Deep learning-based breast region segmentation in raw and processed digital mammograms: generalization across views and vendors.

Purpose: We developed a segmentation method suited for both raw (for processing) and processed (for presentation) digital mammograms (DMs) that is designed to generalize across images acquired with systems from different vendors and across the two standard screening views. Approach: A U-Net was trained to segment mammograms into background, breast, and pectoral muscle. Eight different datasets, including two previously published public sets and six sets of DMs from as many different vendors, were used, totaling 322 screen film mammograms (SFMs) and 4251 DMs (2821 raw/processed pairs and 1430 only processed) from 1077 different women. Three experiments were done: first training on all SFM and processed images, second also including all raw images in training, and finally testing vendor generalization by leaving one dataset out at a time. Results: The model trained on SFM and processed mammograms achieved a good overall performance regardless of projection and vendor, with a mean (±std. dev.) dice score of 0.96±0.06 for all datasets combined. When raw images were included in training, the mean (±std. dev.) dice score for the raw images was 0.95±0.05 and for the processed images was 0.96±0.04. Testing on a dataset with processed DMs from a vendor that was excluded from training resulted in a difference in mean dice varying between -0.23 to +0.02 from that of the fully trained model. Conclusions: The proposed segmentation method yields accurate overall segmentation results for both raw and processed mammograms independent of view and vendor. The code and model weights are made available.

Radiomics in cone-beam breast CT for the prediction of axillary lymph node metastasis in breast cancer: a multi-center multi-device study.

OBJECTIVES: To develop a radiomics model in contrast-enhanced cone-beam breast CT (CE-CBBCT) for preoperative prediction of axillary lymph node (ALN) status and metastatic burden of breast cancer. METHODS: Two hundred and seventy-four patients who underwent CE-CBBCT examination with two scanners between 2012 and 2021 from two institutions were enrolled. The primary tumor was annotated in each patient image, from which 1781 radiomics features were extracted with PyRadiomics. After feature selection, support vector machine models were developed to predict ALN status and metastatic burden. To avoid overfitting on a specific patient subset, 100 randomly stratified splits were made to assign the patients to either training/fine-tuning or test set. Area under the receiver operating characteristic curve (AUC) of these radiomics models was compared to those obtained when training the models only with clinical features and combined clinical-radiomics descriptors. Ground truth was established by histopathology. RESULTS: One hundred and eighteen patients had ALN metastasis (N + (≥ 1)). Of these, 74 had low burden (N + (1~2)) and 44 high burden (N + (≥ 3)). The remaining 156 patients had none (N0). AUC values across the 100 test repeats in predicting ALN status (N0/N + (≥ 1)) were 0.75 ± 0.05 (0.67~0.93, radiomics model), 0.68 ± 0.07 (0.53~0.85, clinical model), and 0.74 ± 0.05 (0.67~0.88, combined model). For metastatic burden prediction (N + (1~2)/N + (≥ 3)), AUC values were 0.65 ± 0.10 (0.50~0.88, radiomics model), 0.55 ± 0.10 (0.40~0.80, clinical model), and 0.64 ± 0.09 (0.50~0.90, combined model), with all the ranges spanning 0.5. In both cases, the radiomics model was significantly better than the clinical model (both p < 0.01) and comparable with the combined model (p = 0.56 and 0.64). CONCLUSIONS: Radiomics features of primary tumors could have potential in predicting ALN metastasis in CE-CBBCT imaging. CLINICAL RELEVANCE STATEMENT: The findings support potential clinical use of radiomics for predicting axillary lymph node metastasis in breast cancer patients and addressing the limited axilla coverage of cone-beam breast CT. KEY POINTS: • Contrast-enhanced cone-beam breast CT-based radiomics could have potential to predict N0 vs. N + (≥ 1) and, to a limited extent, N + (1~2) vs. N + (≥ 3) from primary tumor, and this could help address the limited axilla coverage, pending future verifications on larger cohorts. • The average AUC of radiomics and combined models was significantly higher than that of clinical models but showed no significant difference between themselves. • Radiomics features descriptive of tumor texture were found informative on axillary lymph node status, highlighting a higher heterogeneity for tumor with positive axillary lymph node.

Joint AAPM Task Group 282/EFOMP Working Group Report: Breast dosimetry for standard and contrast-enhanced mammography and breast tomosynthesis.

