Recurrent attention U-Net for segmentation and quantification of breast arterial calcifications on synthesized 2D mammograms.

Breast arterial calcifications (BAC) are a type of calcification commonly observed on mammograms and are generally considered benign and not associated with breast cancer. However, there is accumulating observational evidence of an association between BAC and cardiovascular disease, the leading cause of death in women. We present a deep learning method that could assist radiologists in detecting and quantifying BAC in synthesized 2D mammograms. We present a recurrent attention U-Net model consisting of encoder and decoder modules that include multiple blocks that each use a recurrent mechanism, a recurrent mechanism, and an attention module between them. The model also includes a skip connection between the encoder and the decoder, similar to a U-shaped network. The attention module was used to enhance the capture of long-range dependencies and enable the network to effectively classify BAC from the background, whereas the recurrent blocks ensured better feature representation. The model was evaluated using a dataset containing 2,000 synthesized 2D mammogram images. We obtained 99.8861% overall accuracy, 69.6107% sensitivity, 66.5758% F-1 score, and 59.5498% Jaccard coefficient, respectively. The presented model achieved promising performance compared with related models.

The interrelationship between obesity and race in breast cancer prognosis: a prospective cohort study.

BACKGROUND: Obesity is associated with an increased breast cancer risk in postmenopausal women and may contribute to worse outcomes. Black women experience higher obesity and breast cancer mortality rates than non-Black women. We examined associations between race, obesity, and clinical tumor stage with breast cancer prognosis. METHODS: We conducted a prospective cohort study in 1,110 breast cancer patients, using univariable and multivariable Cox regression analyses to evaluate the effects of obesity, race/ethnicity, and clinical tumor stage on progression-free and overall survival (PFS and OS). RESULTS: 22% of participants were Black, 64% were Hispanic White, and 14% were non-Hispanic White or another race. 39% of participants were obese (body mass index [BMI] ≥ 30 kg/m2). In univariable analyses, tumor stage III-IV was associated with worse PFS and OS compared to tumor stage 0-II (hazard ratio [HR] = 4.68, 95% confidence interval [CI] = 3.52-6.22 for PFS and HR = 5.92, 95% CI = 4.00-8.77 for OS). Multivariable analysis revealed an association between Black race and worse PFS in obese (HR = 2.19, 95% CI = 1.06-4.51) and non-obese (HR = 2.11, 95% CI = 1.05-4.21) women with tumors staged 0-II. Obesity alone was not associated with worse PFS or OS. CONCLUSIONS: Results suggest a complex interrelationship between obesity and race in breast cancer prognosis. The association between the Black race and worse PFS in tumor stages 0-II underscores the importance of early intervention in this group. Future studies are warranted to evaluate whether alternative measures of body composition and biomarkers are better prognostic indicators than BMI among Black breast cancer survivors.

The Interrelationship between Obesity and Race in Breast Cancer Prognosis: A Prospective Cohort Study.

Purpose: Obesity is associated with an increased breast cancer risk in postmenopausal women and may contribute to worse outcomes. Black women experience higher obesity and breast cancer mortality rates than non-Black women. We examined associations between race, obesity, and clinical tumor stage with breast cancer prognosis. Methods: We conducted a prospective cohort study in 1,110 breast cancer patients, using univariable and multivariable Cox regression analyses to evaluate the effects of obesity, race/ethnicity, and clinical tumor stage on progression-free and overall survival (PFS and OS). Results: 22% of participants were Black, 64% were Hispanic White, and 14% were non-Hispanic White or another race. 39% of participants were obese (body mass index [BMI] ≥ 30 kg/m2). In univariable analyses, tumor stage III-IV was associated with worse PFS and OS compared to tumor stage 0-II (hazard ratio [HR] = 4.68, 95% confidence interval [CI] = 3.52-6.22 for PFS and HR = 5.92, 95% CI = 4.00-8.77 for OS). Multivariable analysis revealed an association between Black race and worse PFS in obese (HR = 2.19, 95% CI = 1.06-4.51) and non-obese (HR = 2.11, 95% CI = 1.05-4.21) women with tumors staged 0-II. Obesity alone was not associated with worse PFS or OS. Conclusion: Results suggest a complex interrelationship between obesity and race in breast cancer prognosis. The association between Black race and worse PFS in tumor stages 0-II underscores the importance of early intervention in this group. Future studies are warranted to evaluate whether alternative measures of body composition and biomarkers are better prognostic indicators than BMI among Black breast cancer survivors.

Multimodal Spatiotemporal Deep Learning Framework to Predict Response of Breast Cancer to Neoadjuvant Systemic Therapy.

