Development, standardization, and testing of a lexicon for reporting contrast-enhanced breast magnetic resonance imaging studies

Assessing Axillary Lymph Node Burden and Prognosis in cT1-T2 Stage Breast Cancer Using Machine Learning Methods: A Retrospective Dual-Institutional MRI Study

BACKGROUND

Pathological axillary lymph node (pALN) burden is an important factor for treatment decision-making in clinical T1-T2 (cT1-T2) stage breast cancer. Preoperative assessment of the pALN burden and prognosis aids in the individualized selection of therapeutic approaches.

PURPOSE

To develop and validate a machine learning (ML) model based on clinicopathological and MRI characteristics for assessing pALN burden and survival in patients with cT1-T2 stage breast cancer.

STUDY TYPE

Retrospective.

POPULATION

A total of 506 females (range: 24-83 years) with cT1-T2 stage breast cancer from two institutions, forming the training (N = 340), internal validation (N = 85), and external validation cohorts (N = 81), respectively.

FIELD STRENGTH/SEQUENCE

This study used 1.5-T, axial fat-suppressed T2-weighted turbo spin-echo sequence and axial three-dimensional dynamic contrast-enhanced fat-suppressed T1-weighted gradient echo sequence.

ASSESSMENT

Four ML methods (eXtreme Gradient Boosting [XGBoost], Support Vector Machine, k-Nearest Neighbor, Classification and Regression Tree) were employed to develop models based on clinicopathological and MRI characteristics. The performance of these models was evaluated by their discriminative ability. The best-performing model was further analyzed to establish interpretability and used to calculate the pALN score. The relationships between the pALN score and disease-free survival (DFS) were examined.

STATISTICAL TESTS

Chi-squared test, Fisher's exact test, univariable logistic regression, area under the curve (AUC), Delong test, net reclassification improvement, integrated discrimination improvement, Hosmer-Lemeshow test, log-rank, Cox regression analyses, and intraclass correlation coefficient were performed. A P-value <0.05 was considered statistically significant.

RESULTS

The XGB II model, developed based on the XGBoost algorithm, outperformed the other models with AUCs of 0.805, 0.803, and 0.818 in the three cohorts. The Shapley additive explanation plot indicated that the top variable in the XGB II model was the Node Reporting and Data System score. In multivariable Cox regression analysis, the pALN score was significantly associated with DFS (hazard ratio: 4.013, 95% confidence interval: 1.059-15.207).

DATA CONCLUSION

The XGB II model may allow to evaluate pALN burden and could provide prognostic information in cT1-T2 stage breast cancer patients.

LEVEL OF EVIDENCE

3 TECHNICAL EFFICACY: Stage 2.

Exploring the added value of pretherapeutic MR descriptors in predicting breast cancer pathologic complete response to neoadjuvant chemotherapy

OBJECTIVES

To evaluate the association between pretreatment MRI descriptors and breast cancer (BC) pathological complete response (pCR) to neoadjuvant chemotherapy (NAC).

MATERIALS AND METHODS

Patients with BC treated by NAC with a breast MRI between 2016 and 2020 were included in this retrospective observational single-center study. MR studies were described using the standardized BI-RADS and breast edema score on T2-weighted MRI. Univariable and multivariable logistic regression analyses were performed to assess variables association with pCR according to residual cancer burden. Random forest classifiers were trained to predict pCR on a random split including 70% of the database and were validated on the remaining cases.

RESULTS

Among 129 BC, 59 (46%) achieved pCR after NAC (luminal (n = 7/37, 19%), triple negative (n = 30/55, 55%), HER2 + (n = 22/37, 59%)). Clinical and biological items associated with pCR were BC subtype (p < 0.001), T stage 0/I/II (p = 0.008), higher Ki67 (p = 0.005), and higher tumor-infiltrating lymphocytes levels (p = 0.016). Univariate analysis showed that the following MRI features, oval or round shape (p = 0.047), unifocality (p = 0.026), non-spiculated margins (p = 0.018), no associated non-mass enhancement (p = 0.024), and a lower MRI size (p = 0.031), were significantly associated with pCR. Unifocality and non-spiculated margins remained independently associated with pCR at multivariable analysis. Adding significant MRI features to clinicobiological variables in random forest classifiers significantly increased sensitivity (0.67 versus 0.62), specificity (0.69 versus 0.67), and precision (0.71 versus 0.67) for pCR prediction.

CONCLUSION

Non-spiculated margins and unifocality are independently associated with pCR and can increase models performance to predict BC response to NAC.

CLINICAL RELEVANCE STATEMENT

A multimodal approach integrating pretreatment MRI features with clinicobiological predictors, including tumor-infiltrating lymphocytes, could be employed to develop machine learning models for identifying patients at risk of non-response. This may enable consideration of alternative therapeutic strategies to optimize treatment outcomes.

KEY POINTS

• Unifocality and non-spiculated margins are independently associated with pCR at multivariable logistic regression analysis. • Breast edema score is associated with MR tumor size and TIL expression, not only in TN BC as previously reported, but also in luminal BC. • Adding significant MRI features to clinicobiological variables in machine learning classifiers significantly increased sensitivity, specificity, and precision for pCR prediction.

Using Machine Learning Methods to Assess Lymphovascular Invasion and Survival in Breast Cancer: Performance of Combining Preoperative Clinical and MRI Characteristics

BACKGROUND

Preoperative assessment of lymphovascular invasion (LVI) in invasive breast cancer (IBC) is of high clinical relevance for treatment decision-making and prognosis.

PURPOSE

To investigate the associations of preoperative clinical and magnetic resonance imaging (MRI) characteristics with LVI and disease-free survival (DFS) by using machine learning methods in patients with IBC.

STUDY TYPE

Retrospective.

POPULATION

Five hundred and seventy-five women (range: 24-79 years) with IBC who underwent preoperative MRI examinations at two hospitals, divided into the training (N = 386) and validation datasets (N = 189).

FIELD STRENGTH/SEQUENCE

Axial fat-suppressed T2-weighted turbo spin-echo sequence and dynamic contrast-enhanced with fat-suppressed T1-weighted three-dimensional gradient echo imaging.

ASSESSMENT

MRI characteristics (clinical T stage, breast edema score, MRI axillary lymph node status, multicentricity or multifocality, enhancement pattern, adjacent vessel sign, and increased ipsilateral vascularity) were reviewed independently by three radiologists. Logistic regression (LR), eXtreme Gradient Boosting (XGBoost), k-Nearest Neighbor (KNN), and Support Vector Machine (SVM) algorithms were used to establish the models by combing preoperative clinical and MRI characteristics for assessing LVI status in the training dataset, and the methods were further applied in the validation dataset. The LVI score was calculated using the best-performing of the four models to analyze the association with DFS.

STATISTICAL TESTS

Chi-squared tests, variance inflation factors, receiver operating characteristics (ROC), Kaplan-Meier curve, log-rank, Cox regression, and intraclass correlation coefficient were performed. The area under the ROC curve (AUC) and hazard ratios (HR) were calculated. A P-value <0.05 was considered statistically significant.

RESULTS

The model established by the XGBoost algorithm had better performance than LR, SVM, and KNN models, achieving an AUC of 0.832 (95% confidence interval [CI]: 0.789, 0.876) in the training dataset and 0.838 (95% CI: 0.775, 0.901) in the validation dataset. The LVI score established by the XGBoost model was an independent indicator of DFS (adjusted HR: 2.66, 95% CI: 1.22-5.80).

DATA CONCLUSION

The XGBoost model based on preoperative clinical and MRI characteristics may help to investigate the LVI status and survival in patients with IBC.

LEVEL OF EVIDENCE

3 TECHNICAL EFFICACY: Stage 2.

Robustness of radiomics to variations in segmentation methods in multimodal brain MRI

Radiomics in neuroimaging uses fully automatic segmentation to delineate the anatomical areas for which radiomic features are computed. However, differences among these segmentation methods affect radiomic features to an unknown extent. A scan-rescan dataset (n = 46) of T1-weighted and diffusion tensor images was used. Subjects were split into a sleep-deprivation and a control group. Scans were segmented using four segmentation methods from which radiomic features were computed. First, we measured segmentation agreement using the Dice-coefficient. Second, robustness and reproducibility of radiomic features were measured using the intraclass correlation coefficient (ICC). Last, difference in predictive power was assessed using the Friedman-test on performance in a radiomics-based sleep deprivation classification application. Segmentation agreement was generally high (interquartile range = 0.77–0.90) and median feature robustness to segmentation method variation was higher (ICC > 0.7) than scan-rescan reproducibility (ICC 0.3–0.8). However, classification performance differed significantly among segmentation methods (p < 0.001) ranging from 77 to 84%. Accuracy was higher for more recent deep learning-based segmentation methods. Despite high agreement among segmentation methods, subtle differences significantly affected radiomic features and their predictive power. Consequently, the effect of differences in segmentation methods should be taken into account when designing and evaluating radiomics-based research methods.

