Search

PURPOSE Accurate risk assessment is essential for the success of population screening programs in breast cancer. Models with high sensitivity and specificity would enable programs to target more elaborate screening efforts to high-risk populations, while minimizing overtreatment for the rest. Artificial intelligence (AI)-based risk models have demonstrated a significant advance over risk models used today in clinical practice. However, the responsible deployment of novel AI requires careful validation across diverse populations. To this end, we validate our AI-based model, Mirai, across globally diverse screening populations. METHODS We collected screening mammograms and pathology-confirmed breast cancer outcomes from Massachusetts General Hospital, USA; Novant, USA; Emory, USA; Maccabi-Assuta, Israel; Karolinska, Sweden; Chang Gung Memorial Hospital, Taiwan; and Barretos, Brazil. We evaluated Uno's concordance index for Mirai in predicting risk of breast cancer at one to five years from the mammogram. RESULTS A total of 128,793 mammograms from 62,185 patients were collected across the seven sites, of which 3,815 were followed by a cancer diagnosis within 5 years. Mirai obtained concordance indices of 0.75 (95% CI, 0.72 to 0.78), 0.75 (95% CI, 0.70 to 0.80), 0.77 (95% CI, 0.75 to 0.79), 0.77 (95% CI, 0.73 to 0.81), 0.81 (95% CI, 0.79 to 0.82), 0.79 (95% CI, 0.76 to 0.83), and 0.84 (95% CI, 0.81 to 0.88) at Massachusetts General Hospital, Novant, Emory, Maccabi-Assuta, Karolinska, Chang Gung Memorial Hospital, and Barretos, respectively. CONCLUSION Mirai, a mammography-based risk model, maintained its accuracy across globally diverse test sets from seven hospitals across five countries. This is the broadest validation to date of an AI-based breast cancer model and suggests that the technology can offer broad and equitable improvements in care.

BACKGROUND Artificial intelligence (AI) cancer detectors are showing promising results and may soon be used for clinical breast cancer screening in radiology departments. Validation of such programs in different settings is mandatory. OBJECTIVE Our goal is to corroborate the hypothesis that women with previous benign biopsy and cytology would subsequently show increased AI abnormality scores for screening mammograms even though they remained healthy. METHODS This is a retrospective study after the application of a commercial AI cancer detector program to a cancer-enriched mammography screening dataset where 10 889 women with a median age for the study population of 56 years and a range between 40 and 74 years old where included. The program, is intended for 2D mammography and it generate a prediction score for tumor presence with a decimal number between 0 and 1, where 1 represented the highest level of suspicion. The AI score median and interquartile range for women who were healthy, who had a benign biopsy, and who were diagnosed with breast cancer, was calculated. For women with a previous benign biopsy, the time between mammogram and the biopsy was stratified. The effect of increasing age was examined using stratified analysis and linear regression modelling. RESULTS Two hundred and thirty-four (234) women had had a benign biopsy. The proportion of healthy women above the AI abnormality threshold (0.4) defined at our hospital was 3.5%, 11% and 84% for healthy women without a benign biopsy, healthy women with a benign biopsy, and for women with breast cancer, respectively (p CONCLUSIONS AI cancer detection systems can reliable be used in a screening setting. Adding information about possible prior benign biopsy and other symptoms can probably improve accuracy and subsequently reduce the workload for radiologists.

PURPOSE: In double-reading of screening mammograms, artificial intelligence (AI) algorithms hold promise as a potential replacement for one of the two readers. The choice of operating point, or abnormality threshold, for the AI algorithm will affect cancer detection and workload. In our retrospective study, the baseline approach was based on matching stand-alone reader sensitivity, while the alternative approach was based on matching the combined-reader sensitivity of two humans and of AI plus human. APPROACH: Full-field digital screening mammograms within the Stockholm County area between February 1, 2012, and December 30, 2015, acquired on Philips equipment, were collected. All exams of women with breast cancer within 23 months of screening and a random selection of healthy controls were included. An exam-level continuous AI abnormality score was generated (Insight MMG from Lunit Inc). Sensitivity and abnormal interpretation rates were estimated for operating points defined by the standalone-reader approach and the combined-reader approach. RESULTS: The study population included 1684 exams of women with breast cancer and 5024 exams of healthy women. Observations of healthy women were up-sampled to attain a realistic proportion of cancer. The observed sensitivity for reader 1, 2 and 1+2 was 69.7%, 75.6%, and 78.6%, respectively, at an abnormal interpretation rate of 4.4%, 4.6%, and 6.1%, respectively. For the combination of reader 1 + AI we estimated a sensitivity of 82.4% for standalone-reader matching and 78.6% for combined-reader matching, at an abnormal interpretation rate of 12.6% and 7.0%, respectively. CONCLUSIONS: Setting the operating point by matching stand-alone AI stand-alone with a radiologist will nearly double the downstream workload compared to a modest increase of 15% for the alternative method of matching sensitivity between the combination of AI and a radiologist with two radiologists.

