Outcomes of Breast Cancer Screening Strategies Based on Cancer Intervention and Surveillance Modeling Network Estimates.

Background There is ongoing debate about recommendations for breast cancer screening strategies, specifically regarding the frequency of screening and the age at which to initiate screening. Purpose To compare estimates of breast cancer screening outcomes published by the Cancer Intervention and Surveillance Modeling Network (CISNET) to understand the benefits and risks of different screening scenarios. Materials and Methods Modeling estimates published by CISNET are based on hypothetical cohorts in the United States and compare women, starting at 40 years of age, who do and do not undergo breast cancer screening with mammography. The four scenarios assessed in this study, of multiple possible scenarios, were biennial screening ages 50-74 years (2009 and 2016 U.S. Preventive Services Task Force [USPSTF] recommendations), biennial screening ages 40-74 years (2023 USPSTF draft recommendation), annual screening ages 40-74 years, and annual screening ages 40-79 years. For each scenario, CISNET estimates of median lifetime benefits were compared. Risks that included false-positive screening results per examination and benign biopsies per examination were also calculated and compared. Results Estimates from CISNET 2023 showed that annual screening ages 40-79 years improved breast cancer mortality reduction compared with biennial screening ages 50-74 years and biennial screening ages 40-74 years (41.7%, 25.4%, and 30%, respectively). Annual screening ages 40-79 years averted the most breast cancer deaths (11.5 per 1000) and gained the most life-years (230 per 1000) compared with other screening scenarios (range, 6.7-11.5 per 1000 and 121-230 per 1000, respectively). False-positive screening results per examination were less than 10% for all screening scenarios (range, 6.5%-9.6%) and lowest for annual screening ages 40-79 years (6.5%). Benign biopsies per examination were less than 1.33% for all screening scenarios (range, 0.88%-1.32%) and lowest for annual screening ages 40-79 years (0.88%). Conclusion CISNET 2023 modeling estimates indicate that annual breast cancer screening starting at 40 years of age provides the greatest benefit to women and the least risk per examination. © RSNA, 2024 See also the editorial by Joe in this issue.

Deep learning denoising of digital breast tomosynthesis: Observer performance study of the effect on detection of microcalcifications in breast phantom images.

BACKGROUND: The noise in digital breast tomosynthesis (DBT) includes x-ray quantum noise and detector readout noise. The total radiation dose of a DBT scan is kept at about the level of a digital mammogram but the detector noise is increased due to acquisition of multiple projections. The high noise can degrade the detectability of subtle lesions, specifically microcalcifications (MCs). PURPOSE: We previously developed a deep-learning-based denoiser to improve the image quality of DBT. In the current study, we conducted an observer performance study with breast radiologists to investigate the feasibility of using deep-learning-based denoising to improve the detection of MCs in DBT. METHODS: We have a modular breast phantom set containing seven 1-cm-thick heterogeneous 50% adipose/50% fibroglandular slabs custom-made by CIRS, Inc. (Norfolk, VA). We made six 5-cm-thick breast phantoms embedded with 144 simulated MC clusters of four nominal speck sizes (0.125-0.150, 0.150-0.180, 0.180-0.212, 0.212-0.250 mm) at random locations. The phantoms were imaged with a GE Pristina DBT system using the automatic standard (STD) mode. The phantoms were also imaged with the STD+ mode that increased the average glandular dose by 54% to be used as a reference condition for comparison of radiologists' reading. Our previously trained and validated denoiser was deployed to the STD images to obtain a denoised DBT set (dnSTD). Seven breast radiologists participated as readers to detect the MCs in the DBT volumes of the six phantoms under the three conditions (STD, STD+, dnSTD), totaling 18 DBT volumes. Each radiologist read all the 18 DBT volumes sequentially, which were arranged in a different order for each reader in a counter-balanced manner to minimize any potential reading order effects. They marked the location of each detected MC cluster and provided a conspicuity rating and their confidence level for the perceived cluster. The visual grading characteristics (VGC) analysis was used to compare the conspicuity ratings and the confidence levels of the radiologists for the detection of MCs. RESULTS: The average sensitivities over all MC speck sizes were 65.3%, 73.2%, and 72.3%, respectively, for the radiologists reading the STD, dnSTD, and STD+ volumes. The sensitivity for dnSTD was significantly higher than that for STD (p < 0.005, two-tailed Wilcoxon signed rank test) and comparable to that for STD+. The average false positive rates were 3.9 ± 4.6, 2.8 ± 3.7, and 2.7 ± 3.9 marks per DBT volume, respectively, for reading the STD, dnSTD, and STD+ images but the difference between dnSTD and STD or STD+ did not reach statistical significance. The overall conspicuity ratings and confidence levels by VGC analysis for dnSTD were significantly higher than those for both STD and STD+ (p ≤ 0.001). The critical alpha value for significance was adjusted to be 0.025 with Bonferroni correction. CONCLUSIONS: This observer study using breast phantom images showed that deep-learning-based denoising has the potential to improve the detection of MCs in noisy DBT images and increase radiologists' confidence in differentiating noise from MCs without increasing radiation dose. Further studies are needed to evaluate the generalizability of these results to the wide range of DBTs from human subjects and patient populations in clinical settings.