Breast MRI: does a clinical decision algorithm outweigh reader experience?

Objectives

Due to its high sensitivity, DCE MRI of the breast (MRIb) is increasingly used for both screening and assessment purposes. The Kaiser score (KS) is a clinical decision algorithm, which formalizes and guides diagnosis in breast MRI and is expected to compensate for lesser reader experience. The aim was to evaluate the diagnostic performance of untrained residents using the KS compared to off-site radiologists experienced in breast imaging using only MR BI-RADS.

Methods

Three off-site, board-certified radiologists, experienced in breast imaging, interpreted MRIb according to the MR BI-RADS scale. The same studies were read by three residents in radiology without prior training in breast imaging using the KS. All readers were blinded to clinical information. Histology was used as the gold standard. Statistical analysis was conducted by comparing the AUC of the ROC curves.

Results

A total of 80 women (median age 52 years) with 93 lesions (32 benign, 61 malignant) were included. The individual within-group performance of the three expert readers (AUC 0.723–0.742) as well as the three residents was equal (AUC 0.842–0.928), p > 0.05, respectively. But, the rating of each resident using the KS significantly outperformed the experts’ ratings using the MR BI-RADS scale (p ≤ 0.05).

Conclusion

The KS helped residents to achieve better results in reaching correct diagnoses than experienced radiologists empirically assigning MR BI-RADS categories in a clinical “problem solving MRI” setting. These results support that reporting breast MRI benefits more from using a diagnostic algorithm rather than expert experience.

Key Points

• Reporting breast MRI benefits more from using a diagnostic algorithm rather than expert experience in a clinical “problem solving MRI” setting.

• The Kaiser score, which provides a clinical decision algorithm for structured reporting, helps residents to reach an expert level in breast MRI reporting and to even outperform experienced radiologists using MR BI-RADS without further formal guidance.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00330-022-09015-8.

Evidence-Based and Structured Diagnosis in Breast MRI using the Kaiser Score

BACKGROUND

 Breast MRI is the most sensitive method for the detection of breast cancer and is an integral part of modern breast imaging. On the other hand, interpretation of breast MRI exams is considered challenging due to the complexity of the available information. Clinical decision rules that combine diagnostic criteria in an algorithm can help the radiologist to read breast MRI by supporting objective and largely experience-independent diagnosis.

METHOD

 Narrative review. In this article, the Kaiser Score (KS) as a clinical decision rule for breast MRI is introduced, its diagnostic criteria are defined, and strategies for clinical decision making using the KS are explained and discussed.

RESULTS

 The KS is based on machine learning and has been independently validated by international research. It is largely independent of the examination technique that is used. It allows objective differentiation between benign and malignant contrast-enhancing breast MRI findings using diagnostic BI-RADS criteria taken from T2w and dynamic contrast-enhanced T1w images. A flowchart guides the reader in up to three steps to determine a score corresponding to the probability of malignancy that can be used to assign a BI-RADS category. Individual decision making takes the clinical context into account and is illustrated by typical scenarios.

KEY POINTS

· The KS as an evidence-based decision rule to objectively distinguish benign from malignant breast lesions is based on information contained in T2w und dynamic contrast-enhanced T1w sequences and is largely independent of specific examination protocols.. · The KS diagnostic criteria are in line with the MRI BI-RADS lexicon. We focused on defining a default category to be applied in the case of equivocal imaging criteria.. · The KS reflects increasing probabilities of malignancy and, together with the clinical context, assists individual decision making..

CITATION FORMAT

· Baltzer PA, Krug KB, Dietzel M. Evidence-Based and Structured Diagnosis in Breast MRI using the Kaiser Score. Fortschr Röntgenstr 2022; DOI: 10.1055/a-1829-5985.

Quantification and Classification of Contrast Enhanced Ultrasound Breast Cancer Data: A Preliminary Study

This study aimed to investigate which of the two frequently adopted perfusion models better describes the contrast enhanced ultrasound (CEUS) perfusion signal in order to produce meaningful imaging markers with the goal of developing a machine-learning model that can classify perfusion curves as benign or malignant in breast cancer data. Twenty-five patients with high suspicion of breast cancer were analyzed with exponentially modified Gaussian (EMG) and gamma variate functions (GVF). The adjusted R² metric was the criterion for assessing model performance. Various classifiers were trained on the quantified perfusion curves in order to classify the curves as benign or malignant on a voxel basis. Sensitivity, specificity, geometric mean, and AUROC were the validation metrics. The best quantification model was EMG with an adjusted R² of 0.60 ± 0.26 compared to 0.56 ± 0.25 for GVF. Logistic regression was the classifier with the highest performance (sensitivity, specificity, G_mean, and AUROC = 89.2 ± 10.7, 70.0 ± 18.5, 77.1 ± 8.6, and 91.0 ± 6.6, respectively). This classification method obtained similar results that are consistent with the current literature. Breast cancer patients can benefit from early detection and characterization prior to biopsy.

Apparent Diffusion Coefficient-Based Convolutional Neural Network Model Can Be Better Than Sole Diffusion-Weighted Magnetic Resonance Imaging to Improve the Differentiation of Invasive Breast Cancer From Breast Ductal Carcinoma In Situ

BACKGROUND AND PURPOSE

Breast ductal carcinoma in situ (DCIS) has no metastatic potential, and has better clinical outcomes compared with invasive breast cancer (IBC). Convolutional neural networks (CNNs) can adaptively extract features and may achieve higher efficiency in apparent diffusion coefficient (ADC)-based tumor invasion assessment. This study aimed to determine the feasibility of constructing an ADC-based CNN model to discriminate DCIS from IBC.

METHODS

The study retrospectively enrolled 700 patients with primary breast cancer between March 2006 and June 2019 from our hospital, and randomly selected 560 patients as the training and validation sets (ratio of 3 to 1), and 140 patients as the internal test set. An independent external test set of 102 patients during July 2019 and May 2021 from a different scanner of our hospital was selected as the primary cohort using the same criteria. In each set, the status of tumor invasion was confirmed by pathologic examination. The CNN model was constructed to discriminate DCIS from IBC using the training and validation sets. The CNN model was evaluated using the internal and external tests, and compared with the discriminating performance using the mean ADC. The area under the curve (AUC), sensitivity, specificity, and accuracy were calculated to evaluate the performance of the previous model.

RESULTS

The AUCs of the ADC-based CNN model using the internal and external test sets were larger than those of the mean ADC (AUC: 0.977 vs. 0.866, P = 0.001; and 0.926 vs. 0.845, P = 0.096, respectively). Regarding the internal test set and external test set, the ADC-based CNN model yielded sensitivities of 0.893 and 0.873, specificities of 0.929 and 0.894, and accuracies of 0.907 and 0.902, respectively. Regarding the two test sets, the mean ADC showed sensitivities of 0.845 and 0.818, specificities of 0.821 and 0.829, and accuracies of 0.836 and 0.824, respectively. Using the ADC-based CNN model, the prediction only takes approximately one second for a single lesion.

CONCLUSION

The ADC-based CNN model can improve the differentiation of IBC from DCIS with higher accuracy and less time.

Automatic Detection and Segmentation of Breast Cancer on MRI Using Mask R-CNN Trained on Non-Fat-Sat Images and Tested on Fat-Sat Images

RATIONALE AND OBJECTIVES

Computer-aided methods have been widely applied to diagnose lesions on breast magnetic resonance imaging (MRI). The first step was to identify abnormal areas. A deep learning Mask Regional Convolutional Neural Network (R-CNN) was implemented to search the entire set of images and detect suspicious lesions.

MATERIALS AND METHODS

Two DCE-MRI datasets were used, 241 patients acquired using non-fat-sat sequence for training, and 98 patients acquired using fat-sat sequence for testing. All patients have confirmed unilateral mass cancers. The tumor was segmented using fuzzy c-means clustering algorithm to serve as the ground truth. Mask R-CNN was implemented with ResNet-101 as the backbone. The neural network output the bounding boxes and the segmented tumor for evaluation using the Dice Similarity Coefficient (DSC). The detection performance, and the trade-off between sensitivity and specificity, was analyzed using free response receiver operating characteristic.

RESULTS

When the precontrast and subtraction image of both breasts were used as input, the false positive from the heart and normal parenchymal enhancements could be minimized. The training set had 1469 positive slices (containing lesion) and 9135 negative slices. In 10-fold cross-validation, the mean accuracy = 0.86 and DSC = 0.82. The testing dataset had 1568 positive and 7264 negative slices, with accuracy = 0.75 and DSC = 0.79. When the obtained per-slice results were combined, 240 of 241 (99.5%) lesions in the training and 98 of 98 (100%) lesions in the testing datasets were identified.