Currently, there are multiple breast dosimetry estimation methods for mammography and its variants in use throughout the world. This fact alone introduces uncertainty, since it is often impossible to distinguish which model is internally used by a specific imaging system. In addition, all current models are hampered by various limitations, in terms of overly simplified models of the breast and its composition, as well as simplistic models of the imaging system. Many of these simplifications were necessary, for the most part, due to the need to limit the computational cost of obtaining the required dose conversion coefficients decades ago, when these models were first implemented. With the advancements in computational power, and to address most of the known limitations of previous breast dosimetry methods, a new breast dosimetry method, based on new breast models, has been developed, implemented, and tested. This model, developed jointly by the American Association of Physicists in Medicine and the European Federation for Organizations of Medical Physics, is applicable to standard mammography, digital breast tomosynthesis, and their contrast-enhanced variants. In addition, it includes models of the breast in both the cranio-caudal and the medio-lateral oblique views. Special emphasis was placed on the breast and system models used being based on evidence, either by analysis of large sets of patient data or by performing measurements on imaging devices from a range of manufacturers. Due to the vast number of dose conversion coefficients resulting from the developed model, and the relative complexity of the calculations needed to apply it, a software program has been made available for download or online use, free of charge, to apply the developed breast dosimetry method. The program is available for download or it can be used directly online. A separate User's Guide is provided with the software.

Computed Tomography-Based Radiomics Using Tumor and Vessel Features to Assess Resectability in Cancer of the Pancreatic Head.

The preoperative prediction of resectability pancreatic ductal adenocarcinoma (PDAC) is challenging. This retrospective single-center study examined tumor and vessel radiomics to predict the resectability of PDAC in chemo-naïve patients. The tumor and adjacent arteries and veins were segmented in the portal-venous phase of contrast-enhanced CT scans, and radiomic features were extracted. Features were selected via stability and collinearity testing, and least absolute shrinkage and selection operator application (LASSO). Three models, using tumor features, vessel features, and a combination of both, were trained with the training set (N = 86) to predict resectability. The results were validated with the test set (N = 15) and compared to the multidisciplinary team's (MDT) performance. The vessel-features-only model performed best, with an AUC of 0.92 and sensitivity and specificity of 97% and 73%, respectively. Test set validation showed a sensitivity and specificity of 100% and 88%, respectively. The combined model was as good as the vessel model (AUC = 0.91), whereas the tumor model showed poor performance (AUC = 0.76). The MDT's prediction reached a sensitivity and specificity of 97% and 84% for the training set and 88% and 100% for the test set, respectively. Our clinician-independent vessel-based radiomics model can aid in predicting resectability and shows performance comparable to that of the MDT. With these encouraging results, improved, automated, and generalizable models can be developed that reduce workload and can be applied in non-expert hospitals.

Four-Dimensional Machine Learning Radiomics for the Pretreatment Assessment of Breast Cancer Pathologic Complete Response to Neoadjuvant Chemotherapy in Dynamic Contrast-Enhanced MRI.

BACKGROUND: Breast cancer response to neoadjuvant chemotherapy (NAC) is typically evaluated through the assessment of tumor size reduction after a few cycles of NAC. In case of treatment ineffectiveness, this results in the patient suffering potentially severe secondary effects without achieving any actual benefit. PURPOSE: To identify patients achieving pathologic complete response (pCR) after NAC by spatio-temporal radiomic analysis of dynamic contrast-enhanced (DCE) MRI images acquired before treatment. STUDY TYPE: Single-center, retrospective. POPULATION: A total of 251 DCE-MRI pretreatment images of breast cancer patients. FIELD STRENGTH/SEQUENCE: 1.5 T/3 T, T1-weighted DCE-MRI. ASSESSMENT: Tumor and peritumoral regions were segmented, and 348 radiomic features that quantify texture temporal variation, enhancement kinetics heterogeneity, and morphology were extracted. Based on subsets of features identified through forward selection, machine learning (ML) logistic regression models were trained separately with all images and stratifying on cancer molecular subtype and validated with leave-one-out cross-validation. STATISTICAL TESTS: Feature significance was assessed using the Mann-Whitney U-test. Significance of the area under the receiver operating characteristics (ROC) curve (AUC) of the ML models was assessed using the associated 95% confidence interval (CI). Significance threshold was set to 0.05, adjusted with Bonferroni correction. RESULTS: Nine features related to texture temporal variation and enhancement kinetics heterogeneity were significant in the discrimination of cases achieving pCR vs. non-pCR. The ML models achieved significant AUC of 0.707 (all cancers, n = 251, 59 pCR), 0.824 (luminal A, n = 107, 14 pCR), 0.823 (luminal B, n = 47, 15 pCR), 0.844 (HER2 enriched, n = 25, 11 pCR), 0.803 (triple negative, n = 72, 19 pCR). DATA CONCLUSIONS: Differences in imaging phenotypes were found between complete and noncomplete responders. Furthermore, ML models trained per cancer subtype achieved high performance in classifying pCR vs. non-pCR cases. They may, therefore, have potential to help stratify patients according to the level of response predicted before treatment, pending further validation with larger prospective cohorts. EVIDENCE LEVEL: 4 TECHNICAL EFFICACY: Stage 4.

Deep learning for x-ray scatter correction in dedicated breast CT.

