The evaluation of mammographic breast density, a critical indicator of breast cancer risk, is traditionally performed by radiologists via visual inspection of mammography images, utilizing the Breast Imaging-Reporting and Data System (BI-RADS) breast density categories. However, this method is subject to substantial interobserver variability, leading to inconsistencies and potential inaccuracies in density assessment and subsequent risk estimations. To address this, we present a deep learning-based automatic detection algorithm (DLAD) designed for the automated evaluation of breast density. Our multicentric, multi-reader study leverages a diverse dataset of 122 full-field digital mammography studies (488 images in CC and MLO projections) sourced from three institutions. We invited two experienced radiologists to conduct a retrospective analysis, establishing a ground truth for 72 mammography studies (BI-RADS class A: 18, BI-RADS class B: 43, BI-RADS class C: 7, BI-RADS class D: 4). The efficacy of the DLAD was then compared to the performance of five independent radiologists with varying levels of experience. The DLAD showed robust performance, achieving an accuracy of 0.819 (95% CI: 0.736-0.903), along with an F1 score of 0.798 (0.594-0.905), precision of 0.806 (0.596-0.896), recall of 0.830 (0.650-0.946), and a Cohen\'s Kappa (κ) of 0.708 (0.562-0.841). The algorithm achieved robust performance that matches and in four cases exceeds that of individual radiologists. The statistical analysis did not reveal a significant difference in accuracy between DLAD and the radiologists, underscoring the model\'s competitive diagnostic alignment with professional radiologist assessments. These results demonstrate that the deep learning-based automatic detection algorithm can enhance the accuracy and consistency of breast density assessments, offering a reliable tool for improving breast cancer screening outcomes.

Metaplastic breast cancer (BC-Mp) presents diagnostic and therapeutic complexities, with scant literature available. Correct assessment of tumor size by ultrasound (US) and full-field digital mammography (FFDM) is crucial for treatment planning.
METHODS: A retrospective cohort study was conducted on databases encompassing records of BC patients (2012-2022) at the National Research Institutes of Oncology (Warsaw, Gliwice and Krakow Branches). Inclusion criteria comprised confirmed diagnosis in postsurgical pathology reports with tumor size details (pT) and availability of tumor size from preoperative US and/or FFDM. Patients subjected to neoadjuvant systemic treatment were excluded. Demographics and clinicopathological data were gathered.
RESULTS: Forty-five females were included. A total of 86.7% were triple-negative. The median age was 66 years (range: 33-89). The median pT was 41.63 mm (6-130), and eight patients were N-positive. Median tumor size assessed by US and FFDM was 31.81 mm (9-100) and 34.14 mm (0-120), respectively. Neither technique demonstrated superiority (p > 0.05), but they both underestimated the tumor size (p = 0.002 for US and p = 0.018 for FFDM). Smaller tumors (pT1-2) were statistically more accurately assessed by any technique (p < 0.001). Only pT correlated with overall survival.
CONCLUSIONS: The risk of underestimation in tumor size assessment with US and FFDM has to be taken into consideration while planning surgical procedures for BC-Mp.