CONCLUSION

Deep learning using Mask R-CNN provided a feasible method to search breast MRI, localize, and segment lesions. This may be integrated with other artificial intelligence algorithms to develop a fully automatic breast MRI diagnostic system.

Training Radiologists to Interpret Contrast-enhanced Mammography: Toward a Standardized Lexicon

OBJECTIVE

Using terms adapted from the BI-RADS Mammography and MRI lexicons, we trained radiologists to interpret contrast-enhanced mammography (CEM) and assessed reliability of their description and assessment.

METHODS

A 60-minute presentation on CEM and terminology was reviewed independently by 21 breast imaging radiologist observers. For 21 CEM exams with 31 marked findings, observers recorded background parenchymal enhancement (BPE) (minimal, mild, moderate, marked), lesion type (oval/round or irregular mass, or non-mass enhancement), intensity of enhancement (none, weak, medium, strong), enhancement quality (none, homogeneous, heterogeneous, rim), and BI-RADS assessment category (2, 3, 4A, 4B, 4C, 5). "Expert" consensus of 3 other radiologists experienced in CEM was developed. Kappa statistic was used to assess agreement between radiologists and expert consensus, and between radiologists themselves, on imaging feature categories and final assessments. Reproducibility of specific feature descriptors was assessed as fraction of consensus-concordant responses.

RESULTS

Radiologists demonstrated moderate agreement for BPE, (mean kappa, 0.43; range, 0.05-0.69), and lowest reproducibility for "minimal." Agreement was substantial for lesion type (mean kappa, 0.70; range, 0.47-0.93), moderate for intensity of enhancement (mean kappa, 0.57; range, 0.44-0.76), and moderate for enhancement quality (mean kappa, 0.59; range, 0.20-0.78). Agreement on final assessment was fair (mean kappa, 0.26; range, 0.09-0.44), with BI-RADS category 3 the least reproducible. Decision to biopsy (BI-RADS 2-3 vs 4-5) showed moderate agreement with consensus (mean kappa, 0.54; range, -0.06-0.87).

CONCLUSION

With minimal training, agreement for description of CEM findings by breast imaging radiologists was comparable to other BI-RADS lexicons.

Diffusivity in breast malignancies analyzed for b > 1000 s/mm2 at 1 mm in-plane resolutions: Insight from Gaussian and non-Gaussian behaviors

Diffusion-weighted imaging (DWI) can improve breast cancer characterizations, but often suffers from low image quality -particularly at informative b > 1000 s/mm² values. The aim of this study was to evaluate multishot approaches characterizing Gaussian and non-Gaussian diffusivities in breast cancer. This was a prospective study, in which 15 subjects, including 13 patients with biopsy-confirmed breast cancers, were enrolled. DWI was acquired at 3 T using echo planar imaging (EPI) with and without zoomed excitations, readout-segmented EPI (RESOLVE), and spatiotemporal encoding (SPEN); dynamic contrast-enhanced (DCE) data were collected using three-dimensional gradient-echo T₁ weighting; anatomies were evaluated with T₂ -weighted two-dimensional turbo spin-echo. Congruence between malignancies delineated by DCE was assessed against high-resolution DWI scans with b-values in the 0-1800 s/mm² range, as well as against apparent diffusion coefficient (ADC) and kurtosis maps. Data were evaluated by independent magnetic resonance scientists with 3-20 years of experience, and radiologists with 6 and 20 years of experience in breast MRI. Malignancies were assessed from ADC and kurtosis maps, using paired t tests after confirming that these values had a Gaussian distribution. Agreements between DWI and DCE datasets were also evaluated using Sorensen-Dice similarity coefficients. Cancerous and normal tissues were clearly separable by ADCs: by SPEN their average values were (1.03 ± 0.17) × 10^-3 and (1.69 ± 0.19) × 10^-3  mm² /s (p < 0.0001); by RESOLVE these values were (1.16 ± 0.16) × 10^-3 and (1.52 ± 0.14) × 10^-3 (p = 0.00026). Kurtosis also distinguished lesions (K = 0.64 ± 0.15) from normal tissues (K = 0.45 ± 0.05), but only when measured by SPEN (p = 0.0008). The best statistical agreement with DCE-highlighted regions arose for SPEN-based DWIs recorded with b = 1800 s/mm² (Sorensen-Dice coefficient = 0.67); DWI data recorded with b = 850 and 1200 s/mm² , led to lower coefficients. Both ADC and kurtosis maps highlighted the breast malignancies, with ADCs providing a more significant separation. The most promising alternative for contrast-free delineations of the cancerous lesions arose from b = 1800 s/mm² DWI. LEVEL OF EVIDENCE: 2. TECHNICAL EFFICACY STAGE: 3.

Artificial Intelligence Applied to Breast MRI for Improved Diagnosis

Background Recognition of salient MRI morphologic and kinetic features of various malignant tumor subtypes and benign diseases, either visually or with artificial intelligence (AI), allows radiologists to improve diagnoses that may improve patient treatment. Purpose To evaluate whether the diagnostic performance of radiologists in the differentiation of cancer from noncancer at dynamic contrast material-enhanced (DCE) breast MRI is improved when using an AI system compared with conventionally available software. Materials and Methods In a retrospective clinical reader study, images from breast DCE MRI examinations were interpreted by 19 breast imaging radiologists from eight academic and 11 private practices. Readers interpreted each examination twice. In the "first read," they were provided with conventionally available computer-aided evaluation software, including kinetic maps. In the "second read," they were also provided with AI analytics through computer-aided diagnosis software. Reader diagnostic performance was evaluated with receiver operating characteristic (ROC) analysis, with the area under the ROC curve (AUC) as a figure of merit in the task of distinguishing between malignant and benign lesions. The primary study end point was the difference in AUC between the first-read and the second-read conditions. Results One hundred eleven women (mean age, 52 years ± 13 [standard deviation]) were evaluated with a total of 111 breast DCE MRI examinations (54 malignant and 57 nonmalignant lesions). The average AUC of all readers improved from 0.71 to 0.76 (P = .04) when using the AI system. The average sensitivity improved when Breast Imaging Reporting and Data System (BI-RADS) category 3 was used as the cut point (from 90% to 94%; 95% confidence interval [CI] for the change: 0.8%, 7.4%) but not when using BI-RADS category 4a (from 80% to 85%; 95% CI: -0.9%, 11%). The average specificity showed no difference when using either BI-RADS category 4a or category 3 as the cut point (52% and 52% [95% CI: -7.3%, 6.0%], and from 29% to 28% [95% CI: -6.4%, 4.3%], respectively). Conclusion Use of an artificial intelligence system improves radiologists' performance in the task of differentiating benign and malignant MRI breast lesions. © RSNA, 2020 Online supplemental material is available for this article. See also the editorial by Krupinski in this issue.

Integrated Multiparametric Radiomics and Informatics System for Characterizing Breast Tumor Characteristics with the OncotypeDX Gene Assay

Optimal use of multiparametric magnetic resonance imaging (mpMRI) can identify key MRI parameters and provide unique tissue signatures defining phenotypes of breast cancer. We have developed and implemented a new machine-learning informatic system, termed Informatics Radiomics Integration System (IRIS) that integrates clinical variables, derived from imaging and electronic medical health records (EHR) with multiparametric radiomics (mpRad) for identifying potential risk of local or systemic recurrence in breast cancer patients. We tested the model in patients (n = 80) who had Estrogen Receptor positive disease and underwent OncotypeDX gene testing, radiomic analysis, and breast mpMRI. The IRIS method was trained using the mpMRI, clinical, pathologic, and radiomic descriptors for prediction of the OncotypeDX risk score. The trained mpRad IRIS model had a 95% and specificity was 83% with an Area Under the Curve (AUC) of 0.89 for classifying low risk patients from the intermediate and high-risk groups. The lesion size was larger for the high-risk group (2.9 ± 1.7 mm) and lower for both low risk (1.9 ± 1.3 mm) and intermediate risk (1.7 ± 1.4 mm) groups. The lesion apparent diffusion coefficient (ADC) map values for high- and intermediate-risk groups were significantly (p < 0.05) lower than the low-risk group (1.14 vs. 1.49 × 10-3 mm2/s). These initial studies provide deeper insight into the clinical, pathological, quantitative imaging, and radiomic features, and provide the foundation to relate these features to the assessment of treatment response for improved personalized medicine.

