Discrimination of Breast Cancer with Microcalcifications on Mammography by Deep Learning.
Ontology highlight
ABSTRACT:
SUBMITTER: Wang J
PROVIDER: S-EPMC4895132 | biostudies-literature | 2016 Jun
REPOSITORIES: biostudies-literature
Publications
Discrimination of Breast Cancer with Microcalcifications on Mammography by Deep Learning.
Wang Jinhua J Yang Xi X Cai Hongmin H Tan Wanchang W Jin Cangzheng C Li Li L
Scientific reports 20160607
Microcalcification is an effective indicator of early breast cancer. To improve the diagnostic accuracy of microcalcifications, this study evaluates the performance of deep learning-based models on large datasets for its discrimination. A semi-automated segmentation method was used to characterize all microcalcifications. A discrimination classifier model was constructed to assess the accuracies of microcalcifications and breast masses, either in isolation or combination, for classifying breast ...[more]
Similar Datasets
Project description:BackgroundBreast cancer screening through mammography is crucial for early detection, yet the demand for mammography services surpasses the capacity of radiologists. Artificial intelligence (AI) can assist in evaluating microcalcifications on mammography. We developed and tested an AI model for localizing and characterizing microcalcifications.MethodsThree expert radiologists annotated a dataset of mammograms using histology-based ground truth. The dataset was partitioned for training, validation, and testing. Three neural networks (AlexNet, ResNet18, and ResNet34) were trained and evaluated using specific metrics including receiver operating characteristics area under the curve (AUC), sensitivity, and specificity. The reported metrics were computed on the test set (10% of the whole dataset).ResultsThe dataset included 1,000 patients aged 21-73 years and 1,986 mammograms (180 density A, 220 density B, 380 density C, and 220 density D), with 389 malignant and 611 benign groups of microcalcifications. AlexNet achieved the best performance with 0.98 sensitivity, 0.89 specificity of, and 0.98 AUC for microcalcifications detection and 0.85 sensitivity, 0.89 specificity, and 0.94 AUC of for microcalcifications classification. For microcalcifications detection, ResNet18 and ResNet34 achieved 0.96 and 0.97 sensitivity, 0.91 and 0.90 specificity and 0.98 and 0.98 AUC, retrospectively. For microcalcifications classification, ResNet18 and ResNet34 exhibited 0.75 and 0.84 sensitivity, 0.85 and 0.84 specificity, and 0.88 and 0.92 AUC, respectively.ConclusionsThe developed AI models accurately detect and characterize microcalcifications on mammography.Relevance statementAI-based systems have the potential to assist radiologists in interpreting microcalcifications on mammograms. The study highlights the importance of developing reliable deep learning models possibly applied to breast cancer screening.Key points• A novel AI tool was developed and tested to aid radiologists in the interpretation of mammography by accurately detecting and characterizing microcalcifications. • Three neural networks (AlexNet, ResNet18, and ResNet34) were trained, validated, and tested using an annotated dataset of 1,000 patients and 1,986 mammograms. • The AI tool demonstrated high accuracy in detecting/localizing and characterizing microcalcifications on mammography, highlighting the potential of AI-based systems to assist radiologists in the interpretation of mammograms.
| S-EPMC10630180 | biostudies-literature
Project description:The rapid development of deep learning, a family of machine learning techniques, has spurred much interest in its application to medical imaging problems. Here, we develop a deep learning algorithm that can accurately detect breast cancer on screening mammograms using an "end-to-end" training approach that efficiently leverages training datasets with either complete clinical annotation or only the cancer status (label) of the whole image. In this approach, lesion annotations are required only in the initial training stage, and subsequent stages require only image-level labels, eliminating the reliance on rarely available lesion annotations. Our all convolutional network method for classifying screening mammograms attained excellent performance in comparison with previous methods. On an independent test set of digitized film mammograms from the Digital Database for Screening Mammography (CBIS-DDSM), the best single model achieved a per-image AUC of 0.88, and four-model averaging improved the AUC to 0.91 (sensitivity: 86.1%, specificity: 80.1%). On an independent test set of full-field digital mammography (FFDM) images from the INbreast database, the best single model achieved a per-image AUC of 0.95, and four-model averaging improved the AUC to 0.98 (sensitivity: 86.7%, specificity: 96.1%). We also demonstrate that a whole image classifier trained using our end-to-end approach on the CBIS-DDSM digitized film mammograms can be transferred to INbreast FFDM images using only a subset of the INbreast data for fine-tuning and without further reliance on the availability of lesion annotations. These findings show that automatic deep learning methods can be readily trained to attain high accuracy on heterogeneous mammography platforms, and hold tremendous promise for improving clinical tools to reduce false positive and false negative screening mammography results. Code and model available at: https://github.com/lishen/end2end-all-conv .
