DeepCAT: Deep Computer-Aided Triage of Screening Mammography.

Although much deep learning research has focused on mammographic detection of breast cancer, relatively little attention has been paid to mammography triage for radiologist review. The purpose of this study was to develop and test DeepCAT, a deep learning system for mammography triage based on suspicion of cancer. Specifically, we evaluate DeepCAT's ability to provide two augmentations to radiologists: (1) discarding images unlikely to have cancer from radiologist review and (2) prioritization of images likely to contain cancer. We used 1878 2D-mammographic images (CC & MLO) from the Digital Database for Screening Mammography to develop DeepCAT, a deep learning triage system composed of 2 components: (1) mammogram classifier cascade and (2) mass detector, which are combined to generate an overall priority score. This priority score is used to order images for radiologist review. Of 595 testing images, DeepCAT recommended low priority for 315 images (53%), of which none contained a malignant mass. In evaluation of prioritizing images according to likelihood of containing cancer, DeepCAT's study ordering required an average of 26 adjacent swaps to obtain perfect review order. Our results suggest that DeepCAT could substantially increase efficiency for breast imagers and effectively triage review of mammograms with malignant masses.

Benefits of digital breast tomosynthesis: A lesion-level analysis.

OBJECTIVE: To compare outcome metrics of digital breast tomosynthesis (DBT) breast cancer screening with full-field digital mammogram (FFDM); specifically, to compare recall rates by the type of recalled finding, and to assess if screening with DBT versus FFDM changes biopsy recommendations and if the likelihood of malignancy varied by lesion type, if detected on DBT or FFDM screening mammogram. METHODS: The outcomes of 22,055 FFDM and DBT screening mammograms were retrospectively reviewed. The exams were performed at an academic institution between August 2015 and September 2016. Performance of screening with FFDM versus DBT was compared in terms of recall rate and percentage of recalled lesions resulting in a cancer diagnosis, with subset analyses performed for specific mammographic findings. RESULTS: The recall rate was 10.6% for FFDM and 8.0% for DBT (p < 0.001). Architectural distortion was more likely to be recalled on DBT screening than FFDM (p = 0.002), and was associated with an increased likelihood of malignancy (p = 0.008). Asymmetries were less likely to be recalled on DBT than FFDM (p < 0.001) screening mammogram, but more likely to be recommended for biopsy when detected on DBT. Calcifications more frequently required short-term follow-up or biopsy on both DBT and FFDM. CONCLUSIONS: DBT screening confers an advantage in detection of architectural distortion representing malignancy. Recall rate of asymmetries are reduced with screening DBT, probably due to reduction of tissue superimposition. Calcifications pose a particularly difficult diagnostic challenge for breast imagers, regardless of screening mammogram type.

Image-guided Procedure Versus 2-year Follow-up for a BI-RADS 3 Probably Benign Lesion: A Cost Comparison Analysis.

OBJECTIVE: Perform a comparison between the costs of image-guided breast procedures versus 2-year imaging follow-up for findings classified as BI-RADS assessment category 3-probably benign. METHODS: The national payment amount costs at non-facility locations were obtained from the Centers for Medicare and Medicaid Services physician fee schedule for breast imaging-related Current Procedural Terminology codes. Total costs were calculated and compared for management algorithms of 2-year imaging follow-up of a BI-RADS 3 lesion from 2018 through 2019 versus performing an image-guided procedure of the lesion in 2018 after the initial diagnostic imaging. RESULTS: Two-year mammographic follow-up of a BI-RADS 3 finding costs $484. This was less than a stereotactic-guided breast biopsy, which cost at least $1055. Two-year follow-up for a probably benign US finding cost $615 compared to $1173 for the least expensive US-guided breast biopsy scenario. For breast MRI, 2-year imaging follow-up cost $1510, which was also less than most MRI-guided breast biopsy scenarios. The one exception in which biopsy costs less than 2-year imaging follow-up was in the setting of an MRI-guided biopsy in the average-risk population without a post-benign biopsy follow-up breast MRI; in this setting, MRI biopsy cost $1235. CONCLUSION: In 2018-2019, 2-year imaging follow-up of a BI-RADS 3 finding continues to be less costly than an immediate procedure, except for MRI-guided breast biopsy in the average-risk population without a post-benign biopsy follow-up MRI.