Breast cancer screening in average and high-risk women.

Breast cancer is the most common cancer among females worldwide with rising incidence. In the United States, screening mammography and advances in therapy have lowered mortality by 41% since 1990. Screening mammography is supported by randomized control trials (RCT), observational studies, and computer model data. Digital breast tomosynthesis is a new technology that addresses limitations in mammography resulting from overlapping breast tissue, improving its sensitivity and specificity. Patients at high risk for breast cancer include those with a ≥20% lifetime risk, high-risk germline mutation, or history of thoracic radiation treatment between 10-30 years of age. Such patients are recommended to undergo annual screening mammography and adjunctive annual screening breast MRI. Patients unable to undergo MRI may undergo whole breast ultrasound or contrast-enhanced mammography. Pregnant and lactating patients at average risk for breast cancer are recommended to undergo age-appropriate screening mammography.

Effect of Dose Level on Radiologists' Detection of Microcalcifications in Digital Breast Tomosynthesis: An Observer Study with Breast Phantoms.

OBJECTIVES: To compare radiologists' sensitivity, confidence level, and reading efficiency of detecting microcalcifications in digital breast tomosynthesis (DBT) at two clinically relevant dose levels. MATERIALS AND METHODS: Six 5-cm-thick heterogeneous breast phantoms embedded with a total of 144 simulated microcalcification clusters of four speck sizes were imaged at two dose modes by a clinical DBT system. The DBT volumes at the two dose levels were read independently by six MQSA radiologists and one fellow with 1-33 years (median 12 years) of experience in a fully-crossed counter-balanced manner. The radiologist located each potential cluster and rated its conspicuity and his/her confidence that the marked location contained a cluster. The differences in the results between the two dose modes were analyzed by two-tailed paired t-test. RESULTS: Compared to the lower-dose mode, the average glandular dose in the higher-dose mode for the 5-cm phantoms increased from 1.34 to 2.07 mGy. The detection sensitivity increased for all speck sizes and significantly for the two smaller sizes (p <0.05). An average of 13.8% fewer false positive clusters was marked. The average conspicuity rating and the radiologists' confidence level were higher for all speck sizes and reached significance (p <0.05) for the three larger sizes. The average reading time per detected cluster reduced significantly (p <0.05) by an average of 13.2%. CONCLUSION: For a 5-cm-thick breast, an increase in average glandular dose from 1.34 to 2.07 mGy for DBT imaging increased the conspicuity of microcalcifications, improved the detection sensitivity by radiologists, increased their confidence levels, reduced false positive detections, and increased the reading efficiency.

Annual Screening Mammography Associated With Lower Stage Breast Cancer Compared With Biennial Screening.

