Separation of benign and malignant breast lesions using dynamic contrast enhanced MRI in a biopsy cohort

Digital reference object toolkit of breast DCE MRI for quantitative evaluation of image reconstruction and analysis methods

PURPOSE

To develop a digital reference object (DRO) toolkit to generate realistic breast DCE-MRI data for quantitative assessment of image reconstruction and data analysis methods.

METHODS

A simulation framework in a form of DRO toolkit has been developed using the ultrafast and conventional breast DCE-MRI data of 53 women with malignant (n = 25) or benign (n = 28) lesions. We segmented five anatomical regions and performed pharmacokinetic analysis to determine the ranges of pharmacokinetic parameters for each segmented region. A database of the segmentations and their pharmacokinetic parameters is included in the DRO toolkit that can generate a large number of realistic breast DCE-MRI data. We provide two potential examples for our DRO toolkit: assessing the accuracy of an image reconstruction method using undersampled simulated radial k-space data and assessing the impact of theB 1 + $$ {\mathrm{B}}_1^{+} $$ field inhomogeneity on estimated parameters.

RESULTS

The estimated pharmacokinetic parameters for each region showed agreement with previously reported values. For the assessment of the reconstruction method, it was found that the temporal regularization resulted in significant underestimation of estimated parameters by up to 57% and 10% with the weighting factor λ = 0.1 and 0.01, respectively. We also demonstrated that spatial discrepancy ofv p $$ {v}_p $$ andPS $$ \mathrm{PS} $$ increase to about 33% and 51% without correction forB 1 + $$ {\mathrm{B}}_1^{+} $$ field.

CONCLUSION

We have developed a DRO toolkit that includes realistic morphology of tumor lesions along with the expected pharmacokinetic parameter ranges. This simulation framework can generate many images for quantitative assessment of DCE-MRI reconstruction and analysis methods.

Clinical Breast MRI-based Radiomics for Distinguishing Benign and Malignant Lesions: An Analysis of Sequences and Enhanced Phases

BACKGROUND

Previous studies have used different imaging sequences and different enhanced phases for breast lesion calsification in radiomics. The optimal sequence and contrast enhanced phase is unclear.

PURPOSE

To identify the optimal magnetic resonance imaging (MRI) radiomics model for lesion clarification, and to simulate its incremental value for multiparametric MRI (mpMRI)-guided biopsy.

STUDY TYPE

Retrospective.

POPULATION

329 female patients (138 malignant, 191 benign), divided into a training set (first site, n = 192) and an independent test set (second site, n = 137).

FIELD STRENGTH/SEQUENCE

3.0-T, fast spoiled gradient-echo and fast spin-echo T1-weighted imaging (T1WI), fast spin-echo T2-weighted imaging (T2WI), echo-planar diffusion-weighted imaging (DWI), and fast spoiled gradient-echo contrast-enhanced MRI (CE-MRI).

ASSESSMENT

Two breast radiologists with 3 and 10 years' experience developed radiomics model on CE-MRI, CE-MRI + DWI, CE-MRI + DWI + T2WI, CE-MRI + DWI + T2WI + T1WI at each individual phase (P) and for multiple combinations of phases. The optimal radiomics model (Rad-score) was identified as having the highest area under the receiver operating characteristic curve (AUC) in the test set. Specificity was compared between a traditional mpMRI model and an integrated model (mpMRI + Rad-score) at sensitivity >98%.

STATISTICAL TESTS

Wilcoxon paired-samples signed rank test, Delong test, McNemar test. Significance level was 0.05 and Bonferroni method was used for multiple comparisons (P = 0.007, 0.05/7).

RESULTS

For radiomics models, CE-MRI/P3 + DWI + T2WI achieved the highest performance in the test set (AUC = 0.888, 95% confidence interval: 0.833-0.944). The integrated model had significantly higher specificity (55.3%) than the mpMRI model (31.6%) in the test set with a sensitivity of 98.4%.

DATA CONCLUSION

The CE-MRI/P3 + DWI + T2WI model is the optimized choice for breast lesion classification in radiomics, and has potential to reduce benign biopsies (100%-specificity) from 68.4% to 44.7% while retaining sensitivity >98%.

