Impact of Breast Density Legislation on Breast Cancer Risk Assessment and Supplemental Screening: A Survey of 110 Radiology Facilities.

Breast density notification laws, passed in 19 states as of October 2014, mandate that patients be informed of their breast density. The purpose of this study is to assess the impact of this legislation on radiology practices, including performance of breast cancer risk assessment and supplemental screening studies. A 20-question anonymous web-based survey was emailed to radiologists in the Society of Breast Imaging between August 2013 and March 2014. Statistical analysis was performed using Fisher's exact test. Around 121 radiologists from 110 facilities in 34 USA states and 1 Canadian site responded. About 50% (55/110) of facilities had breast density legislation, 36% of facilities (39/109) performed breast cancer risk assessment (one facility did not respond). Risk assessment was performed as a new task in response to density legislation in 40% (6/15) of facilities in states with notification laws. However, there was no significant difference in performing risk assessment between facilities in states with a law and those without (p < 0.831). In anticipation of breast density legislation, 33% (16/48), 6% (3/48), and 6% (3/48) of facilities in states with laws implemented handheld whole breast ultrasound (WBUS), automated WBUS, and tomosynthesis, respectively. The ratio of facilities offering handheld WBUS was significantly higher in states with a law than in states without (p < 0.001). In response to breast density legislation, more than 33% of facilities are offering supplemental screening with WBUS and tomosynthesis, and many are performing formal risk assessment for determining patient management.

Short-term outcomes of screening mammography using computer-aided detection: a population-based study of medicare enrollees.

BACKGROUND: Computer-aided detection (CAD) has rapidly diffused into screening mammography practice despite limited and conflicting data on its clinical effect. OBJECTIVE: To determine associations between CAD use during screening mammography and the incidence of ductal carcinoma in situ (DCIS) and invasive breast cancer, invasive cancer stage, and diagnostic testing. DESIGN: Retrospective cohort study. SETTING: Medicare program. PARTICIPANTS: Women aged 67 to 89 years having screening mammography between 2001 and 2006 in U.S. SEER (Surveillance, Epidemiology and End Results) regions (409 459 mammograms from 163 099 women). MEASUREMENTS: Incident DCIS and invasive breast cancer within 1 year after mammography, invasive cancer stage, and diagnostic testing within 90 days after screening among women without breast cancer. RESULTS: From 2001 to 2006, CAD prevalence increased from 3.6% to 60.5%. Use of CAD was associated with greater DCIS incidence (adjusted odds ratio [OR], 1.17 [95% CI, 1.11 to 1.23]) but no difference in invasive breast cancer incidence (adjusted OR, 1.00 [CI, 0.97 to 1.03]). Among women with invasive cancer, CAD was associated with greater likelihood of stage I to II versus III to IV cancer (adjusted OR, 1.27 [CI, 1.14 to 1.41]). In women without breast cancer, CAD was associated with increased odds of diagnostic mammography (adjusted OR, 1.28 [CI, 1.27 to 1.29]), breast ultrasonography (adjusted OR, 1.07 [CI, 1.06 to 1.09]), and breast biopsy (adjusted OR, 1.10 [CI, 1.08 to 1.12]). LIMITATION: Short follow-up for cancer stage, potential unmeasured confounding, and uncertain generalizability to younger women. CONCLUSION: Use of CAD during screening mammography among Medicare enrollees is associated with increased DCIS incidence, the diagnosis of invasive breast cancer at earlier stages, and increased diagnostic testing among women without breast cancer. PRIMARY FUNDING SOURCE: Center for Healthcare Policy and Research, University of California, Davis.

Level set segmentation of breast masses in contrast-enhanced dedicated breast CT and evaluation of stopping criteria.

