Purpose:Identifying imaging phenotypes and understanding their relationship with prognostic marke... more 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.
Rationale and Objectives-We investigate associations between measures of mammographic parenchymal... more Rationale and Objectives-We investigate associations between measures of mammographic parenchymal complexity and false-positive (FP) recall from screening with digital mammography (DM) versus digital breast tomosynthesis (DBT). Materials and Methods-We retrospectively analyzed data from 541 women recruited by the ACRIN 4006 trial, designed to evaluate callback and detection rates from screening with DM versus combined DM and DBT. Of these, 68 and 56 were false-positives based on DM alone versus the combined DM/DBT readings, respectively. Mammographic complexity was quantified with computerized texture analysis and percent density (PD). Logistic regression was performed to evaluate associations between extracted features and FP recall, after adjusting for age and number of previous benign biopsies. Odds ratios (OR) and area under the curve (AUC) of the receiver operating characteristic (ROC) were used to assess association strength. Results-For DM, age, previous benign biopsies and texture features of correlation, inverse difference moment, sum average, and sum variance were deemed as significant predictors (p<0.05) of FP recall, with an AUC=0.77. For DBT, age was the only significant predictor of FP recall with AUC=0.64. Using exploratory ROC thresholds for which no true-positives would be missed, a potential FP reduction of 23.5% and 8.9% was demonstrated, respectively, for DM alone versus DM/DBT.
Identifying imaging phenotypes and understanding their relationship with prognostic markers and p... more 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 recurrencefree survival (RFS) across phenotypes. Radiomic phenotypes demonstrated added prognostic value (c ¼ 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.
The National Institutes of Health approved the study design and analysis but had no role in data ... more The National Institutes of Health approved the study design and analysis but had no role in data collection, analysis, interpretation, or drafting or reviewing the article. S.P.W. had full access to all the data in the study and had final responsibility for all decisions to submit for publication.
Journal of The American College of Radiology, Aug 1, 2019
Purpose: To optimize operations for a newly implemented abbreviated MR (AB-MR) breast cancer scre... more Purpose: To optimize operations for a newly implemented abbreviated MR (AB-MR) breast cancer screening protocol, activity times were compared with the full-protocol examination. Methods: Activity times from 70 AB-MR and 736 full MR screening studies were analyzed. Total study time was measured from opening to closing examinations and expected scan time by summing the time of acquisition for each imaging series for either protocol. Actual scan time was obtained from DICOM headers. Total technologist activity time was obtained by subtracting expected scan time from total study time. Technologist activity time included both scan-related and non-scan-related activities. Results: The actual scan time for the AB protocol was 17.5 AE 0.5 min, compared with 28.8 AE 0.3 min (mean AE SE) for the full protocol (difference, 11.3 min; P < .0001). The total study time was 36.0 AE 3.2 min for AB-MRI and 49.7 AE 0.8 min for the full protocol (difference, 13.7 min; P < .0001), implying that the AB-MR protocol had only a 38% greater patient flow rate than the full protocol. Conclusions: The realized gains in patient flow rate were lower than expected based on scan times (65%) because of increased technologist activity time for the AB-MR protocol.
Supplementary data includes appendix of extracted Radiomic features. Figure S1 shows a summary of... more Supplementary data includes appendix of extracted Radiomic features. Figure S1 shows a summary of patient characteristics. Figure S2 shows the radiomic analysis workflow. Figure S3 shows representative cases from heterogeneity phenotypes. Figure S4 shows independent validation of intrinsic imaging phenotypes of tumor heterogeneity.
In this paper, radiomic features are used to validate the textural realism of two anthropomorphic... more 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
Imaging plays a central role in the evaluation of breast tumor response to neoadjuvant chemothera... more Imaging plays a central role in the evaluation of breast tumor response to neoadjuvant chemotherapy. Image-based assessment of tumor change via deformable registration is a powerful, quantitative method potentially to explore novel information of tumor heterogeneity, structure, function, and treatment response. In this study, we continued a previous pilot study to further validate the feasibility of an open source deformable registration algorithm DRAMMS developed within our group as a means to analyze spatio-temporal tumor changes for a set of 14 patients with DCE-MR imaging. Two experienced breast imaging radiologists marked landmarks according to their anatomical meaning on image sets acquired before and during chemotherapy. Yet, chemotherapy remarkably changed the anatomical structure of both tumor and normal breast tissue, leading to significant discrepancies between both raters for landmarks in certain areas. Therefore, we proposed a novel method to grade the manually denoted landmarks into different challenge levels based on the inter-rater agreement, where a high level indicates significant discrepancies and considerable amounts of anatomical structure changes, which would indeed impose giant problem for the following registration algorithm. It is interesting to observe that DRAMMS performed in a similar manner as the human raters: landmark errors increased as inter-rater differences rose. Among all selected six deformable registration algorithms, DRAMMS achieves the highest overall accuracy, which is around 5.5 mm, while the average difference between human raters is 3 mm. Moreover, DRAMMS performed consistently well within both tumor and normal tissue regions. Lastly, we comprehensively tuned the fundamental parameters of DRAMMS to better understand DRAMMS to guide similar works in the future. Overall, we further validated that DRAMMS is a powerful registration tool to accurately quantify tumor changes and potentially predict early tumor response to chemotherapy. Therefore, future studies that aim at examining if DRAMMS can generate valuable biomarkers for tumor response prediction during chemotherapy become feasible.