Item does not contain fulltext

Background Suppression of background parenchymal enhancement (BPE) is commonly observed after neoadjuvant chemotherapy (NAC) at contrast-enhanced breast MRI. It was hypothesized that nonsuppressed BPE may be associated with inferior response to NAC. Purpose To investigate the relationship between lack of BPE suppression and pathologic response. Materials and Methods A retrospective review was performed for women with menopausal status data who were treated for breast cancer by one of 10 drug arms (standard NAC with or without experimental agents) between May 2010 and November 2016 in the Investigation of Serial Studies to Predict Your Therapeutic Response with Imaging and Molecular Analysis 2, or I-SPY 2 TRIAL (NCT01042379). Patients underwent MRI at four points: before treatment (T0), early treatment (T1), interregimen (T2), and before surgery (T3). BPE was quantitatively measured by using automated fibroglandular tissue segmentation. To test the hypothesis effectively, a subset of examinations with BPE with high-quality segmentation was selected. BPE change from T0 was defined as suppressed or nonsuppressed for each point. The Fisher exact test and the Z tests of proportions with Yates continuity correction were used to examine the relationship between BPE suppression and pathologic complete response (pCR) in hormone receptor (HR)-positive and HR-negative cohorts. Results A total of 3528 MRI scans from 882 patients (mean age, 48 years ± 10 [standard deviation]) were reviewed and the subset of patients with high-quality BPE segmentation was determined (T1, 433 patients; T2, 396 patients; T3, 380 patients). In the HR-positive cohort, an association between lack of BPE suppression and lower pCR rate was detected at T2 (nonsuppressed vs suppressed, 11.8% [six of 51] vs 28.9% [50 of 173]; difference, 17.1% [95% CI: 4.7, 29.5]; P = .02) and T3 (nonsuppressed vs suppressed, 5.3% [two of 38] vs 27.4% [48 of 175]; difference, 22.2% [95% CI: 10.9, 33.5]; P = .003). In the HR-negative cohort, patients with nonsuppressed BPE had lower estimated pCR rate at all points, but the P values for the association were all greater than .05. Conclusions In hormone receptor-positive breast cancer, lack of background parenchymal enhancement suppression may indicate inferior treatment response. © RSNA, 2021 Online supplemental material is available for this article. See also the editorial by Philpotts in this issue.

Artificial intelligence (AI) cancer detectors are showing promising results and may soon be used for clinical breast cancer screening in radiology departments. Validation of such programs in different settings is mandatory.This is a retrospective study after the application of a commercial AI cancer detector program to a cancer-enriched mammography screening dataset where 10 889 women with a median age for the study population of 56 years and a range between 40 and 74 years old where included. The program is intended for 2D mammography and it generate a prediction score for tumor presence with a decimal number between 0 and 1, where 1 represented the highest level of suspicion. The AI score median and interquartile range for women who were healthy, who had a benign biopsy, and who were diagnosed with breast cancer, was calculated. For women with a previous benign biopsy, the time between mammogram and the biopsy was stratified. The effect of increasing age was examined using stratified analysis and linear regression modelling. The proportion of healthy women above the AI abnormality threshold (0.4) defined at our hospital was 3.5%, 11% and 84% for healthy women without a benign biopsy, healthy women with a benign biopsy, and for women with breast cancer, respectively (p

Background: We aimed to identify factors associated with false-positive recalls in mammography screening compared with women who were not recalled and those who received true-positive recalls. Methods: We included 29,129 women, aged 40 to 74 years, who participated in the Karolinska Mammography Project for Risk Prediction of Breast Cancer (KARMA) between 2011 and 2013 with follow-up until the end of 2017. Nonmammographic factors were collected from questionnaires, mammographic factors were generated from mammograms, and genotypes were determined using the OncoArray or an Illumina custom array. By the use of conditional and regular logistic regression models, we investigated the association between breast cancer risk factors and risk models and false-positive recalls. Results: Women with a history of benign breast disease, high breast density, masses, microcalcifications, high Tyrer-Cuzick 10-year risk scores, KARMA 2-year risk scores, and polygenic risk scores were more likely to have mammography recalls, including both false-positive and true-positive recalls. Further analyses restricted to women who were recalled found that women with a history of benign breast disease and dense breasts had a similar risk of having false-positive and true-positive recalls, whereas women with masses, microcalcifications, high Tyrer-Cuzick 10-year risk scores, KARMA 2-year risk scores, and polygenic risk scores were more likely to have true-positive recalls than false-positive recalls. Conclusions: We found that risk factors associated with false-positive recalls were also likely, or even more likely, to be associated with true-positive recalls in mammography screening.

ImportanceA discrepancy on current guidelines and clinical practice exists regarding routine imaging surveillance after mastectomy, mainly regarding the lack of adequate evidence for imaging in this setting.ObjectiveTo investigate the usefulness of imaging surveillance in terms of cancer detection and interval cancer rates after mastectomy with or without reconstruction for patients with prior breast cancer.Data SourcesA comprehensive literature search was conducted in 3 electronic databases—PubMed, ISI Web of Science, and Scopus—without year restriction. References from relevant reviews and eligible studies were also manually searched.Study SelectionEligible studies were defined as those conducting surveillance imaging (mammography, ultrasonography, or magnetic resonance imaging [MRI]) of patients with prior breast cancer after mastectomy with or without reconstruction that presented adequate data to calculate cancer detection rates for each surveillance method.Data Extraction and SynthesisIndependent data extraction by 2 investigators with consensus on discrepant results was performed. A quality assessment of studies was performed using the QUADAS-2 (Quality Assessment of Diagnostic Accuracy Studies–2) template. The generalized linear mixed model framework with both fixed-effects and random-effects models was used to meta-analyze the proportion of cases across studies including 3 variables: surveillance method, reconstruction after mastectomy, and surveillance measure.Main Outcomes and MeasuresThree outcome measures were calculated for each eligible study and each surveillance imaging method within studies: overall cancer detection (defined as ipsilateral cancer, both palpable and nonpalpable) rate per 1000 examinations, clinically occult (nonpalpable) cancer detection rate per 1000 examinations, and interval cancer rate per 1000 examinations.ResultsIn total, 16 studies were eligible for the meta-analysis. The pooled overall cancer detection rates per 1000 examinations were 1.86 (95% CI, 1.05-3.30) for mammography, 2.66 (95% CI, 1.48-4.76) for ultrasonography, and 5.17 (95% CI, 1.49-17.75) for MRI. For mastectomy without reconstruction, the rate of clinically occult (nonpalpable) cancer per 1000 examinations (2.96; 95% CI, 1.38-6.32) and the interval cancer rate per 1000 examinations (3.73; 95% CI, 0.84-3.98) were lower than the overall cancer detection rate (including both palpable and nonpalpable lesions) per 1000 examinations (6.41; 95% CI, 3.09-13.25) across all imaging modalities. The interval cancer rate per 1000 examinations for mastectomy with reconstruction (3.73; 95% CI, 0.41-2.73) was comparable to the pooled cancer detection rate per 1000 examinations (4.73; 95% CI, 2.32-9.63) across all imaging modalities. In all clinical scenarios and imaging modalities, lower rates of clinically occult cancer compared with cancer detection rates were observed.Conclusions and RelevanceLower detection rates of clinically occult—compared with overall—cancer across all 3 imaging modalities challenge the use of imaging surveillance after mastectomy, with or without reconstruction. Findings suggest that imaging surveillance in this context is unnecessary in clinical practice, at least until further studies demonstrate otherwise. Future studies should consider using the clinically occult cancer detection rate as a more clinically relevant measure in this setting.