Dedicated breast CT (bCT) produces high-resolution 3D tomographic images of the breast, fully resolving fibroglandular tissue structures within the breast and allowing for breast lesion detection and assessment in 3D. In order to enable quantitative analysis, such as volumetrics, automated lesion segmentation on bCT is highly desirable. In addition, accurate output from CAD (computer-aided detection/diagnosis) methods depends on sufficient segmentation of lesions. Thus, in this study, we present a 3D lesion segmentation method for breast masses in contrast-enhanced bCT images. The segmentation algorithm follows a two-step approach. First, 3D radial-gradient index segmentation is used to obtain a crude initial contour, which is then refined by a 3D level set-based active contour algorithm. The data set included contrast-enhanced bCT images from 33 patients containing 38 masses (25 malignant, 13 benign). The mass centers served as input to the algorithm. In this study, three criteria for stopping the contour evolution were compared, based on (1) the change of region volume, (2) the average intensity in the segmented region increase at each iteration, and (3) the rate of change of the average intensity inside and outside the segmented region. Lesion segmentation was evaluated by computing the overlap ratio between computer segmentations and manually drawn lesion outlines. For each lesion, the overlap ratio was averaged across coronal, sagittal, and axial planes. The average overlap ratios for the three stopping criteria ranged from 0.66 to 0.68 (dice coefficient of 0.80 to 0.81), indicating that the proposed segmentation procedure is promising for use in quantitative dedicated bCT analyses.

The California breast density information group: a collaborative response to the issues of breast density, breast cancer risk, and breast density notification legislation.

In anticipation of breast density notification legislation in the state of California, which would require notification of women with heterogeneously and extremely dense breast tissue, a working group of breast imagers and breast cancer risk specialists was formed to provide a common response framework. The California Breast Density Information Group identified key elements and implications of the law, researching scientific evidence needed to develop a robust response. In particular, issues of risk associated with dense breast tissue, masking of cancers by dense tissue on mammograms, and the efficacy, benefits, and harms of supplementary screening tests were studied and consensus reached. National guidelines and peer-reviewed published literature were used to recommend that women with dense breast tissue at screening mammography follow supplemental screening guidelines based on breast cancer risk assessment. The goal of developing educational materials for referring clinicians and patients was reached with the construction of an easily accessible Web site that contains information about breast density, breast cancer risk assessment, and supplementary imaging. This multi-institutional, multidisciplinary approach may be useful for organizations to frame responses as similar legislation is passed across the United States. Online supplemental material is available for this article.

Anatomical complexity in breast parenchyma and its implications for optimal breast imaging strategies.

PURPOSE: The purpose of this investigation was to assess the anatomical noise in breast images using a mathematically derived parameter ß as a surrogate for detection performance, across the same patient cohort but in different imaging modalities including mammography, tomosynthesis, and breast CT. METHODS: Women who were scheduled for breast biopsy were approached for participation in this IRB and HIPPA-compliant investigation. A total of 23 women had all views of each modality and represent the cohort studied in this investigation. Image data sets across all modalities were analyzed using 1000 regions of interest per image data set, and the anatomical noise power spectrum, NPS(a)(f), was computed and averaged for each breast image data set. After windowing the total noise power spectrum NPS(t)(f) to a specific frequency range corresponding to anatomical noise, the power-law slope (ß) of the NPS(a)(f) was computed where NPS(a)(f) = α f(-) (ß). RESULTS: The value of ß was determined for breast CT data sets, and they were 1.75 (0.424), 1.83 (0.352), and 1.79 (0.397), for the coronal, sagittal, and axial views, respectively. For tomosynthesis, ß was 3.06 (0.361) and 3.10 (0.315) for the craniocaudal (CC) and medial lateral oblique (MLO) views, respectively. For mammography, these values were 3.17 (0.226) and 3.30 (0.236), for the CC and MLO views, respectively. The values of ß for breast CT were significantly different than those for tomosynthesis and mammography (p < 0.001, all 12 comparisons). CONCLUSIONS: Based on the parameter ß which is thought to describe anatomical noise in breast images, breast CT was shown to have a statistically significant lower ß than mammography or tomosynthesis. It has been suggested in the literature that a lower ß may correspond to increased cancer detection performance; however, this has yet to be demonstrated unequivocally.

The characterization of breast anatomical metrics using dedicated breast CT.

