Multivariate testing and effect size measures for batch effect evaluation in radiomic features.

While precision medicine applications of radiomics analysis are promising, differences in image acquisition can cause "batch effects" that reduce reproducibility and affect downstream predictive analyses. Harmonization methods such as ComBat have been developed to correct these effects, but evaluation methods for quantifying batch effects are inconsistent. In this study, we propose the use of the multivariate statistical test PERMANOVA and the Robust Effect Size Index (RESI) to better quantify and characterize batch effects in radiomics data. We evaluate these methods in both simulated and real radiomics features extracted from full-field digital mammography (FFDM) data. PERMANOVA demonstrated higher power than standard univariate statistical testing, and RESI was able to interpretably quantify the effect size of site at extremely large sample sizes. These methods show promise as more powerful and interpretable methods for the detection and quantification of batch effects in radiomics studies.

Breast density quantitative measures and breast cancer risk among screened Black women.

Mammographic density is a strong predictor of breast cancer but only slightly increased the discriminatory ability of existing risk prediction models in previous studies with limited racial diversity. We assessed discrimination and calibration of models consisting of the Breast Cancer Risk Assessment Tool (BCRAT), Breast Imaging-Reporting and Data System density and quantitative density measures. Patients were followed up from the date of first screening mammogram until invasive breast cancer diagnosis or 5-year follow-up. Areas under the curve for White women stayed consistently around 0.59 for all models, whereas the area under the curve increased slightly from 0.60 to 0.62 when adding dense area and area percent density to the BCRAT model for Black women. All women saw underprediction in all models, with Black women having less underprediction. Adding quantitative density to the BCRAT did not statistically significantly improve prediction for White or Black women. Future studies should evaluate whether volumetric breast density improves risk prediction.

Changes in mammographic density and risk of breast cancer among a diverse cohort of women undergoing mammography screening.

PURPOSE: Mammographic density (MD) is a strong breast cancer risk factor. MD may change over time, with potential implications for breast cancer risk. Few studies have assessed associations between MD change and breast cancer in racially diverse populations. We investigated the relationships between MD and MD change over time and breast cancer risk in a large, diverse screening cohort. MATERIALS AND METHODS: We retrospectively analyzed data from 8462 women who underwent ≥ 2 screening mammograms from Sept. 2010 to Jan. 2015 (N = 20,766 exams); 185 breast cancers were diagnosed 1-7 years after screening. Breast percent density (PD) and dense area (DA) were estimated from raw digital mammograms (Hologic Inc.) using LIBRA (v1.0.4). For each MD measure, we modeled breast density change between two sequential visits as a function of demographic and risk covariates. We used Cox regression to examine whether varying degrees of breast density change were associated with breast cancer risk, accounting for multiple exams per woman. RESULTS: PD at any screen was significantly associated with breast cancer risk (hazard ratio (HR) for PD = 1.03 (95% CI [1.01, 1.05], p < 0.0005), but neither change in breast density nor more extreme than expected changes in breast density were associated with breast cancer risk. We found no evidence of differences in density change or breast cancer risk due to density change by race. Results using DA were essentially identical. CONCLUSIONS: Using a large racially diverse cohort, we found no evidence of association between short-term change in MD and risk of breast cancer, suggesting that short-term MD change is not a strong predictor for risk.

Radiomic phenotyping of the lung parenchyma in a lung cancer screening cohort.