Background Background parenchymal enhancement (BPE) of normal tissue at breast magnetic resonance imaging is suggested to be an independent risk factor for breast cancer. Its association with established risk factors for breast cancer is not fully investigated. Purpose To study the association between BPE and risk factors for breast cancer in a healthy, non-high-risk screening population. Material and Methods We measured BPE and mammographic density and used data from self-reported questionnaires in 214 healthy women aged 43–74 years. We estimated odds ratios for the univariable association between BPE and risk factors. We then fitted an adjusted model using logistic regression to evaluate associations between BPE (high vs. low) and risk factors, including mammographic breast density. Results The majority of women had low BPE (84%). In a multivariable model, we found statistically significant associations between BPE and age ( P = 0.002) and BMI ( P = 0.03). We did find a significant association between systemic progesterone medication and BPE, but due to small numbers, the results should be interpreted with caution. The adjusted odds ratio for high BPE was 3.1 among women with density D (compared to B) and 2.1 for density C (compared to B). However, the association between high BPE and density was not statistically significant. We did not find statistically significant associations with any other risk factors. Conclusion Our study confirmed the known association of BPE with age and BMI. Although our results show a higher likelihood for high BPE with increasing levels of mammographic density, the association was not statistically significant.

Background Kidney disease and renal failure are associated with hospital deaths in patients with COVID - 19. We aimed to test if contrast enhancement affects short-term renal function in hospitalized COVID - 19 patients. Methods Plasma creatinine (P-creatinine) was measured on the day of computed tomography (CT) and 24 h, 48 h, and 4-10 days after CT. Contrast-enhanced (n = 142) and unenhanced (n = 24) groups were subdivided, based on estimated glomerular filtration rates (eGFR), > 60 and ≤ 60 ml/min/1.73 m2. Contrast-induced acute renal failure (CI-AKI) was defined as ≥27 μmol/L increase or a > 50% rise in P-creatinine from CT or initiation of renal replacement therapy during follow-up. Patients with renal replacement therapy were studied separately. We evaluated factors associated with a > 50% rise in P-creatinine at 48 h and at 4-10 days after contrast-enhanced CT. Results Median P-creatinine at 24-48 h and days 4-10 post-CT in patients with eGFR> 60 and eGFR≥30-60 in contrast-enhanced and unenhanced groups did not differ from basal values. CI-AKI was observed at 48 h and at 4-10 days post contrast administration in 24 and 36% (n = 5/14) of patients with eGFR≥30-60. Corresponding figures in the eGFR> 60 contrast-enhanced CT group were 5 and 5% respectively, (p Conclusion Patients with COVID - 19 and eGFR≥30-60 had a high frequency of CK-AKI at 48 h and at 4-10 days after contrast administration, which was associated with increased 30-day mortality. For patients with eGFR≥30-60, we recommend strict indications are practiced for contrast-enhanced CT. Contrast-enhanced CT had a modest effect in patients with eGFR> 60.

For AI researchers, access to a large and well-curated dataset is crucial. Working in the field of breast radiology, our aim was to develop a high-quality platform that can be used for evaluation of networks aiming to predict breast cancer risk, estimate mammographic sensitivity, and detect tumors. Our dataset, Cohort of Screen-Aged Women (CSAW), is a population-based cohort of all women 40 to 74 years of age invited to screening in the Stockholm region, Sweden, between 2008 and 2015. All women were invited to mammography screening every 18 to 24 months free of charge. Images were collected from the PACS of the three breast centers that completely cover the region. DICOM metadata were collected together with the images. Screening decisions and clinical outcome data were collected by linkage to the regional cancer center registers. Incident cancer cases, from one center, were pixel-level annotated by a radiologist. A separate subset for efficient evaluation of external networks was defined for the uptake area of one center. The collection and use of the dataset for the purpose of AI research has been approved by the Ethical Review Board. CSAW included 499,807 women invited to screening between 2008 and 2015 with a total of 1,182,733 completed screening examinations. Around 2 million mammography images have currently been collected, including all images for women who developed breast cancer. There were 10,582 women diagnosed with breast cancer; for 8463, it was their first breast cancer. Clinical data include biopsy-verified breast cancer diagnoses, histological origin, tumor size, lymph node status, Elston grade, and receptor status. One thousand eight hundred ninety-one images of 898 women had tumors pixel level annotated including any tumor signs in the prior negative screening mammogram. Our dataset has already been used for evaluation by several research groups. We have defined a high-volume platform for training and evaluation of deep neural networks in the domain of mammographic imaging. Electronic supplementary material The online version of this article (10.1007/s10278-019-00278-0) contains supplementary material, which is available to authorized users.