OBJECTIVE. The purpose of this study was to compare breast cancer characteristics and treatment regimens among women undergoing annual versus nonannual screening mammography. MATERIALS AND METHODS. In this retrospective, institutional review board-approved, HIPAA-compliant cohort study, a breast cancer database was queried for patients who received a mammographic or clinical diagnosis of breast cancer during 2016-2017. Annual versus biennial and annual versus nonannual (biennial and triennial) mammography screening cohorts were compared using t tests or Wilcoxon rank sum tests for continuous variables and chi-square or Fisher exact tests for categoric variables. RESULTS. A total of 490 patients were diagnosed with breast cancer during 2016-2017. Among these women, 245 had an assignable screening frequency and were 40-84 years old (mean, 61.8 ± 9.9 [SD] years; median, 62 years). Screening frequency was annual for 200 of these 245 patients (81.6%), biennial for 32 (13.1%), and triennial for 13 (5.3%). Annual screening resulted in fewer late-stage presentations (AJCC stage II, III, or IV in 48 of 200 patients undergoing annual [24.0%] vs 14 of 32 undergoing biennial [43.8%; p = .02] and vs 20 of 45 undergoing nonannual screening [44.4%; p = .006]), fewer interval cancers (21 of 200 for annual [10.5%] vs 12 of 32 for biennial [37.5%; p < .001] and vs 15 of 45 for nonannual [33.3%; p < .001]), and smaller mean tumor diameter (1.4 ± 1.2 cm for annual vs 1.8 ± 1.6 cm for biennial [p = .04] and vs 1.8 ± 1.5 cm nonannual [p = .03]). Lower AJCC stage, fewer interval cancers, and smaller tumor diameter also persisted among postmenopausal women undergoing annual screening. Patients undergoing biennial and nonannual screening showed nonsignificant greater use of axillary lymph node dissection (annual, 24 of 200 [12.0%]; biennial, 6 of 32 [18.8%]; nonannual, 7 of 45 [15.6%]) and chemotherapy (annual, 55 of 200 [27.5%]; biennial, 12 of 32 [37.5%]; nonannual, 16 of 45 [35.6%]). CONCLUSION. Annual mammographic screening was associated with lower breast cancer stage and fewer interval cancers than biennial or nonannual screening.

Digital Mammography Has Persistently Increased High-Grade and Overall DCIS Detection Without Altering Upgrade Rate.

OBJECTIVE. The purpose of this article is to evaluate whether digital mammography (DM) is associated with persistent increased detection of ductal carcinoma in situ (DCIS) or has altered the upgrade rate of DCIS to invasive cancer. MATERIALS AND METHODS. An institutional review board-approved retrospective search identified DCIS diagnosed in women with mammographic calcifications between 2001 and 2014. Ipsilateral cancer within 2 years, masses, papillary DCIS, and patients with outside imaging were excluded, yielding 484 cases. Medical records were reviewed for mammographic calcifications, technique, and pathologic diagnosis. Mammograms were interpreted by radiologists certified by the Mammography Quality Standards Act. The institution transitioned from film-screen mammography (FSM) to exclusive DM by 2010. Statistical analyses were performed using chi-square test. RESULTS. Of 484 DCIS cases, 158 (33%) were detected by FSM and 326 (67%) were detected by DM. The detection rate was higher with DM than FSM (1.4 and 0.7 per 1000, respectively; p < .001). The detection rate of high-grade DCIS doubled with DM compared with FSM (0.8 and 0.4 per 1000, respectively; p < .001). The prevalent peak of DM-detected DCIS was 2.7 per 1000 in 2008. Incident DM detection remained double FSM (1.4 vs 0.7 per 1000). Similar proportions of high-grade versus low- to intermediate-grade DCIS were detected with both modalities. There was no significant difference in the upgrade rate of DCIS to invasive cancer between DM (10%; 34/326) and FSM (10%; 15/158) (p = .74). High-grade DCIS led to 71% (35/49) of the upgrades to invasive cancer. CONCLUSION. DM was associated with a significant doubling in DCIS and high-grade DCIS detection, which persisted after prevalent peak. The majority of upgrades to invasive cancer arose from high-grade DCIS. DM was not associated with decreased upgrade to invasive cancer.

