Search Results (14)

Background: Synthetic 2D mammography was developed to decrease radiation exposure, but to our knowledge there have been no studies evaluating the impact of implementation of full field synthetic mammography/digital breast tomosynthesis (FFSM/DBT) on indications for stereotactic biopsy. Objective: To compare indications and biopsy outcomes for stereotactic biopsy for full field digital mammography (FFDM/DBT) to those of FFSM/DBT. Methods: Retrospective chart review of stereotactic biopsies performed from July 2014 to September 2018. Reports were reviewed and indication for biopsy, lesion size, and final pathology were recorded. Comparison between the two groups following transition to FFSM/DBT in 2016 was performed. Results: 66 of 361 stereotactic biopsies performed in the FFDM/DBT group were malignant (PPV 18.3%), compared to 60 of the 391 biopsies performed in the FFSM/DBT group (PPV 15.4%) with no significant difference in PPV (p = 0.281). There were statistically significant changes in indications for biopsies after transitioning to FFSM/DBT: with a decrease in calcifications referred for biopsy (68.03% vs. 89.75%; p < 0.001), and a statistically significant increase in referral of masses (10.74% vs. 4.43%; p < 0.001), asymmetries (15.60% vs. 5.26%; p < 0.001), and architectural distortion (5.63% vs. 0.55%; p < 0.001). PPV across all indications (21.8% in FFSM/DBT vs. 20.3% in FFDM; p = 0.213), and invasive cancer yield (5.63% vs. 3.32%; p = 0.129) remained comparable following transition to FFSM/DBT without statistically significant differences. Conclusions: Following transition to FFSM/DBT, statistically significant shifts in indications for biopsies were observed with a decrease in referral of calcifications and an increase for masses, asymmetries and architectural distortions. PPV for stereotactic biopsy was not significantly different and cancer yield across all indications remained similar, with an increase in invasive cancer diagnosis. Full article

Application of machine learning techniques in breast cancer detection has significantly advanced due to the availability of annotated mammography datasets. This paper provides a review of mammography studies using key datasets such as CBIS-DDSM, VinDr-Mammo, and CSAW-CC, which play a critical role in training classification and detection models. The analysis of the studies produces a set of data augmentation techniques in mammography, and their impact and performance improvements in detecting abnormalities in breast tissue are studied. The study discusses the challenges of dataset imbalances and presents methods to address this issue, like synthetic data generation and GAN augmentation as potential solutions. The work underscores the importance of dataset design dedicated for experiments, detailed annotations, and the usage of machine learning models and architectures in improving breast cancer screening models, with a focus on BI-RADS classification. Future directions include refining augmentation methods, addressing class imbalance, and enhancing model interpretability through tools like Grad-CAM. Full article

Background: Breast cancer screening through mammography interpretation is crucial for early detection and improved patient outcomes. However, the manual classification of mammograms using the BIRADS (Breast Imaging-Reporting and Data System) remains challenging due to subtle imaging features, inter-reader variability, and increasing radiologist workload. Traditional computer-aided detection systems often struggle with complex feature extraction and contextual understanding of mammographic abnormalities. To address these limitations, this study proposes MammoViT, a novel hybrid deep learning framework that leverages both ResNet50’s hierarchical feature extraction capabilities and Vision Transformer’s ability to capture long-range dependencies in images. Methods: We implemented a multi-stage approach utilizing a pre-trained ResNet50 model for initial feature extraction from mammogram images. To address the significant class imbalance in our four-class BIRADS dataset, we applied SMOTE (Synthetic Minority Over-sampling Technique) to generate synthetic samples for minority classes. The extracted feature arrays were transformed into non-overlapping patches with positional encodings for Vision Transformer processing. The Vision Transformer employs multi-head self-attention mechanisms to capture both local and global relationships between image patches, with each attention head learning different aspects of spatial dependencies. The model was optimized using Keras Tuner and trained using 5-fold cross-validation with early stopping to prevent overfitting. Results: MammoViT achieved 97.4% accuracy in classifying mammogram images across different BIRADS categories. The model’s effectiveness was validated through comprehensive evaluation metrics, including a classification report, confusion matrix, probability distribution, and comparison with existing studies. Conclusions: MammoViT effectively combines ResNet50 and Vision Transformer architectures while addressing the challenge of imbalanced medical imaging datasets. The high accuracy and robust performance demonstrate its potential as a reliable tool for supporting clinical decision-making in breast cancer screening. Full article