LEVEL OF EVIDENCE

3 TECHNICAL EFFICACY: Stage 2.

High-temporal resolution DCE-MRI improves assessment of intra- and peri-breast lesions categorized as BI-RADS 4

BACKGROUND

BI-RADS 4 breast lesions are suspicious for malignancy with a range from 2 to 95%, indicating that numerous benign lesions are unnecessarily biopsied. Thus, we aimed to investigate whether high-temporal-resolution dynamic contrast-enhanced MRI (H_DCE-MRI) would be superior to conventional low-temporal-resolution DCE-MRI (L_DCE-MRI) in the diagnosis of BI-RADS 4 breast lesions.

METHODS

This single-center study was approved by the IRB. From April 2015 to June 2017, patients with breast lesions were prospectively included and randomly assigned to undergo either H_DCE-MRI, including 27 phases, or L_DCE-MRI, including 7 phases. Patients with BI-RADS 4 lesions were diagnosed by the senior radiologist in this study. Using a two-compartment extended Tofts model and a three-dimensional volume of interest, several pharmacokinetic parameters reflecting hemodynamics, including K^trans, K_ep, V_e, and V_p, were obtained from the intralesional, perilesional and background parenchymal enhancement areas, which were labeled the Lesion, Peri and BPE areas, respectively. Models were developed based on hemodynamic parameters, and the performance of these models in discriminating between benign and malignant lesions was evaluated by receiver operating characteristic (ROC) curve analysis.

RESULTS

A total of 140 patients were included in the study and underwent H_DCE-MRI (n = 62) or L_DCE-MRI (n = 78) scans; 56 of these 140 patients had BI-RADS 4 lesions. Some pharmacokinetic parameters from H_DCE-MRI (Lesion_K^trans, K_ep, and V_p; Peri_K^trans, K_ep, and V_p) and from L_DCE-MRI (Lesion_K_ep, Peri_V_p, BPE_K^trans and BPE_V_p) were significantly different between benign and malignant breast lesions (P < 0.01). ROC analysis showed that Lesion_K^trans (AUC = 0.866), Lesion_K_ep (AUC = 0.929), Lesion_V_p (AUC = 0.872), Peri_K^trans (AUC = 0.733), Peri_K_ep (AUC = 0.810), and Peri_V_p (AUC = 0.857) in the H_DCE-MRI group had good discrimination performance. Parameters from the BPE area showed no differentiating ability in the H_DCE-MRI group. Lesion_K_ep (AUC = 0.767), Peri_V_p (AUC = 0.726), and BPE_K^trans and BPE_V_p (AUC = 0.687 and 0.707) could differentiate between benign and malignant breast lesions in the L_DCE-MRI group. The models were compared with the senior radiologist's assessment for the identification of BI-RADS 4 breast lesions. The AUC, sensitivity and specificity of Lesion_K_ep (0.963, 100.0%, and 88.9%, respectively) in the H_DCE-MRI group were significantly higher than those of the same parameter in the L_DCE-MRI group (0.663, 69.6% and 75.0%, respectively) for the assessment of BI-RADS 4 breast lesions. The DeLong test was conducted, and there was a significant difference only between Lesion_K_ep in the H_DCE-MRI group and the senior radiologist (P = 0.04).

CONCLUSIONS

Pharmacokinetic parameters (K^trans, K_ep and V_p) from the intralesional and perilesional regions on high-temporal-resolution DCE-MRI, especially the intralesional K_ep parameter, can improve the assessment of benign and malignant BI-RADS 4 breast lesions to avoid unnecessary biopsy.

Estimation of the capillary level input function for dynamic contrast-enhanced MRI of the breast using a deep learning approach

PURPOSE

To develop a deep learning approach to estimate the local capillary-level input function (CIF) for pharmacokinetic model analysis of DCE-MRI.

METHODS

A deep convolutional network was trained with numerically simulated data to estimate the CIF. The trained network was tested using simulated lesion data and used to estimate voxel-wise CIF for pharmacokinetic model analysis of breast DCE-MRI data using an abbreviated protocol from women with malignant (n = 25) and benign (n = 28) lesions. The estimated parameters were used to build a logistic regression model to detect the malignancy.

