which could be a novel approach to reduce the strain on radiologists’ case load and visual fatigue. Study results show that DBT alone can be equal to DBT plus SV in both sensitivity and specificity and in detection of different types of lesions, notwithstanding the trend for higher cancer detection in dense breasts with DBT alone among radiology trainees.
In this study, we found that asymmetric density (non-specific density) was the most difficult type of cancer appearance for both radiologists and radiology trainees to detect, with only 14% of readers identifying this cancer appearance on the DBT only read and 26% of readers identifying this cancer type with the DBT and SV (Figure 3). In BI-RADS (ACR), an asymmetry is defined as a finding visible on only one mammographic projection or one breast side. An asymmetry may be a true abnormality that is obscured (masked) by adjacent isodense fibroglandular tissue or not included in the image field on one of the two standard projections or both of the breasts. Alternatively, an asymmetry may represent superimposition of normal breast structures (pseudolesion). Previous studies found that DBT was at least equivalent to additional imaging views, especially for the mammographically detected asymmetric densities. 9,20 DBT has the ability to separate mammographic planes and reduces the masking effect of overlapping tissues, so no further imaging evaluation will be needed. If, on the other hand, a true lesion cannot be excluded with confidence, the female should be recalled for diagnostic imaging evaluation.
Figure 3.
The most difficult case with low sensitivity by radiologists and trainees in both DBT alone (14%) and DBT + SV (26%) in our study. Images are authors own from BREAST. DBT, digital breast tomosynthesis; SV, synthesized view.
The question of how mammographic images should be displayed is fascinating from a perceptual standpoint. A typical FFDM or SV contains four images, MLO and CC of each breast, and readers often use a “scanning” technique in evaluating 2D images in which the eye takes in a rapid, unconscious overview of the entire breast, a holistic approach, followed by a more focused look at a smaller area closer to the region of interest as described in the visual technique by Drew et al. 21 With the DBT images, depending on slice thickness and the thickness of the breast, the radiologist can have 50 or more images to assess for each view. When reading a DBT exam, radiologists alter their accustomed FFDM search strategies to reading a stack of image slices that represent the 3D breast. A previous study 22 showed that readers have used a combination of scanning and drilling visual strategies when interpreting DBT images. “Drillers” kept the eye position limited to one area at a time while scrolling the stack of DBT slices prior to searching another region. In a similar study of lung CT, drillers overall are more likely to obtain a higher nodule detection rate and covered more of the lungs’ volume in their search when compared with scanners. 21
With the decision by the USFDA to accept DBT screening in conjunction with FFDM since 2011, the downside to this dual technique is the increase in radiation exposure to females. The dose equivalence of DBT to FFDM depends on the imaging protocol; nevertheless, when both modalities are used to undertake the females’s breast images, radiation exposure is increased significantly, and it could raise the possibility of increasing radiation-induced breast cancers. 23 Thus, synthesized 2D views reconstructed from the DBT slice data set are developed to reduce overall radiation exposure and obviate the need to take an FFDM image in addition to acquiring the DBT data. DBT performed with SV was found to be approximately dose equivalent to FFDM. 24 It has been shown that the diagnostic accuracy of interpreting SV or SV images is similar to that of FFDM, and that reader performance improved with the addition of tomosynthesis. 8 There are several studies comparing the diagnostic efficacy of SV+DBT and that of FFDM + DBT, the effectiveness of SV as a comparable imaging modality that can replace FFDM in DBT assessment is still being debated. 4,25 A mammographic phantom study by Duke University researchers (2022) reported that image quality, especially the high-contrast and spatial resolution of SVis inferior to that of FFDM. FFDM were found to be able to display more visible features due to its grayscale processing and textural differences in the background noise. 10 This is especially more obvious in small microcalcification visibility and mass circularity measurements, which quickly decrease for SV images as the object size and contrast decrease. These findings were supported by Gur et al’s study which reported lower sensitivity in DBT plus SV vs DBT plus FFDM (0.77 vs 0.83). 9 Due to different results in the effectiveness of synthesized images in previous studies, SV is currently suggested to be used only as a guide during interpretation of the tomosynthesis. 26
