Image-based reconstruction of anthropomorphic breast phantoms for synthetic mammogram generation.

An Institutional Review Board-approved protocol was used to quantify breast tissue inclusion in 52 women, under conditions simulating both craniocaudal (CC) and mediolateral oblique (MLO) views in mammography, dedicated breast CT in the upright subject position, and dedicated breast CT in the prone subject position. Using skin as a surrogate for the underlying breast tissue, the posterior aspect of the breast that is aligned with the chest-wall edge of the breast support in a screen-film mammography system was marked with the study participants positioned for CC and MLO views. The union of skin marks with the study participants positioned for CC and MLO views was considered to represent chest-wall tissue available for imaging with mammography and served as the reference standard. For breast CT, a prone stereotactic breast biopsy unit and a custom-fabricated barrier were used to simulate conditions during prone and upright breast CT, respectively. For the same breast marked on the mammography system, skin marks were made along the breast periphery that was just anterior to the apertures of the prone biopsy unit and the upright barrier. The differences in skin marks between subject positioning simulating breast CT (prone, upright) and mammography were quantified at six anatomic locations. For each location, at least one study participant had a skin mark from breast CT (prone, upright) posterior to mammography. However for all study participants, there was at least one anatomic location where the skin mark from mammography was posterior to that from breast CT (prone, upright) positioning. The maximum amount by which the skin mark from mammography was posterior to breast CT (prone and upright) over all six locations was quantified for each study participant and pair-wise comparison did not exhibit statistically significant difference between prone and upright breast CT (paired t- test, p = 0.4). Quantitatively, for 95% of the study participants the skin mark from mammography was posterior to breast CT (prone or upright) by at the most 9 mm over all six locations. Based on the study observations, geometric design considerations targeting chest-wall coverage with breast CT equivalent to mammography, wherein part of the x-ray beam images through the swale during breast CT are provided. Assuming subjects can extend their chest in to a swale, the optimal swale-depth required to achieve equivalent coverage with breast CT images as mammograms for 95% of the subjects varies in the range of ∼30–50 mm for clinical prototypes and was dependent on the system geometry.

To develop models of compressed breasts undergoing mammography based on objective analysis, that are capable of accurately representing breast shapes in acquired clinical images and generating new, clinically realistic shapes. An automated edge detection algorithm was used to catalogue the breast shapes of clinically acquired cranio-caudal (CC) and medio-lateral oblique (MLO) view mammograms from a large database of digital mammography images. Principal component analysis (PCA) was performed on these shapes to reduce the information contained within the shapes to a small number of linearly independent variables. The breast shape models, one of each view, were developed from the identified principal components, and their ability to reproduce the shape of breasts from an independent set of mammograms not used in the PCA, was assessed both visually and quantitatively by calculating the average distance error (ADE). The PCA breast shape models of the CC and MLO mammographic views based on six principal components, in which 99.2% and 98.0%, respectively, of the total variance of the dataset is contained, were found to be able to reproduce breast shapes with strong fidelity (CC view mean ADE = 0.90 mm, MLO view mean ADE = 1.43 mm) and to generate new clinically realistic shapes. The PCA models based on fewer principal components were also successful, but to a lesser degree, as the two-component model exhibited a mean ADE = 2.99 mm for the CC view, and a mean ADE = 4.63 mm for the MLO view. The four-component models exhibited a mean ADE = 1.47 mm for the CC view and a mean ADE = 2.14 mm for the MLO view. Paired t-tests of the ADE values of each image between models showed that these differences were statistically significant (max p-value = 0.0247). Visual examination of modeled breast shapes confirmed these results. Histograms of the PCA parameters associated with the six principal components were fitted with Gaussian distributions. The six-component model was also used to generate CC and MLO view mammogram breast shapes, using the mean PCA parameter values of these distributions and randomly generated values based on the fitted Gaussian distributions, which resemble clinically encountered breasts. A spreadsheet with the data necessary to apply this model is provided as the supplementary material. Our PCA models of breast shapes in both mammographic views successfully reproduce analyzed breast shapes and generate new clinically relevant shapes. This work can aid in research applications which incorporate breast shape modeling, such as x-ray scatter correction, dosimetry, and image registration.