RESULT

The pharmacokinetic parameters estimated using the network-predicted CIF from our breast DCE data showed significant differences between the malignant and benign groups for all parameters. Testing the diagnostic performance with the estimated parameters, the conventional approach with arterial input function (AIF) showed an area under the curve (AUC) between 0.76 and 0.87, and the proposed approach with CIF demonstrated similar performance with an AUC between 0.79 and 0.81.

CONCLUSION

This study shows the feasibility of estimating voxel-wise CIF using a deep neural network. The proposed approach could eliminate the need to measure AIF manually without compromising the diagnostic performance to detect the malignancy in the clinical setting.

Quantitative transport mapping (QTM) for differentiating benign and malignant breast lesion: Comparison with traditional kinetics modeling and semi-quantitative enhancement curve characteristics

PURPOSE

To test the feasibility of using quantitative transport mapping (QTM) method, which is based on the inversion of transport equation using spatial deconvolution without any arterial input function, for automatically postprocessing dynamic contrast enhanced MRI (DCE-MRI) to differentiate malignant and benign breast tumors.

MATERIALS AND METHODS

Breast DCE-MRI data with biopsy confirmed malignant (n = 13) and benign tumors (n = 13) was used to assess QTM velocity (|u|) and diffusion coefficient (D), volume transfer constant (K^trans), volume fraction of extravascular extracellular space (V_e) from kinetics method, and traditional enhancement curve characteristics (ECC: amplitude A, wash-in rate α, wash-out rate β). A Mann-Whitney U test and receiver operating characteristic curve (ROC) analysis were performed to assess the diagnostic performance of these parameters for distinguishing between benign and malignant tumors.

RESULTS

Between malignant and benign tumors, there was a significant difference in |u| and K^trans, (p = 0.0066, 0.0274, respectively), but not in D, V_e, A, α and β (p = 0.1119, 0.2382, 0.4418,0.2592 and 0.9591, respectively). ROC area-under-the-curve was 0.82, 0.75 (95% confidence level 0.60-0.95, 0.51-0.90) for |u| and K^trans, respectively.

CONCLUSION

QTM postprocesses DCE-MRI automatically through deconvolution in space and time to solve the inverse problem of the transport equation. Comparing with traditional kinetics method and ECC, QTM method showed better diagnostic accuracy in differentiating benign from malignant breast tumors in this study.

Diagnostic performance of dynamic contrast-enhanced magnetic resonance imaging for breast cancer detection: An update meta-analysis

Objective

To evaluate the diagnostic performance of dynamic contrast‐enhanced magnetic resonance imaging (DCE‐MRI) for breast cancer identification.

Methods

A comprehensive electronic systematic searching of Medline, Ovid, EMBASE, Web of Science, CNK, and Cochrane Library databases was performed up to 2 August 2021. Clinical studies associated with DCE‐MRI for breast cancer detection were screened and inlcuded in the meta‐analysis. The data of true positive(tp), false positive(fp), false negative(fn) and true negative(tn) was extracted from includded studies. The sensitivity, specificity, diagnostic odds ratio (DOR), and area under the summary receiver operating characteristic (SROC) were pooled under fixed or random effect models. Publication bias was evaluated by Deek's funnel plot.

Results

A final set of 15 studies with 1321 breast lesions were included in the present work. The pooled diagnostic sensitivity, specificity, and DOR were 0.87 (95% confidence interval [CI] 0.81–0.92), 0.74 (95% CI 0.68–0.80), and 18.83 (95% CI 9.07–36.54), respectively, and the area under the SROC was 0.86 (95% CI 0.82–0.88). Given a pretest probability of 50%, the positive post‐test probability was 77%, and the negative post‐test probability was 14%. Deek's funnel plot indicated low publication bias (p = 0.61).

Conclusion

DCE‐MRI is a noninvasive method of breast cancer diagnosis for suspected malignant breast lesions with relative high diagnostic sensitivity and specificity.

DCE-MRI quantitative transport mapping for noninvasively detecting hypoxia inducible factor-1α, epidermal growth factor receptor overexpression, and Ki-67 in nasopharyngeal carcinoma patients

BACKGROUND

Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) has the potential to noninvasively detect expression of hypoxia inducible factor-1-alpha (HIF-1α), epidermal growth factor receptor (EGFR), and Ki-67 in nasopharyngeal carcinoma (NPC) by quantitatively measuring tumor blood flow, vascularity, and permeability.