| S-EPMC6715802 | biostudies-other
Project description:BackgroundDeep learning breast cancer risk models demonstrate improved accuracy compared with traditional risk models but have not been prospectively tested. We compared the accuracy of a deep learning risk score derived from the patient's prior mammogram to traditional risk scores to prospectively identify patients with cancer in a cohort due for screening.MethodsWe collected data on 119 139 bilateral screening mammograms in 57 617 consecutive patients screened at 5 facilities between September 18, 2017, and February 1, 2021. Patient demographics were retrieved from electronic medical records, cancer outcomes determined through regional tumor registry linkage, and comparisons made across risk models using Wilcoxon and Pearson χ2 2-sided tests. Deep learning, Tyrer-Cuzick, and National Cancer Institute Breast Cancer Risk Assessment Tool (NCI BCRAT) risk models were compared with respect to performance metrics and area under the receiver operating characteristic curves.ResultsCancers detected per thousand patients screened were higher in patients at increased risk by the deep learning model (8.6, 95% confidence interval [CI] = 7.9 to 9.4) compared with Tyrer-Cuzick (4.4, 95% CI = 3.9 to 4.9) and NCI BCRAT (3.8, 95% CI = 3.3 to 4.3) models (P < .001). Area under the receiver operating characteristic curves of the deep learning model (0.68, 95% CI = 0.66 to 0.70) was higher compared with Tyrer-Cuzick (0.57, 95% CI = 0.54 to 0.60) and NCI BCRAT (0.57, 95% CI = 0.54 to 0.60) models. Simulated screening of the top 50th percentile risk by the deep learning model captured statistically significantly more patients with cancer compared with Tyrer-Cuzick and NCI BCRAT models (P < .001).ConclusionsA deep learning model to assess breast cancer risk can support feasible and effective risk-based screening and is superior to traditional models to identify patients destined to develop cancer in large screening cohorts.
| S-EPMC9552206 | biostudies-literature
Project description:The aim of this study was to evaluate the diagnostic performance of contrast-enhanced spectral mammography (CESM) in predicting breast lesion malignancy due to microcalcifications compared to lesions that present with other radiological findings. Three hundred and twenty-one patients with 377 breast lesions that underwent CESM and histological assessment were included. All the lesions were scored using a 4-point qualitative scale according to the degree of contrast enhancement at the CESM examination. The histological results were considered the gold standard. In the first analysis, enhancement degree scores of 2 and 3 were considered predictive of malignity. The sensitivity (SE) and positive predictive value (PPV) were significative lower for patients with lesions with microcalcifications without other radiological findings (SE = 53.3% vs. 82.2%, p-value < 0.001 and PPV = 84.2% vs. 95.2%, p-value = 0.049, respectively). On the contrary, the specificity (SP) and negative predictive value (NPV) were significative higher among lesions with microcalcifications without other radiological findings (SP = 95.8% vs. 84.2%, p-value = 0.026 and NPV = 82.9% vs. 55.2%, p-value < 0.001, respectively). In a second analysis, degree scores of 1, 2, and 3 were considered predictive of malignity. The SE (80.0% vs. 96.8%, p-value < 0.001) and PPV (70.6% vs. 88.3%, p-value: 0.005) were significantly lower among lesions with microcalcifications without other radiological findings, while the SP (85.9% vs. 50.9%, p-value < 0.001) was higher. The enhancement of microcalcifications has low sensitivity in predicting malignancy. However, in certain controversial cases, the absence of CESM enhancement due to its high negative predictive value can help to reduce the number of biopsies for benign lesions.