Improved breast cancer risk models enable targeted screening strategies that achieve earlier detection and less screening harm than existing guidelines. To bring deep learning risk models to clinical practice, we need to further refine their accuracy, validate them across diverse populations, and demonstrate their potential to improve clinical workflows. We developed Mirai, a mammography-based deep learning model designed to predict risk at multiple timepoints, leverage potentially missing risk factor information, and produce predictions that are consistent across mammography machines. Mirai was trained on a large dataset from Massachusetts General Hospital (MGH) in the United States and tested on held-out test sets from MGH, Karolinska University Hospital in Sweden, and Chang Gung Memorial Hospital (CGMH) in Taiwan, obtaining C-indices of 0.76 (95% confidence interval, 0.74 to 0.80), 0.81 (0.79 to 0.82), and 0.79 (0.79 to 0.83), respectively. Mirai obtained significantly higher 5-year ROC AUCs than the Tyrer-Cuzick model ( P < 0.001) and prior deep learning models Hybrid DL ( P < 0.001) and Image-Only DL ( P < 0.001), trained on the same dataset. Mirai more accurately identified high-risk patients than prior methods across all datasets. On the MGH test set, 41.5% (34.4 to 48.5) of patients who would develop cancer within 5 years were identified as high risk, compared with 36.1% (29.1 to 42.9) by Hybrid DL ( P = 0.02) and 22.9% (15.9 to 29.6) by the Tyrer-Cuzick model ( P < 0.001).

Dynamic contrast-enhanced (DCE) MRI provides both morphological and functional information regarding breast tumor response to neoadjuvant chemotherapy (NAC). The purpose of this retrospective study is to test if prediction models combining multiple MRI features outperform models with single features. Four features were quantitatively calculated in each MRI exam: functional tumor volume, longest diameter, sphericity, and contralateral background parenchymal enhancement. Logistic regression analysis was used to study the relationship between MRI variables and pathologic complete response (pCR). Predictive performance was estimated using the area under the receiver operating characteristic curve (AUC). The full cohort was stratified by hormone receptor (HR) and human epidermal growth factor receptor 2 (HER2) status (positive or negative). A total of 384 patients (median age: 49 y/o) were included. Results showed analysis with combined features achieved higher AUCs than analysis with any feature alone. AUCs estimated for the combined versus highest AUCs among single features were 0.81 (95% confidence interval [CI]: 0.76, 0.86) versus 0.79 (95% CI: 0.73, 0.85) in the full cohort, 0.83 (95% CI: 0.77, 0.92) versus 0.73 (95% CI: 0.61, 0.84) in HR-positive/HER2-negative, 0.88 (95% CI: 0.79, 0.97) versus 0.78 (95% CI: 0.63, 0.89) in HR-positive/HER2-positive, 0.83 (95% CI not available) versus 0.75 (95% CI: 0.46, 0.81) in HR-negative/HER2-positive, and 0.82 (95% CI: 0.74, 0.91) versus 0.75 (95% CI: 0.64, 0.83) in triple negatives. Multi-feature MRI analysis improved pCR prediction over analysis of any individual feature that we examined. Additionally, the improvements in prediction were more notable when analysis was conducted according to cancer subtype.

Background Mammography screening reduces breast cancer mortality, but a proportion of breast cancers are missed and are detected at later stages or develop during between-screening intervals. Purpose To develop a risk model based on negative mammograms that identifies women likely to be diagnosed with breast cancer before or at the next screening examination. Materials and Methods This study was based on the prospective screening cohort Karolinska Mammography Project for Risk Prediction of Breast Cancer (KARMA), 2011-2017. An image-based risk model was developed by using the Stratus method and computer-aided detection mammographic features (density, masses, microcalcifications), differences in the left and right breasts, and age. The lifestyle extended model included menopausal status, family history of breast cancer, body mass index, hormone replacement therapy, and use of tobacco and alcohol. The genetic extended model included a polygenic risk score with 313 single nucleotide polymorphisms. Age-adjusted relative risks and tumor subtype specific risks were estimated by using logistic regression, and absolute risks were calculated. Results Of 70 877 participants in the KARMA cohort, 974 incident cancers were sampled from 9376 healthy women (mean age, 54 years ± 10 [standard deviation]). The area under the receiver operating characteristic curve (AUC) for the image-based model was 0.73 (95% confidence interval [CI]: 0.71, 0.74). The AUCs for the lifestyle and genetic extended models were 0.74 (95% CI: 0.72, 0.75) and 0.77 (95% CI: 0.75, 0.79), respectively. There was a relative eightfold difference in risk between women at high risk and those at general risk. High-risk women were more likely to be diagnosed with stage II cancers and with tumors 20 mm or larger and were less likely to have stage I and estrogen receptor-positive tumors. The image-based model was validated in three external cohorts. Conclusion By combining three mammographic features, differences in the left and right breasts, and optionally lifestyle factors and family history and a polygenic risk score, the model identified women at high likelihood of being diagnosed with breast cancer within 2 years of a negative screening examination and in possible need of supplemental screening. © RSNA, 2020 Online supplemental material is available for this article.