BACKGROUND: Accurate correction of x-ray scatter in dedicated breast computed tomography (bCT) imaging may result in improved visual interpretation and is crucial to achieve quantitative accuracy during image reconstruction and analysis. PURPOSE: To develop a deep learning (DL) model to correct for x-ray scatter in bCT projection images. METHODS: A total of 115 patient scans acquired with a bCT clinical system were segmented into the major breast tissue types (skin, adipose, and fibroglandular tissue). The resulting breast phantoms were divided into training (n = 110) and internal validation cohort (n = 5). Training phantoms were augmented by a factor of four by random translation of the breast in the image field of view. Using a previously validated Monte Carlo (MC) simulation algorithm, 12 primary and scatter bCT projection images with a 30-degree step were generated from each phantom. For each projection, the thickness map and breast location in the field of view were also calculated. A U-Net based DL model was developed to estimate the scatter signal based on the total input simulated image and trained single-projection-wise, with the thickness map and breast location provided as additional inputs. The model was internally validated using MC-simulated projections and tested using an external data set of 10 phantoms derived from images acquired with a different bCT system. For this purpose, the mean relative difference (MRD) and mean absolute error (MAE) were calculated. To test for accuracy in reconstructed images, a full bCT acquisition was mimicked with MC-simulations and then assessed by calculating the MAE and the structural similarity (SSIM). Subsequently, scatter was estimated and subtracted from the bCT scans of three patients to obtain the scatter-corrected image. The scatter-corrected projections were reconstructed and compared with the uncorrected reconstructions by evaluating the correction of the cupping artifact, increase in image contrast, and contrast-to-noise ratio (CNR). RESULTS: The mean MRD and MAE across all cases (min, max) for the internal validation set were 0.04% (-1.1%, 1.3%) and 2.94% (2.7%, 3.2%), while for the external test set they were -0.64% (-1.6%, 0.2%) and 2.84% (2.3%, 3.5%), respectively. For MC-simulated reconstruction slices, the computed SSIM was 0.99 and the MAE was 0.11% (range: 0%, 0.35%) with a single outlier slice of 2.06%. For the three patient bCT reconstructed images, the correction increased the contrast by a mean of 25% (range: 20%, 30%), and reduced the cupping artifact. The mean CNR increased by 0.32 after scatter correction, which was not found to be significant (95% confidence interval: [-0.01, 0.65], p = 0.059). The time required to correct the scatter in a single bCT projection was 0.2 s on an NVIDIA GeForce GTX 1080 GPU. CONCLUSION: The developed DL model could accurately estimate scatter in bCT projection images and could enhance contrast and correct for cupping artifact in reconstructed patient images without significantly affecting the CNR. The time required for correction would allow its use in daily clinical practice, and the reported accuracy will potentially allow quantitative reconstructions.

Patient-derived heterogeneous breast phantoms for advanced dosimetry in mammography and tomosynthesis.

BACKGROUND: Understanding the magnitude and variability of the radiation dose absorbed by the breast fibroglandular tissue during mammography and digital breast tomosynthesis (DBT) is of paramount importance to assess risks versus benefits. Although homogeneous breast models have been proposed and used for decades for this purpose, they do not accurately reflect the actual heterogeneous distribution of the fibroglandular tissue in the breast, leading to biases in the estimation of dose from these modalities. PURPOSE: To develop and validate a method to generate patient-derived, heterogeneous digital breast phantoms for breast dosimetry in mammography and DBT. METHODS: The proposed phantoms were developed starting from patient-based models of compressed breasts, generated for multiple thicknesses and representing the two standard views acquired in mammography and DBT, that is, cranio-caudal (CC) and medio-lateral-oblique (MLO). Internally, the breast phantoms were defined as consisting of an adipose/fibroglandular tissue mixture, with a nonspatially uniform relative concentration. The parenchyma distributions were obtained from a previously described model based on patient breast computed tomography data that underwent simulated compression. Following these distributions, phantoms with any glandular fraction (1%-100%) and breast thickness (12-125 mm) can be generated, for both views. The phantoms were validated, in terms of their accuracy for average normalized glandular dose (Dg N) estimation across samples of patient breasts, using 88 patient-specific phantoms involving actual patient distribution of the fibroglandular tissue in the breast, and compared to that obtained using a homogeneous model similar to those currently used for breast dosimetry. RESULTS: The average Dg N estimated for the proposed phantoms was concordant with that absorbed by the patient-specific phantoms to within 5% (CC) and 4% (MLO). These Dg N estimates were over 30% lower than those estimated with the homogeneous models, which overestimated the average Dg N by 43% (CC), and 32% (MLO) compared to the patient-specific phantoms. CONCLUSIONS: The developed phantoms can be used for dosimetry simulations to improve the accuracy of dose estimates in mammography and DBT.

Computer-aided diagnosis of masses in breast computed tomography imaging: deep learning model with combined handcrafted and convolutional radiomic features.