Breast Cancer Mortality Rates Have Stopped Declining in U.S. Women Younger than 40 Years.

Background National Center for Health Statistics (NCHS) data for U.S. women have shown a steady decline in breast cancer mortality rates since 1989. Purpose To analyze U.S. breast cancer mortality rates by age decade in women aged 20-79 years and in women aged 20-39 years and women aged 40-69 years. Materials and Methods The authors conducted a retrospective analysis of (a) female breast cancer mortality rates from NCHS data for 1969-2017 for all races and by race and (b) age- and delay-adjusted invasive breast cancer incidence rates from the Surveillance, Epidemiology, and End Results program. Joinpoint analysis was used to determine trends in breast cancer mortality, invasive breast cancer incidence, and distant-stage (metastatic) breast cancer incidence rates. Results Between 1989 and 2010, breast cancer mortality rates decreased by 1.5%-3.4% per year for each age decade from 20 to 79 years (P < .001 for each). After 2010, breast cancer mortality rates continued to decline by 1.2%-2.2% per year in women in each age decade from 40 to 79 years (P < .001 for each) but stopped declining in women younger than 40 years. After 2010, breast cancer mortality rates demonstrated nonsignificant increases of 2.8% per year in women aged 20-29 years (P = .11) and 0.3% per year in women aged 30-39 years (P = .70), results attributable primarily to changes in mortality rates in White women. A contributing factor is that distant-stage breast cancer incidence rates increased by more than 4% per year after the year 2000 in women aged 20-39 years. Conclusion Female breast cancer mortality rates have stopped declining in women younger than 40 years, ending a trend that existed from 1987 to 2010. Conversely, mortality rates have continued to decline in women aged 40-79 years. Rapidly rising distant-stage breast cancer rates have likely contributed to ending the decline in mortality rates in women younger than 40 years. © RSNA, 2021 Online supplemental material is available for this article.

Risks of feature leakage and sample size dependencies in deep feature extraction for breast mass classification.

PURPOSE: Transfer learning is commonly used in deep learning for medical imaging to alleviate the problem of limited available data. In this work, we studied the risk of feature leakage and its dependence on sample size when using pretrained deep convolutional neural network (DCNN) as feature extractor for classification breast masses in mammography. METHODS: Feature leakage occurs when the training set is used for feature selection and classifier modeling while the cost function is guided by the validation performance or informed by the test performance. The high-dimensional feature space extracted from pretrained DCNN suffers from the curse of dimensionality; feature subsets that can provide excessively optimistic performance can be found for the validation set or test set if the latter is allowed for unlimited reuse during algorithm development. We designed a simulation study to examine feature leakage when using DCNN as feature extractor for mass classification in mammography. Four thousand five hundred and seventy-seven unique mass lesions were partitioned by patient into three sets: 3222 for training, 508 for validation, and 847 for independent testing. Three pretrained DCNNs, AlexNet, GoogLeNet, and VGG16, were first compared using a training set in fourfold cross validation and one was selected as the feature extractor. To assess generalization errors, the independent test set was sequestered as truly unseen cases. A training set of a range of sizes from 10% to 75% was simulated by random drawing from the available training set in addition to 100% of the training set. Three commonly used feature classifiers, the linear discriminant, the support vector machine, and the random forest were evaluated. A sequential feature selection method was used to find feature subsets that could achieve high classification performance in terms of the area under the receiver operating characteristic curve (AUC) in the validation set. The extent of feature leakage and the impact of training set size were analyzed by comparison to the performance in the unseen test set. RESULTS: All three classifiers showed large generalization error between the validation set and the independent sequestered test set at all sample sizes. The generalization error decreased as the sample size increased. At 100% of the sample size, one classifier achieved an AUC as high as 0.91 on the validation set while the corresponding performance on the unseen test set only reached an AUC of 0.72. CONCLUSIONS: Our results demonstrate that large generalization errors can occur in AI tools due to feature leakage. Without evaluation on unseen test cases, optimistically biased performance may be reported inadvertently, and can lead to unrealistic expectations and reduce confidence for clinical implementation.