A novel MRI-based predictive index can identify patients suitable for preservation of the nipple-areola complex in breast reconstructive surgery

INTRODUCTION

Accurately predicting nipple-areola complex (NAC) involvement in breast cancer is necessary for identifying patients who may be candidates for a nipple-sparing mastectomy. Although multiple risk factors are indicated in the guidelines, it is difficult to predict NAC involvement (NAC-i) preoperatively even if these factors are evaluated individually. This study aimed to develop a more accurate and practical preoperative NAC-i prediction model using magnetic resonance imaging (MRI).

MATERIALS AND METHODS

All tumors in 252 patients were evaluated using postcontrast T1-weighted subtraction on MRI.

RESULTS

The receiver operating characteristic curves identified cut-off values for tumor size and tumor-to-nipple distance (TND) as 4 cm and 1.2 cm, respectively. Multivariate analysis demonstrated that TND (p < 0.001), ductal enhancement extending to the nipple (DEEN) (p < 0.001), and nipple enhancement (NE) (p = 0.005) were independent clinical risk factors for pathological NAC-i. A formula was constructed using odds ratios for these three independent preoperative risk factors in multivariate analysis: the MRI-based NAC-i predictive index (mNACPI) = TND × 4 + DEEN × 3 + NE × 1. A total score of ≤4 points was defined as low risk and ≥5 points as high risk. NAC-i rates were 2.4% in the low-risk group and 89.4% in the high-risk group; a significant correlation was observed between the risk group and permanent pathological NAC-i (p < 0.001). Assuming that the NAC was preserved in low-risk patients and resected in high-risk patients, NAC-i was verified using the mNACPI.

CONCLUSION

mNACPI may contribute greatly to the improvement of selecting suitable patients for NAC preservation in breast reconstructive surgery while maintaining oncological safety.

Evaluation of Breast Cancer Morphology Using Diffusion-Weighted and Dynamic Contrast-Enhanced MRI: Intermethod and Interobserver Agreement

BACKGROUND

The capability of diffusion-weighted imaging (DWI) for morphological analysis of breast lesions is underexplored.

PURPOSE

To evaluate the utility of DWI for assessment of morphological features of breast cancer by comparing DWI and dynamic contrast-enhanced (DCE) MRI findings to determine intermethod and interobserver agreement.

STUDY TYPE

Retrospective.

POPULATION

Seventy-eight women with pathohistologically proven breast cancer.

FIELD STRENGTH/SEQUENCE

1.5T. DWI and DCE images.

ASSESSMENT

Diffusion-weighted and DCE images were placed in two separate case sets. Three radiologists, blinded to all other information, independently evaluated each case set on two separate occasions. Lesions were interpreted according to the fifth edition of the ACR BI-RADS lexicon.

STATISTICAL ANALYSIS

Kappa (κ) statistics were calculated in order to assess intermethod and interobserver agreement.

RESULTS

For values that attained statistical significance (P < 0.05), intermethod agreement ranged from fair (κ = 0.22) for nonmass internal patterns to significant (κ = 0.8) for lesion type. On DWI, interobserver agreement varied from fair (κ = 0.34) for mass shape to significant (κ = 0.75) for lesion type. On DCE MRI, interobserver agreement varied from fair (κ = 0.27) for irregular vs. spiculated mass margin to perfect (κ = 1) for circumscribed vs. noncircumscribed mass margin.

DATA CONCLUSION

On the whole, there was moderate intermethod agreement. The values of interobserver agreement were mostly similar between DWI and DCE MRI. This suggests that DWI is applicable for morphological assessment of breast cancer, notwithstanding substantially inferior spatial resolution compared to DCE MRI.

LEVEL OF EVIDENCE

3 Technical Efficacy: Stage 3 J. Magn. Reson. Imaging 2019;49:1381-1390.

Use of Quantitative Morphological and Functional Features for Assessment of Axillary Lymph Node in Breast Dynamic Contrast-Enhanced Magnetic Resonance Imaging

Background

Axillary lymph-node assessment is considered one of the most important prognostic factors concerning breast cancer survival.

Objective

We investigated the discriminative power of morphological and functional features in assessing the axillary lymph node.

Methods

We retrospectively analysed data from 52 consecutive patients who undergone DCE-MRI and were diagnosed with primary breast carcinoma: 94 lymph nodes were identified. Per each lymph node, we extracted morphological features: circularity, compactness, convexity, curvature, elongation, diameter, eccentricity, irregularity, radial length, entropy, rectangularity, roughness, smoothness, sphericity, spiculation, surface, and volume. Moreover, we extracted functional features: time to peak (TTP), maximum signal difference (MSD), wash-in intercept (WII), wash-out intercept (WOI), wash-in slope (WIS), wash-out slope (WOS), area under gadolinium curve (AUGC), area under wash-in (AUWI), and area under wash-out (AUWO). Selection of important features in predicting metastasis has been done by means of receiver operating characteristic (ROC) analysis. Performance of linear discriminant analysis was analysed.

Results

All morphological features but circularity showed a significant difference between median values of metastatic lymph nodes group and nonmetastatic lymph nodes group. All dynamic parameters except for MSD and WOS showed a statistically significant difference between median values of metastatic lymph nodes group and nonmetastatic lymph nodes group. Best results for discrimination of metastatic and nonmetastatic lymph nodes were obtained by AUGC (accuracy 75.8%), WIS (accuracy 71.0%), WOS (accuracy 71.0%), and AUCWO (accuracy 72.6%) for dynamic features and by compactness (accuracy 82.3%), curvature (accuracy 71.0%), radial length (accuracy 71.0%), roughness (accuracy 74.2%), smoothness (accuracy 77.2%), and speculation (accuracy 72.6%) for morphological features. Linear combination of all morphological and/or of all dynamic features did not increase accuracy in metastatic lymph nodes discrimination.

Conclusions

Compactness as morphological feature and area under time-intensity curve as dynamic feature were the best parameters in identifying metastatic lymph nodes on breast MRI.

Breast cancer MRI radiomics: An overview of algorithmic features and impact of inter-reader variability in annotating tumors

PURPOSE

To review features used in MRI radiomics of breast cancer and study the inter-reader stability of the features.

METHODS

We implemented 529 algorithmic features that can be extracted from tumor and fibroglandular tissue (FGT) in breast MRIs. The features were identified based on a review of the existing literature with consideration of their usage, prognostic ability, and uniqueness. The set was then extended so that it comprehensively describes breast cancer imaging characteristics. The features were classified into 10 groups based on the type of data used to extract them and the type of calculation being performed. For the assessment of inter-reader variability, four fellowship-trained readers annotated tumors on preoperative dynamic contrast-enhanced MRIs for 50 breast cancer patients. Based on the annotations, an algorithm automatically segmented the image and extracted all features resulting in one set of features for each reader. For a given feature, the inter-reader stability was defined as the intraclass correlation coefficient (ICC) computed using the feature values obtained through all readers for all cases.

RESULTS

The average inter-reader stability for all features was 0.8474 (95% CI: 0.8068-0.8858). The mean inter-reader stability was lower for tumor-based features (0.6348, 95% CI: 0.5391-0.7257) than FGT-based features (0.9984, 95% CI: 0.9970-0.9992). The feature group with the highest inter-reader stability quantifies breast and FGT volume. The feature group with the lowest inter-reader stability quantifies variations in tumor enhancement.

CONCLUSIONS

Breast MRI radiomics features widely vary in terms of their stability in the presence of inter-reader variability. Appropriate measures need to be taken for reducing this variability in tumor-based radiomics.

Assessment of breast mass morphology with diffusion-weighted MRI: Beyond apparent diffusion coefficient

BACKGROUND

Diffusion-weighted imaging (DWI) is a noncontrast-enhanced MRI technique. There are new promising studies on the use of DWI as a part of the enhanced or unenhanced abbreviated breast MRI protocols.

PURPOSE

To evaluate the ability of breast DWI in the assessment of mass morphology and determine the contribution of this morphologic evaluation in their characterization.

STUDY TYPE

Retrospective.

POPULATION

In all, 213 consecutive women were breast MR imaged and had a later confirmed diagnosis.

FIELD STRENGTH/SEQUENCE

Breast dynamic contrast-enhanced-MRI (DCE-MRI) and DWI at 1.5T.

ASSESSMENT

After Institutional Review Board approval, two radiologists first independently, and later in consensus, evaluated the visibility and morphology of the 143 malignant, 70 benign masses on DWI and DCE-MRI in separate sessions, blindly. Shape, margin, and internal pattern of the masses were evaluated according to BI-RADS lexicon. Apparent diffusion coefficient (ADC) and tumor size were measured by one radiologist.

STATISTICAL TESTS

Consistency between imaging methods and readers was evaluated with Cohen's kappa statistics. Multivariate analysis was applied to find the best predictors of malignancy.