PURPOSE

We aim to explore the utility of DCE-MRI in detecting HIF-1α, EGFR, and Ki-67 expression levels using traditional Kety's/Tofts' modeling and quantitative transport mapping (QTM).

MATERIALS AND METHODS

Eighty-nine NPC patients underwent DCE-MRI before treatment were enrolled. DCE-MRI was processed to generate the following kinetic parameters: |u| and D from the QTM model, tumor blood flow (TBF) from Kety's model, and K^trans, V_e, and K_ep from Tofts' model. Pretreatment levels of HIF-1α, EGFR, and Ki-67 were assessed by immunohistochemistry and classified into low and high expression groups.

RESULTS

|u| (p < 0.001) and TBF (p = 0.015) values were significantly higher in the HIF-1α high-expression group compared to low-expression group. Only K^trans (p = 0.016) was significantly higher in the EGFR high-expression group. Only |u| (p < 0.001) values were significantly higher in the Ki-67 high-expression group compared to low-expression group. Multiple linear regression analyses showed that |u| independently correlated with HIF-1α and Ki-67 expression, and K^trans independently correlated with EGFR. The areas under the ROC curves of |u| for HIF-1α and Ki-67, and K^trans for EGFR were 0.83, 0.74, and 0.70, respectively.

CONCLUSION

|u| and K^trans derived from DCE-MRI may be considered as noninvasive imaging markers for detecting hypoxia and proliferation in NPC patients.

Diagnostic performance of perilesional radiomics analysis of contrast-enhanced mammography for the differentiation of benign and malignant breast lesions

OBJECTIVE

To conduct perilesional region radiomics analysis of contrast-enhanced mammography (CEM) images to differentiate benign and malignant breast lesions.

METHODS AND MATERIALS

This retrospective study included patients who underwent CEM from November 2017 to February 2020. Lesion contours were manually delineated. Perilesional regions were automatically obtained. Seven regions of interest (ROIs) were obtained for each lesion, including the lesion ROI, annular perilesional ROIs (1 mm, 3 mm, 5 mm), and lesion + perilesional ROIs (1 mm, 3 mm, 5 mm). Overall, 4,098 radiomics features were extracted from each ROI. Datasets were divided into training and testing sets (1:1). Seven classification models using features from the seven ROIs were constructed using LASSO regression. Model performance was assessed by the AUC with 95% CI.

RESULTS

Overall, 190 women with 223 breast lesions (101 benign; 122 malignant) were enrolled. In the testing set, the annular perilesional ROI of 3-mm model showed the highest AUC of 0.930 (95% CI: 0.882-0.977), followed by the annular perilesional ROI of 1 mm model (AUC = 0.929; 95% CI: 0.881-0.978) and the lesion ROI model (AUC = 0.909; 95% CI: 0.857-0.961). A new model was generated by combining the predicted probabilities of the lesion ROI and annular perilesional ROI of 3-mm models, which achieved a higher AUC in the testing set (AUC = 0.940).

CONCLUSIONS

Annular perilesional radiomics analysis of CEM images is useful for diagnosing breast cancers. Adding annular perilesional information to the radiomics model built on the lesion information may improve the diagnostic performance.

KEY POINTS

• Radiomics analysis of the annular perilesional region of 3 mm in CEM images may provide valuable information for the differential diagnosis of benign and malignant breast lesions. • The radiomics information from the lesion region and the annular perilesional region may be complementary. Combining the predicted probabilities of the models constructed by the features from the two regions may improve the diagnostic performance of radiomics models.

Apparent diffusion coefficient values in borderline breast lesions upgraded and not upgraded at definitive histopathological examination after surgical excision

Purpose

The study aims were to evaluate if the apparent diffusion coefficient (ADC) value could distinguish between breast lesions classified as B3 at core needle biopsy (CNB) that show or do not show atypia or malignancy at definitive histopathological examination (DHE) after surgical excision.