High-throughput extraction of radiomic features from low-dose CT scans can characterize the heterogeneity of the lung parenchyma and potentially aid in identifying subpopulations that may have higher risk of lung diseases, such as COPD, and lung cancer due to inflammation or obstruction of the airways. We aim to determine the feasibility of a lung radiomics phenotyping approach in a lung cancer screening cohort, while quantifying the effect of different CT reconstruction algorithms on phenotype robustness. We identified low-dose CT scans (n = 308) acquired with Siemens Healthineers scanners from patients who completed low-dose CT within our lung cancer screening program between 2015 and 2018 and had two different sets of image reconstructions kernel available (i.e., medium (I30f.), sharp (I50f.)) for the same acquisition. Following segmentation of the lung field, a total of 26 radiomic features were extracted from the entire 3D lung-field using a previously validated fully-automated lattice-based software pipeline, adapted for low-dose CT scans. The lattice in-house software was used to extract features including gray-level histogram, co-occurrence, and run-length descriptors. The lattice approach uses non-overlapping windows for traversing along pixels of images and calculates different features. Each feature was averaged for each scan within a range of lattice window sizes (W) of 4, 8 and 20 mm. The extracted imaging features from both datasets were harmonized to correct for differences in image acquisition parameters. Subsequently, unsupervised hierarchical clustering was applied on the extracted features to identify distinct phenotypic patterns of the lung parenchyma, where consensus clustering was used to identify the optimal number of clusters (K = 2). Differences between phenotypes for demographic and clinical covariates including sex, age, BMI, pack-years of smoking, Lung-RADS and cancer diagnosis were assessed for each phenotype cluster, and then compared across clusters for the two different CT reconstruction algorithms using the cluster entanglement metric, where a lower entanglement coefficient corresponds to good cluster alignment. Furthermore, an independent set of low-dose CT scans (n = 88) from patients with available pulmonary function data on lung obstruction were analyzed using the identified optimal clusters to assess associations to lung obstruction and validate the lung phenotyping paradigm. Heatmaps generated by radiomic features identified two distinct lung parenchymal phenotype patterns across different feature extraction window sizes, for both reconstruction algorithms (P < 0.05 with K = 2). Associations of radiomic-based clusters with clinical covariates showed significant differences for BMI and pack-years of smoking (P < 0.05) for both reconstruction kernels. Radiomic phenotype patterns were more similar across the two reconstructed kernels, when smaller window sizes (W = 4 and 8 mm) were used for radiomic feature extraction, as deemed by their entanglement coefficient. Validation of clustering approaches using cluster mapping for the independent sample with lung obstruction also showed two statistically significant phenotypes (P < 0.05) with significant difference for BMI and smoking pack-years. Radiomic analysis can be used to characterize lung parenchymal phenotypes from low-dose CT scans, which appear reproducible for different reconstruction kernels. Further work should seek to evaluate the effect of additional CT acquisition parameters and validate these phenotypes in characterizing lung cancer screening populations, to potentially better stratify disease patterns and cancer risk.

External Validation of a Mammography-Derived AI-Based Risk Model in a U.S. Breast Cancer Screening Cohort of White and Black Women.

Despite the demonstrated potential of artificial intelligence (AI) in breast cancer risk assessment for personalizing screening recommendations, further validation is required regarding AI model bias and generalizability. We performed external validation on a U.S. screening cohort of a mammography-derived AI breast cancer risk model originally developed for European screening cohorts. We retrospectively identified 176 breast cancers with exams 3 months to 2 years prior to cancer diagnosis and a random sample of 4963 controls from women with at least one-year negative follow-up. A risk score for each woman was calculated via the AI risk model. Age-adjusted areas under the ROC curves (AUCs) were estimated for the entire cohort and separately for White and Black women. The Gail 5-year risk model was also evaluated for comparison. The overall AUC was 0.68 (95% CIs 0.64−0.72) for all women, 0.67 (0.61−0.72) for White women, and 0.70 (0.65−0.76) for Black women. The AI risk model significantly outperformed the Gail risk model for all women p < 0.01 and for Black women p < 0.01, but not for White women p = 0.38. The performance of the mammography-derived AI risk model was comparable to previously reported European validation results; non-significantly different when comparing White and Black women; and overall, significantly higher than that of the Gail model.

Development of a robust radiomic biomarker of progression-free survival in advanced non-small cell lung cancer patients treated with first-line immunotherapy.