RESULTS

Tumor visibility on DWI was high to moderate in at least 88% of cases. Consistency between DWI and DCE-MRI was substantial (kappa ≥0.757) for shape and margin and moderate (kappa = 0.505) for internal pattern. Interobserver agreement was substantial to moderate for all morphologic parameters (kappa ≥0.596). Morphology evaluated on DWI provided 83-84% accuracy in discriminating malignant from benign masses. ADC alone provided 90-91% accuracy. Both morphologic parameters and ADC were significantly associated with malignancy on multivariate analysis and provided 91-93% accuracy.

DATA CONCLUSION

DWI might be used not only for ADC evaluation but also for the morphological evaluation of breast masses to characterize them.

LEVEL OF EVIDENCE

3 Technical Efficacy: Stage 2 J. Magn. Reson. Imaging 2018;48:1668-1677.

Effects of MRI scanner parameters on breast cancer radiomics

PURPOSE

To assess the impact of varying magnetic resonance imaging (MRI) scanner parameters on the extraction of algorithmic features in breast MRI radiomics studies.

METHODS

In this retrospective study, breast imaging data for 272 patients were analyzed with magnetic resonance (MR) images. From the MR images, we assembled and implemented 529 algorithmic features of breast tumors and fibrograndular tissue (FGT). We divided the features into 10 groups based on the type of data used for the feature extraction and the nature of the extracted information. Three scanner parameters were considered: scanner manufacturer, scanner magnetic field strength, and slice thickness. We assessed the impact of each of the scanner parameters on each of the feature by testing whether the feature values are systematically diverse for different values of these scanner parameters. A two-sample t-test has been used to establish whether the impact of a scanner parameter on values of a feature is significant and receiver operating characteristics have been used for to establish the extent of that effect.

RESULTS

On average, higher proportion (69% FGT versus 20% tumor) of FGT related features were affected by the three scanner parameters. Of all feature groups and scanner parameters, the feature group related to the variation in FGT enhancement was found to be the most sensitive to the scanner manufacturer (AUC = 0.81 ± 0.14).

CONCLUSIONS

Features involving calculations from FGT are particularly sensitive to the scanner parameters.

Prediction of low-risk breast cancer using quantitative DCE-MRI and its pathological basis

Purpose

This study aimed to evaluate the difference of mass in dynamic contrast enhanced magnetic resonance imaging (DCE-MRI) characteristics between low-risk and non-low-risk breast cancers and to explore the possible pathological basis.

Materials and Methods

Approval from the institutional review board and informed consent were acquired for this study. The MR images of 104 patients with pathologically proven breast cancer (104 lesions) were prospectively analyzed. All of included patients were Chinese woman. The DCE-MRI morphologic findings, apparent diffusion coefficient (ADC) values, quantitative DCE-MRI parameters, and pathological biomarkers between the two subtypes of breast cancer were compared. The quantitative DCE-MRI parameters and ADC values were added to the morphologic features in multivariate models to evaluate diagnostic performance in predicting low-risk breast cancer. The values were further subjected to the receiver operating characteristic (ROC) curve analysis.

Results

Low-risk tumors showed significantly lower K^trans and K_epvalues (t = 2.065, P = 0.043 and t = 3.548, P = 0.001, respectively) and higher ADC value (t = 4.713, P = 0.000) than non-low-risk breast cancers. Our results revealed no significant differences in clinic data and conventional imaging findings between the two breast cancer subtypes. Adding the quantitative DCE-MRI parameters and ADC values to conventional MRI improved the diagnostic performance of MRI: The area under the ROC improved from 0.63 to 0.91. Low-risk breast cancers showed significantly lower matrix metalloproteinase (MMP)-2 expression (P = 0.000), lower MMP-9 expression (P = 0.001), and lower microvessel density (MVD) values (P = 0.008) compared with non-low-risk breast cancers. K^trans and K_ep values were positively correlated with pathological biomarkers. The ADC value showed a significant inverse correlation with pathological biomarkers.

Conclusions

The prediction parameter using K^trans, K_ep, and ADC obtained on DCE-MRI and diffusion-weighted imaging could facilitate the identification of low-risk breast cancers. Decreased biological factors, including MVD, vascular endothelial growth factor, MMP-2, and MMP-9, may explain the possible pathological basis.

A simple classification system (the Tree flowchart) for breast MRI can reduce the number of unnecessary biopsies in MRI-only lesions

OBJECTIVES

To assess whether using the Tree flowchart obviates unnecessary magnetic resonance imaging (MRI)-guided biopsies in breast lesions only visible on MRI.

METHODS

This retrospective IRB-approved study evaluated consecutive suspicious (BI-RADS 4) breast lesions only visible on MRI that were referred to our institution for MRI-guided biopsy. All lesions were evaluated according to the Tree flowchart for breast MRI by experienced readers. The Tree flowchart is a decision rule that assigns levels of suspicion to specific combinations of diagnostic criteria. Receiver operating characteristic (ROC) curve analysis was used to evaluate diagnostic accuracy. To assess reproducibility by kappa statistics, a second reader rated a subset of 82 patients.

RESULTS

There were 454 patients with 469 histopathologically verified lesions included (98 malignant, 371 benign lesions). The area under the curve (AUC) of the Tree flowchart was 0.873 (95% CI: 0.839-0.901). The inter-reader agreement was almost perfect (kappa: 0.944; 95% CI 0.889-0.998). ROC analysis revealed exclusively benign lesions if the Tree node was ≤2, potentially avoiding unnecessary biopsies in 103 cases (27.8%).

CONCLUSIONS

Using the Tree flowchart in breast lesions only visible on MRI, more than 25% of biopsies could be avoided without missing any breast cancer.

KEY POINTS

• The Tree flowchart may obviate >25% of unnecessary MRI-guided breast biopsies. • This decrease in MRI-guided biopsies does not cause any false-negative cases. • The Tree flowchart predicts 30.6% of malignancies with >98% specificity. • The Tree's high specificity aids in decision-making after benign biopsy results.

Breast DCE-MRI: lesion classification using dynamic and morphological features by means of a multiple classifier system

Background

In breast magnetic resonance imaging (MRI) analysis for lesion detection and classification, radiologists agree that both morphological and dynamic features are important to differentiate benign from malignant lesions. We propose a multiple classifier system (MCS) to classify breast lesions on dynamic contrast-enhanced MRI (DCE-MRI) combining morphological features and dynamic information.

Methods

The proposed MCS combines the results of two classifiers trained with dynamic and morphological features separately. Twenty-six malignant and 22 benign breast lesions, histologically proven, were analysed. The lesions were subdivided into two groups: training set (14 benign and 18 malignant) and testing set (8 benign and 8 malignant). Volumes of interest were extracted both manually and automatically. We initially considered a feature set including 54 morphological features and 98 dynamic features. These were reduced by means of a selection procedure to delete redundant parameters. The performance of each of the two classifiers and of the overall MCS was compared with pathological classification.

Results

We obtained an accuracy of 91.7% on the testing set using automatic segmentation and combining the best classifier for morphological features (decision tree) and for dynamic information (Bayesian classifier). With implementation of the MCS, an increase in accuracy of 12.5% and of 31.3% was obtained compared with the accuracy of the Bayesian classifier tested with dynamic features and with that of the decision tree tested with morphological parameters, respectively.

Conclusions

An MCS can optimise the accuracy for breast lesion classification combining morphological features and dynamic information.

The Changing World of Breast Cancer: A Radiologist's Perspective

Compared with other fields of medicine, there is hardly an area that has seen such fast development as the world of breast cancer. Indeed, the way we treat breast cancer has changed fundamentally over the past decades. Breast imaging has always been an integral part of this change, and it undergoes constant adjustment to new ways of thinking. This relates not only to the technical tools we use for diagnosing breast cancer but also to the way diagnostic information is used to guide treatment. There is a constant change of concepts for and attitudes toward breast cancer, and a constant flux of new ideas, new treatment approaches, and new insights into the molecular and biological behavior of this disease. Clinical breast radiologists and even more so, clinician scientists, interested in breast imaging need to keep abreast with this rapidly changing world. Diagnostic or treatment approaches that are considered useful today may be abandoned tomorrow. Approaches that seem irrelevant or far too extravagant today may prove clinically useful and adequate next year. Radiologists must constantly question what they do, and align their clinical aims and research objectives with the changing needs of contemporary breast oncology. Moreover, knowledge about the past helps better understand present debates and controversies. Accordingly, in this article, we provide an overview on the evolution of breast imaging and breast cancer treatment, describe current areas of research, and offer an outlook regarding the years to come.