Background There is great interest in developing artificial intelligence (AI)-based computer-aided detection (CAD) systems for use in screening mammography. Comparative performance benchmarks from true screening cohorts are needed. Purpose To determine the range of human first-reader performance measures within a population-based screening cohort of 1 million screening mammograms to gauge the performance of emerging AI CAD systems. Materials and Methods This retrospective study consisted of all screening mammograms in women aged 40-74 years in Stockholm County, Sweden, who underwent screening with full-field digital mammography between 2008 and 2015. There were 110 interpreting radiologists, of whom 24 were defined as high-volume readers (ie, those who interpreted more than 5000 annual screening mammograms). A true-positive finding was defined as the presence of a pathology-confirmed cancer within 12 months. Performance benchmarks included sensitivity and specificity, examined per quartile of radiologists' performance. First-reader sensitivity was determined for each tumor subgroup, overall and by quartile of high-volume reader sensitivity. Screening outcomes were examined based on the first reader's sensitivity quartile with 10 000 screening mammograms per quartile. Linear regression models were fitted to test for a linear trend across quartiles of performance. Results A total of 418 041 women (mean age, 54 years ± 10 [standard deviation]) were included, and 1 186 045 digital mammograms were evaluated, with 972 899 assessed by high-volume readers. Overall sensitivity was 73% (95% confidence interval [CI]: 69%, 77%), and overall specificity was 96% (95% CI: 95%, 97%). The mean values per quartile of high-volume reader performance ranged from 63% to 84% for sensitivity and from 95% to 98% for specificity. The sensitivity difference was very large for basal cancers, with the least sensitive and most sensitive high-volume readers detecting 53% and 89% of cancers, respectively (

Objective Women with advanced HER2− breast cancer have limited treatment options. Breast MRI functional tumor volume (FTV) is used to predict pathologic complete response (pCR) to improve treatment efficacy. In addition to FTV, background parenchymal enhancement (BPE) may predict response and was explored for HER2− patients in the I-SPY-2 TRIAL. Methods Women with HER2− stage II or III breast cancer underwent prospective serial breast MRIs during four neoadjuvant chemotherapy timepoints. BPE was quantitatively calculated using whole-breast manual segmentation. Logistic regression models were systematically explored using pre-specified and optimized predictor selection based on BPE or combined with FTV. Results A total of 352 MRI examinations in 88 patients (29 with pCR, 59 non-pCR) were evaluated. Women with hormone receptor (HR)+HER2− cancers who achieved pCR demonstrated a significantly greater decrease in BPE from baseline to pre-surgery compared to non-pCR patients (odds ratio 0.64, 95% confidence interval (CI): 0.39–0.92, P = 0.04). The associated BPE area under the curve (AUC) was 0.77 (95% CI: 0.56–0.98), comparable to the range of FTV AUC estimates. Among multi-predictor models, the highest cross-validated AUC of 0.81 (95% CI: 0.73–0.90) was achieved with combined FTV+HR predictors, while adding BPE to FTV+HR models had an estimated AUC of 0.82 (95% CI: 0.74–0.92). Conclusion Among women with HER2− cancer, BPE alone demonstrated association with pCR in women with HR+HER2− breast cancer, with similar diagnostic performance to FTV. BPE predictors remained significant in multivariate FTV models, but without added discrimination for pCR prediction. This may be due to small sample size limiting ability to create subtype-specific multivariate models.

Summary Background We examined the potential change in cancer detection when using an artificial intelligence (AI) cancer-detection software to triage certain screening examinations into a no radiologist work stream, and then after regular radiologist assessment of the remainder, triage certain screening examinations into an enhanced assessment work stream. The purpose of enhanced assessment was to simulate selection of women for more sensitive screening promoting early detection of cancers that would otherwise be diagnosed as interval cancers or as next-round screen-detected cancers. The aim of the study was to examine how AI could reduce radiologist workload and increase cancer detection. Methods In this retrospective simulation study, all women diagnosed with breast cancer who attended two consecutive screening rounds were included. Healthy women were randomly sampled from the same cohort; their observations were given elevated weight to mimic a frequency of 0·7% incident cancer per screening interval. Based on the prediction score from a commercially available AI cancer detector, various cutoff points for the decision to channel women to the two new work streams were examined in terms of missed and additionally detected cancer. Findings 7364 women were included in the study sample: 547 were diagnosed with breast cancer and 6817 were healthy controls. When including 60%, 70%, or 80% of women with the lowest AI scores in the no radiologist stream, the proportion of screen-detected cancers that would have been missed were 0, 0·3% (95% CI 0·0–4·3), or 2·6% (1·1–5·4), respectively. When including 1% or 5% of women with the highest AI scores in the enhanced assessment stream, the potential additional cancer detection was 24 (12%) or 53 (27%) of 200 subsequent interval cancers, respectively, and 48 (14%) or 121 (35%) of 347 next-round screen-detected cancers, respectively. Interpretation Using a commercial AI cancer detector to triage mammograms into no radiologist assessment and enhanced assessment could potentially reduce radiologist workload by more than half, and pre-emptively detect a substantial proportion of cancers otherwise diagnosed later. Funding Stockholm City Council.

Breast parenchymal enhancement (BPE) has shown association with breast cancer risk and response to neoadjuvant treatment. However, BPE quantification is challenging, and there is no standardized segmentation method for measurement. We investigated the use of a fully automated breast fibroglandular tissue segmentation method to calculate BPE from dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) for use as a predictor of pathologic complete response (pCR) following neoadjuvant treatment in the I-SPY 2 TRIAL. In this trial, patients had DCE-MRI at baseline (T0), after 3 weeks of treatment (T1), after 12 weeks of treatment and between drug regimens (T2), and after completion of treatment (T3). A retrospective analysis of 2 cohorts was performed: one with 735 patients and another with a final cohort of 340 patients, meeting a high-quality benchmark for segmentation. We evaluated 3 subvolumes of interest segmented from bilateral T1-weighted axial breast DCE-MRI: full stack (all axial slices), half stack (center 50% of slices), and center 5 slices. The differences between methods were assessed, and a univariate logistic regression model was implemented to determine the predictive performance of each segmentation method. The results showed that the half stack method provided the best compromise between sampling error from too little tissue and inclusion of incorrectly segmented tissues from extreme superior and inferior regions. Our results indicate that BPE calculated using the half stack segmentation approach has potential as an early biomarker for response to treatment in the hormone receptor–negative and human epidermal growth factor receptor 2–positive subtype.