Outcomes of Foci on Breast MRI: Features Associated With Malignancy.

OBJECTIVE. The objective of this study was to determine the outcomes of foci seen on breast MRI and to evaluate imaging features associated with malignancy. MATERIALS AND METHODS. In this institutional review board-approved retrospective study, we reviewed 200 eligible foci in 179 patients that were assigned BI-RADS category of 3 or 4 from December 2004 to August 2018. Clinical and imaging features of all eligible foci were collected, and associations with malignant outcomes were evaluated. Malignancy rates were also calculated. RESULTS. Of 200 eligible foci, 64 were assigned BI-RADS category 3 and 136 were assigned BI-RADS category 4. The malignancy rate was 1.6% (1/64) among BI-RADS 3 foci and 17.6% (24/136) for BI-RADS 4 foci. The majority of malignant foci represented invasive breast cancer (68.0%, 17/25). Focus size and washout kinetics were significantly associated with malignant outcome (p < 0.05). CONCLUSION. Despite the high prevalence of foci on breast MRI, data are limited to guide their management. Foci should not be disregarded, because foci undergoing biopsy had a malignancy rate of 17.6%, with the majority of malignant foci representing invasive cancer. Larger size and washout kinetics were associated with malignancy in our study and should raise the suspicion level for a focus on breast MRI.

Impact of Improved Screening Mammography Recall Lay Letter Readability on Patient Follow-Up.

PURPOSE: In the setting of abnormal results on screening mammography, the Mammography Quality Standards Act mandates that patients receive a mailed "recall" lay letter informing them to return for additional follow-up imaging. The language used in this letter should be "easily understood by a lay person." In February 2019, the authors' institution revised the language of its recall lay letter to the sixth grade reading level. The purpose of this study was to analyze the effect of improved readability on patient follow-up rates. METHODS: In this retrospective study, data from all screening mammograms at a single institution with BI-RADS category 0 assessments excluding technical recalls between February 2018 to February 2019 (pre-intervention group) and February 2019 to February 2020 (post-intervention group) were reviewed. The primary outcome measure was the percentage of patients in each intervention group who returned for their diagnostic follow-up examination within 60 days (the standard recommended by the Centers for Disease Control and Prevention). Univariate and multivariate logistic regression was done to estimate odds ratios and 95% confidence intervals for follow-up within 60 days. RESULTS: This study included 1,987 patients in the pre-intervention group and 2,211 patients in the post-intervention group. The patient follow-up rate within 60 days increased from 90.1% (1,790 of 1,987) in the pre-intervention group to 93.9% (2,076 of 2,211) in the post-intervention group (P < .001). When controlling for imaging site, patients in the post-intervention group had 1.96-fold increased odds of returning for a diagnostic follow-up examination within 60 days (95% confidence interval, 1.52-2.53). CONCLUSIONS: Revising an institution's recall lay letter to a lower reading grade level significantly improved timely patient follow-up.

Analysis of Breast Cancer Patients with T1-2 Tumors and 1-3 Positive Lymph Nodes Treated with or without Postmastectomy Radiation Therapy.

The use of postmastectomy radiation therapy (PMRT) has been recommended for patients with 4 or more positive lymph nodes, however, its role in patients with 1-3 positive lymph nodes remains unclear. The purpose of this study is to evaluate oncological outcomes for breast cancer patients with T1-2 tumors and 1-3 positive lymph nodes after undergoing PMRT. We performed a single-institution retrospective investigation that evaluated the association between PMRT and outcomes in breast cancer patients with T1-2 tumors and 1-3 positive lymph nodes, who underwent mastectomy from 2004 to 2015. Multivariable Cox proportional hazards regression was used to evaluate the association of PMRT with disease-free survival and overall survival. A total of 379 patients met inclusion criteria, of which 204 (54%) received PMRT while 175 (46%) did not receive PMRT following mastectomy and were followed over a median of 5.2 years (25th-75th percentile: 2.8-8.4 years). Recurrence was similar in patients receiving PMRT compared to those that did not: locoregional (0 vs 3, P = 0.061), distant (9 vs 3, P = 0.135) and any recurrence (11 vs 7, P = 0.525). After adjustment for potential confounding variables, PMRT was not associated with a statistically significant difference in disease-free survival (HR: 0.93; 95% CI: 0.48, 1.79) or overall survival (HR: 0.91; 95% CI: 0.45, 1.85). PMRT was not associated with improved oncological outcomes in patients with T1-2 breast cancer and 1-3 positive lymph nodes at our institution.