Pattern Recognition Approaches for Breast Cancer DCE-MRI Classification: A Systematic Review

We performed a systematic review of several pattern analysis approaches for classifying breast lesions using dynamic, morphological, and textural features in dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI). Several machine learning approaches, namely artificial neural networks (ANN), support vector machines (SVM), linear discriminant analysis (LDA), tree-based classifiers (TC), and Bayesian classifiers (BC), and features used for classification are described. The findings of a systematic review of 26 studies are presented. The sensitivity and specificity are respectively 91 and 83 % for ANN, 85 and 82 % for SVM, 96 and 85 % for LDA, 92 and 87 % for TC, and 82 and 85 % for BC. The sensitivity and specificity are respectively 82 and 74 % for dynamic features, 93 and 60 % for morphological features, 88 and 81 % for textural features, 95 and 86 % for a combination of dynamic and morphological features, and 88 and 84 % for a combination of dynamic, morphological, and other features. LDA and TC have the best performance. A combination of dynamic and morphological features gives the best performance.

Takeaways from Pre-Contrast T1 and T2 Breast Magnetic Resonance Imaging in Women with Recently Diagnosed Breast Cancer

Background

Dynamic contrast-enhanced magnetic resonance imaging (DCE - MRI) has been widely used in the management of breast cancer, and its diagnostic value in breast imaging has been demonstrated. There have only been a few reports regarding the usefulness of pre-contrast imaging. Knowledge about clinically significant findings of preoperative, pre-contrast T1 and T2 MR images will allow more accurate decisions regarding patient treatment and management.

Objectives

The aim of this study was to evaluate the clinically significant findings of preoperative, pre-contrast T1 and T2 MR images in recently diagnosed breast cancer patients.

Patients and Methods

We analyzed 390 preoperative 3-T MRIs of recently diagnosed breast cancer patients in whom the diagnosis was confirmed by a core needle biopsy.

Results

MRI findings that were correlated with post-core needle-biopsy changes were observed in 27.9% of the pre-contrast T1 and T2 MRIs (n = 109/390). Two of 35 cases that had a subareolar ductal high signal area on the pre-contrast T1 were confirmed by surgery as having nipple-areolar complex involvement.

Conclusion

A subareolar ductal high signal area on a pre-contrast T1 MRI must be carefully assessed in combination with dynamic, contrast-enhanced images for proper surgical management.

The Changing World of Breast Cancer: A Radiologist's Perspective

Compared with other fields of medicine, there is hardly an area that has seen such fast development as the world of breast cancer. Indeed, the way we treat breast cancer has changed fundamentally over the past decades. Breast imaging has always been an integral part of this change, and it undergoes constant adjustment to new ways of thinking. This relates not only to the technical tools we use for diagnosing breast cancer but also to the way diagnostic information is used to guide treatment. There is a constant change of concepts for and attitudes toward breast cancer, and a constant flux of new ideas, new treatment approaches, and new insights into the molecular and biological behavior of this disease. Clinical breast radiologists and even more so, clinician scientists, interested in breast imaging need to keep abreast with this rapidly changing world. Diagnostic or treatment approaches that are considered useful today may be abandoned tomorrow. Approaches that seem irrelevant or far too extravagant today may prove clinically useful and adequate next year. Radiologists must constantly question what they do, and align their clinical aims and research objectives with the changing needs of contemporary breast oncology. Moreover, knowledge about the past helps better understand present debates and controversies. Accordingly, in this article, we provide an overview on the evolution of breast imaging and breast cancer treatment, describe current areas of research, and offer an outlook regarding the years to come.

Incidentally detected enhancing lesions found in breast MRI: analysis of apparent diffusion coefficient and T2 signal intensity significantly improves specificity

OBJECTIVES

To evaluate the value of adding T2- and diffusion-weighted imaging (DWI) to the BI-RADS® classification in MRI-detected lesions.

METHODS

This retrospective study included 112 consecutive patients who underwent 3.0T structural breast MRI with T2- and DWI on the basis of EUSOMA recommendations. Morphological and kinetic features, T2 signal intensity (T2 SI) and apparent diffusion coefficient (ADC) findings were assessed.

RESULTS

Thirty-three (29.5 %) patients (mean age 57.0 ± 12.7 years) had 36 primarily MRI-detected incidental lesions of which 16 (44.4 %) proved to be malignant. No single morphological or kinetic feature was associated with malignancy. Both low T2 SI (P = 0.009) and low ADC values (≤0.87 × 10^-3 mm²s^-1, P < 0.001) yielded high specificity (80.0 %/80.0 %). The BI-RADS classification supplemented with information from DWI and T2-WI improved the diagnostic performance of the BI-RADS classification as sensitivity remained 100 % and specificity improved from 30 % to 65.0 %. The numbers of false positive lesions declined from 39 % (N = 14) to 19 % (N = 7).

CONCLUSION

MRI-detected incidental lesions may be challenging to characterize as they have few specific malignancy indicating features. The specificity of MRI can be improved by incorporating T2 SI and ADC values into the BI-RADS assessment.

KEY POINTS

• MRI-detected incidental lesions have few specific malignancy indicating features. • ≥ 1 suspicious morphologic or kinetic feature may warrant biopsy. • T2 signal intensity and DWI assessment are feasible in primarily MRI-detected lesions. • T2 SI and DWI assessment improve the BI-RADS specificity in MRI-detected lesions.

A simple scoring system for breast MRI interpretation: does it compensate for reader experience?

PURPOSE

To investigate the impact of a scoring system (Tree) on inter-reader agreement and diagnostic performance in breast MRI reading.

MATERIALS AND METHODS

This IRB-approved, single-centre study included 100 patients with 121 consecutive histopathologically verified lesions (52 malignant, 68 benign). Four breast radiologists with different levels of MRI experience and blinded to histopathology retrospectively evaluated all examinations. Readers independently applied two methods to classify breast lesions: BI-RADS and Tree. BI-RADS provides a reporting lexicon that is empirically translated into likelihoods of malignancy; Tree is a scoring system that results in a diagnostic category. Readings were compared by ROC analysis and kappa statistics.

RESULTS

Inter-reader agreement was substantial to almost perfect (kappa: 0.643-0.896) for Tree and moderate (kappa: 0.455-0.657) for BI-RADS. Diagnostic performance using Tree (AUC: 0.889-0.943) was similar to BI-RADS (AUC: 0.872-0.953). Less experienced radiologists achieved AUC: improvements up to 4.7 % using Tree (P-values: 0.042-0.698); an expert's performance did not change (P = 0.526). The least experienced reader improved in specificity using Tree (16 %, P = 0.001). No further sensitivity and specificity differences were found (P > 0.1).

CONCLUSION

The Tree scoring system improves inter-reader agreement and achieves a diagnostic performance similar to that of BI-RADS. Less experienced radiologists, in particular, benefit from Tree.

KEY POINTS

• The Tree scoring system shows high diagnostic accuracy in mass and non-mass lesions. • The Tree scoring system reduces inter-reader variability related to reader experience. • The Tree scoring system improves diagnostic accuracy in non-expert readers.

Subcategorization of Suspicious Breast Lesions (BI-RADS Category 4) According to MRI Criteria: Role of Dynamic Contrast-Enhanced and Diffusion-Weighted Imaging

OBJECTIVE

The purposes of this study were to investigate whether dynamic contrast-enhanced MRI is adequate for subcategorization of suspicious lesions (BI-RADS category 4) and to evaluate whether use of DWI improves diagnostic performance.

MATERIALS AND METHODS

The study group was composed of 103 suspicious lesions found in 83 subjects. Patient ages and lesion sizes were compiled, and two radiologists reanalyzed the images; subcategorized the findings as BI-RADS 4A, 4B, or 4C; and calculated apparent diffusion coefficient (ADC) values. The stratified variables were tested by univariate analysis and inserted in two multivariate predictive models, which were used to generate ROC curves and compare AUCs. Positive predictive values (PPVs) for each subcategory and ADC level were calculated, and interobserver agreement was tested.

RESULTS

Forty-four (42.7%) suspicious findings proved malignant. Except for age (p = 0.08), all stratified predictor variables were significant in univariate analyses (p < 0.01). Logistic regression models did not differ substantially after comparison of the ROC curves (p = 0.09), but the one including ADC values was slightly better: AUC of 0.89 (95% CI, 0.82-0.95) against AUC of 0.85 (95% CI, 0.78-0.93). PPV increased progressively in each BI-RADS 4 subcategory (4A, 0.15; 4B, 0.37; 4C, 0.84). ADC values of 1.10 × 10(-3) mm(2)/s or less had the second highest PPV (0.77). Interobserver agreement was substantial at a kappa value of 0.80 (95% CI, 0.70-0.90; p < 0.01).

CONCLUSION

Risk stratification of suspicious lesions (BI-RADS category 4) can be satisfactorily performed with DCE-MRI and slightly improved when DWI is introduced.

The relationship of breast density in mammography and magnetic resonance imaging in high-risk women and women with breast cancer

PURPOSE

To evaluate the relationship between mammographic breast density (MBD), background parenchymal enhancement (BPE), and fibroglandular tissue (FGT) in women with breast cancer (BC) and at high risk for developing BC.