To retrospectively compare the sensitivity of a computer-aided detection (CAD) system for depicting breast cancer in three digital mammographic views. This study was conducted with institutional review board approval; informed consent was waived. A commercially available CAD system was applied to the craniocaudal, mediolateral oblique, and mediolateral digital mammographic views of 83 women (mean age, 48 years; range, 30-66 years) with 83 histologically proved breast cancers. Findings were 59 masses and 41 microcalcifications (17 lesions showed both findings; 42 lesions, mass only; and 24 lesions, microcalcification only). The paired t test was used to analyze sensitivity of the CAD system for the detection of cancer in these three mammographic views and in combinations of the views. The sensitivities of the CAD system were 92% (76 of 83) in the craniocaudal view, 83% (69 of 83) in the mediolateral oblique view, and 86% (71 of 83) in the mediolateral view; the differences were not significant (P = .07-.62). Sensitivity increased to 96% (80 of 83) in the craniocaudal plus mediolateral oblique views and to 99% (82 of 83) in the craniocaudal plus mediolateral oblique plus mediolateral views. For masses, the sensitivity of the CAD system was 76% (45 of 59) in the craniocaudal view and 75% (44 of 59) in the mediolateral oblique view and increased to 93% (55 of 59) when mediolateral oblique and craniocaudal views were combined (P < .001). For microcalcifications, sensitivity was 98% (40 of 41) in the craniocaudal view and 95% (39 of 41) in the mediolateral oblique view, and this increased to 100% (41 of 41) when the mediolateral oblique and craniocaudal views were combined (P = .31). The sensitivities of the CAD system were not significantly different among these three digital mammographic views. Sensitivity for depicting masses was significantly increased (P < .001) when the craniocaudal view was added to the mediolateral oblique view.

With the advent of three-dimensional (3D) breast imaging modalities such as digital breast tomosynthesis (DBT) and dedicated breast CT (bCT), research into new anthropomorphic breastphantoms has accelerated. These breast phantoms are important for the optimization of new breast imaging systems, assessing new regulatory submissions to prove safety and effectiveness, and for developing new approaches to acceptance and constancy testing of 3D breast imaging systems. This paper provides a review of current research investigating both digital and physical breast phantom development for use in x-ray based imaging. Two approaches for designing anthropomorphic, digital breast phantoms are discussed, procedural model-based phantom generation, where breast features are expressed using mathematical models, and patient-based generation, where breast structures from tissue specimens or patient-based breast MR or CT volumes are segmented. Following this discussion, a review of physical anthropomorphic phantoms is given, with emphasis on the advantages and disadvantages present with each approach. This paper provides a summary of the state-of-the-art in anthropomorphic breast phantom development for x-ray breast imaging. The primary advantage of model-based digital phantoms is that an unlimited number of phantoms with varying size, shape, and density can be generated. Current research on model-based breast phantoms is producing more and more realistic breast models; however, they probably are not yet able to pass the so-called "fool the radiologist" visualization test. Empirical patient-based breast phantoms are typically based on clinical breast CT data and look more realistic. However, clinical breast CT images have limited spatial resolution and thus do not always portray the finer details in the breast. A number of innovative solutions have been proposed for fabricating physical anthropomorphic breast phantoms based on digital phantom models; however, a number of challenges remain, including realistic modeling of x-ray attenuation properties and accurately representing high-frequency structures within breast.