The ability to accurately estimate risk of developing breast cancer would be invaluable for clinical decision-making. One promising new approach is to integrate image-based risk models based on deep neural networks. However, one must take care when using such models, as selection of training data influences the patterns the network will learn to identify. With this in mind, we trained networks using three different criteria to select the positive training data (i.e. images from patients that will develop cancer): an inherent risk model trained on images with no visible signs of cancer, a cancer signs model trained on images containing cancer or early signs of cancer, and a conflated model trained on all images from patients with a cancer diagnosis. We find that these three models learn distinctive features that focus on different patterns, which translates to contrasts in performance. Short-term risk is best estimated by the cancer signs model, whilst long-term risk is best estimated by the inherent risk model. Carelessly training with all images conflates inherent risk with early cancer signs, and yields sub-optimal estimates in both regimes. As a consequence, conflated models may lead physicians to recommend preventative action when early cancer signs are already visible.

Purpose: To test the hypothesis that prior non-adherence to mammography screening could predict subsequent non-adherence to breast cancer treatment. Specifically, we hypothesized that as compared with adherers, screening non-adherers may be more likely to have delayed surgery, discontinue their adjuvant hormone therapy, and consequently have worse breast cancer outcomes. Methods: We conducted a record-linkage study based on data from the Stockholm Mammography Screening Program (1989-2013), Stockholm-Gotland Breast Cancer Register (2001-2017), Swedish Prescribed Drug Register (2005-2017), and Cause of Death Register (2001-2013). Women diagnosed with breast cancer between 2001 and 2008 in Stockholm, Sweden, were prospectively followed for treatment and survival until December 31st, 2013 (N=5106). Screening non-adherers were defined as patients who were invited but did not attend the screening mammography within 2 years before breast cancer diagnosis. Discontinuation of adjuvant hormone therapy was defined as having any interval between two consecutive dispenses exceeding 180 days during follow-up. Disease-free survival was defined as time to local recurrence, distant metastasis, contralateral breast cancer, or death from breast cancer, whichever came first. Results: The proportion of delayed surgery (≥6 weeks) for screening adherers (restricted to interval cancers only) versus non-adherers were 15.5% versus 20.6%, respectively (P Screening AdherersScreening Non-adherers Univariable-analysisMultivariable analysis#Delayed surgery>6 weeks* No897(84.5)624(79.4) 1.00 (Reference)1.00(Reference)Yes†165(15.5)162(20.6) 1.40(1.10-1.78)1.34(1.05-1.72)Discontinuation of adjuvant hormone therapy** No822(50.3)145(42.7) 1.00(Reference)1.00(Reference)Yes††812(49.7)195(57.3) 1.33(1.14-1.56)1.37(1.17-1.61)Disease free survival events No3574(85.8)752(79.8) 1.00(Reference)1.00(Reference)Yes††590(14.2)190(20.2) 1.46(1.24-1.72)1.19(1.01-1.41)* Restricted to patients diagnosed with clinically detected cancers who were primary operated without neoadjuvant therapy; ** Restricted to patients diagnosed after July 1st, 2005, who initiated adjuvant hormone therapy.† Odds ratios (95% CI);† † Hazard ratios (95% CI).# Adjusting for age, tumor size, lymph node involvement, estrogen receptor status, progesterone receptor status, HER2 status, and type of surgery Conclusions: Compared to adherers, screening non-adherers are more likely to have delayed surgery, discontinue their adjuvant hormone therapy, and subsequently have worse breast cancer outcomes even after adjusting for baseline tumor characteristics. Citation Format: He W, Bergman LE, Törnberg S, Strand F, Hall P, Czene K. Non-adherers of mammography screening: Delayed surgery, early discontinuation of adjuvant hormone therapy, and worse breast cancer outcomes [abstract]. In: Proceedings of the 2017 San Antonio Breast Cancer Symposium; 2017 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2018;78(4 Suppl):Abstract nr P6-08-20.

Background If screening participants do not trust computerized decision-making, screening participation may be affected by the introduction of such methods. Purpose To survey breast cancer screening participants’ attitudes towards potential future uses of computerization. Material and Methods A survey was constructed. Women in a breast cancer screening program were invited via the final report letter to participate. Data were collected from February 2018 to March 2019 and 2196 surveys were completed. Questions asked participants to rate propositions using Likert scales. Data analysis was done using χ2 and logistic regression tests. Results The mean age of participants was 61 years. Response rate was 1.3%. Of the submitted surveys, 97.5% were complete; 38% of respondents reported a preference for a computer-only examination. The highest level of confidence was given a computer-only reading followed by a physician reading. Participants with > 12 years of education were more likely to prefer a computer-only reading (odds ratio [OR] 1.655, 95% confidence interval [CI] 1.168–2.344), had a greater trust in letting a computer determine screening intervals and the need for a supplemental MRI (OR 1.606, 95% CI 1.171–2.202 and OR 1.577, 95% CI 1.107–2.247, respectively). Age was not found to be a significant predictor. Conclusion A high level of trust in computerized decision-making was expressed. Higher age was associated with a lower understanding of technology but did not affect attitudes to computerized decision-making. A lower level of education was associated with a lower trust in computerization. This may be valuable knowledge for future studies.