Background. In alloplastic breast reconstruction, the choice of implant positioning and the selection of periprosthetic devices is a critical and challenging decision. Surgeons must navigate between various biologic and synthetic meshes, including acellular dermal matrices (ADM). This study aimed to propose a simple selection tool for periprosthetic devices in prepectoral breast reconstruction. Methods. Patients scheduled for mastectomy followed by implant-based breast reconstruction between September 2019 and December 2023 were included. Preoperative risk assessments were performed using the Pre-Bra Score, and only those deemed suitable for prepectoral implant placement were selected. Mastectomy flap thickness was used as an independent criterion, and only cases with flap thicknesses less than 1 cm were included. Results. A total of 70 cases with an average flap thickness of 0.7 cm (range, 0.4–0.9 cm), as measured by preoperative contrast-enhanced spectral mammography (CESM), underwent prepectoral reconstruction with ADM covering the implant. Of these, 25 patients (35%) received direct-to-implant reconstruction, while 45 (65%) underwent two-stage reconstruction with a temporary tissue expander. Postoperative complications were recorded during a minimum follow-up period of 6 months. Over an average follow-up duration of 17.5 months (range 6–36 months), no major complications were observed. Minor complications occurred in seven patients: infection (1.28%), seroma (3.85%), and superficial skin necrosis (1.28%). Additionally, 21 patients (30%) experienced rippling, and secondary lipofilling was scheduled. Conclusions. The incidence of rippling was reduced by 40% through ADM in this patient subgroup, reducing the need for secondary aesthetic refinements. Full article

Background/Objectives: Digital mammography (DM) has long been the standard for breast cancer screening, while digital breast tomosynthesis (DBT) offers an advanced 3D imaging modality capable of generating 2D Synthetic Mammography (SM) images. Despite SM’s potential to reduce radiation exposure, many clinics favor DM, with DBT and SM, due to its perceived diagnostic reliability. This study investigates whether radiologists can replace DM with SM in breast cancer screening and diagnosis or if both modalities are necessary. Methods: We retrospectively analyzed DM and SM images from 375 women aged 40–65 who underwent DM with DBT at King Khaled University Hospital from 2020–2022. Three radiologists evaluated the images using ACR BI-RADS, assessing diagnostic accuracy via the area under the receiver operating characteristic (ROC) curve (AUC). The agreement in cancer conspicuity, breast density, size, and calcifications were measured using weighted kappa (κ). Results: Among 57 confirmed cancer cases and 290 cancer-free cases, DM demonstrated higher sensitivity (82.5% vs. 78.9%) and diagnostic accuracy (AUC 0.800 vs. 0.783, p < 0.05) compared to SM. However, SM detected more suspicious calcifications in cancer cases (75.6% vs. 51.2%, p < 0.05). Agreement was fair for conspicuity (κ = 0.288) and calcifications (κ = 0.409), moderate for density (κ = 0.591), and poor for size (κ = 0.254). Conclusions: while SM demonstrates enhanced effectiveness in detecting microcalcifications, DM still proves superior in overall diagnostic accuracy and image clarity. Therefore, although SM offers certain advantages, it remains slightly inferior to DM and cannot yet replace DM in breast cancer screening. Full article

Background: Contrast-enhanced mammography is one of the new emerging imaging techniques used for detecting breast tissue lesions. Optimization of imaging protocols and reconstruction techniques for this modality, however, requires the involvement of physical phantoms. Their development is related to the use of radiocontrast agents. This study assesses the X-ray properties of a novel contrast material in clinical settings. This material is intended for experimental use with physical phantoms, offering an alternative to commonly available radiocontrast agents. Materials and Methods: The water-soluble sodium salt of the newly synthesized diiodine-substituted natural eudesmic acid, Sodium 2,6-DiIodo-3,4,5-TriMethoxyBenzoate [NaDITMB], has been investigated with respect to one of the most commonly applied radiocontrast medium in medical practice—Omnipaque^®. For this purpose, simulation and experimental studies were carried out with a computational phantom and a physical counterpart, respectively. Synthetic and experimental X-ray images were subsequently produced under varying beam kilovoltage peaks (kVps), and the proposed contrast material was evaluated. Results and Discussion: Simulation results revealed equivalent absorptions between the two simulated radiocontrast agents. Experimental findings supported these simulations, showing a maximum deviation of 3.7% between the image gray values of contrast materials for NaDITMB and Omnipaque solutions for a 46 kVp X-ray beam. Higher kVp X-ray beams show even smaller deviations in the mean grey values of the imaged contrast agents, with the NaDITMB solution demonstrating less than a 2% deviation compared to Omnipaque. Conclusion: The proposed contrast agent is a suitable candidate for use in experimental work related to contrast-enhanced imaging by utilizing phantoms. It boasts the advantages of easy synthesis and is recognized for its safety, ensuring a secure environment for both the experimenter and the environment. Full article