Background Improved breast cancer risk assessment models are needed to enable personalized screen... more Background Improved breast cancer risk assessment models are needed to enable personalized screening strategies that achieve better harm-to-benefit ratio based on earlier detection and better breast cancer outcomes than existing screening guidelines. Computational mammographic phenotypes have demonstrated a promising role in breast cancer risk prediction. With the recent exponential growth of computational efficiency, the artificial intelligence (AI) revolution, driven by the introduction of deep learning, has expanded the utility of imaging in predictive models. Consequently, AI-based imaging-derived data has led to some of the most promising tools for precision breast cancer screening. Main body This review aims to synthesize the current state-of-the-art applications of AI in mammographic phenotyping of breast cancer risk. We discuss the fundamentals of AI and explore the computing advancements that have made AI-based image analysis essential in refining breast cancer risk assessm...
Digital mammography has seen an explosion in the number of radiomic features used for risk-assess... more 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 scre...
AIM-To compare the utility of breast magnetic resonance imaging (MRI) in determining the extent o... more AIM-To compare the utility of breast magnetic resonance imaging (MRI) in determining the extent of disease in patients with newly diagnosed breast cancer detected on combination digital breast tomosynthesis (DBT) versus digital screening mammography (DM). MATERIALS AND METHODS-Review of 24,563 DBT-screened patients and 10,751 DMscreened patients was performed. Two hundred and thirty-five DBT patients underwent subsequent MRI examinations; 82 to determine extent of disease after newly diagnosed breast cancer. Eightythree DM patients underwent subsequent MRI examinations; 23 to determine extent of disease. MRI examinations performed to assess disease extent were considered true positives if additional disease was discovered in the contralateral breast or >2 cm away from the index malignancy. Differences in cancer subtypes and MRI outcomes between the DM and DBT cohorts were compared using chi-squared tests and post-hoc Bonferroni-adjusted tests for equal proportions. RESULTS-No differences in cancer subtype findings were observed between the two cohorts; however, MRI outcomes were found to differ between the DBT and DM cohorts (p=0.024). Specifically, the DBT cohort had significantly (p=0.013) fewer true-positive findings (7/82, 8.5%) than did the DM cohort (7/23; 30%), whereas the false-positive rate was similar between the cohorts (not statistically significant). When stratifying by breast density, this difference in truepositive rates was primarily observed when evaluating women with non-dense breasts (p=0.001). CONCLUSION-In both the DM-and DBT-screened populations with new cancer diagnoses, MRI is able to detect additional cancer; however, in those patients who have DBT screen-detected cancers the positive impact of preoperative MRI is diminished, particularly in those women with non-dense breasts.
Medical Imaging 2013: Physics of Medical Imaging, 2013
Mammographic breast density is a strong risk factor for breast cancer. Studies on imaging dose in... more Mammographic breast density is a strong risk factor for breast cancer. Studies on imaging dose in mammography have primarily focused on imaging quality and diagnostic accuracy, while little work has been done on understanding its effect on the estimation of breast density. Studies on the effect of dose on mammographic density estimation can be useful in dose reduction for the purpose of density estimation and monitoring. In this study, we investigate the dependence of percent area (PD%) and volumetric (VD%) breast density estimation on imaging dose using an anthropomorphic breast phantom (Rachel, Gammex). A set of digital mammograms were obtained with a GE Senographe 2000D FFDM system, using 220 unique combinations of different imaging physics, namely target/filter, kVp and mAs. Specifically, 8 different mAs settings were defined as corresponding to 10%, 20%, 40%, 70%, 100%, 150%, 200% and 300% of 1.8 mGy reference average glandular dose (AGD) for standard phototimed exposure. Breast density was estimated using fully-automated FDA-cleared software (Quantra v.2.0, Hologic Inc.). The obtained estimates were analyzed to study the effect of the imaging dose, using ANOVA and linear regression. Results show that there is a statistically significant dependence of density estimation on x-ray imaging dose (p-value=0.014 and <0.001 for PD% and VD%, respectively), while the actual variation of the estimation across the different levels of dose is relatively low (standard deviation of 2.87% and 0.66% for PD% and VD% respectively), the differences could be significant when breast density measures are used for risk estimation.