Impact of Telephone Communication on Patient Adherence With Follow-Up Recommendations After an Abnormal Screening Mammogram.

OBJECTIVE: The Mammography Quality Standards Act requires written communication to every patient after an abnormal screening mammogram. Starting in 2013, our institution began telephoning all patients with a Breast Imaging Reporting and Data System (BI-RADS) assessment 0 on the next business day to schedule their diagnostic follow-up. Our aim is to analyze the changes in adherence and time to follow-up of patients recalled from screening mammography after the implementation of telephone communication. METHODS: This retrospective study reviewed data from screening mammograms at our institution with a BI-RADS 0 assessment excluding technical recalls between January 1, and December 31, 2011 (pre-intervention group), 2014 (early post-intervention group), and 2018 (later post-intervention group). We compared patient adherence with timely follow-up diagnostic mammography (within 60 days) in these three groups using univariate and multivariate logistic regression. Cox proportional hazards regression analysis was done to evaluate the impact of telephone communication on time to follow-up. RESULTS: This study included 1899 women in 2011, 2829 women in 2014, and 1999 women in 2018. We found that 73.1% patients in 2011 returned for their diagnostic follow-up within 60 days compared to 87.6% in 2014 and 90.0% in 2018, P < 0.001. Median time to follow-up in 2011 was 28 days (IQR 17-76 days) compared to 15 days (IQR 9-28 days) in 2014 and 9 days (IQR 5-18) in 2018, P < 0.001. CONCLUSION: A higher percentage of women were adherent with a timely diagnostic follow-up examination after an abnormal screening mammogram with the addition of telephone communication.

Improving State-Mandated Breast Density Notifications.

PURPOSE: Effective written communication directly affects health care outcomes. Since 2016, the complex language of state-mandated breast density notifications (BDNs) has been challenged, because it is perceived to be beyond the comprehension of most patients. The aim of this study was to assess whether a revised BDN written at a lower reading grade level improves understanding compared with the current state-mandated BDN. METHODS: A revised notification with similar content to the current state-mandated BDN was developed. Both notifications were presented to patients for direct comparison, using a paper survey asking questions that evaluated patients' perceptions and convictions associated with breast density. Surveys were distributed at four outpatient imaging centers to screening mammography patients. RESULTS: The current BDN's mean readability metric was 13.4, and that of the revised BDN was 6.6. Five hundred surveys were analyzed. Survey data demonstrated that 56.6% of all women perceived that dense breast tissue results indicated a "high" associated lifetime breast cancer risk from the current state-mandated BDN compared with only 2.2% with the revised notification (P < .001). Nearly all women were more likely to initiate discussions with their providers regarding their breast tissue density after reading the revised notification (96.0%) as opposed to the current state-mandated BDN (32.8%; P < .001). CONCLUSIONS: A significant portion of women misinterpret the intended messages of the current state-mandated BDN. Thus, a revised notification at a lower reading grade level may improve understanding of breast density, leading to improved individualized breast cancer screening for women with dense breasts.

Multiparametric deep learning tissue signatures for a radiological biomarker of breast cancer: Preliminary results.