Material and methods

From January 2013 to December 2017, 141 patients with a B3 breast lesion underwent magnetic resonance imaging and were included in the study. The ADC value was assessed drawing a ROI outlining the entire lesion, evaluating the mean (ADC_mean) and minimum ADC values (ADC_min).

Results

Both ADC_mean and ADC_min values showed a statistically significant difference between B3 lesions without and with malignancy or, for B3a lesions, atypia at DHE. They both showed a statistically significant difference also between B3a lesions without or with atypia or malignancy at DHE, but only ADC_min (not ADC_mean) showed statistically significant difference between B3b lesions without or with malignancy at DHE.

Conclusions

The ADC value could help distinguish between B3a lesions without or with atypia/malignancy at DHE after surgical excision and between B3b lesions without or with malignancy at DHE. Therefore, it could be used to help guide the diagnostic-therapeutic pathway of these lesions, particularly of B3a lesions.

Quantitative perfusion parameters of benign inflammatory breast pathologies: A descriptive study

PURPOSE

With this study, we evaluated the perfusion magnetic resonance imaging (MRI) features of benign inflammatory breast lesions for the first time and compared their K_trans, K_ep, V_e values and contrast kinetic curves to benign masses and invasive ductal carcinoma (IDC).

MATERIALS AND METHODS

Perfusion MRIs of the benign masses (n = 42), inflammatory lesions (n = 25), and IDCs (n = 16) were evaluated retrospectively in terms of K_trans, K_ep, V_e values and contrast kinetic curves and compared by the Kruskal-Wallis, Mann-Whitney U, chi-square tests statistically. Cronbach α test was used to measure intraobserver and interobserver reliability.

RESULTS

Mean K_trans values were 0.052 for benign masses, 0.086 for inflammatory lesions and 0.101 for IDC (p < 0.001). Mean K_ep values were 0.241 for benign masses, 0.435 for inflammatory lesions and 0.530 for IDC (p < 0.001). Mean V_e values were 0.476 for benign masses, 0.318 for inflammatory lesions and 0.310 for IDC (p = 0.067). For inflammatory and IDC lesions, K_trans and K_ep values were found to be higher and V_e values were lower than benign masses (p = 0.001 for K_trans, p = 0.001 for K_ep, p = 0.045 for V_e). There were excellent or good intra-interobserver reliabilities. For the kinetic curve pattern, most of the benign lesions showed progressive (81%), inflammatory lesions progressive (64%) and IDC lesions plateau (75%) patterns (p < 0.001).

CONCLUSIONS

On T1 perfusion MRI, similar to IDC lesions, inflammatory lesions demonstrate higher K_trans and K_ep and lower V_e values than benign masses. Quantitative perfusion parameters are not helpful in differentiating them from IDC lesions.

Breast dynamic contrast-enhanced-magnetic resonance imaging and radiomics: State of art

Breast cancer represents the most common malignancy in women, being one of the most frequent cause of cancer-related mortality. Ultrasound, mammography, and magnetic resonance imaging (MRI) play a pivotal role in the diagnosis of breast lesions, with different levels of accuracy. Particularly, dynamic contrast-enhanced MRI has shown high diagnostic value in detecting multifocal, multicentric, or contralateral breast cancers. Radiomics is emerging as a promising tool for quantitative tumor evaluation, allowing the extraction of additional quantitative data from radiological imaging acquired with different modalities. Radiomics analysis may provide novel information through the quantification of lesions heterogeneity, that may be relevant in clinical practice for the characterization of breast lesions, prediction of tumor response to systemic therapies and evaluation of prognosis in patients with breast cancers. Several published studies have explored the value of radiomics with good-to-excellent diagnostic and prognostic performances for the evaluation of breast lesions. Particularly, the integrations of radiomics data with other clinical and histopathological parameters have demonstrated to improve the prediction of tumor aggressiveness with high accuracy and provided precise models that will help to guide clinical decisions and patients management. The purpose of this article in to describe the current application of radiomics in breast dynamic contrast-enhanced MRI.

Assessment of tumor treatment response using active contrast encoding (ACE)-MRI: Comparison with conventional DCE-MRI

Purpose

To investigate the validity of contrast kinetic parameter estimates from Active Contrast Encoding (ACE)-MRI against those from conventional Dynamic Contrast-Enhanced (DCE)-MRI for evaluation of tumor treatment response in mouse tumor models.