OBJECTIVE: To determine whether there are differences in the biopsy outcomes for suspicious calcifications detected with screening mammography using the digital breast tomosynthesis and synthetic 2D (DBT/SM) technique compared to calcifications detected using the full-field digital (DM) technique.
METHODS: This retrospective study was IRB approved. The records for all stereotactic biopsies performed for suspicious calcifications detected on screening mammograms using DM in 2011-2014 and DBT/SM in 2017-2020 were reviewed. We collected patient, imaging, and pathology data from the breast imaging database and from retrospective review of a subset of mammograms. The biopsy outcome results were categorized as benign, benign with upgrade potential (BWUP), and malignant based on final pathology. Frequencies and proportions of outcomes were calculated and compared using Mann-Whitney U tests and Wilcoxson signed-rank tests with P-values and 95% confidence intervals (95% CIs).
RESULTS: From 2011 to 2014 (DM), 1274 stereotactic biopsies of calcifications yielded 74.2% (945/1274) benign, 11.5% (147/1274) BWUP, and 14.3% (182/1274) malignant outcomes. From 2017 to 2020 (DBT/SM), 1049 stereotactic biopsies yielded 65.2% (684/1049) benign, 15.6% (164/1049) BWUP, and 19.2% (201/1049) malignant outcomes. With DBT/SM, benign biopsy outcomes decreased (9.0%, 95% CI 0.87-11.53, P < 0.05), whereas malignant biopsy outcomes increased (4.9%, 95% CI 0.94-8.36, P < 0.05). There was no significant difference in BWUP biopsy outcomes and total biopsy rates between techniques (P > 0.05).
CONCLUSIONS: Calcifications detected with screening DBT/SM technique were significantly more likely to be malignant than those found using DM. These results support using the DBT/SM technique without obtaining concurrent DM images.

OBJECTIVE: Accurate pectoral muscle removal is critical in mammographic breast density estimation and many other computer-aided algorithms. We propose a novel approach to remove pectoral muscles form mediolateral oblique (MLO) view mammograms and compare accuracy and computational efficiency with existing method (Libra).
METHODS: A pectoral muscle identification pipeline was developed. The image is first binarized to enhance contrast and then the Canny algorithm was applied for edge detection. Robust interpolation is used to smooth out the pectoral muscle region. Accuracy and computational speed of pectoral muscle identification was assessed using 951 women (1,902 MLO mammograms) from the Joanne Knight Breast Health Cohort at Washington University School of Medicine.
RESULTS: Our proposed algorithm exhibits lower mean error of 12.22% in comparison to Libra\'s estimated error of 20.44%. This 40% gain in accuracy was statistically significant (p < 0.001). The computational time for the proposed algorithm is 5.4 times faster when compared to Libra (5.1 s for proposed vs. 27.7 s for Libra per mammogram).
CONCLUSIONS: We present a novel approach for pectoral muscle removal in mammogram images that demonstrates significant improvement in accuracy and efficiency compared to existing method. Our findings have important implications for the development of computer-aided systems and other automated tools in this field.

Digital Breast Tomosynthesis (DBT) is a cutting-edge technology introduced in recent years as an in-depth analysis of breast cancer diagnostics. Compared with 2D Full-Field Digital Mammography, DBT has demonstrated greater sensitivity and specificity in detecting breast tumors. This work aims to quantitatively evaluate the impact of the systematic introduction of DBT in terms of Biopsy Rate and Positive Predictive Values for the number of biopsies performed (PPV-3). For this purpose, we collected 69,384 mammograms and 7894 biopsies, of which 6484 were Core Biopsies and 1410 were stereotactic Vacuum-assisted Breast Biopsies (VABBs), performed on female patients afferent to the Breast Unit of the Istituto Tumori \"Giovanni Paolo II\" of Bari from 2012 to 2021, thus, in the period before, during and after the systematic introduction of DBT. Linear regression analysis was then implemented to investigate how the Biopsy Rate had changed over the 10 year screening. The next step was to focus on VABBs, which were generally performed during in-depth examinations of mammogram detected lesions. Finally, three radiologists from the institute\'s Breast Unit underwent a comparative study to ascertain their performances in terms of breast cancer detection rates before and after the introduction of DBT. As a result, it was demonstrated that both the overall Biopsy Rate and the VABBs Biopsy Rate significantly decreased following the introduction of DBT, with the diagnosis of an equal number of tumors. Besides, no statistically significant differences were observed among the three operators evaluated. In conclusion, this work highlights how the systematic introduction of DBT has significantly impacted the breast cancer diagnostic procedure, by improving the diagnostic quality and thereby reducing needless biopsies, resulting in a consequent reduction in costs.