We aim to determine the feasibility of a novel radiomic biomarker that can integrate with other established clinical prognostic factors to predict progression-free survival (PFS) in patients with non-small cell lung cancer (NSCLC) undergoing first-line immunotherapy. Our study includes 107 patients with stage 4 NSCLC treated with pembrolizumab-based therapy (monotherapy: 30%, combination chemotherapy: 70%). The ITK-SNAP software was used for 3D tumor volume segmentation from pre-therapy CT scans. Radiomic features (n = 102) were extracted using the CaPTk software. Impact of heterogeneity introduced by image physical dimensions (voxel spacing parameters) and acquisition parameters (contrast enhancement and CT reconstruction kernel) was mitigated by resampling the images to the minimum voxel spacing parameters and harmonization by a nested ComBat technique. This technique was initialized with radiomic features, clinical factors of age, sex, race, PD-L1 expression, ECOG status, body mass index (BMI), smoking status, recurrence event and months of progression-free survival, and image acquisition parameters as batch variables. Two phenotypes were identified using unsupervised hierarchical clustering of harmonized features. Prognostic factors, including PDL1 expression, ECOG status, BMI and smoking status, were combined with radiomic phenotypes in Cox regression models of PFS and Kaplan Meier (KM) curve-fitting. Cox model based on clinical factors had a c-statistic of 0.57, which increased to 0.63 upon addition of phenotypes derived from harmonized features. There were statistically significant differences in survival outcomes stratified by clinical covariates, as measured by the log-rank test (p = 0.034), which improved upon addition of phenotypes (p = 0.00022). We found that mitigation of heterogeneity by image resampling and nested ComBat harmonization improves prognostic value of phenotypes, resulting in better prediction of PFS when added to other prognostic variables.

Impact of deformable registration methods for prediction of recurrence free survival response to neoadjuvant chemotherapy in breast cancer: Results from the ISPY 1/ACRIN 6657 trial.

PURPOSE: Image registration plays a vital role in spatially aligning multiple MRI scans for better longitudinal assessment of tumor morphological features. The objective was to evaluate the effect of registration accuracy of six established deformable registration methods(ANTs, DRAMMS, ART, NiftyReg, SSD-FFD, and NMI-FFD) on the predictive value of extracted radiomic features when modeling recurrence-free-survival(RFS) for women after neoadjuvant chemotherapy(NAC) for locally advanced breast cancer. METHODS: 130 women had DCE-MRI scans available from the first two visits in the ISPY1/ACRIN-6657 cohort. We calculated the transformation field from each of the different deformable registration methods, and used it to compute voxel-wise parametric-response-maps(PRM) for established four kinetic features.104-radiomic features were computed from each PRM map to characterize intra-tumor heterogeneity. We evaluated performance for RFS using Cox-regression, C-statistic, and Kaplan-Meier(KM) plots. RESULTS: A baseline model(F1:Age, Race, and Hormone-receptor-status) had a 0.54 C-statistic, and model F2(baseline + functional-tumor-volume at early treatment visit(FTV2)) had 0.63. The F2+ANTs had the highest C-statistic(0.72) with the smallest landmark differences(5.40±4.40mm) as compared to other models. The KM curve for model F2 gave p=0.004 for separation between women above and below the median hazard compared to the model F1(p=0.31). A models augmented with radiomic features, also achieved significant KM curve separation(p<0.001) except the F2+ART model. CONCLUSION: Incorporating image registration in quantifying changes in tumor heterogeneity during NAC can improve prediction of RFS. Radiomic features of PRM maps derived from warping the DCE-MRI kinetic maps using ANTs registration method further improved the early prediction of RFS as compared to other methods.

Incorporating Robustness to Imaging Physics into Radiomic Feature Selection for Breast Cancer Risk Estimation.

Digital mammography has seen an explosion in the number of radiomic features used for risk-assessment modeling. However, having more features is not necessarily beneficial, as some features may be overly sensitive to imaging physics (contrast, noise, and image sharpness). To measure the effects of imaging physics, we analyzed the feature variation across imaging acquisition settings (kV, mAs) using an anthropomorphic phantom. We also analyzed the intra-woman variation (IWV), a measure of how much a feature varies between breasts with similar parenchymal patterns-a woman's left and right breasts. From 341 features, we identified "robust" features that minimized the effects of imaging physics and IWV. We also investigated whether robust features offered better case-control classification in an independent data set of 575 images, all with an overall BI-RADS® assessment of 1 (negative) or 2 (benign); 115 images (cases) were of women who developed cancer at least one year after that screening image, matched to 460 controls. We modeled cancer occurrence via logistic regression, using cross-validated area under the receiver-operating-characteristic curve (AUC) to measure model performance. Models using features from the most-robust quartile of features yielded an AUC = 0.59, versus 0.54 for the least-robust, with p < 0.005 for the difference among the quartiles.