Purpose: A computer-aided diagnosis (CADx) system for breast masses is proposed, which incorporates both handcrafted and convolutional radiomic features embedded into a single deep learning model. Approach: The model combines handcrafted and convolutional radiomic signatures into a multi-view architecture, which retrieves three-dimensional (3D) image information by simultaneously processing multiple two-dimensional mass patches extracted along different planes through the 3D mass volume. Each patch is processed by a stream composed of two concatenated parallel branches: a multi-layer perceptron fed with automatically extracted handcrafted radiomic features, and a convolutional neural network, for which discriminant features are learned from the input patches. All streams are then concatenated together into a final architecture, where all network weights are shared and the learning occurs simultaneously for each stream and branch. The CADx system was developed and tested for diagnosis of breast masses ( N = 284 ) using image datasets acquired with independent dedicated breast computed tomography systems from two different institutions. The diagnostic classification performance of the CADx system was compared against other machine and deep learning architectures adopting handcrafted and convolutional approaches, and three board-certified breast radiologists. Results: On a test set of 82 masses (45 benign, 37 malignant), the proposed CADx system performed better than all other model architectures evaluated, with an increase in the area under the receiver operating characteristics curve (AUC) of 0.05 ± 0.02 , and achieving a final AUC of 0.947, outperforming the three radiologists ( AUC = 0.814 - 0.902 ). Conclusions: In conclusion, the system demonstrated its potential usefulness in breast cancer diagnosis by improving mass malignancy assessment.

Deep learning reconstruction of digital breast tomosynthesis images for accurate breast density and patient-specific radiation dose estimation.

The two-dimensional nature of mammography makes estimation of the overall breast density challenging, and estimation of the true patient-specific radiation dose impossible. Digital breast tomosynthesis (DBT), a pseudo-3D technique, is now commonly used in breast cancer screening and diagnostics. Still, the severely limited 3rd dimension information in DBT has not been used, until now, to estimate the true breast density or the patient-specific dose. This study proposes a reconstruction algorithm for DBT based on deep learning specifically optimized for these tasks. The algorithm, which we name DBToR, is based on unrolling a proximal-dual optimization method. The proximal operators are replaced with convolutional neural networks and prior knowledge is included in the model. This extends previous work on a deep learning-based reconstruction model by providing both the primal and the dual blocks with breast thickness information, which is available in DBT. Training and testing of the model were performed using virtual patient phantoms from two different sources. Reconstruction performance, and accuracy in estimation of breast density and radiation dose, were estimated, showing high accuracy (density <±3%; dose <±20%) without bias, significantly improving on the current state-of-the-art. This work also lays the groundwork for developing a deep learning-based reconstruction algorithm for the task of image interpretation by radiologists.

Fibroglandular tissue distribution in the breast during mammography and tomosynthesis based on breast CT data: A patient-based characterization of the breast parenchyma.

PURPOSE: To develop a patient-based breast density model by characterizing the fibroglandular tissue distribution in patient breasts during compression for mammography and digital breast tomosynthesis (DBT) imaging. METHODS: In this prospective study, 88 breast images were acquired using a dedicated breast computed tomography (CT) system. The breasts in the images were classified into their three main tissue components and mechanically compressed to mimic the positioning for mammographic acquisition of the craniocaudal (CC) and mediolateral oblique (MLO) views. The resulting fibroglandular tissue distribution during these compressions was characterized by dividing the compressed breast volume into small regions, for which the median and the 25th and 75th percentile values of local fibroglandular density were obtained in the axial, coronal, and sagittal directions. The best fitting function, based on the likelihood method, for the median distribution was obtained in each direction. RESULTS: The fibroglandular tissue tends to concentrate toward the caudal (about 15% below the midline of the breast) and anterior regions of the breast, in both the CC- and MLO-view compressions. A symmetrical distribution was found in the MLO direction in the case of the CC-view compression, while a shift of about 12% toward the lateral direction was found in the MLO-view case. CONCLUSIONS: The location of the fibroglandular tissue in the breast under compression during mammography and DBT image acquisition is a major factor for determining the actual glandular dose imparted during these examinations. A more realistic model of the parenchyma in the compressed breast, based on patient image data, was developed. This improved model more accurately reflects the fibroglandular tissue spatial distribution that can be found in patient breasts, and therefore might aid in future studies involving radiation dose and/or cancer development risk estimation.

Multi-marker quantitative radiomics for mass characterization in dedicated breast CT imaging.