Generative models are used as an alternative data augmentation technique to alleviate the data scarcity problem faced in the medical imaging field. Diffusion models have gathered special attention due to their innovative generation approach, the high quality of the generated images, and their relatively less complex training process compared with Generative Adversarial Networks. Still, the implementation of such models in the medical domain remains at an early stage. In this work, we propose exploring the use of diffusion models for the generation of high-quality, full-field digital mammograms using state-of-the-art conditional diffusion pipelines. Additionally, we propose using stable diffusion models for the inpainting of synthetic mass-like lesions on healthy mammograms. We introduce MAM-E, a pipeline of generative models for high-quality mammography synthesis controlled by a text prompt and capable of generating synthetic mass-like lesions on specific regions of the breast. Finally, we provide quantitative and qualitative assessment of the generated images and easy-to-use graphical user interfaces for mammography synthesis. Full article

Objective: to determine the positive predictive value (PPV) of tomosynthesis (DBT)-detected architectural distortions (ADs) and evaluate correlations between AD’s imaging characteristics and histopathologic outcomes. Methods: biopsies performed between 2019 and 2021 on ADs were included. Images were interpreted by dedicated breast imaging radiologists. Pathologic results after DBT-vacuum assisted biopsy (DBT-VAB) and core needle biopsy were compared with AD detected by DBT, synthetic2D (synt2D) and ultrasound (US). Results: US was performed to assess a correlation for ADs in all 123 cases and a US correlation was identified in 12/123 (9.7%) cases, which underwent US-guided core needle biopsy (CNB). The remaining 111/123 (90.2%) ADs were biopsied under DBT guidance. Among the 123 ADs included, 33/123 (26.8%) yielded malignant results. The overall PPV for malignancy was 30.1% (37/123). The imaging-specific PPV for malignancy was 19.2% (5/26) for DBT-only ADs, 28.2% (24/85) for ADs visible on DBT and synth2D mammography and 66.7% (8/12) for ADs with a US correlation with a statistically significant difference among the three groups (p = 0.01). Conclusions: DBT-only ADs demonstrated a lower PPV of malignancy when compared with syntD mammography, and DBT detected ADs but not low enough to avoid biopsy. As the presence of a US correlate was found to be related with malignancy, it should increase the radiologist’s level of suspicion, even when CNB returned a B3 result. Full article

Breast cancer is a serious medical condition that requires early detection for successful treatment. Mammography is a commonly used imaging technique for breast cancer screening, but its analysis can be time-consuming and subjective. This study explores the use of deep learning-based methods for mammogram analysis, with a focus on improving the performance of the analysis process. The study is focused on applying different computer vision models, with both CNN and ViT architectures, on a publicly available dataset. The innovative approach is represented by the data augmentation technique based on synthetic images, which are generated to improve the performance of the models. The results of the study demonstrate the importance of data pre-processing and augmentation techniques for achieving high classification performance. Additionally, the study utilizes explainable AI techniques, such as class activation maps and centered bounding boxes, to better understand the models’ decision-making process. Full article