Racial Disparities in 30-Day Outcomes Following Index Admission for COVID-19.

We investigated racial disparities in a 30-day composite outcome of readmission and death among patients admitted across a 5-hospital health system following an index COVID-19 admission. A dataset of 1,174 patients admitted between March 1, 2020 and August 21, 2020 for COVID-19 was retrospectively analyzed for odds of readmission among Black patients compared to all other patients, with sequential adjustment for demographics, index admission characteristics, type of post-acute care, and comorbidities. Tabulated results demonstrated a significantly greater odds of 30-day readmission or death among Black patients (18.0% of Black patients vs. 11.3% of all other patients; Univariate Odds Ratio: 1.71, p = 0.002). Sequential adjustment via logistic regression revealed that the odds of 30-day readmission or death were significantly greater among Black patients after adjustment for demographics, index admission characteristics, and type of post-acute care, but not comorbidities. Stratification by type of post-acute care received on discharge revealed that the same disparity in odds of 30-day readmission or death existed among patients discharged home without home services, but not those discharged to home with home services or to a skilled nursing facility or acute rehab facility. Collectively, the findings suggest that weighing comorbidity burdens in post-acute care decisions may be relevant in addressing racial disparities in 30-day outcomes following discharge from an index COVID-19 admission.

Fully Automated Volumetric Breast Density Estimation from Digital Breast Tomosynthesis.

Background While digital breast tomosynthesis (DBT) is rapidly replacing digital mammography (DM) in breast cancer screening, the potential of DBT density measures for breast cancer risk assessment remains largely unexplored. Purpose To compare associations of breast density estimates from DBT and DM with breast cancer. Materials and Methods This retrospective case-control study used contralateral DM/DBT studies from women with unilateral breast cancer and age- and ethnicity-matched controls (September 19, 2011-January 6, 2015). Volumetric percent density (VPD%) was estimated from DBT using previously validated software. For comparison, the publicly available Laboratory for Individualized Breast Radiodensity Assessment software package, or LIBRA, was used to estimate area-based percent density (APD%) from raw and processed DM images. The commercial Quantra and Volpara software packages were applied to raw DM images to estimate VPD% with use of physics-based models. Density measures were compared by using Spearman correlation coefficients (r), and conditional logistic regression was performed to examine density associations (odds ratios [OR]) with breast cancer, adjusting for age and body mass index. Results A total of 132 women diagnosed with breast cancer (mean age ± standard deviation [SD], 60 years ± 11) and 528 controls (mean age, 60 years ± 11) were included. Moderate correlations between DBT and DM density measures (r = 0.32-0.75; all P < .001) were observed. Volumetric density estimates calculated from DBT (OR, 2.3 [95% CI: 1.6, 3.4] per SD for VPD%DBT) were more strongly associated with breast cancer than DM-derived density for both APD% (OR, 1.3 [95% CI: 0.9, 1.9] [P < .001] and 1.7 [95% CI: 1.2, 2.3] [P = .004] per SD for LIBRA raw and processed data, respectively) and VPD% (OR, 1.6 [95% CI: 1.1, 2.4] [P = .01] and 1.7 [95% CI: 1.2, 2.6] [P = .04] per SD for Volpara and Quantra, respectively). Conclusion The associations between quantitative breast density estimates and breast cancer risk are stronger for digital breast tomosynthesis compared with digital mammography. © RSNA, 2021 Online supplemental material is available for this article. See also the editorial by Yaffe in this issue.

Deep-LIBRA: An artificial-intelligence method for robust quantification of breast density with independent validation in breast cancer risk assessment.