PURPOSE: To develop and evaluate the diagnostic performance of an algorithm for multi-marker radiomic-based classification of breast masses in dedicated breast computed tomography (bCT) images. METHODS: Over 1000 radiomic descriptors aimed at quantifying mass and border heterogeneity, morphology, and margin sharpness were developed and implemented. These included well-established texture and shape feature descriptors, which were supplemented with additional approaches for contour irregularity quantification, spicule and lobe detection, characterization of degree of infiltration, and differences in peritumoral compartments. All descriptors were extracted from a training set of 202 bCT masses (133 benign and 69 malignant), and their individual diagnostic performance was investigated in terms of area under the receiver operating characteristics (ROC) curve (AUC) of single-feature-based linear discriminant analysis (LDA) classifiers. Subsequently, the most relevant descriptors were selected through a multiple-step feature selection process (including stability analysis, statistical significance, evaluation of feature interaction, and dimensionality reduction), and used to develop a final LDA radiomic model for classification of benign and malignant masses, which was then tested on an independent test set of 82 cases (45 benign and 37 malignant). RESULTS: The majority of the individual radiomic descriptors showed, on the training set, an AUC value deriving from a linear decision boundary higher than 0.65, with the lower limit of the associated 95% confidence interval (C.I.) not overlapping with random chance (AUC = 0.5). The final LDA radiomic model resulted in a test set AUC of 0.90 (95% C.I. 0.80-0.96). CONCLUSIONS: The proposed multi-marker radiomic approach achieved high diagnostic accuracy in bCT mass classification, using a radiomic signature based on different feature types. While future studies with larger datasets are needed to further validate these results, quantitative radiomics applied to bCT shows potential to improve the breast cancer diagnosis pipeline.

Deep learning-based segmentation of breast masses in dedicated breast CT imaging: Radiomic feature stability between radiologists and artificial intelligence.

A deep learning (DL) network for 2D-based breast mass segmentation in unenhanced dedicated breast CT images was developed and validated, and its robustness in radiomic feature stability and diagnostic performance compared to manual annotations of multiple radiologists was investigated. 93 mass-like lesions were extensively augmented and used to train the network (n = 58 masses), which was then tested (n = 35 masses) against manual ground truth of a qualified breast radiologist with experience in breast CT imaging using the Conformity coefficient (with a value equal to 1 indicating a perfect performance). Stability and diagnostic power of 672 radiomic descriptors were investigated between the computerized segmentation, and 4 radiologists' annotations for the 35 test set cases. Feature stability and diagnostic performance in the discrimination between benign and malignant cases were quantified using intraclass correlation (ICC) and multivariate analysis of variance (MANOVA), performed for each segmentation case (4 radiologists and DL algorithm). DL-based segmentation resulted in a Conformity of 0.85 ± 0.06 against the annotated ground truth. For the stability analysis, although modest agreement was found among the four annotations performed by radiologists (Conformity 0.78 ± 0.03), over 90% of all radiomic features were found to be stable (ICC>0.75) across multiple segmentations. All MANOVA analyses were statistically significant (p ≤ 0.05), with all dimensions equal to 1, and Wilks' lambda ≤0.35. In conclusion, DL-based mass segmentation in dedicated breast CT images can achieve high segmentation performance, and demonstrated to provide stable radiomic descriptors with comparable discriminative power in the classification of benign and malignant tumors to expert radiologist annotation.

Multi-Scale deep learning framework for cochlea localization, segmentation and analysis on clinical ultra-high-resolution CT images.

BACKGROUND AND OBJECTIVE: Performing patient-specific, pre-operative cochlea CT-based measurements could be helpful to positively affect the outcome of cochlear surgery in terms of intracochlear trauma and loss of residual hearing. Therefore, we propose a method to automatically segment and measure the human cochlea in clinical ultra-high-resolution (UHR) CT images, and investigate differences in cochlea size for personalized implant planning. METHODS: 123 temporal bone CT scans were acquired with two UHR-CT scanners, and used to develop and validate a deep learning-based system for automated cochlea segmentation and measurement. The segmentation algorithm is composed of two major steps (detection and pixel-wise classification) in cascade, and aims at combining the results of a multi-scale computer-aided detection scheme with a U-Net-like architecture for pixelwise classification. The segmentation results were used as an input to the measurement algorithm, which provides automatic cochlear measurements (volume, basal diameter, and cochlear duct length (CDL)) through the combined use of convolutional neural networks and thinning algorithms. Automatic segmentation was validated against manual annotation, by the means of Dice similarity, Boundary-F1 (BF) score, and maximum and average Hausdorff distances, while measurement errors were calculated between the automatic results and the corresponding manually obtained ground truth on a per-patient basis. Finally, the developed system was used to investigate the differences in cochlea size within our patient cohort, to relate the measurement errors to the actual variation in cochlear size across different patients. RESULTS: Automatic segmentation resulted in a Dice of 0.90 ± 0.03, BF score of 0.95 ± 0.03, and maximum and average Hausdorff distance of 3.05 ± 0.39 and 0.32 ± 0.07 against manual annotation. Automatic cochlear measurements resulted in errors of 8.4% (volume), 5.5% (CDL), 7.8% (basal diameter). The cochlea size varied broadly, ranging between 0.10 and 0.28 ml (volume), 1.3 and 2.5 mm (basal diameter), and 27.7 and 40.1 mm (CDL). CONCLUSIONS: The proposed algorithm could successfully segment and analyze the cochlea on UHR-CT images, resulting in accurate measurements of cochlear anatomy. Given the wide variation in cochlear size found in our patient cohort, it may find application as a pre-operative tool in cochlear implant surgery, potentially helping elaborate personalized treatment strategies based on patient-specific, image-based anatomical measurements.