Overdiagnosis and Risks of Breast Cancer Screening.

Since its widespread introduction 30 years ago, screening mammography has contributed to substantial reduction in breast cancer-associated mortality, ranging from 15% to 50% in observational trials. It is currently the best examination available for the early diagnosis of breast cancer, when survival and treatment options are most favorable. However, like all medical tests and procedures, screening mammography has associated risks, including overdiagnosis and overtreatment, false-positive examinations, false-positive biopsies, and radiation exposure. Women should be aware of the benefits and risks of screening mammography in order to make the most appropriate care decisions for themselves.

Digital Breast Tomosynthesis Slab Thickness: Impact on Reader Performance and Interpretation Time.

Background Digital breast tomosynthesis (DBT) helps reduce recall rates and improve cancer detection compared with two-dimensional (2D) mammography but has a longer interpretation time. Purpose To evaluate the effect of DBT slab thickness and overlap on reader performance and interpretation time in the absence of 1-mm slices. Materials and Methods In this retrospective HIPAA-compliant multireader study of DBT examinations performed between August 2013 and July 2017, four fellowship-trained breast imaging radiologists blinded to final histologic findings interpreted DBT examinations by using a standard protocol (10-mm slabs with 5-mm overlap, 1-mm slices, synthetic 2D mammogram) and an experimental protocol (6-mm slabs with 3-mm overlap, synthetic 2D mammogram) with a crossover design. Among the 122 DBT examinations, 74 mammographic findings had final histologic findings, including 31 masses (26 malignant), 20 groups of calcifications (12 malignant), 18 architectural distortions (15 malignant), and five asymmetries (two malignant). Durations of reader interpretations were recorded. Comparisons were made by using receiver operating characteristic curves for diagnostic performance and paired t tests for continuous variables. Results Among 122 women, mean age was 58.6 years ± 10.1 (standard deviation). For detection of malignancy, areas under the receiver operating characteristic curves were similar between protocols (range, 0.83-0.94 vs 0.84-0.92; P ≥ .63). Mean DBT interpretation time was shorter with the experimental protocol for three of four readers (reader 1, 5.6 minutes ± 1.7 vs 4.7 minutes ± 1.4 [P < .001]; reader 2, 2.8 minutes ± 1.1 vs 2.3 minutes ± 1.0 [P = .001]; reader 3, 3.6 minutes ± 1.4 vs 3.3 minutes ± 1.3 [P = .17]; reader 4, 4.3 minutes ± 1.0 vs 3.8 minutes ± 1.1 [P ≤ .001]), with 72% reduction in both mean number of images and mean file size (P < .001 for both). Conclusion A digital breast tomosynthesis reconstruction protocol that uses 6-mm slabs with 3-mm overlap, without 1-mm slices, had similar diagnostic performance compared with the standard protocol and led to a reduced interpretation time for three of four readers. © RSNA, 2020 See also the editorial by Chang in this issue.

Generalization error analysis for deep convolutional neural network with transfer learning in breast cancer diagnosis.

Deep convolutional neural network (DCNN), now popularly called artificial intelligence (AI), has shown the potential to improve over previous computer-assisted tools in medical imaging developed in the past decades. A DCNN has millions of free parameters that need to be trained, but the training sample set is limited in size for most medical imaging tasks so that transfer learning is typically used. Automatic data mining may be an efficient way to enlarge the collected data set but the data can be noisy such as incorrect labels or even a wrong type of image. In this work we studied the generalization error of DCNN with transfer learning in medical imaging for the task of classifying malignant and benign masses on mammograms. With a finite available data set, we simulated a training set containing corrupted data or noisy labels. The balance between learning and memorization of the DCNN was manipulated by varying the proportion of corrupted data in the training set. The generalization error of DCNN was analyzed by the area under the receiver operating characteristic curve for the training and test sets and the weight changes after transfer learning. The study demonstrates that the transfer learning strategy of DCNN for such tasks needs to be designed properly, taking into consideration the constraints of the available training set having limited size and quality for the classification task at hand, to minimize memorization and improve generalizability.