Improving the modeling of the breast shapes during mechanical compression in both cranio-caudal (CC) and medio-lateral oblique (MLO) views can enhance the development of image processing and dosimetric estimates in digital mammography and digital breast tomosynthesis (DBT). In previous work, a CC model was created using a pair of optical structured light scanning systems, but acquiring similar data for an MLO view model during clinical practice proved impractical with these devices. The present work instead uses two smartphone infrared cameras with 3D-printed holders to obtain surface scans for the MLO view during a DBT acquisition. The study compared the average distance between the MLO breast shape information recorded by the smartphone-based scans to the corresponding DBT exam-based surface scan for 20 patient breasts. Results showed that there was close overlap between the smartphone-based scanned surfaces of the breast and the corresponding DBT images. The agreement between the breast shape represented by these surfaces was dependent on the smartphone-based scanner precision, the segmentation procedure used to obtain the DBT surface, and the manual alignment of the smartphone-based left and right-side view of the breast. The agreement was however of sufficiently good quality for the data to be used for the development of an MLO breast shape model.

To develop a set of accurate 2D models of compressed breasts undergoing mammography or breast tomosynthesis, based on objective analysis, to accurately characterize mammograms with few linearly independent parameters, and to generate novel clinically realistic paired cranio-caudal (CC) and medio-lateral oblique (MLO) views of the breast. We seek to improve on an existing model of compressed breasts by overcoming detector size bias, removing the nipple and non-mammary tissue, pairing the CC and MLO views from a single breast, and incorporating the pectoralis major muscle contour into the model. The outer breast shapes in 931 paired CC and MLO mammograms were automatically detected with an in-house developed segmentation algorithm. From these shapes three generic models (CC-only, MLO-only, and joint CC/MLO) with linearly independent components were constructed via principal component analysis (PCA). The ability of the models to represent mammograms not used for PCA was tested via leave-one-out cross-validation, by measuring the average distance error (ADE). The individual models based on six components were found to depict breast shapes with accuracy (mean ADE-CC=0.81mm, ADE-MLO=1.64mm, ADE-Pectoralis=1.61mm), outperforming the joint CC/MLO model (P≤0.001). The joint model based on 12 principal components contains 99.5% of the total variance of the data, and can be used to generate new clinically realistic paired CC and MLO breast shapes. This is achieved by generating random sets of 12 principal components, following the Gaussian distributions of the histograms of each component, which were obtained from the component values determined from the images in the mammography database used. Our joint CC/MLO model can successfully generate paired CC and MLO view shapes of the same simulated breast, while the individual models can be used to represent with high accuracy clinical acquired mammograms with a small set of parameters. This is the first step toward objective 3D compressed breast models, useful for dosimetry and scatter correction research, among other applications.

To investigate the glandular dose magnitudes and characteristics resulting from image acquisition using a dedicated breast computed tomography (BCT) clinical prototype imaging system. The x-ray spectrum and output characteristics of a BCT clinical prototype (Koning Corporation, West Henrietta, NY) were determined using empirical measurements, breast phantoms, and an established spectrum model. The geometry of the BCT system was replicated in a Monte Carlo-based computer simulation using the GEANT4 toolkit and was validated by comparing the simulated results for exposure distribution in a standard 16 cm CT head phantom with those empirically determined using a 10 cm CT pencil ionization chamber and dosimeter. The computer simulation was further validated by replicating the results of a previous BCT dosimetry study. Upon validation, the computer simulation was modified to include breasts of varying sizes and homogeneous compositions spanning those encountered clinically, and the normalized mean glandular dose resulting from BCT was determined. Using the system's measured exposure output determined automatically for breasts of different size and density, the mean glandular dose for these breasts was computed and compared to the glandular dose resulting from mammography. Finally, additional Monte Carlo simulations were performed to study how the glandular dose values vary within the breast tissue during acquisition with both this BCT prototype and a typical craniocaudal (CC) mammographic acquisition. This BCT prototype uses an x-ray spectrum with a first half-value layer of 1.39 mm Al and a mean x-ray energy of 30.3 keV. The normalized mean glandular dose for breasts of varying size and composition during BCT acquisition with this system ranges from 0.278 to 0.582 mGy/mGy air kerma with the reference air kerma measured in air at the center of rotation. Using the measured exposure outputs for the tube currents automatically selected by the system for the breasts of different sizes and densities, the mean glandular dose for a BCT acquisition with this prototype system varies from 5.6 to 17.5 mGy, with the value for a breast of mean size and composition being 17.06 mGy. The glandular dose throughout the breast tissue of this mean breast varies by up to +/- 50% of the mean value. During a typical CC view mammographic acquisition of an equivalent mean breast, which typically results in a mean glandular dose of 2.0-2.5 mGy, the glandular dose throughout the breast tissue varies from approximately 15% to approximately 400% of the mean value. Acquisition of a BCT image with the automated tube output settings for a mean breast with the Koning Corp. clinical prototype results in mean glandular dose values approximately equivalent to three to five two-view mammographic examinations for a similar breast. For all breast sizes and compositions studied, this glandular dose ratio between acquisition with this BCT prototype and two-view mammography ranges from 1.4 to 7.2. In mammography, portions of the mean-sized breast receive a considerably higher dose than the mean value for the whole breast. However, only a small portion of a breast undergoing mammography would receive a glandular dose similar to that from BCT.