Towards 4D dedicated breast CT perfusion imaging of cancer: development and validation of computer simulated images.

Dedicated breast CT is a fully tomographic breast imaging modality with potential for various applications throughout breast cancer care. If implemented to perform dynamic contrast-enhanced (CE) imaging (4D breast CT), it could be useful to obtain functional information at high combined spatio-temporal resolution. Before developing a 4D dedicated breast CT system, a computer simulation method for breast CT perfusion imaging is proposed. The simulation uses previously developed patient-based 4D digital breast phantoms, and generates realistic images with the selected acquisition parameters, allowing to investigate the effect of different acquisition settings on image quality. The simulation pipeline includes all steps of the image generation process, from ray tracing and scatter map generation, to the addition of realistic resolution losses and noise models. The pipeline was validated against experimental measurements performed on physical phantoms with a dedicated breast CT system, in terms of average error compared to ground truth projections (6.0% ± 1.65%), Hounsfield unit (HU) values in a homogeneous phantom (acquired: -149 HU ± 2 HU; simulated: -140 HU ± 2 HU), signal-to-noise ratio (SNR) (average error 6.7% ± 4.2%), noise power spectra (NPS) (average error 4.3% ± 2.5%), modulation transfer function (MTF) (average error 8.4% ± 7.2%), and attenuation of different adipose/glandular equivalent mixtures (average error 6.9% ± 4.0%) and glandular plus iodinated contrast medium concentrations equivalent mixtures (average error of 9.1% ± 9.0%). 4D patient images were then simulated for different 4D digital breast phantoms at different air kerma levels to determine the effect of noise on the extracted tumor perfusion curves. In conclusion, the proposed pipeline could simulate images with a good level of realism, resulting in a tool that can be used for the design, development, and optimization of a 4D dedicated breast CT system.

Monte Carlo study on optimal breast voxel resolution for dosimetry estimates in digital breast tomosynthesis.

Digital breast tomosynthesis (DBT) is currently used as an adjunct technique to digital mammography (DM) for breast cancer imaging. Being a quasi-3D image, DBT is capable of providing depth information on the internal breast glandular tissue distribution, which may be enough to obtain an accurate patient-specific radiation dose estimate. However, for this, information regarding the location of the glandular tissue, especially in the vertical direction (i.e. x-ray source to detector), is needed. Therefore, a dedicated reconstruction algorithm designed to localize the amount of glandular tissue, rather than for optimal diagnostic value, could be desirable. Such a reconstruction algorithm, or, alternatively, a reconstructed DBT image classification algorithm, could benefit from the use of larger voxels, rather than the small sizes typically used for the diagnostic task. In addition, the Monte Carlo (MC) based dose estimates would be accelerated by the representation of the breast tissue with fewer and larger voxels. Therefore, in this study we investigate the optimal DBT reconstructed voxel size that allows accurate dose evaluations (i.e. within 5%) using a validated Geant4-based MC code. For this, sixty patient-based breast models, previously acquired using dedicated breast computed tomography (BCT) images, were deformed to reproduce the breast during compression under a given DBT scenario. Two re-binning approaches were applied to the compressed phantoms, leading to isotropic and anisotropic voxels of different volumes. MC DBT simulations were performed reproducing the acquisition geometry of a SIEMENS Mammomat Inspiration system. Results show that isotropic cubic voxels of 2.73 mm size provide a dose estimate accurate to within 5% for 51/60 patients, while a comparable accuracy is obtained with anisotropic voxels of dimension 5.46 × 5.46 × 2.73 mm3. In addition, the MC simulation time is reduced by more than half in respect to the original voxel dimension of 0.273 × 0.273 × 0.273 mm3 when either of the proposed re-binning approaches is used. No significant differences in the effect of binning on the dose estimates are observed (Wilcoxon-Mann-Whitney test, p-value > 0.4) between the 0° the 23° (i.e. the widest angular range) exposure.

Development of 3D patient-based super-resolution digital breast phantoms using machine learning.