Breast density is an important risk factor for breast cancer that also affects the specificity and sensitivity of screening mammography. Current federal legislation mandates reporting of breast density for all women undergoing breast cancer screening. Clinically, breast density is assessed visually using the American College of Radiology Breast Imaging Reporting And Data System (BI-RADS) scale. Here, we introduce an artificial intelligence (AI) method to estimate breast density from digital mammograms. Our method leverages deep learning using two convolutional neural network architectures to accurately segment the breast area. An AI algorithm combining superpixel generation and radiomic machine learning is then applied to differentiate dense from non-dense tissue regions within the breast, from which breast density is estimated. Our method was trained and validated on a multi-racial, multi-institutional dataset of 15,661 images (4,437 women), and then tested on an independent matched case-control dataset of 6368 digital mammograms (414 cases; 1178 controls) for both breast density estimation and case-control discrimination. On the independent dataset, breast percent density (PD) estimates from Deep-LIBRA and an expert reader were strongly correlated (Spearman correlation coefficient = 0.90). Moreover, in a model adjusted for age and BMI, Deep-LIBRA yielded a higher case-control discrimination performance (area under the ROC curve, AUC = 0.612 [95% confidence interval (CI): 0.584, 0.640]) compared to four other widely-used research and commercial breast density assessment methods (AUCs = 0.528 to 0.599). Our results suggest a strong agreement of breast density estimates between Deep-LIBRA and gold-standard assessment by an expert reader, as well as improved performance in breast cancer risk assessment over state-of-the-art open-source and commercial methods.

Fully automatic quantification of fibroglandular tissue and background parenchymal enhancement with accurate implementation for axial and sagittal breast MRI protocols.

PURPOSE: To propose and evaluate a fully automated technique for quantification of fibroglandular tissue (FGT) and background parenchymal enhancement (BPE) in breast MRI. METHODS: We propose a fully automated method, where after preprocessing, FGT is segmented in T1-weighted, nonfat-saturated MRI. Incorporating an anatomy-driven prior probability for FGT and robust texture descriptors against intensity variations, our method effectively addresses major image processing challenges, including wide variations in breast anatomy and FGT appearance among individuals. Our framework then propagates this segmentation to dynamic contrast-enhanced (DCE)-MRI to quantify BPE within the segmented FGT regions. Axial and sagittal image data from 40 cancer-unaffected women were used to evaluate our proposed method vs a manually annotated reference standard. RESULTS: High spatial correspondence was observed between the automatic and manual FGT segmentation (mean Dice similarity coefficient 81.14%). The FGT and BPE quantifications (denoted FGT% and BPE%) indicated high correlation (Pearson's r = 0.99 for both) between automatic and manual segmentations. Furthermore, the differences between the FGT% and BPE% quantified using automatic and manual segmentations were low (mean differences: -0.66 ± 2.91% for FGT% and -0.17 ± 1.03% for BPE%). When correlated with qualitative clinical BI-RADS ratings, the correlation coefficient for FGT% was still high (Spearman's ρ = 0.92), whereas that for BPE was lower (ρ = 0.65). Our proposed approach also performed significantly better than a previously validated method for sagittal breast MRI. CONCLUSIONS: Our method demonstrated accurate fully automated quantification of FGT and BPE in both sagittal and axial breast MRI. Our results also suggested the complexity of BPE assessment, demonstrating relatively low correlation between segmentation and clinical rating.

Calculation of Radiomic Features to Validate the Textural Realism of Physical Anthropomorphic Phantoms for Digital Mammography.

In this paper, radiomic features are used to validate the textural realism of two anthropomorphic phantoms for digital mammography. One phantom was based off a computational breast model; it was 3D printed by CIRS (Computerized Imaging Reference Systems, Inc., Norfolk, VA) under license from the University of Pennsylvania. We investigate how the textural realism of this phantom compares against a phantom derived from an actual patient's mammogram ("Rachel", Gammex 169, Madison, WI). Images of each phantom were acquired at three kV in 1 kV increments using auto-time technique settings. Acquisitions at each technique setting were repeated twice, resulting in six images per phantom. In the raw ("FOR PROCESSING") images, 341 features were calculated; i.e., gray-level histogram, co-occurrence, run length, fractal dimension, Gabor Wavelet, local binary pattern, Laws, and co-occurrence Laws features. Features were also calculated in a negative screening population. For each feature, the middle 95% of the clinical distribution was used to evaluate the textural realism of each phantom. A feature was considered realistic if all six measurements in the phantom were within the middle 95% of the clinical distribution. Otherwise, a feature was considered unrealistic. More features were actually found to be realistic by this definition in the CIRS phantom (305 out of 341 features or 89.44%) than in the phantom derived from a specific patient's mammogram (261 out of 341 features or 76.54%). We conclude that the texture is realistic overall in both phantoms.