METHODS

Our institutional database was queried for patients who underwent mammography and MRI.

RESULTS

Four hundred three (85%) had BC and 72 (15%) were at high risk. MBD (P=.0005), BPE (P<.0001), and FGT (P=.02) were all higher in high-risk women compared to the BC group.

CONCLUSIONS

Higher levels of MBD, BPE and FGT are seen in women at higher risk for developing BC when compared to women with BC.

Combined reading of Contrast Enhanced and Diffusion Weighted Magnetic Resonance Imaging by using a simple sum score

OBJECTIVE

To improve specificity of breast MRI by integrating Apparent Diffusion Coefficient (ADC) values with contrast enhanced MRI (CE-MRI) using a simple sum score.

METHODS

Retrospective analysis of a consecutive series of patients referred to breast MRI at 1.5 T for further workup of breast lesions. Reading results of CE-MRI were dichotomized into score 1 (suspicious) or 0 (benign). Lesion's ADC-values (in *10-3 mm2/s) were assigned two different scores: ADC2: likely malignant (score +1, ADC ≤ 1), indeterminate (score 0, ADC >1- ≤ 1.4) and likely benign (score -1, ADC > 1.4) and ADC1: indeterminate (score 0, ADC ≤ 1.4) and likely benign (score -1, ADC > 1.4). Final added CE-MRI and ADC scores >0 were considered suspicious. Reference standard was histology and imaging follow-up of >24 months. Diagnostic parameters were compared using McNemar tests.

RESULTS

A total of 150 lesions (73 malignant) were investigated. Reading of CE-MRI showed a sensitivity of 100 % (73/73) and a specificity of 81.8 % (63/77). Additional integration of ADC scores increased specificity (ADC2/ADC1, P = 0.008/0.001) without causing false negative results.

CONCLUSION

Using a simple sum score, ADC-values can be integrated with CE-MRI of the breast, improving specificity. The best approach is using one threshold to exclude cancer.

KEY POINTS

ADC is used to assign levels of suspicion to breast lesions. ADC values >1.4 *10 (-3)   mm (2) /s are likely benign and effectively rule out malignancy. ADC values below ≤1*10 (-3)   mm (2) /s) are likely malignant but may be false positive. CE-MRI (+1: suspicious, 0: benign) and ADC (0: indeterminate, -1: benign) scores are added. Sum scores >0 should be biopsied.

Association of Different MRI BIRADS Descriptors With Malignancy in Non Mass-Like Breast Lesions

Background:

Several studies on the diagnostic efficacy of MRI has not real consensus for the accuracy of MRI characteristics in non mass like breast lesions, and the number of malignant lesions in different studies is insufficient.

Objectives:

In this study we aimed to analyze the diagnostic role of MRI BIRADS features for diagnosis of malignancy in non mass like breast lesions.

Patients and Methods:

All patients with positive findings (BIRADS 3, 4, 5), which had either biopsy proved pathology or follow-up MRI data at least for 12 months were included in the study. Finally, 213 breasts MRI that showed non mass like enhancing lesions among our patients were assessed in study. One experienced breast radiologist who was unaware of any clinical information or the histopathologic diagnosis evaluated all images retrospectively. The morphologic parameters evaluated consisted of distribution modifiers and pattern of internal enhancement. The kinetic enhancement parameters were assessed as showing washout, plateau, or persistent patterns. In the enhancement kinetic analysis, thew most worrisome curve type in each lesion was considered for interpretation, if it was more than 2% enhancement. We have evaluated the visual findings by comparison of the signal intensity on the first and third dynamic series. Data for the study were extracted from the breast MRI database and analyzed using SPSS version 16 statistical software.

Results:

Totally 188 patients had 213 non mass like lesions. Mean age of the patients was 44.9 ± 8.3 years (24-63). Totally 46 of lesions were malignant (21.6%). The most common BIRADS score was 4 (116; 54.5%). The most prevalent feature of distribution, internal enhancement and curve type were focal (59.2%), clumped (27.2%) and washout (34.3%). Distribution of different subgroups of MR BIRADS features was different among benign and malignant lesions (All Pvalues < 0.05). Regarding association with malignancy, odds ratio of lesions with segmental or ductal linear distribution was 3.4 (95% CI = 1.7-6.8), Clumped, Reticular and Dendritic internal enhancement was 2.5 (95% CI = 1.3-5) and wash out curve type was 5.4 (95% CI = 2.7-10.9). Sensitivity of higher MR BIRADS (4,5) for diagnosis of malignancy was 100%. Specificity of segmental or ductal linear distribution in diagnosis of malignancy was 81%. Specificity of BIRADS 5 for diagnosis of malignancy was 98%. In a multivariate logistic regression analysis for diagnosis of malignancy in which distribution, internal enhancement and curve type were considered as independent variables, distribution and curve type remained significant in the model while the internal enhancement showed a borderline P-value.

Conclusions:

Although in our study washout pattern was the most powerful indicator for malignant pathology in non mass like enhancing lesions, more studies with larger sample size needs in this regard.

Adjacent vessel sign and breast imaging reporting and data system are valuable for diagnosis of benign and malignant breast lesions

The purpose of this study is to investigate whether an adjacent vessel sign (AVS) observed on the maximum intensity projections (MIPs) from the subtracted images can help distinguish between malignant and benign breast lesions and to test whether the combination of breast imaging reporting and data system (BI-RADS) category and AVS can increase the specificity and diagnostic accuracy of breast magnetic resonance imaging (MRI). The study included 63 histologically verified lesions which underwent dynamic breast MRI before biopsy. All magnetic resonance (MR) images were evaluated by two radiologists in consensus, who were unaware of the histopathological outcome. The MR images of all cases were analyzed according to BI-RADS-MRI assessment category. Levels of suspicion were reported as categories of I-V. The presence of vessels either entering the enhancing lesion or in contact with the lesion edge on MIP images was considered as the presence of AVS. Final analysis of 63 masses revealed 41 malignant lesions (65.1%) and 22 benign lesions (34.9%). Thirty seven out of 41 malignant lesions and 3 out of 22 benign lesions were associated with adjacent vessel, with highly significant difference between benign and malignant lesions (P < 0.001), especially for lesions smaller than 2.0 cm. The corresponding specificity, sensitivity and accuracy of contrast-enhanced 3.0-T breast were 86.4%, 82.9% and 84.1%, respectively. Based on BI-RADS-MRI category, the specificity, sensitivity and accuracy of breast MRI were 54.5%, 100% and 84.1%, respectively. After combining BI-RADS category and AVS, the specificity, sensitivity and accuracy of breast MRI were 90.9%, 82.9% and 85.7%, respectively. AVS can help differentiate malignant from benign breast lesions, especially for the lesions smaller than 2.0 cm. The combination of BI-RADS category and AVS can increase the specificity and the diagnostic accuracy of breast MRI.

Breast imaging reporting and data system (BI-RADS) lexicon for breast MRI: interobserver variability in the description and assignment of BI-RADS category

PURPOSE

To retrospectively evaluate interobserver variability between breast radiologists when describing abnormal enhancement on breast MR examinations and assigning a BI-RADS category using the Breast Imaging Reporting and Data System (BI-RADS) terminology.

MATERIALS AND METHODS

Five breast radiologists blinded to patients' medical history and pathologic results retrospectively and independently reviewed 257 abnormal areas of enhancement on breast MRI performed in 173 women. Each radiologist described the focal enhancement using BI-RADS terminology and assigned a final BI-RADS category. Krippendorff's α coefficient of agreement was used to asses interobserver variability.

RESULTS

All radiologists agreed on the morphology of enhancement in 183/257 (71%) lesions, yielding a substantial agreement (Krippendorff's α=0.71). Moderate agreement was obtained for mass descriptors - shape, margins and internal enhancement - (α=0.55, 0.51 and 0.45 respectively) and NME (non-mass enhancement) descriptors - distribution and internal enhancement - (α=0.54 and 0.43). Overall substantial agreement was obtained for BI-RADS category assignment (α=0.71). It was however only moderate (α=0.38) for NME compared to mass (α=0.80).

CONCLUSION

Our study shows good agreement in describing mass and NME on a breast MR examination but a better agreement in predicting malignancy for mass than NME.