UNASSIGNED: Architectural distortion (AD) is a common imaging manifestation of breast cancer, but is also seen in benign lesions. This study aimed to construct deep learning models using mask regional convolutional neural network (Mask-RCNN) for AD identification in full-field digital mammography (FFDM) and evaluate the performance of models for malignant AD diagnosis.
UNASSIGNED: This retrospective diagnostic study was conducted at the Second Affiliated Hospital of Guangzhou University of Chinese Medicine between January 2011 and December 2020. Patients with AD in the breast in FFDM were included. Machine learning models for AD identification were developed using the Mask RCNN method. Receiver operating characteristics (ROC) curves, their areas under the curve (AUCs), and recall/sensitivity were used to evaluate the models. Models with the highest AUCs were selected for malignant AD diagnosis.
UNASSIGNED: A total of 349 AD patients (190 with malignant AD) were enrolled. EfficientNetV2, EfficientNetV1, ResNext, and ResNet were developed for AD identification, with AUCs of 0.89, 0.87, 0.81 and 0.79. The AUC of EfficientNetV2 was significantly higher than EfficientNetV1 (0.89 vs. 0.78, P=0.001) for malignant AD diagnosis, and the recall/sensitivity of the EfficientNetV2 model was 0.93.
UNASSIGNED: The Mask-RCNN-based EfficientNetV2 model has a good diagnostic value for malignant AD.

Accurate characterization of microcalcifications (MCs) in 2D digital mammography is a necessary step toward reducing the diagnostic uncertainty associated with the callback of indeterminate MCs. Quantitative analysis of MCs can better identify MCs with a higher likelihood of ductal carcinoma in situ or invasive cancer. However, automated identification and segmentation of MCs remain challenging with high false positive rates. We present a two-stage multiscale approach to MC segmentation in 2D full-field digital mammograms (FFDMs) and diagnostic magnification views. Candidate objects are first delineated using blob detection and Hessian analysis. A regression convolutional network, trained to output a function with a higher response near MCs, chooses the objects which constitute actual MCs. The method was trained and validated on 435 screening and diagnostic FFDMs from two separate datasets. We then used our approach to segment MCs on magnification views of 248 cases with amorphous MCs. We modeled the extracted features using gradient tree boosting to classify each case as benign or malignant. Compared to state-of-the-art comparison methods, our approach achieved superior mean intersection over the union (0.670 ± 0.121 per image versus 0.524 ± 0.034 per image), intersection over the union per MC object (0.607 ± 0.250 versus 0.363 ± 0.278) and true positive rate of 0.744 versus 0.581 at 0.4 false positive detections per square centimeter. Features generated using our approach outperformed the comparison method (0.763 versus 0.710 AUC) in distinguishing amorphous calcifications as benign or malignant.

To compare digital breast tomosynthesis (DBT) and MRI as an adjunct to full-field digital mammography (FFDM) for the preoperative evaluation of women with breast cancer based on mammographic density.
This retrospective study enrolled 280 patients with breast cancer who had undergone FFDM, DBT, and MRI for preoperative local tumor staging. Three radiologists independently sought the index cancer and additional ipsilateral and contralateral breast cancers using either FFDM alone, DBT plus FFDM, or MRI plus FFDM. Diagnostic performances across the three radiologists were compared among the reading modes in all patients and subgroups with dense (n = 186) and non-dense breasts (n = 94) according to mammographic density.
Of 280 patients, 46 (16.4%) had 48 additional (39 ipsilateral and nine contralateral) cancers in addition to the index cancer. For index cancers, both DBT plus FFDM and MRI plus FFDM showed sensitivities of 100% in the non-dense group. In the dense group, DBT plus FFDM showed lower sensitivity than that of MRI plus FFDM (94.6% vs. 99.6%, p < 0.001). For additional ipsilateral cancers, DBT plus FFDM showed specificity and positive predictive value (PPV) of 100% in the non-dense group, but sensitivity and negative predictive value (NPV) were not statistically different from those of MRI plus FFDM (p > 0.05). In the dense group, DBT plus FFDM showed higher specificity (98.2% vs. 94.1%, p = 0.005) and PPV (83.1% vs. 65.4%; p = 0.036) than those of MRI plus FFDM, but lower sensitivity (59.9% vs. 75.3%; p = 0.049). For contralateral cancers, DBT plus FFDM showed higher specificity than that of MRI plus FFDM (99.0% vs. 96.7%, p = 0.014), however, the other values did not differ (all p > 0.05) in the dense group.
DBT plus FFDM showed an overall higher specificity than that of MRI plus FFDM regardless of breast density, perhaps without substantial loss in sensitivity and NPV in the diagnosis of additional cancers. Thus, DBT may have the potential to be used as a preoperative breast cancer staging tool.