The production of thermoluminescence (TL) dosimeters fabricated from B₂O₃-CaF₂-Al₂O₃-SiO₂ doped with Cu and Pr for use in diagnostic radiology is the main goal of this research. The TL samples were synthesized via the melt-quench technique processed by melting the mixture at 1200 °C for 1 h, and, after cooling, the sample thus created was divided into two samples and retreated by heating for 2 h (referred to as TLV30) and for 15 h (referred to as TLV17). SEM and EDS analyses were performed on the TL samples to confirm the preparation process and to investigate the effects of irradiation dosimetry on the TL samples. Furthermore, the TL samples were irradiated with γ-rays using a 450 Ci ¹³⁷Cs irradiator and variable X-ray beams (5–70 mGy). Two important diagnostic radiology applications were considered: CT (6–24 mGy) and mammography (2.72–10.8 mGy). Important dosimetric properties, such as the glow curves, reproducibility, dose–response linearity, energy dependence, minimum dose detectability and fading, were investigated for the synthetized samples (TLV17 and TLV30), the results of which were compared with the Harshaw TLD-100. The TLV17 dosimeter showed higher sensitivity than TLV30 in all applied irradiation procedures. The dose–response linearity coefficients of determination R² for TLV17 were higher than TLD-100 and TLV30 in some applications and were almost equal in others. The reproducibility results of TLV17, TLV30 and TLD-100 were less than 5%, which is acceptable. On the other hand, the results of the fading investigations showed that, in general, TLV17 showed less fading than TLV30. Both samples showed a significant decrease in this regard after the first day, and then the signal variation became essentially stable though with a slight decrease until the eighth day. Therefore, it is recommended to read the TL dosimeters after 24 h, as with TLD-100. The SEM images confirmed the existence of crystallization, whilst the EDS spectra confirmed the presence of the elements used for preparation. Furthermore, we noticed that TLV17 had grown dense crystals that were larger in size compared to those of TLV30, which explains the higher sensitivity in TLV17. Overall, despite the fading, TLV17 showed greater radiation sensitivity and dose–response linearity compared with TLD-100. The synthetized TL samples showed their suitability for use as dosimeters in diagnostic radiology radiation dosimetry. Full article

Tools based on deep learning models have been created in recent years to aid radiologists in the diagnosis of breast cancer from mammograms. However, the datasets used to train these models may suffer from class imbalance, i.e., there are often fewer malignant samples than benign or healthy cases, which can bias the model towards the healthy class. In this study, we systematically evaluate several popular techniques to deal with this class imbalance, namely, class weighting, over-sampling, and under-sampling, as well as a synthetic lesion generation approach to increase the number of malignant samples. These techniques are applied when training on three diverse Full-Field Digital Mammography datasets, and tested on in-distribution and out-of-distribution samples. The experiments show that a greater imbalance is associated with a greater bias towards the majority class, which can be counteracted by any of the standard class imbalance techniques. On the other hand, these methods provide no benefit to model performance with respect to Area Under the Curve of the Recall Operating Characteristic (AUC-ROC), and indeed under-sampling leads to a reduction of 0.066 in AUC in the case of a 19:1 benign to malignant imbalance. Our synthetic lesion methodology leads to better performance in most cases, with increases of up to 0.07 in AUC on out-of-distribution test sets over the next best experiment. Full article

Purpose: To evaluate radiomics features in order to: differentiate malignant versus benign lesions; predict low versus moderate and high grading; identify positive or negative hormone receptors; and discriminate positive versus negative human epidermal growth factor receptor 2 related to breast cancer. Methods: A total of 182 patients with known breast lesions and that underwent Contrast-Enhanced Mammography were enrolled in this retrospective study. The reference standard was pathology (118 malignant lesions and 64 benign lesions). A total of 837 textural metrics were extracted by manually segmenting the region of interest from both craniocaudally (CC) and mediolateral oblique (MLO) views. Non-parametric Wilcoxon–Mann–Whitney test, receiver operating characteristic, logistic regression and tree-based machine learning algorithms were used. The Adaptive Synthetic Sampling balancing approach was used and a feature selection process was implemented. Results: In univariate analysis, the classification of malignant versus benign lesions achieved the best performance when considering the original_gldm_DependenceNonUniformity feature extracted on CC view (accuracy of 88.98%). An accuracy of 83.65% was reached in the classification of grading, whereas a slightly lower value of accuracy (81.65%) was found in the classification of the presence of the hormone receptor; the features extracted were the original_glrlm_RunEntropy and the original_gldm_DependenceNonUniformity, respectively. The results of multivariate analysis achieved the best performances when using two or more features as predictors for classifying malignant versus benign lesions from CC view images (max test accuracy of 95.83% with a non-regularized logistic regression). Considering the features extracted from MLO view images, the best test accuracy (91.67%) was obtained when predicting the grading using a classification-tree algorithm. Combinations of only two features, extracted from both CC and MLO views, always showed test accuracy values greater than or equal to 90.00%, with the only exception being the prediction of the human epidermal growth factor receptor 2, where the best performance (test accuracy of 89.29%) was obtained with the random forest algorithm. Conclusions: The results confirm that the identification of malignant breast lesions and the differentiation of histological outcomes and some molecular subtypes of tumors (mainly positive hormone receptor tumors) can be obtained with satisfactory accuracy through both univariate and multivariate analysis of textural features extracted from Contrast-Enhanced Mammography images. Full article