Methods

The ACE-MRI method that incorporates measurement of T₁ and B₁ into the enhancement curve washout region, was implemented on a 7T MRI scanner to measure tracer kinetic model parameters of 4T1 and GL261 tumors with treatment using bevacizumab and 5FU. A portion of the same ACE-MRI data was used for conventional DCE-MRI data analysis with a separately measured pre-contrast T₁ map. Tracer kinetic model parameters, such as K^trans (permeability area surface product) and v_e (extracellular space volume fraction), estimated from ACE-MRI were compared with those from DCE-MRI, in terms of correlation and Bland-Altman analyses.

Results

A three-fold increase of the median K^trans by treatment was observed in the flank 4T1 tumors by both ACE-MRI and DCE-MRI. In contrast, the brain tumors did not show a significant change by the treatment in either ACE-MRI or DCE-MRI. K^trans and v_e values of the tumors from ACE-MRI were strongly correlated with those from DCE-MRI methods with correlation coefficients of 0.92 and 0.78, respectively, for the median values of 17 tumors. The Bland-Altman plot analysis showed a mean difference of -0.01 min^-1 for K^trans with the 95% limits of agreement of -0.12 min^-1 to 0.09 min^-1, and -0.05 with -0.37 to 0.26 for v_e.

Conclusion

The tracer kinetic model parameters estimated from ACE-MRI and their changes by treatment closely matched those of DCE-MRI, which suggests that ACE-MRI can be used in place of conventional DCE-MRI for tumor progression monitoring and treatment response evaluation with a reduced scan time.

Magnetic resonance imaging semantic and quantitative features analyses: an additional diagnostic tool for breast phyllodes tumors

OBJECTIVE

This study aimed to differentiate benign and non-benign (borderline/malignant) phyllodes tumors of the breast by the semantic and quantitative features in magnetic resonance imaging (MRI).

METHODS

The female patients, diagnosed with phyllodes tumors by MRI and pathological test, were retrospectively selected from December, 2006 to April, 2019. The MRI of benign, borderline and malignant phyllodes tumors was analyzed using 8 semantic features and 20 computed quantitative features from diffuse contrast-enhanced magnetic resonance imaging (DCE-MRI). The semantic features were analyzed by univariate analysis. The least absolute shrinkage and selection operator (LASSO) method was used to identify the optimal subset of MRI quantitative features. According to the results from multivariate logistic regression for the semantic and quantitative features, the model was constructed to differentiate benign and non-benign (borderline/malignant) phyllodes tumors.

RESULTS

Thirty-two benign (58.18%), 13 borderline (23.64%) and 10 malignant (18.18%) phyllodes tumors were identified in 54 patients. Five semantic features were proved to be significantly correlated with pathologic grade, including size, the T1 weighted image signal intensity, fat-saturated T2-weighted image signal intensity, enhanced signal intensity, and kinetic curve pattern. With the analysis of LASSO method, three quantitative texture features with significant predictive ability were selected. The model combining both the semantic and quantitative features was proved to have good performance in differentiation on phyllodes tumors, yielding an area under receiver operating characteristic curve, accuracy, sensitivity and specificity of 0.893, 0.933, 1.000, and 0.818, respectively.

CONCLUSION

The constructed model based on the semantic and quantitative features of DCE-MRI can significantly improve the differential diagnosis of phyllodes tumors in breast.

Advances in Diffusion and Perfusion MRI for Quantitative Cancer Imaging

PURPOSE OF REVIEW

This article is to review recent technical developments and their clinical applications in cancer imaging quantitative measurement of cellular and vascular properties of the tumors.

RECENT FINDINGS

Rapid development of fast Magnetic Resonance Imaging (MRI) technologies over last decade brought new opportunities in quantitative MRI methods to measure both cellular and vascular properties of tumors simultaneously.