Current approaches to breast cancer therapy include neoadjuvant systemic therapy (NST). The efficacy of NST is measured by pathologic complete response (pCR). A patient who attains pCR has significantly enhanced disease-free survival progress. The accurate prediction of pCR in response to a given treatment regimen could increase the likelihood of achieving pCR and prevent toxicities caused by treatments that are not effective. Th early prediction of response to NST can increase the likelihood of survival and help with decisions regarding breast-conserving surgery. An automated NST prediction framework that is able to precisely predict which patient undergoing NST will achieve a pathological complete response (pCR) at an early stage of treatment is needed. Here, we propose an end-to-end efficient multimodal spatiotemporal deep learning framework (deep-NST) framework to predict the outcome of NST prior or at an early stage of treatment. The deep-NST model incorporates imaging data captured at different timestamps of NST regimens, a tumor's molecular data, and a patient's demographic data. The efficacy of the proposed work is validated on the publicly available ISPY-1 dataset, in terms of accuracy, area under the curve (AUC), and computational complexity. In addition, seven ablation experiments were carried out to evaluate the impact of each design module in the proposed work. The experimental results show that the proposed framework performs significantly better than other recent methods.

Natural Language Processing for Breast Imaging: A Systematic Review.

Natural Language Processing (NLP) has gained prominence in diagnostic radiology, offering a promising tool for improving breast imaging triage, diagnosis, lesion characterization, and treatment management in breast cancer and other breast diseases. This review provides a comprehensive overview of recent advances in NLP for breast imaging, covering the main techniques and applications in this field. Specifically, we discuss various NLP methods used to extract relevant information from clinical notes, radiology reports, and pathology reports and their potential impact on the accuracy and efficiency of breast imaging. In addition, we reviewed the state-of-the-art in NLP-based decision support systems for breast imaging, highlighting the challenges and opportunities of NLP applications for breast imaging in the future. Overall, this review underscores the potential of NLP in enhancing breast imaging care and offers insights for clinicians and researchers interested in this exciting and rapidly evolving field.

Axillary lymph node "bubbly" calcifications and body tattoo: A case series and proposed algorithm to minimize lymph node biopsies.

The presence of mammographically evident hyperdense foci within axillary lymph nodes elicits concern for calcium deposits, which in turn have a wide differential diagnosis including both benign and malignant entities. Tissue sampling, most commonly by way of image-guided core needle biopsy, is needed in many cases when a definite etiology cannot be clinically established. In this case series we present history, imaging findings, and pathology results (or long term follow-up stability as biopsy surrogate) of several women with body tattoos who at mammography were noted to have a characteristic pattern of "bubbly" pseudo-calcifications within axillary lymph nodes, and absence of other mammographic, sonographic and clinical abnormalities.

Towards a better understanding of annotation tools for medical imaging: a survey.

Medical imaging refers to several different technologies that are used to view the human body to diagnose, monitor, or treat medical conditions. It requires significant expertise to efficiently and correctly interpret the images generated by each of these technologies, which among others include radiography, ultrasound, and magnetic resonance imaging. Deep learning and machine learning techniques provide different solutions for medical image interpretation including those associated with detection and diagnosis. Despite the huge success of deep learning algorithms in image analysis, training algorithms to reach human-level performance in these tasks depends on the availability of large amounts of high-quality training data, including high-quality annotations to serve as ground-truth. Different annotation tools have been developed to assist with the annotation process. In this survey, we present the currently available annotation tools for medical imaging, including descriptions of graphical user interfaces (GUI) and supporting instruments. The main contribution of this study is to provide an intensive review of the popular annotation tools and show their successful usage in annotating medical imaging dataset to guide researchers in this area.

Convolutional neural networks for breast cancer detection in mammography: A survey.

Despite its proven record as a breast cancer screening tool, mammography remains labor-intensive and has recognized limitations, including low sensitivity in women with dense breast tissue. In the last ten years, Neural Network advances have been applied to mammography to help radiologists increase their efficiency and accuracy. This survey aims to present, in an organized and structured manner, the current knowledge base of convolutional neural networks (CNNs) in mammography. The survey first discusses traditional Computer Assisted Detection (CAD) and more recently developed CNN-based models for computer vision in mammography. It then presents and discusses the literature on available mammography training datasets. The survey then presents and discusses current literature on CNNs for four distinct mammography tasks: (1) breast density classification, (2) breast asymmetry detection and classification, (3) calcification detection and classification, and (4) mass detection and classification, including presenting and comparing the reported quantitative results for each task and the pros and cons of the different CNN-based approaches. Then, it offers real-world applications of CNN CAD algorithms by discussing current Food and Drug Administration (FDA) approved models. Finally, this survey highlights the potential opportunities for future work in this field. The material presented and discussed in this survey could serve as a road map for developing CNN-based solutions to improve mammographic detection of breast cancer further.