However, recent advances in technology have significantly improved synthesized image quality and a single technology assessment of Hologic in the Norwegian screening program has indicated that DBT with the addition of the SV does not increased dose and does increase screen-detected cancers but there is limited evidence to show any clinical effectiveness in reducing recall rates or false-positive decisions. 27 A further complementary was conducted by a group of researchers from Japan 28 comparing between SV and FFDM images, using an image set (Siemens) from 136 patients with breast cancer show that the clinical performance of SV images, even without DBT information, was not inferior to that of FFDM images in terms of detectability, probability of malignancy, or lesion conspicuity. Here, the authors found that visibility scores of abnormal calcifications in SV images were significantly higher than those in FFDM images while in dense breasts the cancer detection rate using SV images was similar as in FFDM images. It is possible that SV has the advantage of using images less affected by tissue overlapping in DBT than FFDM, without adding to the radiation dose burden. In addition, some studies showed that the diagnostic accuracy of radiologists was similar or, in some cases, significantly higher in DBT than in FFDM and implied the possibility of using DBT as a sole examination. 11,12 The findings from our study are in line with the previous finding as the performance of radiology fellows in reading DBT alone was not significantly different from the DBT plus SV group.
This study however has a number of limitations. In the first instance, the results represent a relatively small sample size with data collected through the introduction of a DBT module in the BREAST program. Further work will be required to confirm findings in a larger population with different levels of breast density and lesion types to determine the efficacy of the DBT alone compared with DBT plus 2D SV. For example, there was one case with abnormal calcifications in this set and therefore it was grouped with architectural distortion and stellate lesions to increase the statistical power. Since this study focused on DBT and SV, no prior comparison mammograms were provided to readers. The effect of prior images on the observer’s ability to detect change in the perception of developing asymmetries and in assessing probability of malignancyon DBT will require further investigation. Finally, we acknowledge that it would be a meaningful baseline if the performances of radiologists could be benchmarked with FFDM. Since there were a great number of studies previously investigating the performances of clinicians in reading FFDM compared with DBT and SVs as discussed above, our study was conducted with the aim to explore the performances of radiologists and radiology trainees in DBT compared with DBT plus SV which has not been explored before and therefore FFDM was not included in the test set. The findings from our study will be useful for a comprehensive evaluation of the mammography protocol for breast cancer screening with state-of-the-art modalities.
In summary, this study showed that reading DBT alone had equivalent diagnostic accuracy as DBT plus SV for radiologists and radiology trainees. The results imply the robustness of DBT for a consideration to use as a sole modality without SV. Further investigation with larger sample size will be needed to confirm the findings and explore factors related to viewing and perception affect readers' ability to detect breast cancer.
Footnotes
Acknowledgment: The authors would like to acknowledge the BREAST (BreastscreenREader Assessment STrategy) project through the New South Wales Cancer Institute grant (2017-2020) and Department of Health grant (2018-2024) for support and facilitating the data collection opportunity.
Competing interests: All authors have no conflicts of interest related to the work described in the paper.
Funding: Open access publishing facilitated by The University of Sydney, as part of the Wiley - The University of Sydney agreement via the Council of Australian University Librarians.
Contributor Information
Phuong Dung (Yun) Trieu, Email: phuong.trieu@sydney.edu.au.
Jennie Noakes, Email: jennie.noakes@sydney.edu.au.
Tong Li, Email: t.li@sydney.edu.au.
Natacha Borecky, Email: natacha.borecky@sydney.edu.au.
Patrick C Brennan, Email: patrick.brennan@sydney.edu.au.
Melissa L Barron, Email: melissa.robinson@sydney.edu.au.
Sarah J Lewis, Email: sarah.lewis@sydney.edu.au.