According to World Cancer Research Fund International, breast cancer is the 2nd common cancer type worldwide and 1st type among women. To diagnose breast cancer, mammography (MG) and digital breast tomosynthesis (DBT) are commonly preferred imaging methods. Radiotherapy is one of the treatment methods of breast cancer and in case of radiotherapy is preferred, breast computed tomography (B-CT) is taken for breast contouring which is essential for radiotherapy. Mammography (MG), digital breast tomosynthesis (DBT) and breast computed tomography (B-CT) are medical imaging techniques using ionizing radiation. During the MG, DBT and B-CT scan, the patient is exposure to ionizing radiation and determining the correct radiation dose is crucial for both correct imaging of breast and avoiding side effects of radiation to the patient. Determination of the correct dose for diagnosis is called dose optimization. For dose optimization, medical imaging devices are calibrated via phantoms. Phantom is a device that mimicking real patient tissues on radiosensitivity. Formerly, phantoms were made of water, as the human body mainly consists of water. Then, epoxy or PMMA based phantoms are used due to the water is not perfect material for mimicking human tissue. Epoxy or PMMA based phantoms are better at mimicking soft tissue and bone tissue but not enough for breast due to the breast is formed by heterogeneous distribution of glandular and adipose tissues. Heterogeneity of glandular and adipose tissues, glandular and adipose tissue percentage (%G/A), vary with race and age of women. Due to the radiation dose absorbed by glandular tissue is dependent on the glandular tissue percentage, it is essential to develop breast phantom specific to glandular and adipose tissue percentage (%G/A) for better radiation dose optimization. The aim of this study is to develop anthropomorphic breast phantoms specific to Turkish women at different age intervals. In previous study, %G/A ratios for Turkish women at different age intervals are determined and epoxy-based phantoms were produced for each age interval. Those phantoms assumed the glandular and adipose tissues are distributed homogeneously contrary to real breast. The aimed more realistic, anthropomorphic phantoms will consist of three different materials mimicking three different tissues. A FDM type 3D printer with dual nozzle is going to be used for producing glandular and adipose tissues; an SLA type 3D printer is going to be used for skin. By this way, development of phantom mimicking breast better than epoxy-based one is the goal of the study. At first step, the materials and filling ratios mimicking best each three tissues will be determined. By using PVA, TPU, ABS, PLA and PET type filaments, cubic samples will be produced with filling ratios between %50 and %100. Physical density, HU value, effective atomic number, HVL values for each sample will be determined and the filament types and filling ratios closest to real tissues will be chosen to produce phantom. Anonymous B-CT images of breast with specific %G/A ratios will be base for design of phantoms. The phantoms will have cavities for placing point detectors to measure the glandular dose during the CT scan. After all the phantoms will be produced, the phantoms will undergo CT scan and during the scan, glandular dose and tissue entrance dose will be measured via appropriate detectors. Via finding the relation between skin entrance dose and glandular dose, it is aimed to determine more accurate factors for Dance’s equation for calculation of glandular dose. This study is being supported by TÜBİTAK with project code 224S843 in the scope of 1002-A Fast Support Program.