BACKGROUND: Pectoral muscle removal is a fundamental preliminary step in computer-aided diagnosis systems for full-field digital mammography (FFDM). Currently, two open-source publicly available packages (LIBRA and OpenBreast) provide algorithms for pectoral muscle removal within Matlab environment.
OBJECTIVE: To compare performance of the two packages on a single database of FFDM images.
METHODS: Only mediolateral oblique (MLO) FFDM was considered because of large presence of pectoral muscle on this type of projection. For obtaining ground truth, pectoral muscle has been manually segmented by two radiologists in consensus. Both LIBRA\'s and OpenBreast\'s removal performance with respect to ground truth were compared using Dice similarity coefficient and Cohen-kappa reliability coefficient; Wilcoxon signed-rank test has been used for assessing differences in performances; Kruskal-Wallis test has been used to verify possible dependence of the performance from the breast density or image laterality.
RESULTS: FFDMs from 168 consecutive women at our institution have been included in the study. Both LIBRA\'s Dice-index and Cohen-kappa were significantly higher than OpenBreast (Wilcoxon signed-rank test P < 0.05). No dependence on breast density or laterality has been found (Kruskal-Wallis test P > 0.05).
CONCLUSIONS: Libra has a better performance than OpenBreast in pectoral muscle delineation so that, although our study has not a direct clinical application, these results are useful in the choice of packages for the development of complex systems for computer-aided breast evaluation.

The purpose of this study was to test for superiority of wide-angle digital breast tomosynthesis plus synthetic mammography (Insight 2D) in comparison to full-field digital mammography (FFDM).
In this study, twenty readers interpreted 350 screening and diagnostic cases of wide-angle digital breast tomosynthesis (DBT) plus Insight 2D and FFDM in two separate reading sessions separated by at least a 6-week washout period. Breast-level estimates of the area under the curve and sensitivity along with subject-level recall rate were measured and compared between wide-angle DBT plus Insight 2D and FFDM. The same measures were also assessed for dense breasts. A hierarchical analysis plan was used to control the study\'s type I error rate at 0.05.
The mean breast-level area under the curve for distinguishing breasts with cancer from non-cancer breasts was 0.893 with DBT plus Insight 2D versus 0.837 with FFDM, showing superiority of DBT plus Insight 2D (p < 0.001). Breast-level sensitivity was significantly superior for DBT plus Insight 2D in comparison to FFDM (0.852 vs. 0.805, p = 0.043). Subject-level recall rate for DBT plus Insight 2D was significantly lower in comparison to FFDM (0.344 vs. 0.473, p < 0.001). For dense breasts, the readers\' accuracy with DBT plus Insight 2D was superior to their accuracy with FFDM (0.875 vs. 0.830, p = 0.026), and their recall rate was significantly lower for DBT plus Insight 2D in comparison to FFDM (0.338 vs. 0.441, p = 0.003).
Reader performance with wide-angle DBT plus Insight 2D is superior to that with FFDM, showing significantly higher breast-level accuracy and sensitivity and significantly lower recall rates.