| S-EPMC9956946 | biostudies-literature
Project description:PurposeTo develop a Breast Imaging Reporting and Data System (BI-RADS) breast density deep learning (DL) model in a multisite setting for synthetic two-dimensional mammographic (SM) images derived from digital breast tomosynthesis examinations by using full-field digital mammographic (FFDM) images and limited SM data.Materials and methodsA DL model was trained to predict BI-RADS breast density by using FFDM images acquired from 2008 to 2017 (site 1: 57 492 patients, 187 627 examinations, 750 752 images) for this retrospective study. The FFDM model was evaluated by using SM datasets from two institutions (site 1: 3842 patients, 3866 examinations, 14 472 images, acquired from 2016 to 2017; site 2: 7557 patients, 16 283 examinations, 63 973 images, 2015 to 2019). Each of the three datasets were then split into training, validation, and test. Adaptation methods were investigated to improve performance on the SM datasets, and the effect of dataset size on each adaptation method was considered. Statistical significance was assessed by using CIs, which were estimated by bootstrapping.ResultsWithout adaptation, the model demonstrated substantial agreement with the original reporting radiologists for all three datasets (site 1 FFDM: linearly weighted Cohen κ [κw] = 0.75 [95% CI: 0.74, 0.76]; site 1 SM: κw = 0.71 [95% CI: 0.64, 0.78]; site 2 SM: κw = 0.72 [95% CI: 0.70, 0.75]). With adaptation, performance improved for site 2 (site 1: κw = 0.72 [95% CI: 0.66, 0.79], 0.71 vs 0.72, P = .80; site 2: κw = 0.79 [95% CI: 0.76, 0.81], 0.72 vs 0.79, P < .001) by using only 500 SM images from that site.ConclusionA BI-RADS breast density DL model demonstrated strong performance on FFDM and SM images from two institutions without training on SM images and improved by using few SM images.Supplemental material is available for this article.Published under a CC BY 4.0 license.
| S-EPMC8082294 | biostudies-literature
Project description:BackgroundBreast microcalcifications are key diagnostically significant radiological features for localisation of malignancy. This study explores the hypothesis that breast calcification composition is directly related to the local tissue pathological state.MethodsA total of 236 human breast calcifications from 110 patients were analysed by mid-Fouries transform infrared (FTIR) spectroscopy from three different pathology types (112 invasive carcinoma (IC), 64 in-situ carcinomas and 60 benign). The biochemical composition and the incorporation of carbonate into the hydroxyapatite lattice of the microcalcifications were studied by infrared microspectroscopy. This allowed the spectrally identified composition to be directly correlated with the histopathology grading of the surrounding tissue.ResultsThe carbonate content of breast microcalcifications was shown to significantly decrease when progressing from benign to malignant disease. In this study, we report significant correlations (P<0.001) between microcalcification chemical composition (carbonate content and protein matrix : mineral ratios) and distinct pathology grades (benign, in-situ carcinoma and ICs). Furthermore, a significant correlation (P<0.001) was observed between carbonate concentrations and carcinoma in-situ sub-grades. Using the two measures of pathology-specific calcification composition (carbonate content and protein matrix : mineral ratios) as the inputs to a two-metric discriminant model sensitivities of 79, 84 and 90% and specificities of 98, 82 and 96% were achieved for benign, ductal carcinoma in situ and invasive malignancies, respectively.ConclusionsWe present the first demonstration of a direct link between the chemical nature of microcalcifications and the grade of the pathological breast disease. This suggests that microcalcifications have a significant association with cancer progression, and could be used for future objective analytical classification of breast pathology. A simple two-metric model has been demonstrated, more complex spectral analysis may yeild greater discrimination performance. Furthermore there appears to be a sequential progression of calcification composition.
| S-EPMC2965876 | biostudies-literature
Project description:BackgroundMammographic microcalcifications are considered early signs of breast cancer (BC). We examined the association between microcalcification clusters and the risk of overall and subtype-specific BC. Furthermore, we studied how mammographic density (MD) influences the association between microcalcification clusters and BC risk.MethodsWe used a prospective cohort (n = 53,273) of Swedish women with comprehensive information on BC risk factors and mammograms. The total number of microcalcification clusters and MD were measured using a computer-aided detection system and the STRATUS method, respectively. Cox regressions and logistic regressions were used to analyse the data.ResultsOverall, 676 women were diagnosed with BC. Women with ≥3 microcalcification clusters had a hazard ratio [HR] of 2.17 (95% confidence interval [CI] = 1.57-3.01) compared to women with no clusters. The estimated risk was more pronounced in premenopausal women (HR = 2.93; 95% CI = 1.67-5.16). For postmenopausal women, microcalcification clusters and MD had a similar influence on BC risk. No interaction was observed between microcalcification clusters and MD. Microcalcification clusters were significantly associated with in situ breast cancer (odds ratio: 2.03; 95% CI = 1.13-3.63).ConclusionsMicrocalcification clusters are an independent risk factor for BC, with a higher estimated risk in premenopausal women. In postmenopausal women, microcalcification clusters have a similar association with BC as baseline MD.