PURPOSE: Women with radiographically dense or texturally complex breasts are at increased risk for interval cancer, defined as cancers diagnosed after a normal screening examination. The purpose of this study was to create masking measures and apply them to identify interval risk in a population of women who experienced either screen-detected or interval cancers after controlling for breast density. METHODS: We examined full-field digital screening mammograms acquired from 2006–2015. Examinations associated with 182 interval cancers were matched to 173 screen-detected cancers on age, race, exam date and time since last imaging examination. Local Image Quality Factor (IQF) values were calculated and used to create IQF maps that represented mammographic masking. We used various statistics to define global masking measures of these maps. Association of these masking measures with interval cancer vs. screen-detected cancer was estimated using conditional logistic regression in a univariate and adjusted model for Breast Imaging-Reporting and Data System (BI-RADS) density. Receiver operator curves were calculated in each case to compare specificity versus sensitivity, and area under those curves were generated. Proportion of screen-detected cancer was estimated for stratifications of IQF features. RESULTS: Several masking features showed significant association with interval compared to screen-detected cancers after adjusting for BI-RADS density (up to p = 2.52E-6), and the 10(th) percentile of the IQF value (p = 1.72E-3) showed the strongest improvement in the area under the receiver operator curve, increasing from 0.65 using only BI-RADS density to 0.69. The highest masking group had a 32% proportion of screen-detected cancers while the low masking group had a 69% proportion. CONCLUSIONS: We conclude that computer vision methods using model observers may improve quantifying the probability of breast cancer detection beyond using breast density alone.

Background High mammographic density is associated with breast cancer and with delayed detection. We have examined whether localized density, at the site of the subsequent cancer, is independently associated with being diagnosed with a large-sized or interval breast cancer. Methods Within a prospective cohort of 63,130 women, we examined 891 women who were diagnosed with incident breast cancer. For 386 women, retrospective localized density assessment was possible. The main outcomes were interval cancer vs. screen-detected cancer and large (> 2 cm) vs. small cancer. In negative screening mammograms, overall and localized density were classified reflecting the BI-RADS standard. Density concordance probabilities were estimated through multinomial regression. The associations between localized density and the two outcomes were modeled through logistic regression, adjusted for overall density, age, body mass index, and other characteristics. Results The probabilities of concordant localized density were 0.35, 0.60, 0.38, and 0.32 for overall categories “A,” “B,” “C,” and “D.” Overall density was associated with large cancer, comparing density category D to A with OR 4.6 (95%CI 1.8–11.6) and with interval cancer OR 31.5 (95%CI 10.9–92) among all women. Localized density was associated with large cancer at diagnosis with OR 11.8 (95%CI 2.7–51.8) among all women and associated with first-year interval cancer with OR 6.4 (0.7 to 58.7) with a significant linear trend p = 0.027. Conclusions Overall density often misrepresents localized density at the site where cancer subsequently arises. High localized density is associated with interval cancer and with large cancer. Our findings support the continued effort to develop and examine computer-based measures of localized density for use in personalized breast cancer screening. Electronic supplementary material The online version of this article (10.1186/s13058-019-1099-y) contains supplementary material, which is available to authorized users.

Importance A computer algorithm that performs at or above the level of radiologists in mammography screening assessment could improve the effectiveness of breast cancer screening. Objective To perform an external evaluation of 3 commercially available artificial intelligence (AI) computer-aided detection algorithms as independent mammography readers and to assess the screening performance when combined with radiologists. Design, Setting, and Participants This retrospective case-control study was based on a double-reader population-based mammography screening cohort of women screened at an academic hospital in Stockholm, Sweden, from 2008 to 2015. The study included 8805 women aged 40 to 74 years who underwent mammography screening and who did not have implants or prior breast cancer. The study sample included 739 women who were diagnosed as having breast cancer (positive) and a random sample of 8066 healthy controls (negative for breast cancer). Main Outcomes and Measures Positive follow-up findings were determined by pathology-verified diagnosis at screening or within 12 months thereafter. Negative follow-up findings were determined by a 2-year cancer-free follow-up. Three AI computer-aided detection algorithms (AI-1, AI-2, and AI-3), sourced from different vendors, yielded a continuous score for the suspicion of cancer in each mammography examination. For a decision of normal or abnormal, the cut point was defined by the mean specificity of the first-reader radiologists (96.6%). Results The median age of study participants was 60 years (interquartile range, 50-66 years) for 739 women who received a diagnosis of breast cancer and 54 years (interquartile range, 47-63 years) for 8066 healthy controls. The cases positive for cancer comprised 618 (84%) screen detected and 121 (16%) clinically detected within 12 months of the screening examination. The area under the receiver operating curve for cancer detection was 0.956 (95% CI, 0.948-0.965) for AI-1, 0.922 (95% CI, 0.910-0.934) for AI-2, and 0.920 (95% CI, 0.909-0.931) for AI-3. At the specificity of the radiologists, the sensitivities were 81.9% for AI-1, 67.0% for AI-2, 67.4% for AI-3, 77.4% for first-reader radiologist, and 80.1% for second-reader radiologist. Combining AI-1 with first-reader radiologists achieved 88.6% sensitivity at 93.0% specificity (abnormal defined by either of the 2 making an abnormal assessment). No other examined combination of AI algorithms and radiologists surpassed this sensitivity level. Conclusions and Relevance To our knowledge, this study is the first independent evaluation of several AI computer-aided detection algorithms for screening mammography. The results of this study indicated that a commercially available AI computer-aided detection algorithm can assess screening mammograms with a sufficient diagnostic performance to be further evaluated as an independent reader in prospective clinical trials. Combining the first readers with the best algorithm identified more cases positive for cancer than combining the first readers with second readers.