Breast Implant-Associated Anaplastic Large Cell lymphoma: Brief overview of current data and imaging findings.

In 2016, the World Health Organization added Breast Implant-Associated Anaplastic Large Cell lymphoma as a provisionally recognized lymphoma to the family of existing Anaplastic Large Cell lymphomas. Current estimates of the lifetime risk of the disease in women with textured breast implants range from 1:1,000 to 1:30,000. The mean interval from implant placement to diagnosis is 10.7 ± 4.6 years and the most common clinical symptom at presentation is breast swelling. A high level of clinical suspicion is recommended in patients presenting with breast symptoms and/or peri-implant fluid collection occurring more than 1 year after breast implant placement. Ultrasound is the imaging modality of choice, with a high sensitivity for peri-implant fluid and a high specificity for peri-implant mass. When ultrasound is inconclusive, breast MRI is indicated. As of today, all confirmed cases have tested positive for CD30 immunohistochemistry and the disease has shown to have an excellent prognosis when it is diagnosed earlier (localized disease), and when complete surgery, consisting of explantation, capsulectomy, and removal of any associated capsule mass, is performed. This overview summarizes the available epidemiological and clinical data of Breast Implant-Associated Anaplastic Large Cell lymphoma, with an emphasis on imaging features.

Breast Arterial Calcifications on Mammography: A Survey of Practicing Radiologists.

OBJECTIVE: To explore current practice patterns of reporting and issuing recommendations based on the presence of breast arterial calcifications on mammography and existing knowledge of their prevalence and associated factors. METHODS: An online anonymous 19-question survey was distributed to 2583 practicing radiologists who were members of the Society of Breast Imaging. Questions covered demographics, breast imaging training, practice type, and knowledge regarding the epidemiology and potential clinical significance of breast arterial calcifications detected on mammograms. Differences between groups were calculated using the chi-square test or Fisher exact test. An α level of 0.05 was used to determine statistical significance. RESULTS: Response rate was 22% (364/1662). The median age of respondents was 51 years (range: 29-76) and most were female (248/323, 77%). The most prevalent characteristics among respondents were as follows: 69% (223/323) had completed a breast imaging fellowship, 55% (179/323) were in private practice, 49% (158/323) practiced dedicated breast imaging, and 38% (124/323) had been in practice for more than 20 years. The prevalence of breast arterial calcifications was correctly estimated to be 1%-30% by 39% (125/323) of respondents. Most respondents correctly recognized the growing evidence of an association between breast arterial calcifications and coronary artery disease (275/323, 85%). However, only 15% (48/323) always reported the presence of these calcifications, and of those who report them at any time, only 0.7% (2/274) always issued recommendations. CONCLUSION: There are differences in both knowledge of the epidemiology of breast arterial calcifications and practices around their reporting amongst breast radiologists.

DU-Net: Convolutional Network for the Detection of Arterial Calcifications in Mammograms.

Breast arterial calcifications (BACs) are part of several benign findings present on some mammograms. Previous studies have indicated that BAC may provide evidence of general atherosclerotic vascular disease, and potentially be a useful marker of cardiovascular disease (CVD). Currently, there is no technique in use for the automatic detection of BAC in mammograms. Since a majority of women over the age of 40 already undergo breast cancer screening with mammography, detecting BAC may offer a method to screen women for CVD in a way that is effective, efficient, and broad reaching, at no additional cost or radiation. In this paper, we present a deep learning approach for detecting BACs in mammograms. Inspired by the promising results achieved using the U-Net model in many biomedical segmentation problems and the DenseNet in semantic segmentation, we extend the U-Net model with dense connectivity to automatically detect BACs in mammograms. The presented model helps to facilitate the reuse of computation and improve the flow of gradients, leading to better accuracy and easier training of the model. We evaluate the performance using a set of full-field digital mammograms collected and prepared for this task from a publicly available dataset. Experimental results demonstrate that the presented model outperforms human experts as well as the other related deep learning models. This confirms the effectiveness of our model in the BACs detection task, which is a promising step in providing a cost-effective risk assessment tool for CVD.

The Role of Artificial Intelligence in Diagnostic Radiology: A Survey at a Single Radiology Residency Training Program.