Robust Radiomic Feature Selection in Digital Mammography: Understanding the Effect of Imaging Acquisition Physics Using Phantom and Clinical Data Analysis.

Studies have shown that combining calculations of radiomic features with estimates of mammographic density results in an even better assessment of breast cancer risk than density alone. However, to ensure that risk assessment calculations are consistent across different imaging acquisition settings, it is important to identify features that are not overly sensitive to changes in these settings. In this study, digital mammography (DM) images of an anthropomorphic phantom ("Rachel", Gammex 169, Madison, WI) were acquired at various technique settings. We varied kV and mAs, which control contrast and noise, respectively. DM images in women with negative screening exams were also analyzed. Radiomic features were calculated in the raw ("FOR PROCESSING") DM images; i.e., grey-level histogram, co-occurrence, run length, fractal dimension, Gabor Wavelet, local binary pattern, Laws, and co-occurrence Laws features. For each feature, the range of variation across technique settings in phantom images was calculated. This range was scaled against the range of variation in the clinical distribution (specifically, the range corresponding to the middle 90% of the distribution). In order for a radiomic feature to be considered robust, this metric of imaging acquisition variation (IAV) should be as small as possible (approaching zero). An IAV threshold of 0.25 was proposed for the purpose of this study. Out of 341 features, 284 features (83%) met the threshold IAV ≤ 0.25. In conclusion, we have developed a method to identify robust radiomic features in DM.

O-Net: An Overall Convolutional Network for Segmentation Tasks.

Convolutional neural networks (CNNs) have recently been popular for classification and segmentation through numerous network architectures offering a substantial performance improvement. Their value has been particularly appreciated in the domain of biomedical applications, where even a small improvement in the predicted segmented region (e.g., a malignancy) compared to the ground truth can potentially lead to better diagnosis or treatment planning. Here, we introduce a novel architecture, namely the Overall Convolutional Network (O-Net), which takes advantage of different pooling levels and convolutional layers to extract more deeper local and containing global context. Our quantitative results on 2D images from two distinct datasets show that O-Net can achieve a higher dice coefficient when compared to either a U-Net or a Pyramid Scene Parsing Net. We also look into the stability of results for training and validation sets which can show the robustness of model compared with new datasets. In addition to comparison to the decoder, we use different encoders including simple, VGG Net, and ResNet. The ResNet encoder could help to improve the results in most of the cases.

Imaging Phenotypes of Breast Cancer Heterogeneity in Preoperative Breast Dynamic Contrast Enhanced Magnetic Resonance Imaging (DCE-MRI) Scans Predict 10-Year Recurrence.

PURPOSE: Identifying imaging phenotypes and understanding their relationship with prognostic markers and patient outcomes can allow for a noninvasive assessment of cancer. The purpose of this study was to identify and validate intrinsic imaging phenotypes of breast cancer heterogeneity in preoperative breast dynamic contrast enhanced magnetic resonance imaging (DCE-MRI) scans and evaluate their prognostic performance in predicting 10 years recurrence. EXPERIMENTAL DESIGN: Pretreatment DCE-MRI scans of 95 women with primary invasive breast cancer with at least 10 years of follow-up from a clinical trial at our institution (2002-2006) were retrospectively analyzed. For each woman, a signal enhancement ratio (SER) map was generated for the entire segmented primary lesion volume from which 60 radiomic features of texture and morphology were extracted. Intrinsic phenotypes of tumor heterogeneity were identified via unsupervised hierarchical clustering of the extracted features. An independent sample of 163 women diagnosed with primary invasive breast cancer (2002-2006), publicly available via The Cancer Imaging Archive, was used to validate phenotype reproducibility. RESULTS: Three significant phenotypes of low, medium, and high heterogeneity were identified in the discovery cohort and reproduced in the validation cohort (P < 0.01). Kaplan-Meier curves showed statistically significant differences (P < 0.05) in recurrence-free survival (RFS) across phenotypes. Radiomic phenotypes demonstrated added prognostic value (c = 0.73) predicting RFS. CONCLUSIONS: Intrinsic imaging phenotypes of breast cancer tumor heterogeneity at primary diagnosis can predict 10-year recurrence. The independent and additional prognostic value of imaging heterogeneity phenotypes suggests that radiomic phenotypes can provide a noninvasive characterization of tumor heterogeneity to augment personalized prognosis and treatment.