Importance: Mammography screening currently relies on subjective human interpretation. Artificial intelligence (AI) advances could be used to increase mammography screening accuracy by reducing missed cancers and false positives. Objective: To evaluate whether AI can overcome human mammography interpretation limitations with a rigorous, unbiased evaluation of machine learning algorithms. Design, Setting, and Participants: In this diagnostic accuracy study conducted between September 2016 and November 2017, an international, crowdsourced challenge was hosted to foster AI algorithm development focused on interpreting screening mammography. More than 1100 participants comprising 126 teams from 44 countries participated. Analysis began November 18, 2016. Main Outcomes and Measurements: Algorithms used images alone (challenge 1) or combined images, previous examinations (if available), and clinical and demographic risk factor data (challenge 2) and output a score that translated to cancer yes/no within 12 months. Algorithm accuracy for breast cancer detection was evaluated using area under the curve and algorithm specificity compared with radiologists' specificity with radiologists' sensitivity set at 85.9% (United States) and 83.9% (Sweden). An ensemble method aggregating top-performing AI algorithms and radiologists' recall assessment was developed and evaluated. Results: Overall, 144 231 screening mammograms from 85 580 US women (952 cancer positive ≤12 months from screening) were used for algorithm training and validation. A second independent validation cohort included 166 578 examinations from 68 008 Swedish women (780 cancer positive). The top-performing algorithm achieved an area under the curve of 0.858 (United States) and 0.903 (Sweden) and 66.2% (United States) and 81.2% (Sweden) specificity at the radiologists' sensitivity, lower than community-practice radiologists' specificity of 90.5% (United States) and 98.5% (Sweden). Combining top-performing algorithms and US radiologist assessments resulted in a higher area under the curve of 0.942 and achieved a significantly improved specificity (92.0%) at the same sensitivity. Conclusions and Relevance: While no single AI algorithm outperformed radiologists, an ensemble of AI algorithms combined with radiologist assessment in a single-reader screening environment improved overall accuracy. This study underscores the potential of using machine learning methods for enhancing mammography screening interpretation. CA extern

BACKGROUND: Background parenchymal enhancement (BPE) describes the natural phenomenon observed on breast MRI in which normal breast tissue demonstrates signal enhancement from uptake of intravenous contrast. BPE may provide independent and additive value for prediction of pathologic complete response (pCR) using MRI measured functional tumor volume (FTV), which has only moderate discrimination (FTV AUC ~ 0.7). We evaluated the additive value of quantitative whole breast BPE to a FTV model for prediction of pCR to neoadjuvant chemotherapy in the ISPY-2 trial. METHODS: In this HIPAA-compliant/IRB-approved study, women 18 years of age and older diagnosed with stage II or III breast cancer and with tumor size measured ≥ 2.5 cm were eligible to enroll in the I-SPY 2 TRIAL. Participants received a weekly dose of paclitaxel alone (control) or in combination with Veliparib and Carboplatin for 12 weekly cycles followed by four (every 2-3 weeks) cycles of anthracycline-cyclophosphamide prior to surgery. All breast cancers in these drug arms were Her2 negative. MRI was performed before the initiation of neoadjuvant therapy or “baseline” (T0), after three weeks of therapy or “early treatment” (T1), after twelve weeks between drug regimens or “inter-regimen” (T2), and after neoadjuvant therapy completion and prior to surgery or “pre-surgery” (T3). MRI segmentation was manually performed of the whole unaffected contralateral breast, and tissue classification was performed using fuzzy c-means clustering. BPE was calculated as the average enhancement of all tissue voxels at the first postcontrast acquisition. Predictor variables were parameterized as absolute values of BPE/FTV for each treatment time point or relative change values for each treatment time period. Logistic regression, stratified by hormone receptor (HR) subtype, was performed using 1) univariate models of FTV/BPE predictors alone and 2) multivariate models using all possible combinations of FTV/BPE predictors and HR status. Additive benefit for multivariate models was evaluated by estimating change in the area under the curve (AUC) for overall diagnostic performance with 10-repeat 5-fold cross validation. The 95% confidence interval (CI) of cross-validated AUC was estimated using 1,000 bootstrap resamples. RESULTS: A total of 88 patients (29 pCR, 59 non-pCR) were evaluated with serial breast MRIs to assess neoadjuvant response. Among univariate models, women with HR+ cancers who had PCR demonstrated a significantly greater decrease in BPE from baseline to pre-surgery compared to non-PCR (OR = 0.64, 95% CI = 0.39-0.92, p-value = 0.04). The associated AUC was 0.77 (95% CI 0.56-0.98), comparable to the range of univariate FTV AUC values (0.57-0.87). Among optimized multivariate models, the highest cross-validated AUC for FTV and HR predictors was 0.81 (95% CI 0.73-0.90), while adding BPE slightly increased AUC to 0.82 (95% CI 0.74-0.92). CONCLUSION: Changes in BPE in response to neoadjuvant therapy, which represents normal breast tissue changes measurable on any breast MRI, demonstrated significant association with pCR in women with HR+ breast cancer. Moreover, it had a similar diagnostic performance to univariate prediction with tumor volume. However, additive prediction of BPE to multivariate FTV models was only marginal. Citation Format: Vignesh A. Arasu, Paul Kim, Wen Li, Fredrik Strand, Cody McHargue, Ella Jones, David Newitt, John Kornak, Laura Esserman, Nola Hylton, ACRIN 6657 Trial TeamI-SPY Investigators Network. Additive benefit of MRI background parenchymal enhancement (BPE) to an MRI tumor volume model for predicting pathologic complete response (pCR) to neoadjuvant treatment in ISPY2 [abstract]. In: Proceedings of the 2019 San Antonio Breast Cancer Symposium; 2019 Dec 10-14; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2020;80(4 Suppl):Abstract nr PD9-06.