Prospective Imaging Trial Assessing Gadoteridol Retention in the Deep Brain Nuclei of Women Undergoing Breast MRI.

RATIONALE AND OBJECTIVES: To assess for indirect evidence of gadoteridol retention in the deep brain nuclei of women undergoing serial screening breast MRI. METHODS: This HIPAA-compliant prospective observational noninferiority imaging trial was approved by the IRB. From December 2016 to March 2018, 12 consented subjects previously exposed to 0-1 doses of gadoteridol (group 1) and 7 consented subjects previously exposed to ≥4 doses of gadoteridol (group 2) prospectively underwent research-specific unenhanced brain MRI including T1w spin echo imaging and T1 mapping. Inclusion criteria were: (1) planned breast MRI with gadoteridol, (2) no gadolinium exposure other than gadoteridol, (3) able to undergo MRI, (4) no neurological illness, (5) no metastatic disease, (6) no chemotherapy. Regions of interest were manually drawn in the globus pallidus, thalamus, dentate nucleus, and pons. Globus pallidus/thalamus and dentate nucleus/pons signal intensities and T1-time ratios were calculated using established methods and correlated with cumulative gadoteridol dose (mL). RESULTS: All subjects were female (mean age: 50 ± 12 years) and previously had received an average of 0.5 ± 0.5 (group 1) and 5.9 ± 2.1 (group 2) doses of gadoteridol (cumulative dose: 8 ± 8 and 82 ± 31 mL, respectively), with the last dose an average of 492 ± 299 days prior to scanning. There was no significant correlation between cumulative gadoteridol dose (mL) and deep brain nuclei signal intensity at T1w spin echo imaging (p = 0.365-0.512) or T1 mapping (p = 0.197-0.965). CONCLUSION: We observed no indirect evidence of gadolinium retention in the deep brain nuclei of women undergoing screening breast MRI with gadoteridol.

Prostate MRI and quality: lessons learned from breast imaging rad-path correlation.

Radiologic-pathologic correlation of Prostate Imaging Reporting & Data System (PI-RADS) scores ensures local quality and offers opportunities to improve future iterations of the reporting system. Tracking positive predictive values of lesion-targeted biopsies helps provide generalizable population-level risks associated with each PI-RADS category and can highlight the sources of variation. While this process of formalized pathologic correlation is somewhat new to abdominal radiology, we are fortunate to have a model to follow which was developed by our colleagues in breast imaging. If the success and multi-national adoption of BI-RADS is an indicator, building a scoring system anchored on a histologic reference is an important step to ensuring diagnostic quality and reliability.

Breast Cancer Diagnosis in Digital Breast Tomosynthesis: Effects of Training Sample Size on Multi-Stage Transfer Learning Using Deep Neural Nets.