PURPOSE: Accurate anatomical characterization of the breast is useful in breast phantom development and computer modeling of breast imaging technologies. Capitalizing on the three-dimensional capabilities of dedicated breast CT (bCT), a number of parameters which describe breast shape and fibroglandular distribution are defined. METHODS: Among 219 bCT data sets, the effective diameter and length of the pendant breast as well as the breast volume were measured and characterized for each bra cup size. The volume glandular fraction (VGF) was determined as a function of patient age, BIRADS density, bra cup size, and breast diameter. The glandular fraction was examined in coronal and sagittal planes of the breast, and the radial distribution of breast glandular fraction within a coronal bCT image was examined for three breast regions. The areal glandular fraction (AGF) was estimated from two-dimensional projections of the breast (simulated by projecting bCT data sets) and was compared to the corresponding VGF. RESULTS: The effective breast diameter and length increase with increasing bra cup size. The mean breast diameters (+/- standard error) of bra cup sizes A/AA, B, C, and D/DD were 11.1 +/- 0.5, 11.4 +/- 0.3, 13.0 +/- 0.2, and 13.7 +/- 0.2 cm, respectively. VGF was lower among older women and those with larger breast diameter and larger bra cup size. VGF increased as a function of the reported BIRADS density. AGF increased with VGF. Fibroglandular tissue was distributed primarily in the central portion of the breast. CONCLUSIONS: Breast metrics were examined and a number of parameters were defined which may be useful for breast modeling. The reported data may provide researchers with useful information for characterizing the breast for various imaging or dosimetry tasks.

Contrast-enhanced dedicated breast CT: initial clinical experience.

PURPOSE: To quantify contrast material enhancement of breast lesions scanned with dedicated breast computed tomography (CT) and to compare their conspicuity with that at unenhanced breast CT and mammography. MATERIALS AND METHODS: Approval of the institutional review board and the Radiation Use Committee and written informed consent were obtained for this HIPAA-compliant study. Between September 2006 and April 2009, 46 women (mean age, 53.2 years; age range, 35-72 years) with Breast Imaging Reporting and Data System category 4 or 5 lesions underwent unenhanced breast CT and contrast material-enhanced breast CT before biopsy. Two radiologists independently scored lesion conspicuity for contrast-enhanced breast CT versus mammography and for contrast-enhanced breast CT versus unenhanced breast CT. Mean lesion voxel intensity was measured in Hounsfield units and normalized to adipose tissue intensity on manually segmented images obtained before and after administration of contrast material. Regression models focused on conspicuity and quantified enhancement were used to estimate the effect of pathologic diagnosis (benign vs malignant), lesion type (mass vs calcifications), breast density, and interradiologist variability. RESULTS: Fifty-four lesions (25 benign, 29 malignant) in 46 subjects were analyzed. Malignant lesions were seen significantly better at contrast-enhanced breast CT than at unenhanced breast CT (P < .001) or mammography (P < .001). Malignant calcifications (malignant lesions manifested mammographically as microcalcifications only, n = 7) were seen better at contrast-enhanced breast CT than at unenhanced breast CT (P < .001) and were seen similarly at contrast-enhanced breast CT and mammography. Malignant lesions enhanced 55.9 HU +/- 4.0 (standard error), whereas benign lesions enhanced 17.6 HU +/- 6.1 (P < .001). Ductal carcinoma in situ (n = 5) enhanced a mean of 59.6 HU +/- 2.8. Receiver operating characteristic curve analysis of lesion enhancement yielded an area under the receiver operating characteristic curve of 0.876. CONCLUSION: Conspicuity of malignant breast lesions, including ductal carcinoma in situ, is significantly improved at contrast-enhanced breast CT. Quantifying lesion enhancement may aid in the detection and diagnosis of breast cancer.

Initial characterization of a dedicated breast PET/CT scanner during human imaging.

UNLABELLED: We have constructed a dedicated breast PET/CT scanner capable of high-resolution functional and anatomic imaging. Here, we present an initial characterization of scanner performance during patient imaging. METHODS: The system consisted of a lutetium oxyorthosilicate-based dual-planar head PET camera (crystal size, 3 x 3 x 20 mm) and 768-slice cone-beam CT. The position of the PET heads (separation and height) could be adjusted for varying breast dimensions. For scanning, the patient lay prone on a specialized bed and inserted a single pendent breast through an aperture in the table top. Compression of the breast as used in mammography is not required. PET and CT systems rotate in the coronal plane underneath the patient sequentially to collect fully tomographic datasets. PET images were reconstructed with the fully 3-dimensional maximum a posteriori method, and CT images were reconstructed with the Feldkamp algorithm, then spatially registered and fused for display. Phantom scans were obtained to assess the registration accuracy between PET and CT images and the influence of PET electronics and activity on CT image quality. We imaged 4 women with mammographic findings highly suggestive of breast cancer (breast imaging reporting and data system, category 5) in an ongoing clinical trial. Patients were injected with (18)F-FDG and imaged for 12.5 min per breast. From patient data, noise-equivalent counting rates and the singles-to-trues ratio (a surrogate for the randoms fraction) were calculated. RESULTS: The average registration error between PET and CT images was 0.18 mm. PET electronics and activity did not significantly affect CT image quality. For the patient trial, biopsy-confirmed cancers were visualized on dedicated breast PET/CT on all patient scans, including the detection of ductal carcinoma in situ in 1 case. The singles-to-trues ratio was found to be inversely correlated with breast volume in the field of view, suggesting that larger breasts trend toward increased noise-equivalent counting rates for all other things equal. CONCLUSION: Scanning of the uncompressed breast with dedicated breast PET/CT can accurately visualize suspected lesions in 3 dimensions.