The purpose of this study was to compare the thickness of the compressed breast between mediolateral oblique and craniocaudal mammograms and to relate these differences in thickness to image quality and radiation dose. These differences may partially explain why some subtle tumors are better visualized on the craniocaudal view. The study population consisted of 250 paired mediolateral oblique and craniocaudal mammograms obtained on one mammographic unit by seven certified mammography technologists during a 2-month period. Only women with breast implants, prior lumpectomy and radiotherapy, or chest wall deformity were excluded. The digital readout of compressed breast thickness and applied compression force was recorded. Mammographic positioning was assessed using standard criteria. Absorbed radiation dose at different thicknesses was measured with a BR-12 breast phantom. Image quality differences for geometric unsharpness and contrast were calculated for the observed breast thickness differences between mediolateral oblique and craniocaudal mammograms. The mean thickness of the compressed breast on the craniocaudal view was less than the mean thickness on the mediolateral oblique view (4.4 versus 4.8 cm, p < .0001) despite the greater force used to compress the breast for mediolateral oblique than for craniocaudal views (93 versus 86 newtons, p < .0001). The breast thickness on the mediolateral oblique view exceeded that on the craniocaudal view in 98 (84%) of 117 pairs that differed in thickness by 5 mm or more and 46 (94%) of 49 pairs that differed by 10 mm or more (p < .0001). Geometric unsharpness increased by 8% and 19% when a 4.4-cm-thick breast was compared to a 4.8- and 5.4-cm-thick breast, respectively. A 5% and 12% loss of contrast was noted when a 4.4-cm-thick breast was compared to a 4.8- and 5.4-cm-thick breast. Mean glandular radiation dose at 4.4, 4.8, and 5.4 cm was 1.40, 1.70, and 2.33 mGy, respectively. The compressed breast is 8% thicker on mediolateral oblique than on craniocaudal mammograms, a small but statistically significant difference. This difference results in a small loss of spatial and contrast resolution on the mediolateral oblique views and an increase in radiation dose. These image quality differences may partially explain why some subtle carcinomas are better visualized on the craniocaudal view.

The aim of this study is to investigate any difference in the shape and location of infiltrating lobular carcinoma (ILC) with respect to the parenchymal density between the cranio-caudal (CC) and medio-lateral oblique (MLO) mammographic views. Six film-readers independently re-read 59 ILC mammograms and a matched sample of 59 infiltrating ductal carcinoma (IDC) mammograms from one 3-year screening round to quantify lesion characteristics. There is fair to moderate reader agreement for parenchymal pattern, lesion shape and location (kappa = 0.41-0.60). Both ILC (33/60, 55%) and IDC (22/65, 37%) appear as a spiculate mass more often on the CC view than on the MLO view. 41% (25/60) of the ILC spiculate masses become architectural distortions or asymmetric densities on the MLO view. No more ILC lesions (4/60, 7%) are seen in dense breasts than IDC (5/65, 8%), but ILC is mainly associated with (58/60, 97%), and rarely isolated from (2/60, 3%), the main glandular density. The appearance of ILC is significantly different between the MLO and CC views (paired Wilcoxon test: z = -17.059; significance level <or=0.0005). IDC appearance is not significantly different between these two views (z = -1.244; significance level 0.213). In conclusion, the CC view is optimum for distinctly visualizing ILC as a spiculate mass, as it appears as a more subtle distortion or asymmetry on the MLO view. ILC is not often isolated from the main glandular density and so optimizing visualization of this area of the breast is key to perception.