PURPOSE: Advances in artificial intelligence applied to diagnostic radiology are predicted to have a major impact on this medical specialty. With the goal of establishing a baseline upon which to build educational activities on this topic, a survey was conducted among trainees and attending radiologists at a single residency program. METHODS: An anonymous questionnaire was distributed. Comparisons of categorical data between groups (trainees and attending radiologists) were made using Pearson χ2 analysis or an exact analysis when required. Comparisons were made using the Wilcoxon rank sum test when the data were not normally distributed. An α level of 0.05 was used. RESULTS: The overall response rate was 66% (69 of 104). Thirty-six percent of participants (n = 25) reported not having read a scientific medical article on the topic of artificial intelligence during the past 12 months. Twenty-nine percent of respondents (n = 12) reported using artificial intelligence tools during their daily work. Trainees were more likely to express doubts on whether they would have pursued diagnostic radiology as a career had they known of the potential impact artificial intelligence is predicted to have on the specialty (P = .0254) and were also more likely to plan to learn about the topic (P = .0401). CONCLUSIONS: Radiologists lack exposure to current scientific medical articles on artificial intelligence. Trainees are concerned by the implications artificial intelligence may have on their jobs and desire to learn about the topic. There is a need to develop educational resources to help radiologists assume an active role in guiding and facilitating the development and implementation of artificial intelligence tools in diagnostic radiology.

Utility of supplemental screening with breast ultrasound in asymptomatic women with dense breast tissue who are not at high risk for breast cancer.

OBJECTIVE: To assess the results of an initial round of supplemental screening with hand-held bilateral breast ultrasound following a negative screening mammogram in asymptomatic women with dense breast tissue who are not at high risk for breast cancer. MATERIALS AND METHODS: A retrospective, Health Insurance Portability and Accountability Act compliant, Institutional Research Board approved study was performed at a single academic tertiary breast center. Informed consent was waived. A systematic review of the breast imaging center database was conducted to identify and retrieve data for all asymptomatic women, who were found to have heterogeneously dense or extremely dense breast tissue on screening bilateral mammograms performed from July 1, 2010 through June 30, 2012 and who received a mammographic final assessment American College of Radiology's (ACR) Breast Imaging Reporting and Data System (BI-RADS) category 1 or BI-RADS category 2. Hand-held screening ultrasound was performed initially by a technologist followed by a radiologist. Chi-square and t-test were used and statistical significance was considered at P < 0.05. RESULTS: A total of 1210 women were identified. Of these, 394 underwent the offered supplemental screening ultrasound. BI-RADS category 1 or 2 was assigned to 323 women (81.9%). BI-RADS category 3 was assigned to 50 women (12.9%). A total of 26 biopsies/aspirations were recommended and performed in 26 women (6.6%). The most common finding for which biopsy was recommended was a solid mass (88.5%) with an average size of 0.9 cm (0.5-1.7 cm). Most frequent pathology result was fibroadenoma (60.8%). No carcinoma was found. CONCLUSION: Our data support the reported occurrence of a relatively high number of false positives at supplemental screening with breast ultrasound following a negative screening mammogram in asymptomatic women with dense breast tissue, who are not at a high risk of developing breast cancer, and suggests that caution is necessary in establishing wide implementation of this type of supplemental screening for all women with dense breast tissue without considering other risk factors for breast cancer.

Time Spent by Breast Imaging Radiologists to Perform Value-Added Activities at an Academic Cancer Center.

BACKGROUND: Health care reform in the United States has generated a paradigm shift in the practice of radiology aimed at increasing the degree of patient-centered care. We conducted a study to quantify the amount of time breast imaging radiologists spend on value-added activities at an academic comprehensive cancer center located in Miami, Florida, and accredited by the American College of Radiology as a Breast Imaging Center of Excellence. METHODS: A prospective, observational study was conducted during a period of 20 consecutive workdays. Three participating breast imaging radiologists maintained a real-time log of each activity performed. A generalized linear model was used to perform a 1-way analysis of variance. An alpha level of .05 was used to determine statistical significance. RESULTS: The average daily time dedicated to these activities was 92.1 minutes (range, 56.4-132.2). The amount of time significantly differed among breast imaging radiologists and correlated with their assigned daily role (P < .001 for both) but was independent of their years of experience. The daily role that required the most time was the interpretation of diagnostic imaging studies, which is when most interactions with patients, their relatives, and referring physicians occurred. The specific activity that required the most time was preparing for and participating in tumor boards. CONCLUSIONS: Our findings suggest that the breast imaging radiologists who participated in this study dedicated a significant amount of their time to value-added activities to help improve patients' experience across the continuity of their care. We propose that similar studies be conducted at other institutions to better assess the magnitude of this finding across different breast imaging care settings.