Prediction of Treatment Response to Neoadjuvant Chemotherapy for Breast Cancer via Early Changes in Tumor Heterogeneity Captured by DCE-MRI Registration.

We analyzed DCE-MR images from 132 women with locally advanced breast cancer from the I-SPY1 trial to evaluate changes of intra-tumor heterogeneity for augmenting early prediction of pathologic complete response (pCR) and recurrence-free survival (RFS) after neoadjuvant chemotherapy (NAC). Utilizing image registration, voxel-wise changes including tumor deformations and changes in DCE-MRI kinetic features were computed to characterize heterogeneous changes within the tumor. Using five-fold cross-validation, logistic regression and Cox regression were performed to model pCR and RFS, respectively. The extracted imaging features were evaluated in augmenting established predictors, including functional tumor volume (FTV) and histopathologic and demographic factors, using the area under the curve (AUC) and the C-statistic as performance measures. The extracted voxel-wise features were also compared to analogous conventional aggregated features to evaluate the potential advantage of voxel-wise analysis. Voxel-wise features improved prediction of pCR (AUC = 0.78 (±0.03) vs 0.71 (±0.04), p < 0.05 and RFS (C-statistic = 0.76 ( ± 0.05), vs 0.63 ( ± 0.01)), p < 0.05, while models based on analogous aggregate imaging features did not show appreciable performance changes (p > 0.05). Furthermore, all selected voxel-wise features demonstrated significant association with outcome (p < 0.05). Thus, precise measures of voxel-wise changes in tumor heterogeneity extracted from registered DCE-MRI scans can improve early prediction of neoadjuvant treatment outcomes in locally advanced breast cancer.

Effect of Mammographic Screening Modality on Breast Density Assessment: Digital Mammography versus Digital Breast Tomosynthesis.

Background Breast Imaging Reporting and Data System (BI-RADS) breast density categories assigned by interpreting radiologists often influence decisions surrounding supplemental breast cancer screening and risk assessment. The landscape of mammographic screening continuously evolves, and different mammographic screening modalities may result in different perception of density, reflected in different assignment of BI-RADS density categories. Purpose To investigate the effect of screening mammography modality on BI-RADS breast density assessments. Materials and Methods Data were retrospectively analyzed from 24 736 individual women (42.3% [10 455 of 24 736] white women, 57.7% [14 281 of 24 736] black women; mean age, 56.3 years; age range, 40.0-74.9 years) who underwent from one to seven mammographic screening examinations from September 2010 through February 2017 (60 766 examinations). Three screening modalities were used: digital mammography alone (8935 examinations); digital mammography with digital breast tomosynthesis (DBT; 30 779 examinations); and synthetic mammography with DBT (21 052 examinations). Random-effects logistic regression analysis was performed to estimate the likelihood of assignment to high versus low BI-RADS density category according to each modality, adjusted for ethnicity, age, body mass index (BMI), and radiologist. The interactions of modality with ethnicity and BMI on density categorization were also tested with the model. Results Women screened with DBT versus digital mammography alone had lower likelihood regarding categorization of high density breasts (digital mammography and DBT vs digital mammography: odds ratio, 0.69 [95% confidence interval: 0.61, 0.80], P < .001; synthetic mammography and DBT vs digital mammography: odds ratio, 0.43 [95% confidence interval: 0.37, 0.50], P < .001). Lower likelihood of high density was also observed at synthetic mammography and DBT compared with digital mammography and DBT (odds ratio, 0.62; 95% confidence interval: 0.56, 0.69; P < .001). There were interactions of modality with ethnicity (P = .007) and BMI (P = .003) on breast density assessment, with greater differences in density categorization according to modality observed for black women than for white women and groups with higher BMI. Conclusion Breast density categorization may vary by screening mammographic modality, and this effect appears to vary by ethnicity and body mass index. © RSNA, 2019 Online supplemental material is available for this article. See also the editorial by Philpotts in this issue.