Conspicuity of suspicious breast lesions on contrast enhanced breast CT compared to digital breast tomosynthesis and mammography.

OBJECTIVE: Compare conspicuity of suspicious breast lesions on contrast-enhanced dedicated breast CT (CEbCT), tomosynthesis (DBT) and digital mammography (DM). METHODS: 100 females with BI-RADS 4/5 lesions underwent CEbCT and/or DBT prior to biopsy in this IRB approved, HIPAA compliant study. Two breast radiologists adjudicated lesion conspicuity scores (CS) for each modality independently. Data are shown as mean CS ±standard deviation. Two-sided t-test was used to determine significance between two modalities within each subgroup. Multiple comparisons were controlled by the false-discovery rate set to 5%. RESULTS: 50% of studied lesions were biopsy-confirmed malignancies. Malignant masses were more conspicuous on CEbCT than on DBT or DM (9.7 ±0.5, n = 25; 6.8 ± 3.1, n = 15; 6.7 ± 3.0, n = 27; p < 0.05). Malignant calcifications were equally conspicuous on all three modalities (CEbCT 8.7 ± 0.8, n = 18; DBT 8.5 ± 0.6, n = 15; DM 8.8 ± 0.7, n = 23; p = NS). Benign masses were equally conspicuous on CEbCT (6.6 ± 4.1, n = 22); DBT (6.4 ± 3.8, n = 17); DM (5.9 ± 3.6, n = 24; p = NS). Benign calcifications CS were similar between DBT (8.5 ± 1.0, n = 17) and DM (8.8 ± 0.8, n = 26; p = NS) but less conspicuous on CEbCT (4.0 ± 2.9, n = 25, p < 0.001). 55 females were imaged with all modalities. Results paralleled the entire cohort. 69%(n = 62) of females imaged by CEbCT had dense breasts. Benign/malignant lesion CSs in dense/non-dense categories were 4.8 ± 3.7, n = 33, vs 6.0 ± 3.9, n = 14, p = 0.35; 9.2 ± 0.9, n = 29 vs. 9.4 ± 0.7, n = 14; p = 0.29, respectively. CONCLUSION: Malignant masses are more conspicuous on CEbCT than DM or DBT. Malignant microcalcifications are equally conspicuous on all three modalities. Benign calcifications remain better visualized by DM and DBT than with CEbCT. We observed no differences in benign masses on all modalities. CS of both benign and malignant lesions were independent of breast density. ADVANCES IN KNOWLEDGE: CEbCT is a promising diagnostic imaging modality for suspicious breast lesions.

Dedicated breast CT: initial clinical experience.

PURPOSE: To prospectively and intraindividually compare dedicated breast computed tomographic (CT) images with screen-film mammograms. MATERIALS AND METHODS: All patient studies were performed according to protocols approved by the institutional review board and Radiation Use Committee; informed consent was obtained. A breast CT scanner prototype was used to individually scan uncompressed breasts in 10 healthy volunteers (mean age, 52.1 years) and 69 women with Breast Imaging Reporting and Data System category 4 and 5 lesions (mean age, 54.4 years). In women with lesions, breast CT images were compared with screen-film mammograms by an experienced mammographer and ranked with a continuous scale of 1-10 (score 1, excellent lesion visualization with CT and poor visualization with mammography; score 5.5, equal visualization with both modalities; and score 10, poor visualization with CT and excellent visualization with mammography). A Wilcoxon signed rank procedure was used to test the null hypothesis that ratings were symmetric at about a score of 5.5 for the entire group and for distinguishing microcalcifications versus masses and other findings and benign versus malignant lesions and for effect of breast density on lesion visualization. Women were asked to compare their comfort during CT with that during mammography on a continuous scale of 1-10. With a Wilcoxon signed rank procedure, the null hypothesis that comfort ratings were symmetric about a score of 5.5 (equal comfort with CT and mammography) was tested. RESULTS: Overall, CT was equal to mammography for visualization of breast lesions. Breast CT was significantly better than mammography for visualization of masses (P = .002); mammography outperformed CT for visualization of microcalcifications (P = .006). No significant differences between CT and mammography were seen among benign versus malignant lesions or for effect of breast density on lesion visualization. Subjects found CT significantly more comfortable than mammography (P < .001). CONCLUSION: Some technical challenges remain, but breast CT is promising and may have potential clinical applications.