Primary neuroendocrine carcinoma of the breast: report of 2 cases and literature review.

Neuroendocrine tumors of the breast are very rare accounting for less than 0.1% of all breast cancers and less than 1% of all neuroendocrine tumors. Focal neuroendocrine differentiation can be found in different histologic types of breast carcinoma including in situ and invasive ductal or invasive lobular. However, primary neuroendocrine carcinoma of the breast requires the expression of neuroendocrine markers in more than 50% of the cell population, the presence of ductal carcinoma in situ, and the absence of clinical evidence of concurrent primary neuroendocrine carcinoma of any other organ. Reports discussing the imaging characteristics of this rare carcinoma in different breast imaging modalities are scarce. We present 2 cases of primary neuroendocrine carcinoma of the breast for which mammography, ultrasound, and magnetic resonance imaging findings and pathology findings are described. A review of the medical literature on this particular topic was performed, and the results are presented.

Can mammographic and sonographic imaging features predict the Oncotype DX™ recurrence score in T1 and T2, hormone receptor positive, HER2 negative and axillary lymph node negative breast cancers?

To determine whether mammographic or sonographic features can predict the Oncotype DX™ recurrence scores (RS) in patients with TI-II, hormone receptor (HR) positive, HER2/neu negative and node negative breast cancers. Institutional board review was obtained and informed consent was waived for this retrospective study. Seventy-eight patients with stage I-II invasive breast cancer that was HR positive, HER2 negative, and lymph node negative for whom mammographic and or sonographic imaging and Oncotype DX™ assay scores were available were included in the study Four breast dedicated radiologists blinded to the RS retrospectively described the lesions according to BI-RADS lexicon descriptors. Multivariable logistic regression was used to test for significant independent predictors of low (<18) versus intermediate to high range (≥18). Two imaging features reached statistical significance in predicting low from intermediate or high risk RS: pleomorphic microcalcifications within a mass (P = 0.017); OR 8.37, 95 % CI (1.47-47.79) on mammography and posterior acoustic enhancement in a mass on ultrasound (P = 0.048); OR 4.35, 95 % CI (1.01-18.73) on multivariable logistic regression. A mass with pleomorphic microcalcifications on mammography or the presence of posterior acoustic enhancement on ultrasound may predict an intermediate to high RS as determined by the Oncotype DX(TM) assay in patients with stage I-II HR positive, HER2 negative, and lymph node negative invasive breast cancer.

Diagnostic performance of CT attenuation values of focal 18F-FDG avid breast lesions detected on whole-body PET-CT in postoperative breast cancer patients.

To assess whether CT attenuation values help in differentiating benign from malignant etiology of focal (18) F-FDG avid breast lesions detected on whole-body PET/CT exam in postoperative breast cancer patients. Institutional review board approval and waived informed consent were obtained for this HIPAA-compliant retrospective study. Between January 2009 and July 2011, a total of 85 patients had 97 focal (18) F-FDG avid breast lesions on whole-body PET/CT. Of these, 54 (56%) lesions were biopsy-proven primary invasive breast carcinoma that had not undergone treatment at the time of PET/CT, 35 (36%) were benign lesions, and 8 were locally recurrent breast carcinoma. Mean attenuation values were retrospectively measured in Hounsfield units (HU) for the correlative lesion on the CT portion of the exam. Receiver-operating characteristic curves (ROC) were calculated to determine the optimal cutoff values of HU that would best discriminate between benign and malignant lesions. Interobserver agreement for measured mean attenuation values was assessed by calculating the intraclass correlation coefficient (ICC). Mean HU for the benign lesions group and the local recurrence lesions group was -11.0 ± 30.3 versus 32.9 ± 6.87 (p < 0.0002). ROC curve analysis comparing benign breast lesions to local recurrence lesions found an optimal cutoff value of 17 HU (area under curve = 0.982, p < 0.0001, Sensitivity = 100%, Specificity = 89%). ICC with regard to interobserver agreement in measuring the mean HU of the benign lesions was 0.84 (95% confidence interval 0.64-0.93) and for the malignant lesions was 0.88 (95% confidence interval 0.77-0.94). A CT attenuation threshold value of less than 17 HU suggests benign etiology of focal (18) FDG avid breast lesions in postoperative breast cancer patients. If confirmed by additional studies, these findings may provide additional information to guide the treating physician regarding decisions for supplementary imaging or the need to biopsy.