Purpose To evaluate agreement between automated estimates of breast density made from standard-do... more Purpose To evaluate agreement between automated estimates of breast density made from standard-dose versus synthetic digital mammograms in a large cohort of women undergoing screening. Materials and Methods This study received institutional review board approval with waiver of consent. A total of 3668 negative (Breast Imaging Reporting and Data System category 1 or 2) digital breast tomosynthesis (DBT) screening examinations consecutively performed over a 4-month period at one institution for which both standard-dose and synthetic mammograms were available for analysis were retrospectively analyzed. All mammograms were acquired with a Selenia Dimensions system (Hologic, Bedford, Mass), and synthetic mammograms were generated by using the U.S. Food and Drug Administration-approved "C-View" software module. The "For Presentation" standard-dose mammograms and synthetic images were analyzed by using a fully automated algorithm. Agreement between density estimates was...
AJR. American journal of roentgenology, Jan 2, 2017
Synthesized 2D (s2D) mammography is rapidly replacing digital mammography in breast imaging with ... more Synthesized 2D (s2D) mammography is rapidly replacing digital mammography in breast imaging with digital breast tomosynthesis (DBT) to reduce radiation dose and maintain screening outcomes. We illustrate variations in the appearance of s2D and digital mammograms to aid in implementation of this technology. Despite subjective differences in the appearance of s2D and digital mammograms, early outcomes of screening using s2D mammography and DBT are not inferior to those achieved with digital mammography and DBT. Understanding these variations may aid in implementing this technique and improving patient outcomes.
Radiographics : a review publication of the Radiological Society of North America, Inc
Digital breast tomosynthesis (DBT) represents a valuable addition to breast cancer screening by d... more Digital breast tomosynthesis (DBT) represents a valuable addition to breast cancer screening by decreasing recall rates while increasing cancer detection rates. The increased accuracy achieved with DBT is due to the quasi-three-dimensional format of the reconstructed images and the ability to "scroll through" breast tissue in the reconstructed images, thereby reducing the effect of tissue superimposition found with conventional planar digital mammography. The margins of both benign and malignant lesions are more conspicuous at DBT, which allows improved lesion characterization, increased reader confidence, and improved screening outcomes. However, even with the improvements in accuracy achieved with DBT, there remain differences in breast cancer conspicuity by mammographic view. Early data suggest that breast cancers may be more conspicuous on craniocaudal (CC) views than on mediolateral oblique (MLO) views. While some very laterally located breast cancers may be visualize...
Rationale and Objectives-We investigate associations between measures of mammographic parenchymal... more Rationale and Objectives-We investigate associations between measures of mammographic parenchymal complexity and false-positive (FP) recall from screening with digital mammography (DM) versus digital breast tomosynthesis (DBT). Materials and Methods-We retrospectively analyzed data from 541 women recruited by the ACRIN 4006 trial, designed to evaluate callback and detection rates from screening with DM versus combined DM and DBT. Of these, 68 and 56 were false-positives based on DM alone versus the combined DM/DBT readings, respectively. Mammographic complexity was quantified with computerized texture analysis and percent density (PD). Logistic regression was performed to evaluate associations between extracted features and FP recall, after adjusting for age and number of previous benign biopsies. Odds ratios (OR) and area under the curve (AUC) of the receiver operating characteristic (ROC) were used to assess association strength. Results-For DM, age, previous benign biopsies and texture features of correlation, inverse difference moment, sum average, and sum variance were deemed as significant predictors (p<0.05) of FP recall, with an AUC=0.77. For DBT, age was the only significant predictor of FP recall with AUC=0.64. Using exploratory ROC thresholds for which no true-positives would be missed, a potential FP reduction of 23.5% and 8.9% was demonstrated, respectively, for DM alone versus DM/DBT.