Radiomic Phenotypes of Mammographic Parenchymal Complexity: Toward Augmenting Breast Density in Breast Cancer Risk Assessment.

Purpose To identify phenotypes of mammographic parenchymal complexity by using radiomic features and to evaluate their associations with breast density and other breast cancer risk factors. Materials and Methods Computerized image analysis was used to quantify breast density and extract parenchymal texture features in a cross-sectional sample of women screened with digital mammography from September 1, 2012, to February 28, 2013 (n = 2029; age range, 35-75 years; mean age, 55.9 years). Unsupervised clustering was applied to identify and reproduce phenotypes of parenchymal complexity in separate training (n = 1339) and test sets (n = 690). Differences across phenotypes by age, body mass index, breast density, and estimated breast cancer risk were assessed by using Fisher exact, χ2, and Kruskal-Wallis tests. Conditional logistic regression was used to evaluate preliminary associations between the detected phenotypes and breast cancer in an independent case-control sample (76 women diagnosed with breast cancer and 158 control participants) matched on age. Results Unsupervised clustering in the screening sample identified four phenotypes with increasing parenchymal complexity that were reproducible between training and test sets (P = .001). Breast density was not strongly correlated with phenotype category (R2 = 0.24 for linear trend). The low- to intermediate-complexity phenotype (prevalence, 390 of 2029 [19%]) had the lowest proportion of dense breasts (eight of 390 [2.1%]), whereas similar proportions were observed across other phenotypes (from 140 of 291 [48.1%] in the high-complexity phenotype to 275 of 511 [53.8%] in the low-complexity phenotype). In the independent case-control sample, phenotypes showed a significant association with breast cancer (P = .001), resulting in higher discriminatory capacity when added to a model with breast density and body mass index (area under the curve, 0.84 vs 0.80; P = .03 for comparison). Conclusion Radiomic phenotypes capture mammographic parenchymal complexity beyond conventional breast density measures and established breast cancer risk factors. © RSNA, 2018 Online supplemental material is available for this article. See also the editorial by Pinker in this issue.

Three-Dimensional Whole Breast Segmentation in Sagittal and Axial Breast MRI With Dense Depth Field Modeling and Localized Self-Adaptation for Chest-Wall Line Detection.

OBJECTIVE: Whole breast segmentation is an essential task in quantitative analysis of breast MRI for cancer risk assessment. It is challenging, mainly, because the chest-wall line (CWL) can be very difficult to locate due to its spatially varying appearance-caused by both nature and imaging artifacts-and neighboring distracting structures. This paper proposes an automatic three-dimensional (3-D) segmentation method, termed DeepSeA, of whole breast for breast MRI. METHODS: DeepSeA distinguishes itself from previous methods in three aspects. First, it reformulates the challenging problem of CWL localization as an equivalent problem that optimizes a smooth depth field and so fully utilizes the CWL's 3-D continuity. Second, it employs a localized self-adapting algorithm to adjust to the CWL's spatial variation. Third, it applies to breast MRI data in both sagittal and axial orientations equally well without training. RESULTS: A representative set of 99 breast MRI scans with varying imaging protocols is used for evaluation. Experimental results with expert-outlined reference standard show that DeepSeA can segment breasts accurately: the average Dice similarity coefficients, sensitivity, specificity, and CWL deviation error are 96.04%, 97.27%, 98.77%, and 1.63 mm, respectively. In addition, the configuration of DeepSeA is generalized based on experimental findings, for application to broad prospective data. CONCLUSION: A fully automatic method-DeepSeA-for whole breast segmentation in sagittal and axial breast MRI is reported. SIGNIFICANCE: DeepSeA can facilitate cancer risk assessment with breast MRI.