Classification of breast computed tomography data.

Differences in breast tissue composition are important determinants in assessing risk, identifying disease in images and following changes over time. This paper presents an algorithm for tissue classification that separates breast tissue into its three primary constituents of skin, fat and glandular tissue. We have designed and built a dedicated breast CT scanner. Fifty-five normal volunteers and patients with mammographically identified breast lesions were scanned. Breast CT voxel data were filtered using a 5 pt median filter and the image histogram was computed. A two compartment Gaussian fit of histogram data was used to provide an initial estimate of tissue compartments. After histogram analysis, data were input to region-growing algorithms and classified as to belonging to skin, fat or gland based on their value and architectural features. Once tissues were classified, a more detailed analysis of glandular tissue patterns and a more quantitative analysis of breast composition was made. Algorithm performance assessment demonstrated very good or excellent agreement between algorithm and radiologist observers in 97.7% of the segmented data. We observed that even in dense breasts the fraction of glandular tissue seldom exceeded 50%. For most individuals the composition is better characterized as being a 70% (fat)-30% (gland) composition than a 50% (fat)-50% (gland) composition.

Observer efficiency in discrimination tasks simulating malignant and benign breast lesions imaged with ultrasound.

We investigate and extend the ideal observer methodology developed by Smith and Wagner to detection and discrimination tasks related to breast sonography. We provide a numerical approach for evaluating the ideal observer acting on radio frequency (RF) frame data, which involves inversion of large nonstationary covariance matrices, and we describe a power-series approach to computing this inverse. Considering a truncated power series suggests that the RF data be Wiener-filtered before forming the final envelope image. We have compared human performance for Wiener-filtered and conventional B-mode envelope images using psychophysical studies for 5 tasks related to breast cancer classification. We find significant improvements in visual detection and discrimination efficiency in four of these five tasks. We also use the Smith-Wagner approach to distinguish between human and processing inefficiencies, and find that generally the principle limitation comes from the information lost in computing the final envelope image.

Differentiation of ductal carcinoma in-situ from benign micro-calcifications by dedicated breast computed tomography.

PURPOSE: Compare conspicuity of ductal carcinoma in-situ (DCIS) to benign calcifications on unenhanced (bCT), contrast-enhanced dedicated breast CT (CEbCT) and mammography (DM). METHODS AND MATERIALS: The institutional review board approved this HIPAA-compliant study. 42 women with Breast Imaging Reporting and Data System 4 or 5 category micro-calcifications had breast CT before biopsy. Three subjects with invasive disease at surgery were excluded. Two breast radiologists independently compared lesion conspicuity scores (CS) for CEbCT, to bCT and DM. Enhancement was measured in Hounsfield units (HU). Mean CS ± standard deviations are shown. Receiver operating characteristic analysis (ROC) measured radiologists' discrimination performance by comparing CS to enhancement alone. Statistical measurements were made using ANOVA F-test, Wilcoxon rank-sum test and robust linear regression analyses. RESULTS: 39 lesions (17 DCIS, 22 benign) were analyzed. DCIS (8.5 ± 0.9, n=17) was more conspicuous than benign micro-calcifications (3.6 ± 2.9, n=22; p<0.0001) on CEbCT. DCIS was equally conspicuous on CEbCT and DM (8.5 ± 0.9, 8.7 ± 0.8, n=17; p=0.85) and more conspicuous when compared to bCT (5.3 ± 2.6, n=17; p<0.001). All DCIS enhanced; mean enhancement (90HU ± 53HU, n=17) was higher compared to benign lesions (33 ± 30HU, n=22) (p<0.0001). ROC analysis of the radiologists' CS showed high discrimination performance (AUC=0.94) compared to enhancement alone (AUC=0.85) (p<0.026). CONCLUSION: DCIS is more conspicuous than benign micro-calcifications on CEbCT. DCIS visualization on CEbCT is equal to mammography but improved compared to bCT. Radiologists' discrimination performance using CEBCT is significantly higher than enhancement values alone. CEbCT may have an advantage over mammography by reducing false positive examinations when calcifications are analyzed.