The objective of this study was to investigate the impact of decreasing breast compression during digital mammography and breast tomosynthesis (DBT) on perceived pain and image quality. In this two-part study, two groups of women with prior mammograms were recruited. In part 1, subjects were positioned for craniocaudal (CC) and mediolateral oblique (MLO) views, and four levels of compression force were applied to evaluate changes in breast thickness, perceived pain, and relative tissue coverage. No imaging was performed. In part 2, two MLO DBT images of one breast of each patient were acquired at standard and reduced compression. Blurring artifacts and tissue coverage were judged by three breast imaging radiologists, and compression force, breast thickness, relative tissue coverage, and perceived pain were recorded. Only the first reduction in force was feasible because further reduction resulted in inadequate breast immobilization. Mean force reductions of 48% and 47% for the CC and MLO views, respectively, resulted in a significantly reduced perceived pain level, whereas the thickness of the compressed breast increased by 0.02 cm (CC view) and 0.09 (MLO view, part 1 of the study) and 0.38 cm (MLO view, part 2 of the study), respectively, with no change in tissue coverage or increase in motion blurring. Mammography and DBT acquisitions may be possible using half of the compression force used currently, with a significant and substantial reduction in perceived pain with no clinically significant change in breast thickness and tissue coverage.

Screening and diagnosis of breast cancer with Digital Breast Tomosynthesis (DBT) and Mammography are increasingly supported by algorithms for automatic post-processing. The pectoral muscle, which dorsally delineates the breast tissue towards the chest wall, is an important anatomical structure for navigation. Along with the nipple and the skin, the pectoral muscle boundary is often used for reporting the location of breast lesions. It is visible in mediolateral oblique (MLO) views where it is well approximated by a straight line. Here, we propose two machine learning-based algorithms to robustly detect the pectoral muscle in MLO views from DBT and mammography. Embedded into the Marginal Space Learning framework, the algorithms involve the evaluation of multiple candidate boundaries in a hierarchical manner. To this end, we propose a novel method for candidate generation using a Hough-based approach. Experiments were performed on a set of 100 DBT volumes and 95 mammograms from different clinical cases. Our novel combined approach achieves competitive accuracy and robustness. In particular, for the DBT data, we achieve significantly lower deviation angle error and mean distance error than the standard approach. The proposed algorithms run within a few seconds.

Objective. This work describes an approach for producing physical anthropomorphic breast phantoms from clinical patient data using three-dimensional (3D) fused-deposition modelling (FDM) printing. Approach. The source of the anthropomorphic model was a clinical Magnetic Resonance Imaging (MRI) patient image set, which was segmented slice by slice into adipose and glandular tissues, skin and tumour formations; thus obtaining a four component computational breast model. The segmented tissues were mapped to specific Hounsfield Units (HU) values, which were derived from clinical breast Computed Tomography (CT) data. The obtained computational model was used as a template for producing a physical anthropomorphic breast phantom using 3D printing. FDM technology with only one polylactic acid filament was used. The physical breast phantom was scanned at Siemens SOMATOM Definition CT. Quantitative and qualitative evaluation were carried out to assess the clinical realism of CT slices of the physical breast phantom. Main results. The comparison between selected slices from the computational breast phantom and CT slices of the physical breast phantom shows similar visual x-ray appearance of the four breast tissue structures: adipose, glandular, tumour and skin. The results from the task-based evaluation, which involved three radiologists, showed a high degree of realistic clinical radiological appearance of the modelled breast components. Measured HU values of the printed structures are within the range of HU values used in the computational phantom. Moreover, measured physical parameters of the breast phantom, such as weight and linear dimensions, agreed very well with the corresponding ones of the computational breast model. Significance. The presented approach, based on a single FDM material, was found suitable for manufacturing of a physical breast phantom, which mimics well the 3D spatial distribution of the different breast tissues and their x-ray absorption properties. As such, it could be successfully exploited in advanced x-ray breast imaging research applications.