Purpose:Identifying imaging phenotypes and understanding their relationship with prognostic marke... more 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 &lt; 0.01). Kaplan–Meier curves showed statistically significant differences (P &lt; 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.
Rationale and Objectives-We investigate associations between measures of mammographic parenchymal... more Rationale and Objectives-We investigate associations between measures of mammographic parenchymal complexity and false-positive (FP) recall from screening with digital mammography (DM) versus digital breast tomosynthesis (DBT). Materials and Methods-We retrospectively analyzed data from 541 women recruited by the ACRIN 4006 trial, designed to evaluate callback and detection rates from screening with DM versus combined DM and DBT. Of these, 68 and 56 were false-positives based on DM alone versus the combined DM/DBT readings, respectively. Mammographic complexity was quantified with computerized texture analysis and percent density (PD). Logistic regression was performed to evaluate associations between extracted features and FP recall, after adjusting for age and number of previous benign biopsies. Odds ratios (OR) and area under the curve (AUC) of the receiver operating characteristic (ROC) were used to assess association strength. Results-For DM, age, previous benign biopsies and texture features of correlation, inverse difference moment, sum average, and sum variance were deemed as significant predictors (p<0.05) of FP recall, with an AUC=0.77. For DBT, age was the only significant predictor of FP recall with AUC=0.64. Using exploratory ROC thresholds for which no true-positives would be missed, a potential FP reduction of 23.5% and 8.9% was demonstrated, respectively, for DM alone versus DM/DBT.
Identifying imaging phenotypes and understanding their relationship with prognostic markers and p... more 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 recurrencefree survival (RFS) across phenotypes. Radiomic phenotypes demonstrated added prognostic value (c ¼ 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.
The National Institutes of Health approved the study design and analysis but had no role in data ... more The National Institutes of Health approved the study design and analysis but had no role in data collection, analysis, interpretation, or drafting or reviewing the article. S.P.W. had full access to all the data in the study and had final responsibility for all decisions to submit for publication.
Journal of The American College of Radiology, Aug 1, 2019
Purpose: To optimize operations for a newly implemented abbreviated MR (AB-MR) breast cancer scre... more Purpose: To optimize operations for a newly implemented abbreviated MR (AB-MR) breast cancer screening protocol, activity times were compared with the full-protocol examination. Methods: Activity times from 70 AB-MR and 736 full MR screening studies were analyzed. Total study time was measured from opening to closing examinations and expected scan time by summing the time of acquisition for each imaging series for either protocol. Actual scan time was obtained from DICOM headers. Total technologist activity time was obtained by subtracting expected scan time from total study time. Technologist activity time included both scan-related and non-scan-related activities. Results: The actual scan time for the AB protocol was 17.5 AE 0.5 min, compared with 28.8 AE 0.3 min (mean AE SE) for the full protocol (difference, 11.3 min; P < .0001). The total study time was 36.0 AE 3.2 min for AB-MRI and 49.7 AE 0.8 min for the full protocol (difference, 13.7 min; P < .0001), implying that the AB-MR protocol had only a 38% greater patient flow rate than the full protocol. Conclusions: The realized gains in patient flow rate were lower than expected based on scan times (65%) because of increased technologist activity time for the AB-MR protocol.
Supplementary data includes appendix of extracted Radiomic features. Figure S1 shows a summary of... more Supplementary data includes appendix of extracted Radiomic features. Figure S1 shows a summary of patient characteristics. Figure S2 shows the radiomic analysis workflow. Figure S3 shows representative cases from heterogeneity phenotypes. Figure S4 shows independent validation of intrinsic imaging phenotypes of tumor heterogeneity.