REFERENCES
- 1. The benefits and harms of breast cancer screening: an independent review. The Lancet 2012; 380: 1778–86. doi: 10.1016/S0140-6736(12)61611-0 [DOI] [PubMed] [Google Scholar]
- 2. Carney PA, Miglioretti DL, Yankaskas BC, Kerlikowske K, Rosenberg R, Rutter CM, et al. Individual and combined effects of age, breast density, and hormone replacement therapy use on the accuracy of screening mammography. Ann Intern Med 2003; 138: 168–75. doi: 10.7326/0003-4819-138-3-200302040-00008 [DOI] [PubMed] [Google Scholar]
- 3. Gilbert FJ, Tucker L, Young KC. Digital breast tomosynthesis (DBT): a review of the evidence for use as a screening tool. Clin Radiol 2016; 71: 141–50: S0009-9260(15)00445-6. doi: 10.1016/j.crad.2015.11.008 [DOI] [PubMed] [Google Scholar]
- 4. Skaane P, Bandos AI, Eben EB, Jebsen IN, Krager M, Haakenaasen U, et al. Two-View digital breast tomosynthesis screening with synthetically reconstructed projection images: comparison with digital breast tomosynthesis with full-field digital mammographic images. Radiology 2014; 271: 655–63. doi: 10.1148/radiol.13131391 [DOI] [PubMed] [Google Scholar]
- 5. Marinovich ML, Hunter KE, Macaskill P, Houssami N. Breast cancer screening using tomosynthesis or mammography: a meta-analysis of cancer detection and recall. JNCI 2018; 110: 942–49. doi: 10.1093/jnci/djy121 [DOI] [PubMed] [Google Scholar]
- 6. Skaane P, Bandos AI, Gullien R, Eben EB, Ekseth U, Haakenaasen U, et al. Comparison of digital mammography alone and digital mammography plus tomosynthesis in a population-based screening program.Radiology 2013; 267(1):47-56. [DOI] [PubMed] [Google Scholar]
- 7. Olgar T, Kahn T, Gosch D.. Average glandular dose in digital mammography and breast tomosynthesis.Rofo 2012; 184(10):911-8. [DOI] [PubMed] [Google Scholar]
- 8. Zuley ML, Guo B, Catullo VJ, Chough DM, Kelly AE, Lu AH, et al. Comparison of two-dimensional synthesized mammograms versus original digital mammograms alone and in combination with tomosynthesis images. Radiology 2014; 271: 664–71. doi: 10.1148/radiol.13131530 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Gur D, Zuley ML, Anello MI, Rathfon GY, Chough DM, Ganott MA, et al. Dose reduction in digital breast tomosynthesis (DBT) screening using synthetically reconstructed projection images: an observer performance study. Acad Radiol 2012; 19: 166–71. doi: 10.1016/j.acra.2011.10.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. How does c-view image quality compare with conventional 2D FFDM?med phys. 2016; 43(5). [DOI] [PubMed] [Google Scholar]
- 11. Giess CS, Pourjabbar S, Ip IK, Lacson R, Alper E, Khorasani R. Comparing diagnostic performance of digital breast tomosynthesis and full-field digital mammography in a hybrid screening environment. AJR Am J Roentgenol 2017; 209: 929–34. doi: 10.2214/AJR.17.17983 [DOI] [PubMed] [Google Scholar]
- 12. Georgian-Smith D, Obuchowski NA, Lo JY, Brem RF, Baker JA, Fisher PR, et al. Can digital breast tomosynthesis replace full-field digital mammography? A multireader, multicase study of wide-angle tomosynthesis. AJR Am J Roentgenol 2019; 2019: 1–7. doi: 10.2214/AJR.18.20294 [DOI] [PubMed] [Google Scholar]
- 13. Brennan PC, Trieu PD, Tapia K, Ryan J, Mello-Thoms C, Lee W. BREAST: A Novel Strategy to Improve the Detection of BREAST Cancer. In: TH HF, Muramatsu C, eds. Lecture Notes in Computer Science - The 12th International Workshop on Breast Imaging. 8539: Springer; 2014, pp. 438–43. [Google Scholar]
- 14. Trieu PDY, Tapia K, Frazer H, Lee W, Brennan P. Improvement of cancer detection on mammograms via breast test sets. Acad Radiol 2019; 26: e341–47: S1076-6332(19)30010-8. doi: 10.1016/j.acra.2018.12.017 [DOI] [PubMed] [Google Scholar]
