Detecting Breast Arterial Calcifications in Mammograms with Transfer Learning

Khan, Rimsha; Masala, G. L.

doi:10.3390/electronics12010231

Cited by 6 publications

(3 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our performances are similar to those reported by previous studies: Khan and Masala [ 40 ] recently published a study on BAC detection using transfer learning, comparing the results obtained from different deep learning architectures trained on a small population of just 104 mammograms from 26 patients. They reported an accuracy of 0.96 of VGG19, marginally lower than that yielded by deeper CNNs such as ResNet50 or DenseNet-121, which showed an accuracy of 0.97 and 0.98, respectively.…”

Section: Discussionsupporting

confidence: 88%

Detection and quantification of breast arterial calcifications on mammograms: a deep learning approach

et al. 2023

View full text Add to dashboard Cite

Objective Breast arterial calcifications (BAC) are a sex-specific cardiovascular disease biomarker that might improve cardiovascular risk stratification in women. We implemented a deep convolutional neural network for automatic BAC detection and quantification. Methods In this retrospective study, four readers labelled four-view mammograms as BAC positive (BAC+) or BAC negative (BAC−) at image level. Starting from a pretrained VGG16 model, we trained a convolutional neural network to discriminate BAC+ and BAC− mammograms. Accuracy, F1 score, and area under the receiver operating characteristic curve (AUC-ROC) were used to assess the diagnostic performance. Predictions of calcified areas were generated using the generalized gradient-weighted class activation mapping (Grad-CAM++) method, and their correlation with manual measurement of BAC length in a subset of cases was assessed using Spearman ρ. Results A total 1493 women (198 BAC+) with a median age of 59 years (interquartile range 52–68) were included and partitioned in a training set of 410 cases (1640 views, 398 BAC+), validation set of 222 cases (888 views, 89 BAC+), and test set of 229 cases (916 views, 94 BAC+). The accuracy, F1 score, and AUC-ROC were 0.94, 0.86, and 0.98 in the training set; 0.96, 0.74, and 0.96 in the validation set; and 0.97, 0.80, and 0.95 in the test set, respectively. In 112 analyzed views, the Grad-CAM++ predictions displayed a strong correlation with BAC measured length (ρ = 0.88, p < 0.001). Conclusion Our model showed promising performances in BAC detection and in quantification of BAC burden, showing a strong correlation with manual measurements. Clinical relevance statement Integrating our model to clinical practice could improve BAC reporting without increasing clinical workload, facilitating large-scale studies on the impact of BAC as a biomarker of cardiovascular risk, raising awareness on women’s cardiovascular health, and leveraging mammographic screening. Key Points • We implemented a deep convolutional neural network (CNN) for BAC detection and quantification. • Our CNN had an area under the receiving operator curve of 0.95 for BAC detection in the test set composed of 916 views, 94 of which were BAC+ . • Furthermore, our CNN showed a strong correlation with manual BAC measurements (ρ = 0.88) in a set of 112 views.

show abstract

Section: Discussionsupporting

confidence: 88%

Detection and quantification of breast arterial calcifications on mammograms: a deep learning approach

et al. 2023

View full text Add to dashboard Cite

show abstract

“…10 Such studies have assessed methods of BAC quantification, though await association with clinical outcomes. [11][12][13] The findings in our study support the efficacy of assessing both BAC presence and quantitative BAC Bradley score to improve risk assessment for mortality and CVD outcomes in women undergoing screening mammography.…”

Section: Discussionsupporting

confidence: 72%

“…10 Such studies have assessed methods of BAC quantification, though await association with clinical outcomes. 11–13 The findings in our study support the efficacy of assessing both BAC presence and quantitative BAC Bradley score to improve risk assessment for mortality and CVD outcomes in women undergoing screening mammography. With the advent of AI in medical imaging, automated, quantitative BAC assessment may facilitate seamless integration into clinical workflow and allow personalized risk assessment.…”

Section: Discussionsupporting

confidence: 72%

Association of an Automated Breast Arterial Calcification Score with Cardiovascular Outcomes and Mortality

Bui,

Allen,

Mantey

et al. 2023

Preprint

View full text Add to dashboard Cite

BackgroundBreast arterial calcification (BAC), visible on mammograms, has emerged as a biomarker of cardiovascular disease (CVD) in women. Barriers to clinical implementation of BAC include limited studies with clinical outcomes and lack of quantification tools.MethodsThis single-center, retrospective study included women with a screening digital mammogram from 2008-2016. BAC was quantified using an automated, artificial intelligence (AI)-generated Bradley score, as a binary (Bradley score ≥5) and continuous variable. Clinical outcomes were determined via electronic medical records. Regression analyses were used to evaluate the association between BAC and outcomes of mortality and a composite of acute myocardial infarction, heart failure, stroke, and mortality. Models were adjusted for age, race, diabetes, smoking status, blood pressure, cholesterol, and history of CVD and chronic kidney disease.ResultsA total of 18,092 women were included with a mean age 56.8±11.0 years. Prevalence of comorbidities included diabetes (13%), hypertension (36%), hyperlipidemia (40%) and smoking (5%). BAC was present in 4,223 (23%). Over a median follow-up of 6 years, death occurred in 7.8% women with and 2.3% women without BAC. The composite outcome occurred in 12.4% of women with and 4.3% of women without BAC. Compared to those without, women with BAC had adjusted hazard ratios (aHR) of 1.49 (95% CI 1.33-1.67) for mortality and 1.56 (1.41-1.72) for the composite endpoint, after accounting for traditional risk factors. With a continuous BAC score, each 10-point increase was associated with higher risk of mortality (1.08 [1.06-1.11]) and the composite endpoint (1.08 [1.06-1.10]). BAC was especially predictive of future events among younger women.ConclusionBAC is significantly and independently associated with mortality and incident CVD, especially among younger women. Measuring BAC using an AI algorithm is feasible and clinically relevant. Further studies are needed to confirm these findings and to evaluate whether interventions guided by BAC improve outcomes.CLINICAL PERSPECTIVEWhat is new?Breast arterial calcification (BAC) on mammograms can be reliably quantified using a novel software based on an artificial intelligence (AI) algorithm.BAC is independently associated with an increased risk of all-cause mortality and cardiovascular outcomes. These associations held true when looking at BAC as presence, score quartile, and as a continuous value as well as after accounting for traditional cardiovascular risk factors.In stratification analysis, BAC was most predictive of all-cause mortality and cardiovascular outcomes among younger women (age 40-59 years), but still independently predictive in women aged 60-74 years.What are the clinical implications?Our data provide support for the inclusion of BAC findings on mammogram reports.Automated quantification tools and reporting methods of BAC will be critical to engagement of radiologists and implementation of reporting.While additional studies are needed to determine the appropriate clinical response, the presence of BAC should at the minimum stimulate patient-provider conversations on lifestyle changes to mitigate cardiovascular risk, especially among younger women.

show abstract

Advancing precision in breast cancer detection: a fusion of vision transformers and CNNs for calcification mammography classification

Boudouh,

Bouakkaz

2024

Appl Intell

View full text Add to dashboard Cite

Detecting Breast Arterial Calcifications in Mammograms with Transfer Learning

Cited by 6 publications

References 34 publications

Detection and quantification of breast arterial calcifications on mammograms: a deep learning approach

Detection and quantification of breast arterial calcifications on mammograms: a deep learning approach

Association of an Automated Breast Arterial Calcification Score with Cardiovascular Outcomes and Mortality

Advancing precision in breast cancer detection: a fusion of vision transformers and CNNs for calcification mammography classification

Contact Info

Product

Resources

About