Ductal carcinoma in situ (DCIS) of male breast is a rare lesion, often associated with invasive carcinoma. When the in situ component is present in pure form, histological grade is usually low or intermediate. Imaging is difficult as gynaecomastia is often present and can mask underlying findings. We report a rare case of pure high-grade DCIS in a young male patient, with associated intraductal papilloma and atypical ductal hyperplasia. Digital breast tomosynthesis (DBT) showed an area of architectural distortion at the union of outer quadrants of the left breast without gynaecomastia. Triple assessment suggested performing a nipple-sparing mastectomy, which revealed the presence of a focal area of high-grade DCIS of 2 mm. DCIS, even of high grade, is difficult to detect with mammography and even more rare, especially when associated with other proliferative lesions. DBT with 2D synthetic reconstruction is useful as the imaging step of a triple assessment and it should be performed in both symptomatic and asymptomatic high-risk men to differentiate between malignant and benign lesions. We propose a diagnostic model to early detect breast cancer in men, optimizing resources according to efficiency, effectiveness and economy, and look forward to radiomics as a powerful tool to help radiologists. Full article

The aim of the study was to estimate the diagnostic accuracy of textural features extracted by dual-energy contrast-enhanced mammography (CEM) images, by carrying out univariate and multivariate statistical analyses including artificial intelligence approaches. In total, 80 patients with known breast lesion were enrolled in this prospective study according to regulations issued by the local Institutional Review Board. All patients underwent dual-energy CEM examination in both craniocaudally (CC) and double acquisition of mediolateral oblique (MLO) projections (early and late). The reference standard was pathology from a surgical specimen for malignant lesions and pathology from a surgical specimen or fine needle aspiration cytology, core or Tru-Cut needle biopsy, and vacuum assisted breast biopsy for benign lesions. In total, 104 samples of 80 patients were analyzed. Furthermore, 48 textural parameters were extracted by manually segmenting regions of interest. Univariate and multivariate approaches were performed: non-parametric Wilcoxon–Mann–Whitney test; receiver operating characteristic (ROC), linear classifier (LDA), decision tree (DT), k-nearest neighbors (KNN), artificial neural network (NNET), and support vector machine (SVM) were utilized. A balancing approach and feature selection methods were used. The univariate analysis showed low accuracy and area under the curve (AUC) for all considered features. Instead, in the multivariate textural analysis, the best performance considering the CC view (accuracy (ACC) = 0.75; AUC = 0.82) was reached with a DT trained with leave-one-out cross-variation (LOOCV) and balanced data (with adaptive synthetic (ADASYN) function) and a subset of three robust textural features (MAD, VARIANCE, and LRLGE). The best performance (ACC = 0.77; AUC = 0.83) considering the early-MLO view was reached with a NNET trained with LOOCV and balanced data (with ADASYN function) and a subset of ten robust features (MEAN, MAD, RANGE, IQR, VARIANCE, CORRELATION, RLV, COARSNESS, BUSYNESS, and STRENGTH). The best performance (ACC = 0.73; AUC = 0.82) considering the late-MLO view was reached with a NNET trained with LOOCV and balanced data (with ADASYN function) and a subset of eleven robust features (MODE, MEDIAN, RANGE, RLN, LRLGE, RLV, LZLGE, GLV_GLSZM, ZSV, COARSNESS, and BUSYNESS). Multivariate analyses using pattern recognition approaches, considering 144 textural features extracted from all three mammographic projections (CC, early MLO, and late MLO), optimized by adaptive synthetic sampling and feature selection operations obtained the best results (ACC = 0.87; AUC = 0.90) and showed the best performance in the discrimination of benign and malignant lesions. Full article