- 15. Breastscreen_Australia . Breastscreen National Accreditation Standards. In: In: Health Do, editor. Canberra: Department of Health; 2019. [Google Scholar]
- 16. Dorfman DD, Berbaum KS, Metz CE.. Receiver operating characteristic rating analysis. Generalization to the population of readers and patients with the jackknife method.Invest Radiol 1992; 27(9):723-31. [PubMed] [Google Scholar]
- 17. Sharma N, McMahon M, Haigh I, Chen Y, Dall BJG. The potential impact of digital breast tomosynthesis on the benign biopsy rate in women recalled within the UK breast screening programme. Radiology 2019; 291: 310–17. doi: 10.1148/radiol.2019180809 [DOI] [PubMed] [Google Scholar]
- 18. Lowry KP, Coley RY, Miglioretti DL, Kerlikowske K, Henderson LM, Onega T, et al. Screening performance of digital breast tomosynthesis vs digital mammography in community practice by patient age, screening round, and breast density. JAMA Netw Open 2020; 3(7): e2011792. doi: 10.1001/jamanetworkopen.2020.11792 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Houssami N, Lockie D, Clemson M, Pridmore V, Taylor D, Marr G, et al. Pilot trial of digital breast tomosynthesis (3D mammography) for population-based screening in breastscreen victoria. Med J Aust 2019; 211: 357–62. doi: 10.5694/mja2.50320 [DOI] [PubMed] [Google Scholar]
- 20. Noroozian M, Hadjiiski L, Rahnama-Moghadam S, Klein KA, Jeffries DO, Pinsky RW, et al. Digital breast tomosynthesis is comparable to mammographic spot views for mass characterization.Radiology 2012; 262(1):61-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Drew T, Vo ML, Olwal A, Jacobson F, Seltzer SE, Wolfe JM.. Scanners and drillers: characterizing expert visual search through volumetric images.J Vis 2013; 13(10). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Aizenman A, Drew T, Ehinger KA, Georgian-Smith D, Wolfe JM. Comparing search patterns in digital breast tomosynthesis and full-field digital mammography: an eye tracking study.J med imaging (bellingham). 2017; 4(4). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Svahn TM, Houssami N, Sechopoulos I, Mattsson S.. Review of radiation dose estimates in digital breast tomosynthesis relative to those in two-view full-field digital mammography.Breast 2015; 24(2):93-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Nelson JS, Wells JR, Baker JA, Samei E. How does c-view image quality compare with conventional 2D FFDM? Med Phys 2016; 43: 2538–47: 2538. doi: 10.1118/1.4947293 [DOI] [PubMed] [Google Scholar]
- 25. Gilbert FJ, Tucker L, Gillan MG, Willsher P, Cooke J, Duncan KA, et al. The TOMMY trial: a comparison of tomosynthesis with digital mammography in the UK NHS breast screening programme--a multicentre retrospective reading study comparing the diagnostic performance of digital breast tomosynthesis and digital mammography with digital mammography alone.health technol assess. 2015; 19: 1–136. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. U.S.-Food-Drug-Administration . Hologic Selenia Dimensions C-view Software Module. In: In: Summary SE, editor. US: FDA; 2013. [Google Scholar]
- 27. Movik E, Dalsbo TK, Fagelund BC, Friberg EG, Haheim LL, Skar A. n.d.).( Digital breast tomosynthesis with hologic 3D mammography selenia dimensions system for use in breast cancer screening: A single technology assessment. NIPH Systematic Reviews Oslo. [PubMed] [Google Scholar]
- 28. Murakami R, Uchiyama N, Tani H, Yoshida T, Kumita S. Comparative analysis between synthetic mammography reconstructed from digital breast tomosynthesis and full-field digital mammography for breast cancer detection and visibility. Eur J Radiol Open 2020; 7: 100207. doi: 10.1016/j.ejro.2019.12.001 [DOI] [PMC free article] [PubMed] [Google Scholar]