Technique factors and their relationship to radiation dose in pendant geometry breast CT.

The use of breast computed tomography (CT) as an alternative to mammography in some patients is being studied at several institutions. However, the radiation dosimetry issues associated with breast CT are markedly different than in the case of mammography. In this study, the spectral properties of an operational breast CT scanner were characterized both by physical measurement and computer modeling of the kVp-dependent spectra, from 40 to 110 kVp (Be window W anode with 0.30 mm added Cu filtration). Previously reported conversion factors, normalized glandular dose for CT-DgN(ct), derived from Monte Carlo methods, were used in concert with the output spectra of the breast scanner to compute the mean glandular dose to the breast based upon different combinations of x-ray technique factors (kVp and mAs). The mean glandular dose (MGD) was measured as a function of the compressed breast thickness (2-8 cm) and three different breast compositions (0%, 50%, and 100% glandular fractions) in four clinical mammography systems in our institution. The average MGD from these four systems was used to compute the technique factors for breast CT systems that would match the two-view mammographic dose levels. For a 14 cm diameter breast (equivalent to a 5 cm thick compressed breast in mammography), air kerma levels at the breast CT scanner's isocenter (468 mm from the source) of 4.4, 6.4, and 9.0 mGy were found to deliver equivalent mammography doses for 0%, 50%, and 100% glandular breasts (respectively) at 80 kVp. At 80 kVp (where air kerma was 11.3 mGy/100 mAs at the isocenter), 57 mAs (integrated over the entire scan) was required to match the mammography dose for a 14 cm 50% glandular breast. At 50 kVp, 360 mAs is required to match mammographic dose levels. Tables are provided for both air kerma at the isocenter and mAs for 0%, 50%, and 100% glandular breasts. Other issues that impact breast CT technique factors are also discussed.

Impact of the California breast density law on primary care physicians.

PURPOSE: To investigate primary physician awareness of the California Breast Density Notification Law and its impact on primary care practice. METHODS: An online survey was distributed to 174 physicians within a single primary care network system 10 months after California's breast density notification law took effect. The survey assessed physicians' awareness of the law, perceived changes in patient levels of concern about breast density, and physician comfort levels in handling breast density management issues. RESULTS: The survey was completed by 77 physicians (45%). Roughly half of those surveyed (49%) reported no knowledge of the breast density notification legislation. Only 32% of respondents noted an increase in patient levels of concern about breast density compared to prior years. The majority were only "somewhat comfortable" (55%) or "not comfortable" (12%) with breast density questions, and almost one-third (32%) had referred patients to a breast health clinic for these discussions. A total of 75% of those surveyed would be interested in more specific education on the subject. CONCLUSIONS: Awareness among primary care clinicians of the California Breast Density Notification Law is low, and many do not feel comfortable answering breast density-related patient questions. Breast imagers and institutions may need to devote additional time and resources to primary physician education in order for density notification laws to have significant impact on patient care.

Local curvature analysis for classifying breast tumors: Preliminary analysis in dedicated breast CT.