PURPOSE: Deep learning is emerging in radiology due to the increased computational capabilities available to reading rooms. These computational developments have the ability to mimic the radiologist and may allow for more accurate tissue characterization of normal and pathological lesion tissue to assist radiologists in defining different diseases. We introduce a novel tissue signature model based on tissue characteristics in breast tissue from multiparametric magnetic resonance imaging (mpMRI). The breast tissue signatures are used as inputs in a stacked sparse autoencoder (SSAE) multiparametric deep learning (MPDL) network for segmentation of breast mpMRI. METHODS: We constructed the MPDL network from SSAE with 5 layers with 10 nodes at each layer. A total cohort of 195 breast cancer subjects were used for training and testing of the MPDL network. The cohort consisted of a training dataset of 145 subjects and an independent validation set of 50 subjects. After segmentation, we used a combined SAE-support vector machine (SAE-SVM) learning method for classification. Dice similarity (DS) metrics were calculated between the segmented MPDL and dynamic contrast enhancement (DCE) MRI-defined lesions. Sensitivity, specificity, and area under the curve (AUC) metrics were used to classify benign from malignant lesions. RESULTS: The MPDL segmentation resulted in a high DS of 0.87 ± 0.05 for malignant lesions and 0.84 ± 0.07 for benign lesions. The MPDL had excellent sensitivity and specificity of 86% and 86% with positive predictive and negative predictive values of 92% and 73%, respectively, and an AUC of 0.90. CONCLUSIONS: Using a new tissue signature model as inputs into the MPDL algorithm, we have successfully validated MPDL in a large cohort of subjects and achieved results similar to radiologists.

Breast MRI BI-RADS 3: Impact of Patient-Level Factors on Compliance With Short-Term Follow-Up.

PURPOSE: A BI-RADS 3 assessment on breast MRI is given when a finding is estimated to have less than 2% chance of breast cancer. Patients in this category are typically recommended to return for a 6-month follow-up MRI. Compliance with this recommendation is low, and we aim to understand which factors are associated with compliance. MATERIALS AND METHODS: All patients with an MRI examination given a BI-RADS category 3 between February 1, 2011, and June 30, 2016, were retrospectively reviewed. Patient demographics and breast-related medical history were extracted from the electronic medical record. Patients presenting for follow-up MRI between 3 and 10 months were considered compliant. Univariate and multivariate analysis was performed to identify which patient-level factors were associated with compliance with follow-up MRI. RESULTS: Overall, 190 women with a BI-RADS 3 assessment on MRI were included in the study. Of these women, 106 were compliant with the recommended follow-up MRI (57.3%), 34 had delayed follow-up (18.4%), and 45 were noncompliant (24.3%). Reason for examination, personal history of breast cancer, and family history of breast cancer were significantly associated with compliance. CONCLUSIONS: We found that 75.7% of patients had a follow-up MRI after a BI-RADS 3 assessment, but only 57.3% were timely in their follow-up. Our data suggest that there may be subsets of patients who would benefit from additional support and resources to help increase overall compliance and timely compliance.

A Global Perspective on Screening.

Breast cancer is emerging as a major global public health problem. Incidence and mortality continues to rise in low- and middle-income countries (LMICs). A significant and growing disparity exists between high-income countries and LMICs in the availability of screening services and associated preventable mortality. However, population imaging-based screening programs are not appropriate for all settings. Planners should perform a thorough assessment of the target setting prior to implementing any breast cancer detection program, as appropriate guidelines vary according to the resources available. Financial, social, and cultural barriers to breast cancer care need to be addressed to sustainably improve the morbidity and mortality of the populations and make efficient use of available services. Creative approaches, such as mobile and portable imaging and bundling of services, can facilitate the installation of early breast cancer detection programs in LMICs. While image-based screening programs are not initially resource-appropriate in many LMICs, planners can work towards this goal as part of their comprehensive breast cancer detection strategy.

Ultrasound for Breast Cancer Detection Globally: A Systematic Review and Meta-Analysis.

PURPOSE: Mammography is not always available or feasible. The purpose of this systematic review and meta-analysis is to assess the diagnostic performance of ultrasound as a primary tool for early detection of breast cancer. MATERIALS AND METHODS: For this systematic review and meta-analysis, we comprehensively searched PubMed and SCOPUS to identify articles from January 2000 to December 2018 that included data on the performance of ultrasound for detection of breast cancer. Studies evaluating portable, handheld ultrasound as an independent detection modality for breast cancer were included. Quality assessment and bias analysis were performed with the Quality Assessment of Diagnostic Accuracy Studies-2 tool. Sensitivity analyses and meta-regression were used to explore heterogeneity. The study protocol has been registered with the international prospective register of systematic reviews (PROSPERO identifier: CRD42019127752). RESULTS: Of the 526 identified studies, 26 were eligible for inclusion. Ultrasound had an overall pooled sensitivity and specificity of 80.1% (95% CI, 72.2% to 86.3%) and 88.4% (95% CI, 79.8% to 93.6%), respectively. When only low- and middle-income country data were considered, ultrasound maintained a diagnostic sensitivity of 89.2% and specificity of 99.1%. Meta-analysis of the included studies revealed heterogeneity. The high sensitivity of ultrasound for the detection of breast cancer was not statistically significantly different in subgroup analyses on the basis of mean age, risk, symptoms, study design, bias level, and study setting. CONCLUSION: Given the increasing burden of breast cancer and infeasibility of mammography in certain settings, we believe these results support the potential use of ultrasound as an effective primary detection tool for breast cancer, which may be beneficial in low-resource settings where mammography is unavailable.