Digital phantoms are important tools for optimization and evaluation of x-ray imaging systems, and should ideally reflect the 3D structure of human anatomy and its potential variability. In addition, they need to include a good level of detail at a high enough spatial resolution to accurately model the continuous nature of the human anatomy. A pipeline to increase the spatial resolution of patient-based digital breast phantoms that can be used for computer simulations of breast imaging is proposed. Given a tomographic breast image of finite resolution, the proposed methods can generate a phantom and increase its resolution at will, not only simply through super-sampling, but also by generating additional random glandular details to account for glandular edges and strands to compensate for those that may have not been detected in the original image due to the limited spatial resolution of the imaging system used. The proposed algorithms use supervised learning to predict the loss in glandularity due to limited resolution, and then to realistically recover this loss by learning the mapping between low and high resolution images. They were trained on high-resolution synchrotron images (detector pixel size 60 µm) reconstructed at seven voxel dimensions (60 µm-480 µm), and applied to patient images acquired with a clinical breast CT system (detector pixel size 194 µm) to generate super-resolution phantoms (voxel sizes 68 µm). Several evaluations were made to assess the appropriateness of the developed methods, both with the synchrotron (relative prediction error 0.010 ± 0.004, recovering accuracy 0.95 ± 0.04), and with the clinical images (average glandularity error at 194 µm: 0.15% ± 0.12%). Finally, a breast radiologist assessed the realism of the developed phantoms by blindly comparing original and phantom images, resulting in not being able to distinguish the real from the phantom images. In conclusion, the proposed method can generate super-resolution phantoms from tomographic breast patient images that can be used for future computer simulations for optimization of new breast imaging technologies.

Patient-based 4D digital breast phantom for perfusion contrast-enhanced breast CT imaging.

PURPOSE: The purpose of this study was to develop a realistic patient-based 4D digital breast phantom including time-varying contrast enhancement for simulation of dedicated breast CT perfusion imaging. METHODS: A 3D static phantom is first created by segmenting a breast CT image from a healthy patient into skin, fibroglandular tissue, adipose tissue, and vasculature. For the creation of abnormal cases, a breast lesion model was developed and can be added to the phantom. After defining the necessary perfusion parameters for each tissue (e.g., arterial input function for vasculature, blood volume and blood flow for the other normal tissues) based on contrast-enhanced dynamic breast MRI data, the corresponding time-enhancement curves are computed for each voxel in the phantom, according to tissue type. These curves are calculated by convolution between the arterial input function and a shifted exponential function. This exponential depends on the perfusion parameters associated with each tissue voxel, and, to incorporate normal biological variability, a uniform random distribution is used to vary the perfusion parameters on a voxel-basis. Finally, a 4D array is produced by sampling the continuous time-enhancement curves at the desired sampling rate. Beside modeling different enhancement dynamics according to the given input perfusion parameters, the phantom also includes the possibility to realistically simulate different spatial enhancement patterns for the breast parenchyma, taking into account the arterial sources supplying the breast. Finally, different patterns of contrast medium uptake can also be simulated for the tumor models (homogeneous and rim enhancement). RESULTS: As an example, a typical 4D phantom has dimensions of 426 × 421 × 260 × 559 (x, y, z, t), with a voxel size of 273 µm and a sampling time of 1 s. The characteristics of the tumor model can be modified at will to evaluate perfusion in different types of breast lesions. Results show the expected enhancement of tissues, consistent with the given input parameters. Moreover, the tumor models evaluated in this work show different enhancement dynamics according to the tumor type (defined by different input perfusion parameters), and also present a higher enhancement compared to the other healthy tissues, as expected. CONCLUSIONS: The proposed digital phantom can model the breast tissue perfusion during 4D breast CT image acquisition, displaying the different enhancement dynamics that could be found in a real patient breast. This phantom can be used during the development of dynamic contrast-enhanced dedicated breast CT imaging, for optimization of image acquisition, image reconstruction, and image analysis. This modality could provide functional information of the breast, resulting in detection, diagnosis, and treatment improvements of breast cancer with breast CT.

Internal breast dosimetry in mammography: Monte Carlo validation in homogeneous and anthropomorphic breast phantoms with a clinical mammography system.

PURPOSE: To validate Monte Carlo (MC)-based breast dosimetry estimations using both a homogeneous and a 3D anthropomorphic breast phantom under polyenergetic irradiation for internal breast dosimetry purposes. METHODS: Experimental measurements were performed with a clinical digital mammography system (Mammomat Inspiration, Siemens Healthcare), using the x-ray spectrum selected by the automatic exposure control and a tube current-exposure time product of 360 mAs. A homogeneous 50% glandular breast phantom and a 3D anthropomorphic breast phantom were used to investigate the dose at different depths (range 0-4 cm with 1 cm steps) for the homogeneous case and at a depth of 2.25 cm for the anthropomorphic case. Local dose deposition was measured using thermoluminescent dosimeters (TLD), metal oxide semiconductor field-effect transistor dosimeters (MOSFET), and GafChromic™ films. A Geant4-based MC simulation was modified to match the clinical experimental setup. Thirty sensitive volumes (3.2 × 3.2 × 0.38 mm3 ) on the axial-phantom plane were included at each depth in the simulation to characterize the internal dose variation and compare it to the experimental TLD and MOSFET measurements. The experimental 2D dose maps obtained with the GafChromic™ films were compared to the simulated 2D dose distributions. RESULTS: Due to the energy dependence of the dosimeters and due to x-ray beam hardening, dosimeters based on these three technologies have to be calibrated at each depth of the phantom. As expected, the dose was found to decrease with increasing phantom depth, with the reduction being ~93% after 4 cm for the homogeneous breast phantom. The 2D dose map showed nonuniformities in the dose distribution in the axial plane of the phantom. The mean combined standard uncertainty increased with phantom depth by up to 5.3% for TLD, 6.3% for MOSFET, and 9.6% for GafChromic™ film. In the case of a heterogeneous phantom, the dosimeters are able to detect local dose gradient variations. In particular, GafChromic™ film showed local dose variations of about 46% at the boundaries of two materials. CONCLUSIONS: Results showed a good agreement between experimental measurements (with TLD and MOSFET) and MC data for both homogeneous and anthropomorphic breast phantoms. Larger discrepancies are found when comparing the GafChromic™ dose values to the MC results due to the inherent less precise nature of the former. MC validations in a heterogeneous background at the level of local dose deposition and in absolute terms play a fundamental role in the development of an accurate method to estimate radiation dose. The potential introduction of a breast dosimetry model involving a nonhomogeneous glandular/adipose tissue composition makes the validation of internal dose distributions estimates crucial.