PURPOSE: The purpose of this study is to measure the effectiveness of local curvature measures as novel image features for classifying breast tumors. METHODS: A total of 119 breast lesions from 104 noncontrast dedicated breast computed tomography images of women were used in this study. Volumetric segmentation was done using a seed-based segmentation algorithm and then a triangulated surface was extracted from the resulting segmentation. Total, mean, and Gaussian curvatures were then computed. Normalized curvatures were used as classification features. In addition, traditional image features were also extracted and a forward feature selection scheme was used to select the optimal feature set. Logistic regression was used as a classifier and leave-one-out cross-validation was utilized to evaluate the classification performances of the features. The area under the receiver operating characteristic curve (AUC, area under curve) was used as a figure of merit. RESULTS: Among curvature measures, the normalized total curvature (CT) showed the best classification performance (AUC of 0.74), while the others showed no classification power individually. Five traditional image features (two shape, two margin, and one texture descriptors) were selected via the feature selection scheme and its resulting classifier achieved an AUC of 0.83. Among those five features, the radial gradient index (RGI), which is a margin descriptor, showed the best classification performance (AUC of 0.73). A classifier combining RGI and CT yielded an AUC of 0.81, which showed similar performance (i.e., no statistically significant difference) to the classifier with the above five traditional image features. Additional comparisons in AUC values between classifiers using different combinations of traditional image features and CT were conducted. The results showed that CT was able to replace the other four image features for the classification task. CONCLUSIONS: The normalized curvature measure contains useful information in classifying breast tumors. Using this, one can reduce the number of features in a classifier, which may result in more robust classifiers for different datasets.

Determining sensitivity of mammography from screening data, cancer incidence, and receiver-operating characteristic curve parameters.

OBJECTIVES: A mathematical model is presented that allows the computation of the sensitivity and specificity of breast screening based on receiver-operating characteristic (ROC) curve shape, the positive predictive value (PPV) of screening mammography, and the cancer incidence, f. METHODS: The normal and cancer populations are modeled as normal distributions with independent means and standard deviations. The distributions are scaled such that the area of the normal population is equal to 1-f and that of the cancer population is f. The PPV for screening mammography is used to determine the operating point on the ROC curve. Knowing this leads directly to the computation of sensitivity and specificity. The derivation is general and is applicable to both symmetrical and asymmetrical ROC curves. RESULTS: For symmetric ROC curves and typical values for the PPV of mammography (about 8%) and cancer incidence (f= 0.003), an A(z) value of 0.95 was required to achieve 63% sensitivity and an A. value of 0.98 led to 86% sensitivity. CONCLUSION: A model was developed that should allow researchers to deduce sensitivity and specificity for screening mammography based on ROC curve measurements and using realistic values of PPV and f. This model allows A. values to be related to the probability of breast cancer detection.

Impact of lesion segmentation metrics on computer-aided diagnosis/detection in breast computed tomography.

Evaluation of segmentation algorithms usually involves comparisons of segmentations to gold-standard delineations without regard to the ultimate medical decision-making task. We compare two segmentation evaluations methods-a Dice similarity coefficient (DSC) evaluation and a diagnostic classification task-based evaluation method using lesions from breast computed tomography. In our investigation, we use results from two previously developed lesion-segmentation algorithms [a global active contour model (GAC) and a global with local aspects active contour model]. Although similar DSC values were obtained (0.80 versus 0.77), we show that the global + local active contour (GLAC) model, as compared with the GAC model, is able to yield significantly improved classification performance in terms of area under the receivers operating characteristic (ROC) curve in the task of distinguishing malignant from benign lesions. [Area under the [Formula: see text] compared to 0.63, [Formula: see text]]. This is mainly because the GLAC model yields better detailed information required in the calculation of morphological features. Based on our findings, we conclude that the DSC metric alone is not sufficient for evaluating segmentation lesions in computer-aided diagnosis tasks.

Segmentation of breast masses on dedicated breast computed tomography and three-dimensional breast ultrasound images.

We present and evaluate a method for the three-dimensional (3-D) segmentation of breast masses on dedicated breast computed tomography (bCT) and automated 3-D breast ultrasound images. The segmentation method, refined from our previous segmentation method for masses on contrast-enhanced bCT, includes two steps: (1) initial contour estimation and (2) active contour-based segmentation to further evolve and refine the initial contour by adding a local energy term to the level-set equation. Segmentation performance was assessed in terms of Dice coefficients (DICE) for 129 lesions on noncontrast bCT, 38 lesions on contrast-enhanced bCT, and 98 lesions on 3-D breast ultrasound (US) images. For bCT, DICE values of 0.82 and 0.80 were obtained on contrast-enhanced and noncontrast images, respectively. The improvement in segmentation performance with respect to that of our previous method was statistically significant ( p = 0.002 ). Moreover, segmentation appeared robust with respect to the presence of glandular tissue. For 3-D breast US, the DICE value was 0.71. Hence, our method obtained promising results for both 3-D imaging modalities, laying a solid foundation for further quantitative image analysis and potential future expansion to other 3-D imaging modalities.