SUMMARY

Diffusion MRI (dMRI) and dynamic contrast enhanced (DCE)-MRI have become widely used to assess the tissue structural and vascular properties, respectively. However, the ultimate potential of these advanced imaging modalities has not been fully exploited. The dependency of dMRI on the diffusion weighting gradient strength and diffusion time can be utilized to measure tumor perfusion, cellular structure, and cellular membrane permeability. Similarly, DCE-MRI can be used to measure vascular and cellular membrane permeability along with cellular compartment volume fractions. To facilitate the understanding of these potentially important methods for quantitative cancer imaging, we discuss the basic concepts and recent developments, as well as future directions for further development.

Machine learning in breast MRI

Machine-learning techniques have led to remarkable advances in data extraction and analysis of medical imaging. Applications of machine learning to breast MRI continue to expand rapidly as increasingly accurate 3D breast and lesion segmentation allows the combination of radiologist-level interpretation (eg, BI-RADS lexicon), data from advanced multiparametric imaging techniques, and patient-level data such as genetic risk markers. Advances in breast MRI feature extraction have led to rapid dataset analysis, which offers promise in large pooled multiinstitutional data analysis. The object of this review is to provide an overview of machine-learning and deep-learning techniques for breast MRI, including supervised and unsupervised methods, anatomic breast segmentation, and lesion segmentation. Finally, it explores the role of machine learning, current limitations, and future applications to texture analysis, radiomics, and radiogenomics. Level of Evidence: 3 Technical Efficacy Stage: 2 J. Magn. Reson. Imaging 2019. J. Magn. Reson. Imaging 2020;52:998-1018.

Correlation between DCE-MRI radiomics features and Ki-67 expression in invasive breast cancer

The aim of the present study was to investigate the association between Ki-67 expression and radiomics features of dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) in patients with invasive breast cancer. A total of 53 cases with low-Ki-67 expression (Ki-67 proliferation index <14%) and 106 cases with high-Ki-67 expression (Ki-67 proliferation index >14%) were investigated. A systematic approach was applied that focused on the automated segmentation of lesions and extraction of radiomics features. For each lesion 5 morphology, 4 gray-scale histogram and 6 texture features were obtained, and statistical analyzes were performed to assess the differences in these features between the low- and high-Ki-67 expressions. One morphology metric (area), 3 gray-scale histogram indexes (standard deviation, skewness and kurtosis) and 3 texture features (contrast, homogeneity and inverse differential moment) demonstrated a significant difference (P<0.05), with low-Ki-67 expression lesions tending to be smaller, clearer and heterogeneous when compared with the high-Ki-67 expressed cases. These results may provide a noninvasive means to better understand the proliferation of breast cancer.

Accuracy and precision of quantitative DCE-MRI parameters: How should one estimate contrast concentration?

INTRODUCTION

Pharmacokinetic parameters derived from dynamic contrast-enhanced MRI (DCE-MRI) data are sensitive to acquisition and post-processing techniques, which makes it difficult to compare results obtained using different methods. In particular, one of the most important factors affecting estimation of model parameters is how to convert MRI signal intensities to contrast agent concentration. The purpose of our study was to quantitatively compare a linear signal-to-concentration conversion (LC) as an approximation and a non-linear conversion (NLC) based on the MRI signal equation, in terms of the accuracy and precision of the pharmacokinetic parameters in T₁-weighted DCE-MRI.

MATERIALS AND METHODS

Numerical simulation studies were conducted to compare LC and NLC in terms of the accuracy and precision in contrast kinetic parameter estimation, and to evaluate their dependency on flip angle (FA), pre-contrast T₁ (T₁₀) and arterial input function (AIF). In addition, the effect of the conversion method on the diagnostic accuracy was evaluated with 36 breast lesions (19 benign and 17 malignant).

RESULTS

The transfer rate (K^trans) estimated using LC and measured AIF (mAIF) were up to 38% higher than the true K^trans values, while the LC K^trans estimates with the presumed AIF (pAIF) were up to 7% lower than the true K^trans values, when FA = 45°. When using a small FA, such as 12°, the LC K^trans with pAIF had least sensitivity to the error in T₁₀ compared to the K^trans estimated using LC with mAIF, and NLC with pAIF or mAIF. The breast DCE-MRI study showed that both LC and NLC K^trans were significantly different (p < 0.05) between the malignant and benign lesions. The effect size between benign and malignant values as measured by Cohen's d was 1.06 for LC K^trans and 1.02 for NLC K^trans.