In this paper, radiomic features are used to validate the textural realism of two anthropomorphic... more 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
Imaging plays a central role in the evaluation of breast tumor response to neoadjuvant chemothera... more Imaging plays a central role in the evaluation of breast tumor response to neoadjuvant chemotherapy. Image-based assessment of tumor change via deformable registration is a powerful, quantitative method potentially to explore novel information of tumor heterogeneity, structure, function, and treatment response. In this study, we continued a previous pilot study to further validate the feasibility of an open source deformable registration algorithm DRAMMS developed within our group as a means to analyze spatio-temporal tumor changes for a set of 14 patients with DCE-MR imaging. Two experienced breast imaging radiologists marked landmarks according to their anatomical meaning on image sets acquired before and during chemotherapy. Yet, chemotherapy remarkably changed the anatomical structure of both tumor and normal breast tissue, leading to significant discrepancies between both raters for landmarks in certain areas. Therefore, we proposed a novel method to grade the manually denoted landmarks into different challenge levels based on the inter-rater agreement, where a high level indicates significant discrepancies and considerable amounts of anatomical structure changes, which would indeed impose giant problem for the following registration algorithm. It is interesting to observe that DRAMMS performed in a similar manner as the human raters: landmark errors increased as inter-rater differences rose. Among all selected six deformable registration algorithms, DRAMMS achieves the highest overall accuracy, which is around 5.5 mm, while the average difference between human raters is 3 mm. Moreover, DRAMMS performed consistently well within both tumor and normal tissue regions. Lastly, we comprehensively tuned the fundamental parameters of DRAMMS to better understand DRAMMS to guide similar works in the future. Overall, we further validated that DRAMMS is a powerful registration tool to accurately quantify tumor changes and potentially predict early tumor response to chemotherapy. Therefore, future studies that aim at examining if DRAMMS can generate valuable biomarkers for tumor response prediction during chemotherapy become feasible.
Background Improved breast cancer risk assessment models are needed to enable personalized screen... more Background Improved breast cancer risk assessment models are needed to enable personalized screening strategies that achieve better harm-to-benefit ratio based on earlier detection and better breast cancer outcomes than existing screening guidelines. Computational mammographic phenotypes have demonstrated a promising role in breast cancer risk prediction. With the recent exponential growth of computational efficiency, the artificial intelligence (AI) revolution, driven by the introduction of deep learning, has expanded the utility of imaging in predictive models. Consequently, AI-based imaging-derived data has led to some of the most promising tools for precision breast cancer screening. Main body This review aims to synthesize the current state-of-the-art applications of AI in mammographic phenotyping of breast cancer risk. We discuss the fundamentals of AI and explore the computing advancements that have made AI-based image analysis essential in refining breast cancer risk assessm...
Digital mammography has seen an explosion in the number of radiomic features used for risk-assess... more 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 scre...
AIM-To compare the utility of breast magnetic resonance imaging (MRI) in determining the extent o... more AIM-To compare the utility of breast magnetic resonance imaging (MRI) in determining the extent of disease in patients with newly diagnosed breast cancer detected on combination digital breast tomosynthesis (DBT) versus digital screening mammography (DM). MATERIALS AND METHODS-Review of 24,563 DBT-screened patients and 10,751 DMscreened patients was performed. Two hundred and thirty-five DBT patients underwent subsequent MRI examinations; 82 to determine extent of disease after newly diagnosed breast cancer. Eightythree DM patients underwent subsequent MRI examinations; 23 to determine extent of disease. MRI examinations performed to assess disease extent were considered true positives if additional disease was discovered in the contralateral breast or >2 cm away from the index malignancy. Differences in cancer subtypes and MRI outcomes between the DM and DBT cohorts were compared using chi-squared tests and post-hoc Bonferroni-adjusted tests for equal proportions. RESULTS-No differences in cancer subtype findings were observed between the two cohorts; however, MRI outcomes were found to differ between the DBT and DM cohorts (p=0.024). Specifically, the DBT cohort had significantly (p=0.013) fewer true-positive findings (7/82, 8.5%) than did the DM cohort (7/23; 30%), whereas the false-positive rate was similar between the cohorts (not statistically significant). When stratifying by breast density, this difference in truepositive rates was primarily observed when evaluating women with non-dense breasts (p=0.001). CONCLUSION-In both the DM-and DBT-screened populations with new cancer diagnoses, MRI is able to detect additional cancer; however, in those patients who have DBT screen-detected cancers the positive impact of preoperative MRI is diminished, particularly in those women with non-dense breasts.
Medical Imaging 2013: Physics of Medical Imaging, 2013
Mammographic breast density is a strong risk factor for breast cancer. Studies on imaging dose in... more Mammographic breast density is a strong risk factor for breast cancer. Studies on imaging dose in mammography have primarily focused on imaging quality and diagnostic accuracy, while little work has been done on understanding its effect on the estimation of breast density. Studies on the effect of dose on mammographic density estimation can be useful in dose reduction for the purpose of density estimation and monitoring. In this study, we investigate the dependence of percent area (PD%) and volumetric (VD%) breast density estimation on imaging dose using an anthropomorphic breast phantom (Rachel, Gammex). A set of digital mammograms were obtained with a GE Senographe 2000D FFDM system, using 220 unique combinations of different imaging physics, namely target/filter, kVp and mAs. Specifically, 8 different mAs settings were defined as corresponding to 10%, 20%, 40%, 70%, 100%, 150%, 200% and 300% of 1.8 mGy reference average glandular dose (AGD) for standard phototimed exposure. Breast density was estimated using fully-automated FDA-cleared software (Quantra v.2.0, Hologic Inc.). The obtained estimates were analyzed to study the effect of the imaging dose, using ANOVA and linear regression. Results show that there is a statistically significant dependence of density estimation on x-ray imaging dose (p-value=0.014 and <0.001 for PD% and VD%, respectively), while the actual variation of the estimation across the different levels of dose is relatively low (standard deviation of 2.87% and 0.66% for PD% and VD% respectively), the differences could be significant when breast density measures are used for risk estimation.