DNA Methylation Markers for Breast Cancer Detection in the Developing World.

PURPOSE: An unmet need in low-resource countries is an automated breast cancer detection assay to prioritize women who should undergo core breast biopsy and pathologic review. Therefore, we sought to identify and validate a panel of methylated DNA markers to discriminate between cancer and benign breast lesions using cells obtained by fine-needle aspiration (FNA).Experimental Design: Two case-control studies were conducted comparing cancer and benign breast tissue identified from clinical repositories in the United States, China, and South Africa for marker selection/training (N = 226) and testing (N = 246). Twenty-five methylated markers were assayed by Quantitative Multiplex-Methylation-Specific PCR (QM-MSP) to select and test a cancer-specific panel. Next, a pilot study was conducted on archival FNAs (49 benign, 24 invasive) from women with mammographically suspicious lesions using a newly developed, 5-hour, quantitative, automated cartridge system. We calculated sensitivity, specificity, and area under the receiver-operating characteristic curve (AUC) compared with histopathology for the marker panel. RESULTS: In the discovery cohort, 10 of 25 markers were selected that were highly methylated in breast cancer compared with benign tissues by QM-MSP. In the independent test cohort, this panel yielded an AUC of 0.937 (95% CI = 0.900-0.970). In the FNA pilot, we achieved an AUC of 0.960 (95% CI = 0.883-1.0) using the automated cartridge system. CONCLUSIONS: We developed and piloted a fast and accurate methylation marker-based automated cartridge system to detect breast cancer in FNA samples. This quick ancillary test has the potential to prioritize cancer over benign tissues for expedited pathologic evaluation in poorly resourced countries.

Validation of a low-cost, carbon dioxide-based cryoablation system for percutaneous tumor ablation.

Breast cancer rates are rising in low- and middle-income countries (LMICs), yet there is a lack of accessible and cost-effective treatment. As a result, the cancer burden and death rates are highest in LMICs. In an effort to meet this need, our work presents the design and feasibility of a low-cost cryoablation system using widely-available carbon dioxide as the only consumable. This system uses an 8-gauge outer-diameter needle and Joule-Thomson expansion to percutaneously necrose tissue with cryoablation. Bench top experiments characterized temperature dynamics in ultrasound gel demonstrated that isotherms greater than 2 cm were formed. Further, this system was applied to mammary tumors in an in vivo rat model and necrosis was verified by histopathology. Finally, freezing capacity under a large heat load was assessed with an in vivo porcine study, where volumes of necrosis greater than 1.5 cm in diameter confirmed by histopathology were induced in a highly perfused liver after two 7-minute freeze cycles. These results demonstrate the feasibility of a carbon-dioxide based cryoablation system for improving solid tumor treatment options in resource-constrained environments.

Improving Patient Comprehension of Screening Mammography Recall Lay Letters.

PURPOSE: In accordance with the Mammography Quality Standards Act, recall lay letters should be written using language that is "easily understood by a lay person." The authors hypothesized that the readability of their institution's current recall lay letter may contribute to the misinterpretation of intended time for patient return. Thus, the aim of this study was to evaluate the comprehension of the current recall lay letter statement. METHODS: The Flesch-Kincaid grade level was used to assess readability of the lay letter currently used at the authors' institution. A revised statement was formulated. A single paper-based survey presented both current and revised statements and asked questions to probe patients' understanding of expected time to return. Surveys were provided to screening mammography patients at four outpatient imaging centers. RESULTS: The Flesch-Kincaid grade levels of the current and revised statements were 12th and 4th grades, respectively. Five hundred ninety-nine surveys were analyzed. Survey data demonstrated that only 49.6% of all patients understood that the current statement requests return within 1 month specifically with the breast center, compared with 95.2% of patients with the revised statement (P < .001). Of the patients who misunderstood when to follow up with the breast center, about 80% had achieved less than a college degree (P < .001). CONCLUSIONS: The grade level of the authors' institution's current lay letter statement is greater than the national recommendation for the average patient, which is between 6th and 8th grade. These data suggest that revising the language of recall statements to the recommended grade level would potentially improve patient understanding and compliance.