CONCLUSION

The present study results show that, when precontrast T₁ measurement is not available and a low FA is used for DCE-MRI, the uncertainty in the contrast kinetic parameter estimation can be reduced by using the LC method with pAIF, without compromising the diagnostic accuracy.

Preoperative predicting malignancy in breast mass-like lesions: value of adding histogram analysis of apparent diffusion coefficient maps to dynamic contrast-enhanced magnetic resonance imaging for improving confidence level

OBJECTIVE

This study aims to find out the benefits of adding histogram analysis of apparent diffusion coefficient (ADC) maps onto dynamic contrast-enhanced MRI (DCE-MRI) in predicting breast malignancy.

METHODS

This study included 95 patients who were found with breast mass-like lesions from January 2014 to March 2016 (47 benign and 48 malignant). These patients were estimated by both DCE-MRI and diffusion-weighted imaging (DWI) and classified into two groups, namely, the benign and the malignant. Between these groups, the DCE-MRI parameters, including morphology, enhancement homogeneity, maximum slope of increase (MSI) and time-signal intensity curve (TIC) type, as well as histogram parameters generated from ADC maps were compared. Then, univariate and multivariate logistic regression analyses were conducted to determine the most valuable variables in predicting malignancy. Receiver operating characteristic curve analyses were taken to assess their clinical values.

RESULTS

The lesion morphology, MSI and TIC Type (p < 0.05) were significantly different between the two groups. Multivariate logistic regression analyses revealed that irregular morphology, TIC Type II/III and ADC₁₀ were important predictors for breast malignancy. Increased area under curve (AUC) and specificity can be achieved with Model 2 (irregular morphology + TIC Type II/III + ADC₁₀ < 1.047 ×10^-3 mm² s^-1) as the criterion than Model 1 (irregular morphology + TIC Type II/III) only (Model 2 vs Model 1; AUC, 0.822 vs 0.705; sensitivity, 68.8 vs 75.0%; specificity, 95.7 vs 66.0%).

CONCLUSION

Irregular morphology, TIC Type II/III and ADC₁₀ are indicators for predicting breast malignancy. Histogram analysis of ADC maps can provide additional value in predicting breast malignancy. Advances in knowledge: The morphology, MSI and TIC types in DCE-MRI examination have significant difference between the benign and malignant groups. A higher AUC can be achieved by using ADC₁₀ as the diagnostic index than other ADC parameters, and the difference in AUC based on ADC₁₀ and ADC_mean was statistically significant. The irregular morphology, TIC Type II/III and ADC₁₀ were significant predictors for malignant lesions.

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.

Automation of pattern recognition analysis of dynamic contrast-enhanced MRI data to characterize intratumoral vascular heterogeneity

PURPOSE

To automate dynamic contrast-enhanced MRI (DCE-MRI) data analysis by unsupervised pattern recognition (PR) to enable spatial mapping of intratumoral vascular heterogeneity.

METHODS

Three steps were automated. First, the arrival time of the contrast agent at the tumor was determined, including a calculation of the precontrast signal. Second, four criteria-based algorithms for the slice-specific selection of number of patterns (NP) were validated using 109 tumor slices from subcutaneous flank tumors of five different tumor models. The criteria were: half area under the curve, standard deviation thresholding, percent signal enhancement, and signal-to-noise ratio (SNR). The performance of these criteria was assessed by comparing the calculated NP with the visually determined NP. Third, spatial assignment of single patterns and/or pattern mixtures was obtained by way of constrained nonnegative matrix factorization.

RESULTS

The determination of the contrast agent arrival time at the tumor slice was successfully automated. For the determination of NP, the SNR-based approach outperformed other selection criteria by agreeing >97% with visual assessment. The spatial localization of single patterns and pattern mixtures, the latter inferring tumor vascular heterogeneity at subpixel spatial resolution, was established successfully by automated assignment from DCE-MRI signal-versus-time curves.

CONCLUSION

The PR-based DCE-MRI analysis was successfully automated to spatially map intratumoral vascular heterogeneity. Magn Reson Med 79:1736-1744, 2018. © 2017 International Society for Magnetic Resonance in Medicine.