Purpose To evaluate agreement between automated estimates of breast density made from standard-do... more Purpose To evaluate agreement between automated estimates of breast density made from standard-dose versus synthetic digital mammograms in a large cohort of women undergoing screening. Materials and Methods This study received institutional review board approval with waiver of consent. A total of 3668 negative (Breast Imaging Reporting and Data System category 1 or 2) digital breast tomosynthesis (DBT) screening examinations consecutively performed over a 4-month period at one institution for which both standard-dose and synthetic mammograms were available for analysis were retrospectively analyzed. All mammograms were acquired with a Selenia Dimensions system (Hologic, Bedford, Mass), and synthetic mammograms were generated by using the U.S. Food and Drug Administration-approved "C-View" software module. The "For Presentation" standard-dose mammograms and synthetic images were analyzed by using a fully automated algorithm. Agreement between density estimates was...
AJR. American journal of roentgenology, Jan 2, 2017
Synthesized 2D (s2D) mammography is rapidly replacing digital mammography in breast imaging with ... more Synthesized 2D (s2D) mammography is rapidly replacing digital mammography in breast imaging with digital breast tomosynthesis (DBT) to reduce radiation dose and maintain screening outcomes. We illustrate variations in the appearance of s2D and digital mammograms to aid in implementation of this technology. Despite subjective differences in the appearance of s2D and digital mammograms, early outcomes of screening using s2D mammography and DBT are not inferior to those achieved with digital mammography and DBT. Understanding these variations may aid in implementing this technique and improving patient outcomes.
Radiographics : a review publication of the Radiological Society of North America, Inc
Digital breast tomosynthesis (DBT) represents a valuable addition to breast cancer screening by d... more Digital breast tomosynthesis (DBT) represents a valuable addition to breast cancer screening by decreasing recall rates while increasing cancer detection rates. The increased accuracy achieved with DBT is due to the quasi-three-dimensional format of the reconstructed images and the ability to "scroll through" breast tissue in the reconstructed images, thereby reducing the effect of tissue superimposition found with conventional planar digital mammography. The margins of both benign and malignant lesions are more conspicuous at DBT, which allows improved lesion characterization, increased reader confidence, and improved screening outcomes. However, even with the improvements in accuracy achieved with DBT, there remain differences in breast cancer conspicuity by mammographic view. Early data suggest that breast cancers may be more conspicuous on craniocaudal (CC) views than on mediolateral oblique (MLO) views. While some very laterally located breast cancers may be visualize...
Rationale and Objectives-We investigate associations between measures of mammographic parenchymal... more Rationale and Objectives-We investigate associations between measures of mammographic parenchymal complexity and false-positive (FP) recall from screening with digital mammography (DM) versus digital breast tomosynthesis (DBT). Materials and Methods-We retrospectively analyzed data from 541 women recruited by the ACRIN 4006 trial, designed to evaluate callback and detection rates from screening with DM versus combined DM and DBT. Of these, 68 and 56 were false-positives based on DM alone versus the combined DM/DBT readings, respectively. Mammographic complexity was quantified with computerized texture analysis and percent density (PD). Logistic regression was performed to evaluate associations between extracted features and FP recall, after adjusting for age and number of previous benign biopsies. Odds ratios (OR) and area under the curve (AUC) of the receiver operating characteristic (ROC) were used to assess association strength. Results-For DM, age, previous benign biopsies and texture features of correlation, inverse difference moment, sum average, and sum variance were deemed as significant predictors (p<0.05) of FP recall, with an AUC=0.77. For DBT, age was the only significant predictor of FP recall with AUC=0.64. Using exploratory ROC thresholds for which no true-positives would be missed, a potential FP reduction of 23.5% and 8.9% was demonstrated, respectively, for DM alone versus DM/DBT.
Uploads
Papers by Emily Conant