Deep-Learning-Based Semantic Labeling for 2D Mammography and Comparison of Complexity for Machine Learning Tasks.

Machine learning has several potential uses in medical imaging for semantic labeling of images to improve radiologist workflow and to triage studies for review. The purpose of this study was to (1) develop deep convolutional neural networks (DCNNs) for automated classification of 2D mammography views, determination of breast laterality, and assessment and of breast tissue density; and (2) compare the performance of DCNNs on these tasks of varying complexity to each other. We obtained 3034 2D-mammographic images from the Digital Database for Screening Mammography, annotated with mammographic view, image laterality, and breast tissue density. These images were used to train a DCNN to classify images for these three tasks. The DCNN trained to classify mammographic view achieved receiver-operating-characteristic (ROC) area under the curve (AUC) of 1. The DCNN trained to classify breast image laterality initially misclassified right and left breasts (AUC 0.75); however, after discontinuing horizontal flips during data augmentation, AUC improved to 0.93 (p < 0.0001). Breast density classification proved more difficult, with the DCNN achieving 68% accuracy. Automated semantic labeling of 2D mammography is feasible using DCNNs and can be performed with small datasets. However, automated classification of differences in breast density is more difficult, likely requiring larger datasets.

Review of axillary web syndrome: What the radiologist should know.

Axillary web syndrome is common after axillary surgery, frequently affecting breast cancer patients. In this condition, patients develop one or more linear bands of firm tissue, also known as "cords", in the axilla and arm, associated with pain and limited range of motion of the shoulder and arm. Radiologists may encounter this syndrome in patients referred for axillary or upper extremity ultrasound, and should be aware of the physical examination and ultrasound findings for accurate diagnosis. However, there are currently limited articles about this syndrome published in radiology journals, suggesting that radiologists may be unaware of this entity. In this work, axillary web syndrome will be discussed, including background knowledge, incidence, clinical presentation, possible etiology, and ultrasound appearance.

Factors Impacting False Positive Recall in Screening Mammography.

RATIONALE AND OBJECTIVES: Our objective was to identify factors impacting false positive recalls in screening mammography. MATERIALS AND METHODS: We retrospectively reviewed our screening mammography database from August 31, 2015 to September 30, 2016, including full field digital mammograms (FFDM) and digital breast tomosynthesis (DBT) mammograms. False positive (FP) exams were defined as Breast Imaging-Reporting and Data System (BI-RADS) 1 or 2 assessments at diagnostic imaging with 1 year cancer-free follow-up, Breast Imaging-Reporting and Data System 3 assessment at diagnostic imaging with 2 years cancer free follow-up, or biopsy with benign pathology. True positives were defined as malignant pathology on biopsy or surgical excision. We evaluated the association of FP recalls with multiple patient-level factors and imaging features. RESULTS: A total of 22,055 screening mammograms were performed, and 1887 patients were recalled (recall rate 8.6%). Recall rate was lower for DBT than full field digital mammograms (8.0% vs 10.6%, p < 0.001). FP results were lower if prior mammograms were available (90.8% vs 95.8%, p = 0.02), and if there was a previous benign breast biopsy (87.6% vs 92.9%, p = 0.01). Mean age for the FP group was lower than the true positive group (56.1 vs 62.9 years, p < 0.001). There were no significant differences in FP recalls based on history of high-risk lesions, family history of breast or ovarian cancer, hormone use, breast density, race, or body mass index. CONCLUSION: FP recalls were significantly less likely with DBT, in older women, in patients with prior mammograms available for comparison, and in patients with histories of benign breast biopsy. This study supports the importance of using DBT in the screening setting and obtaining prior mammograms for comparison.