In this paper, we developed a deep convolutional neural network (CNN) for the classification of malignant and benign masses in digital breast tomosynthesis (DBT) using a multi-stage transfer learning approach that utilized data from similar auxiliary domains for intermediate-stage fine-tuning. Breast imaging data from DBT, digitized screen-film mammography, and digital mammography totaling 4039 unique regions of interest (1797 malignant and 2242 benign) were collected. Using cross validation, we selected the best transfer network from six transfer networks by varying the level up to which the convolutional layers were frozen. In a single-stage transfer learning approach, knowledge from CNN trained on the ImageNet data was fine-tuned directly with the DBT data. In a multi-stage transfer learning approach, knowledge learned from ImageNet was first fine-tuned with the mammography data and then fine-tuned with the DBT data. Two transfer networks were compared for the second-stage transfer learning by freezing most of the CNN structures versus freezing only the first convolutional layer. We studied the dependence of the classification performance on training sample size for various transfer learning and fine-tuning schemes by varying the training data from 1% to 100% of the available sets. The area under the receiver operating characteristic curve (AUC) was used as a performance measure. The view-based AUC on the test set for single-stage transfer learning was 0.85 ± 0.05 and improved significantly (p <; 0.05$ ) to 0.91 ± 0.03 for multi-stage learning. This paper demonstrated that, when the training sample size from the target domain is limited, an additional stage of transfer learning using data from a similar auxiliary domain is advantageous.

Breast cancer deaths averted over 3 decades.

BACKGROUND: From 1975 to 1990, female breast cancer mortality rates in the United States increased by 0.4% per year. Since 1990, breast cancer mortality rates have fallen between 1.8% and 3.4% per year, a decrease that is attributed to increased mammography screening and improved treatment. METHODS: The authors used age-adjusted female breast cancer mortality rate and population data from the Surveillance, Epidemiology, and End Results (SEER) program to estimate the number of breast cancer deaths averted by screening mammography and improved treatment since 1989. Four different assumptions regarding background mortality rates (in the absence of screening mammography and improved treatment) were used to estimate deaths averted for women aged 40 to 84 years by taking the difference between SEER-reported mortality rates and background mortality rates for each 5-year age group, multiplied by the population for each 5-year age group. SEER data were used to estimate annual and cumulative breast cancer deaths averted in 2012 and 2015 and extrapolated SEER data were used to estimate deaths averted in 2018. RESULTS: The number of single-year breast cancer deaths averted ranged from 20,860 to 33,842 in 2012, from 23,703 to 39,415 in 2015, and from 27,083 to 45,726 in 2018. Breast cancer mortality reductions ranged from 38.6% to 50.5% in 2012, from 41.5% to 54.2% in 2015, and from 45.3% to 58.3% in 2018. Cumulative breast cancer deaths averted since 1989 ranged from 237,234 to 370,402 in 2012, from 305,934 to 483,435 in 2015, and from 384,046 to 614,484 in 2018. CONCLUSIONS: Since 1989, between 384,000 and 614,500 breast cancer deaths have been averted through the use of mammography screening and improved treatment.

Automated pectoral muscle identification on MLO-view mammograms: Comparison of deep neural network to conventional computer vision.

OBJECTIVES: The aim of this study was to develop a fully automated deep learning approach for identification of the pectoral muscle on mediolateral oblique (MLO) view mammograms and evaluate its performance in comparison to our previously developed texture-field orientation (TFO) method using conventional image feature analysis. Pectoral muscle segmentation is an important step for automated image analyses such as breast density or parenchymal pattern classification, lesion detection, and multiview correlation. MATERIALS AND METHODS: Institutional Review Board (IRB) approval was obtained before data collection. A dataset of 729 MLO-view mammograms including 637 digitized film mammograms (DFM) and 92 digital mammograms (DM) from our previous study was used for the training and validation of our deep convolutional neural network (DCNN) segmentation method. In addition, we collected an independent set of 203 DMs from 131 patients for testing. The film mammograms were digitized at a pixel size of 50 µm × 50 µm with a Lumiscan digitizer. All DMs were acquired with GE systems at a pixel size of 100 µm × 100 µm. An experienced MQSA radiologist manually drew the pectoral muscle boundary on each mammogram as the reference standard. We trained the DCNN to estimate a probability map of the pectoral muscle region on mammograms. The DCNN consisted of a contracting path to capture multiresolution image context and a symmetric expanding path for prediction of the pectoral muscle region. Three DCNN structures were compared for automated identification of pectoral muscles. Tenfold cross-validation was used in training of the DCNNs. After training, we applied the ten trained models during cross-validation to the independent DM test set. The predicted pectoral muscle region of each test DM was obtained as the mean probability map by averaging the ensemble of probability maps from the ten models. The DCNN-segmented pectoral muscle was evaluated by three performance measures relative to the reference standard: (a) the percent overlap area (POA) of the pectoral muscle regions, (b) the Hausdorff distance (Hdist), and (c) the average Euclidean distance (AvgDist) between the boundaries. The results were compared to those obtained with the TFO method, used as our baseline. A two-tailed paired t test was performed to examine the significance in the differences between the DCNN and the baseline. RESULTS: In the ten test partitions of the cross-validation set, the DCNN achieved a mean POA of 96.5 ± 2.9%, a mean Hdist of 2.26 ± 1.31 mm, and a mean AvgDist of 0.78 ± 0.58 mm, while the corresponding measures by the baseline method were 94.2 ± 4.8%, 3.69 ± 2.48 mm, and 1.30 ± 1.22 mm, respectively. For the independent DM test set, the DCNN achieved a mean POA of 93.7% ± 6.9%, a mean Hdist of 3.80 ± 3.21 mm, and a mean AvgDist of 1.49 ± 1.62 mm comparing to 86.9% ± 16.0%, 7.18 ± 14.22 mm, and 3.98 ± 14.13 mm, respectively, by the baseline method. CONCLUSION: In comparison to the TFO method, DCNN significantly improved the accuracy of pectoral muscle identification on mammograms (P < 0.05).