| S-EPMC8405644 | biostudies-literature
Project description:ObjectiveTo evaluate if a deep learning model can be used to characterise breast cancers on contrast-enhanced spectral mammography (CESM).MethodsThis retrospective mono-centric study included biopsy-proven invasive cancers with an enhancement on CESM. CESM images include low-energy images (LE) comparable to digital mammography and dual-energy subtracted images (DES) showing tumour angiogenesis. For each lesion, histologic type, tumour grade, estrogen receptor (ER) status, progesterone receptor (PR) status, HER-2 status, Ki-67 proliferation index, and the size of the invasive tumour were retrieved. The deep learning model used was a CheXNet-based model fine-tuned on CESM dataset. The area under the curve (AUC) of the receiver operating characteristic (ROC) curve was calculated for the different models: images by images and then by majority voting combining all the incidences for one tumour.ResultsIn total, 447 invasive breast cancers detected on CESM with pathological evidence, in 389 patients, which represented 2460 images analysed, were included. Concerning the ER, the deep learning model on the DES images had an AUC of 0.83 with the image-by-image analysis and of 0.85 for the majority voting. For the triple-negative analysis, a high AUC was observable for all models, in particularity for the model on LE images with an AUC of 0.90 for the image-by-image analysis and 0.91 for the majority voting. The AUC for the other histoprognostic factors was lower.ConclusionDeep learning analysis on CESM has the potential to determine histoprognostic tumours makers, notably estrogen receptor status, and triple-negative receptor status.Key points• A deep learning model developed for chest radiography was adapted by fine-tuning to be used on contrast-enhanced spectral mammography. • The adapted models allowed to determine for invasive breast cancers the status of estrogen receptors and triple-negative receptors. • Such models applied to contrast-enhanced spectral mammography could provide rapid prognostic and predictive information.
| S-EPMC8800426 | biostudies-literature
Project description:Breast density and microcalcifications are strongly associated with the risk of breast cancer. However, few studies have evaluated the combined association between these two factors and breast cancer risk. We investigated the association between breast density, microcalcifications, and risk of breast cancer. This cohort study included 3,910,815 women aged 40-74 years who were screened for breast cancer between 2009 and 2010 and followed up until 2020. The National Health Insurance Service database includes national health-screening results from the national breast cancer screening program, which is an organized screening program provided every 2 years for all women aged 40 years or older. Breast density was assessed based on the Breast Imaging Reporting and Data System (BI-RADS) 4th edition, mostly through visual assessment by radiologists. The presence or absence of microcalcifications was obtained from the mammographic results. Cox proportional hazard regression for breast cancer risk was used to estimate hazard ratios (aHRs) adjusted for breast cancer risk factors. A total of 58,315 women developed breast cancer during a median follow-up of 10.8 years. Women with breast cancer had a higher proportion of microcalcifications than women without breast cancer (0.9% vs. 0.3%). After adjusting for breast density, women with microcalcification had a 3.07-fold (95% confidence interval [CI] 2.82-3.35) increased risk of breast cancer compared to women without microcalcification. The combined association between microcalcification and breast density dramatically increased the risk of breast cancer, corresponding to a higher level of breast density. Among postmenopausal women, the highest risk group was women with BI-RADS 4 and microcalcification. These women had more than a sevenfold higher risk than women with BI-RADS 1 and non-microcalcification (aHR, 7.26; 95% CI 5.01-10.53). Microcalcification is an independent risk factor for breast cancer, and its risk is elevated when combined with breast density.
| S-EPMC9773568 | biostudies-literature
Project description:Microcalcifications, primarily biogenic apatite, occur in cancerous and benign breast pathologies and are key mammographic indicators. Outside the clinic, numerous microcalcification compositional metrics (e.g., carbonate and metal content) are linked to malignancy, yet microcalcification formation is dependent on microenvironmental conditions, which are notoriously heterogeneous in breast cancer. We interrogate multiscale heterogeneity in 93 calcifications from 21 breast cancer patients using an omics-inspired approach: For each microcalcification, we define a "biomineralogical signature" combining metrics derived from Raman microscopy and energy-dispersive spectroscopy. We observe that (i) calcifications cluster into physiologically relevant groups reflecting tissue type and local malignancy; (ii) carbonate content exhibits substantial intratumor heterogeneity; (iii) trace metals including zinc, iron, and aluminum are enhanced in malignant-localized calcifications; and (iv) the lipid-to-protein ratio within calcifications is lower in patients with poor composite outcome, suggesting that there is potential clinical value in expanding research on calcification diagnostic metrics to include "mineral-entrapped" organic matrix.
| S-EPMC9946357 | biostudies-literature