Quantitative analysis of thyroid tumors vascularity: A comparison between 3-D contrast-enhanced ultrasound and 3-D Power Doppler on benign and malignant thyroid nodules.

PURPOSE: To perform a comparative quantitative analysis of Power Doppler ultrasound (PDUS) and Contrast-Enhancement ultrasound (CEUS) for the quantification of thyroid nodules vascularity patterns, with the goal of identifying biomarkers correlated with the malignancy of the nodule with both imaging techniques. METHODS: We propose a novel method to reconstruct the vascular architecture from 3-D PDUS and CEUS images of thyroid nodules, and to automatically extract seven quantitative features related to the morphology and distribution of vascular network. Features include three tortuosity metrics, the number of vascular trees and branches, the vascular volume density, and the main spatial vascularity pattern. Feature extraction was performed on 20 thyroid lesions (ten benign and ten malignant), of which we acquired both PDUS and CEUS. MANOVA (multivariate analysis of variance) was used to differentiate benign and malignant lesions based on the most significant features. RESULTS: The analysis of the extracted features showed a significant difference between the benign and malignant nodules for both PDUS and CEUS techniques for all the features. Furthermore, by using a linear classifier on the significant features identified by the MANOVA, benign nodules could be entirely separated from the malignant ones. CONCLUSIONS: Our early results confirm the correlation between the morphology and distribution of blood vessels and the malignancy of the lesion, and also show (at least for the dataset used in this study) a considerable similarity in terms of findings of PDUS and CEUS imaging for thyroid nodules diagnosis and classification.

An unsupervised automatic segmentation algorithm for breast tissue classification of dedicated breast computed tomography images.

PURPOSE: To develop and evaluate a new automatic classification algorithm to identify voxels containing skin, vasculature, adipose, and fibroglandular tissue in dedicated breast CT images. METHODS: The proposed algorithm combines intensity- and region-based segmentation methods with energy minimizing splines and unsupervised data mining approaches for classifying and segmenting the different tissue types. Breast skin segmentation is achieved by a region-growing method which uses constraints from the previously extracted skin centerline to add robustness to the model and to reduce the false positive rate. An energy minimizing active contour model is then used to classify adipose tissue voxels by including gradient flow and region-based features. Finally, blood vessels are separated from fibroglandular tissue by a k-means clustering algorithm based on automatically extracted shape-based features. To evaluate the accuracy of the algorithm, two sets of 15 different patient breast CT scans, each acquired with different breast CT systems and acquisition settings were obtained. Three slices from each scan were manually segmented under the supervision of an experienced breast radiologist and considered the gold standard. Comparisons with manual segmentation were quantified using five similarity metrics: Dice similarity coefficient (DSC), sensitivity, conformity coefficient, and two Hausdorff distance measures. To evaluate the robustness to image noise, the segmentation was repeated after separately adding Gaussian noise with increasing standard deviation (in four steps, from 0.01 to 0.04) to an additional 15 slices from the first dataset. In addition, to evaluate vasculature classification, three different pre- and postcontrast injection patient breast CT images were classified and compared. Finally, DSC was also used for quantitative comparisons with previously proposed approaches for breast CT tissue classification using 10 images from the first dataset. RESULTS: The algorithm showed a high accuracy in classifying the different tissue types for both breast CT systems, with an average DSC of 95% and 90% for the first and second image dataset, respectively. Furthermore, it demonstrated to be robust to image noise with a robustness to image noise of 85%, 83%, 79%, and 71% for the images corrupted with the four increasing noise levels. Previous methods for breast tissues classification resulted, for the tested dataset, in an average global DSC of 87%, while our approach resulted in a global average DSC of 94.5%. CONCLUSIONS: The proposed algorithm resulted in accurate and robust breast tissue classification, with no prior training or threshold setting. Potential applications include breast density quantification and tissue pattern characterization (both biomarkers of cancer development), simulation-based radiation dose analysis, and patient data-based phantom design, which could be used for further breast imaging research.