Features of occult invasion in biopsy-proven DCIS at breast MRI

The purpose of this study is to determine if MRI BI-RADS criteria or radiologist perception correlate with presence of invasive cancer after initial core biopsy of ductal carcinoma in situ (DCIS). Retrospective search spanning 2000-2007 identified all core-biopsy diagnoses of pure DCIS that coincided with preoperative MRI. Two radiologists fellowship-trained in breast imaging categorized lesions according to ACR MRI BI-RADS lexicon and estimated likelihood of occult invasion. Semiquantitative signal enhancement ratio (SER) kinetic analysis was also performed. Results were compared with histopathology. 51 consecutive patients with primary core biopsy-proven DCIS and concurrent MRI were identified. Of these, 13 patients (25%) had invasion at excision. Invasion correlated significantly with presence of a mass for both readers (p = 0.012 and 0.001), rapid initial enhancement for Reader 1 (p = 0.001), and washout kinetics for Reader 2 (p = 0.012). Significant correlation between washout and invasion was confirmed by SER (p = 0.006) when threshold percent enhancement was sufficiently high (130%), corresponding to rapidly enhancing portions of the lesion. Radiologist perception of occult invasion was strongly correlated with true presence of invasion. These results provide evidence that certain BI-RADS MRI criteria, as well as radiologist perception, correlate with occult invasion after an initial core biopsy of DCIS.

Diffusion-weighted MRI: association between patient characteristics and apparent diffusion coefficients of normal breast fibroglandular tissue at 3 T

OBJECTIVE

The purpose of this study is to assess associations between patient characteristics and apparent diffusion coefficient (ADC) values of normal breast fibroglandular tissue on diffusion-weighted imaging (DWI) at 3 T.

MATERIALS AND METHODS

The retrospective study included 103 women with negative bilateral findings on 3-T breast MRI examinations (BI-RADS category 1). DWI was acquired during clinical breast MRI scans using b = 0 and b = 800 s/mm(2). Mean ADC of normal breast fibroglandular tissue was calculated for each breast using a semiautomated software tool in which parenchyma pixels were selected by interactive thresholding of the b = 0 s/mm(2) image to exclude fat. Intrasubject right- and left-breast ADC values were compared and averaged together to evaluate the association of mean breast ADC with age, mammographic breast density, and background parenchymal enhancement.

RESULTS

Overall mean ± SD breast ADC was 1.62 ± 0.30 × 10(-3) mm(2)/s. Intrasubject right- and left-breast ADC measurements were highly correlated (R(2) = 0.89; p < 0.0001). Increased breast density was strongly associated with increased ADC (p ≤ 0.0001). Age and background parenchymal enhancement were not associated with ADC.

CONCLUSION

Normal breast parenchymal ADC values increase with mammographic density but are independent of age and background parenchymal enhancement. Because breast malignancies have been shown to have low ADC values, DWI may be particularly valuable in women with dense breasts owing to greater contrast between lesion and normal tissue.

Provision of the DDSM mammography metadata in an accessible format

PURPOSE

The Digital Database for Screening Mammography (DDSM) is the largest publicly available resource for mammographic image analysis research and has been used extensively in the past for computer assisted diagnosis (CADx) studies. However, the database has not been searchable for a specific kind of lesion, which rendered the case selection process in past studies often times arbitrary. Therefore, the authors want to provide the complete metadata of the DDSM in an accessible format.

METHODS

The authors semiautomatically transformed the data available athttp://marathon.csee.usf.edu/Mammography/Database.html into table format. The 1769 cases (914 from cancer volumes, 855 from benign volumes) comprise 1220 mass lesions (578 benign, 642 malignant) and 859 calcifications (433 benign, 426 malignant). Additionally, 694 normal cases were processed to allow for matching according to age and breast density.

RESULTS

The authors provide the entire DDSM metadata (for benign, malignant, and normal cases) as tab-delimited text files[see supplementary material at http://dx.doi.org/10.1118/1.4870379E-MPHYA6-41-006405 for DDSM metadata].

CONCLUSIONS

The data provided make the case selection for future studies using the DDSM reproducible. Furthermore, it may serve as a validation dataset for CADx approaches using the BI-RADS lexicon.

The added diagnostic value of dynamic contrast-enhanced MRI at 3.0 T in nonpalpable breast lesions

Objective

To investigate the added diagnostic value of 3.0 Tesla breast MRI over conventional breast imaging in the diagnosis of in situ and invasive breast cancer and to explore the role of routine versus expert reading.

Materials and Methods

We evaluated MRI scans of patients with nonpalpable BI-RADS 3–5 lesions who underwent dynamic contrast-enhanced 3.0 Tesla breast MRI. Initially, MRI scans were read by radiologists in a routine clinical setting. All histologically confirmed index lesions were re-evaluated by two dedicated breast radiologists. Sensitivity and specificity for the three MRI readings were determined, and the diagnostic value of breast MRI in addition to conventional imaging was assessed. Interobserver reliability between the three readings was evaluated.

Results

MRI examinations of 207 patients were analyzed. Seventy-eight of 207 (37.7%) patients had a malignant lesion, of which 33 (42.3%) patients had pure DCIS and 45 (57.7%) invasive breast cancer. Sensitivity of breast MRI was 66.7% during routine, and 89.3% and 94.7% during expert reading. Specificity was 77.5% in the routine setting, and 61.0% and 33.3% during expert reading. In the routine setting, MRI provided additional diagnostic information over clinical information and conventional imaging, as the Area Under the ROC Curve increased from 0.76 to 0.81. Expert MRI reading was associated with a stronger improvement of the AUC to 0.87. Interobserver reliability between the three MRI readings was fair and moderate.

Conclusions

3.0 T breast MRI of nonpalpable breast lesions is of added diagnostic value for the diagnosis of in situ and invasive breast cancer.

Rate of malignancy in MRI-detected probably benign (BI-RADS 3) lesions

OBJECTIVE

The purpose of this study was to evaluate the malignancy rate in MRI-detected probably benign (BI-RADS 3) lesions in women without a history of breast cancer.

MATERIALS AND METHODS

In this study, 1265 patients underwent breast MRI during a 7-year period. One hundred and eight (8.5%) patients with a nonpalpable breast lesion classified as BI-RADS 3 at MRI and with a needle biopsy or adequate follow-up of at least 24 months were included. Statistical analysis included calculation of the negative predictive value with its 95% CI.

RESULTS

Of 108 lesions, 107 (99.1%) were correctly assessed as probably benign, resulting in a negative predictive value of 99.1% (95% CI, 94.99-99.98%). Histopathology was requested by the patient or referring physician in 44 patients. Of these, 43 (39.8%) lesions were classified as benign and one (0.9%) as malignant. There were no changes evident in any of the remaining 64 (59.2%) lesions during follow-up (range, 2-9 years).

CONCLUSION

In MRI-detected probably benign (BI-RADS 3) lesions, the malignancy rate is low and within the accepted cancer rate for mammographically or sonographically detected BI-RADS 3 lesions. Short-term follow-up MRI at intervals of 6, 12, and 24 months in MRI BI-RADS 3 lesions remains a strong tool with which to detect suspicious lesions. Interval changes in size, morphology, or enhancement are regarded as indicative of malignancy.

MRI and CT imaging for preoperative target volume delineation in breast-conserving therapy

BACKGROUND

Accurate tumor bed delineation after breast-conserving surgery is important. However, consistency among observers on standard postoperative radiotherapy planning CT is low and volumes can be large due to seroma formation. A preoperative delineation of the tumor might be more consistent. Therefore, the purpose of this study was to determine the consistency of preoperative target volume delineation on CT and MRI for breast-conserving radiotherapy.

METHODS

Tumors were delineated on preoperative contrast-enhanced (CE) CT and newly developed 3D CE-MR images, by four breast radiation oncologists. Clinical target volumes (CTVs) were created by addition of a 1.5 cm margin around the tumor, excluding skin and chest wall. Consistency in target volume delineation was expressed by the interobserver variability. Therefore, the conformity index (CI), center of mass distance (dCOM) and volumes were calculated. Tumor characteristics on CT and MRI were scored by an experienced breast radiologist.

RESULTS

Preoperative tumor delineation resulted in a high interobserver agreement with a high median CI for the CTV, for both CT (0.80) and MRI (0.84). The tumor was missed on CT in 2/14 patients (14%). Leaving these 2 patients out of the analysis, CI was higher on MRI compared to CT for the GTV (p<0.001) while not for the CTV (CT (0.82) versus MRI (0.84), p=0.123). The dCOM did not differ between CT and MRI. The median CTV was 48 cm3 (range 28-137 cm3) on CT and 59 cm3 (range 30-153 cm3) on MRI (p<0.001). Tumor shapes and margins were rated as more irregular and spiculated on CE-MRI.

CONCLUSIONS

This study showed that preoperative target volume delineation resulted in small target volumes with a high consistency among observers. MRI appeared to be necessary for tumor detection and the visualization of irregularities and spiculations. Regarding the tumor delineation itself, no clinically relevant differences in interobserver variability were observed. These results will be used to study the potential for future MRI-guided and neoadjuvant radiotherapy.

TRIAL REGISTRATION

International Clinical Trials Registry Platform NTR3198.