Synthesizing mammogram from digital breast tomosynthesis.

The purpose of this study is to develop a new method for generating synthesized mammogram (SM) from digital breast tomosynthesis (DBT) and to assess its potential as an adjunct to DBT. We first applied multiscale bilateral filtering to the reconstructed DBT slices to enhance the high-frequency features and reduce noise. A maximum intensity projection (MIP) image was then obtained from the high-frequency components of the DBT slices. A multiscale image fusion method was designed to combine the MIP image and the central DBT projection view into an SM and further enhance the high-frequency features. We conducted a pilot reader study to visually assess the image quality of SM in comparison to full field digital mammograms (FFDM). For each DBT craniocaudal or mediolateral view, a clinical FFDM of the corresponding view was retrospectively collected. Three MQSA radiologists, blinded to the pathological and other clinical information, independently interpreted the SM and the corresponding FFDM side by side marked with the lesion locations. The differences in the BI-RADS assessments of both MCs and masses between SM and FFDM did not achieve statistical significance for all three readers. The conspicuity of MCs on SM was superior to that on FFDM and the BI-RADS assessments of MCs were comparable while the conspicuity of masses on SM was degraded and interpretation on SM was less accurate than that on FFDM. The SM may be useful for efficient prescreening of MCs in DBT but the DBT should be used for detection and characterization of masses.

Perspectives on the Overdiagnosis of Breast Cancer Associated with Mammographic Screening.

Overdiagnosis of breast cancer refers to the screen detection and diagnosis of breast cancer that would not have progressed to symptomatic cancer during a woman's lifetime. Screening mammography, like all screening tests, will result in some overdiagnosis that is attributable to competing causes of death occurring during the lead time (the time period between asymptomatic screen detection and clinical detection) and detection of very indolent cancer. The primary harm of overdiagnosis relates to subsequent (unnecessary) treatment. Importantly, overdiagnosis concerns must be balanced with the lifesaving and morbidity benefits of screening mammography and the prevention of some invasive cancer by detection and treatment of ductal carcinoma in situ. Reasonable estimates of overdiagnosis of women aged 40-80 years are in the order of 1%-10%, with lower values when overdiagnosis is restricted to invasive cancer and among younger women. Prospective identification of an overdiagnosed invasive cancer is not currently possible. Delaying screening until age 50 years or screening biennially rather than annually will not substantially reduce the amount of overdiagnosis of invasive cancer. The clinical significance of overdiagnosis will continue to be minimized as advances in personalized medicine further reduce treatment-associated morbidity.