The purpose of this study was to develop a fully automated algorithm for mammographic breast density estimation using deep learning. Our algorithm used a fully convolutional network, which is a deep learning framework for image segmentation, to segment both the breast and the dense fibroglandular areas on mammographic images. Using the segmented breast and dense areas, our algorithm computed the breast percent density (PD), which is the faction of dense area in a breast. Our dataset included full-field digital screening mammograms of 604 women, which included 1208 mediolateral oblique (MLO) and 1208 craniocaudal (CC) views. We allocated 455, 58, and 91 of 604 women and their exams into training, testing, and validation datasets, respectively. We established ground truth for the breast and the dense fibroglandular areas via manual segmentation and segmentation using a simple thresholding based on BI-RADS density assessments by radiologists, respectively. Using the mammograms and ground truth, we fine-tuned a pretrained deep learning network to train the network to segment both the breast and the fibroglandular areas. Using the validation dataset, we evaluated the performance of the proposed algorithm against radiologists' BI-RADS density assessments. Specifically, we conducted a correlation analysis between a BI-RADS density assessment of a given breast and its corresponding PD estimate by the proposed algorithm. In addition, we evaluated our algorithm in terms of its ability to classify the BI-RADS density using PD estimates, and its ability to provide consistent PD estimates for the left and the right breast and the MLO and CC views of the same women. To show the effectiveness of our algorithm, we compared the performance of our algorithm against a state of the art algorithm, laboratory for individualized breast radiodensity assessment (LIBRA). The PD estimated by our algorithm correlated well with BI-RADS density ratings by radiologists. Pearson's rho values of our algorithm for CC view, MLO view, and CC-MLO-averaged were 0.81, 0.79, and 0.85, respectively, while those of LIBRA were 0.58, 0.71, and 0.69, respectively. For CC view and CC-MLO averaged cases, the difference in rho values between the proposed algorithm and LIBRA showed statistical significance (P<0.006). In addition, our algorithm provided reliable PD estimates for the left and the right breast (Pearson's ρ>0.87) and for the MLO and CC views (Pearson's ρ=0.76). However, LIBRA showed a lower Pearson's rho value (0.66) for both the left and right breasts for the CC view. In addition, our algorithm showed an excellent ability to separate each sub BI-RADS breast density class (statistically significant, p-values = 0.0001 or less); only one comparison pair, density 1 and density 2 in the CC view, was not statistically significant (P=0.54). However, LIBRA failed to separate breasts in density 1 and 2 for both the CC and MLO views (P>0.64). We have developed a new deep learning based algorithm for breast density segmentation and estimation. We showed that the proposed algorithm correlated well with BI-RADS density assessments by radiologists and outperformed an existing state of the art algorithm.

The authors are developing a series of computational breast phantoms based on breast CT data for imaging research. In this work, the authors develop a program that will allow a user to alter the phantoms to simulate the effect of gravity and compression of the breast (craniocaudal or mediolateral oblique) making the phantoms applicable to multimodality imaging. This application utilizes a template finite-element (FE) breast model that can be applied to their presegmented voxelized breast phantoms. The FE model is automatically fit to the geometry of a given breast phantom, and the material properties of each element are set based on the segmented voxels contained within the element. The loading and boundary conditions, which include gravity, are then assigned based on a user-defined position and compression. The effect of applying these loads to the breast is computed using a multistage contact analysis in FEBio, a freely available and well-validated FE software package specifically designed for biomedical applications. The resulting deformation of the breast is then applied to a boundary mesh representation of the phantom that can be used for simulating medical images. An efficient script performs the above actions seamlessly. The user only needs to specify which voxelized breast phantom to use, the compressed thickness, and orientation of the breast. The authors utilized their FE application to simulate compressed states of the breast indicative of mammography and tomosynthesis. Gravity and compression were simulated on example phantoms and used to generate mammograms in the craniocaudal or mediolateral oblique views. The simulated mammograms show a high degree of realism illustrating the utility of the FE method in simulating imaging data of repositioned and compressed breasts. The breast phantoms and the compression software can become a useful resource to the breast imaging research community. These phantoms can then be used to evaluate and compare imaging modalities that involve different positioning and compression of the breast.