Objectives
Due to its high sensitivity, DCE MRI of the breast (bMRI) is increasingly used for both screening and assessment purposes. The high number of detected lesions poses a significant logistic ...challenge in clinical practice. The aim was to evaluate a temporally and spatially resolved (4D) radiomics approach to distinguish benign from malignant enhancing breast lesions and thereby avoid unnecessary biopsies.
Methods
This retrospective study included consecutive patients with MRI-suspicious findings (BI-RADS 4/5). Two blinded readers analyzed DCE images using a commercially available software, automatically extracting BI-RADS curve types and pharmacokinetic enhancement features. After principal component analysis (PCA), a neural network–derived A.I. classifier to discriminate benign from malignant lesions was constructed and tested using a random split simple approach. The rate of avoidable biopsies was evaluated at exploratory cutoffs (C
1
, 100%, and C
2
, ≥ 95% sensitivity).
Results
Four hundred seventy (295 malignant) lesions in 329 female patients (mean age 55.1 years, range 18–85 years) were examined. Eighty-six DCE features were extracted based on automated volumetric lesion analysis. Five independent component features were extracted using PCA. The A.I. classifier achieved a significant (
p
< .001) accuracy to distinguish benign from malignant lesion within the test sample (AUC: 83.5%; 95% CI: 76.8–89.0%). Applying identified cutoffs on testing data not included in training dataset showed the potential to lower the number of unnecessary biopsies of benign lesions by 14.5% (C
1
) and 36.2% (C
2
).
Conclusion
The investigated automated 4D radiomics approach resulted in an accurate A.I. classifier able to distinguish between benign and malignant lesions. Its application could have avoided unnecessary biopsies.
Key Points
• Principal component analysis of the extracted volumetric and temporally resolved (4D) DCE markers favored pharmacokinetic modeling derived features.
• An A.I. classifier based on 86 extracted DCE features achieved a good to excellent diagnostic performance as measured by the area under the ROC curve with 80.6% (training dataset) and 83.5% (testing dataset).
• Testing the resulting A.I. classifier showed the potential to lower the number of unnecessary biopsies of benign breast lesions by up to 36.2%, p < .001 at the cost of up to 4.5% (n = 4) false negative low-risk cancers.
Available data proving the value of DWI for breast cancer diagnosis is mainly for enhancing masses; DWI may be less sensitive and specific in non-mass enhancement (NME) lesions. The objective of this ...study was to assess the diagnostic accuracy of DWI using different ROI measurement approaches and ADC metrics in breast lesions presenting as NME lesions on dynamic contrast-enhanced (DCE) MRI.
In this retrospective study, 95 patients who underwent multiparametric MRI with DCE and DWI from September 2007 to July 2013 and who were diagnosed with a suspicious NME (BI-RADS 4/5) were included. Twenty-nine patients were excluded for lesion non-visibility on DWI (n = 24: 12 benign and 12 malignant) and poor DWI quality (n = 5: 1 benign and 4 malignant). Two readers independently assessed DWI and DCE-MRI findings in two separate randomized readings using different ADC metrics and ROI approaches. NME lesions were classified as either benign (> 1.3 × 10
mm
/s) or malignant (≤ 1.3 × 10
mm
/s). Histopathology was the standard of reference. ROC curves were plotted, and AUCs were determined. Concordance correlation coefficient (CCC) was measured.
There were 39 malignant (59%) and 27 benign (41%) lesions in 66 (65 women, 1 man) patients (mean age, 51.8 years). The mean ADC value of the darkest part of the tumor (Dptu) achieved the highest diagnostic accuracy, with AUCs of up to 0.71. Inter-reader agreement was highest with Dptu ADC max (CCC 0.42) and lowest with the point tumor (Ptu) ADC min (CCC = - 0.01). Intra-reader agreement was highest with Wtu ADC mean (CCC = 0.44 for reader 1, 0.41 for reader 2), but this was not associated with the highest diagnostic accuracy.
Diagnostic accuracy of DWI with ADC mapping is limited in NME lesions. Thirty-one percent of lesions presenting as NME on DCE-MRI could not be evaluated with DWI, and therefore, DCE-MRI remains indispensable. Best results were achieved using Dptu 2D ROI measurement and ADC mean.
Purpose
To assess the additional value of quantitative tCho evaluation to diagnose malignancy and lymph node metastases in suspicious lesions on multiparametric breast MRI (mpMRI, BI-RADS 4, and ...BI-RADS 5).
Methods
One hundred twenty-one patients that demonstrated suspicious multiparametric breast MRI lesions using DCE, T2w, and diffusion-weighted (DW) images were prospectively enrolled in this IRB-approved study. All underwent single-voxel proton MR spectroscopy (
1
H-MRS, point-resolved spectroscopy sequence, TR 2000 ms, TE 272 ms) with and without water suppression. The total choline (tCho) amplitude was measured and normalized to millimoles/liter according to established methodology by two independent readers (R1, R2). ROC-analysis was employed to predict malignancy and lymph node status by tCho results.
Results
One hundred three patients with 74 malignant and 29 benign lesions had full
1
H-MRS data. The area under the ROC curve (AUC) for prediction of malignancy was 0.816 (R1) and 0.809 (R2). A cutoff of 0.8 mmol/l tCho could diagnose malignancy with a sensitivity of > 95%. For prediction of lymph node metastases, tCho measurements achieved an AUC of 0.760 (R1) and 0.788 (R2). At tCho levels < 2.4 mmol/l, no metastatic lymph nodes were found.
Conclusion
Quantitative tCho evaluation from
1
H-MRS allowed diagnose malignancy and lymph node status in breast lesions suspicious on multiparametric breast MRI. tCho therefore demonstrated the potential to downgrade suspicious mpMRI lesions and stratify the risk of lymph node metastases for improved patient management.
Key Points
• Quantitative tCho evaluation can distinguish benign from malignant breast lesions suspicious after multiparametric MRI assessment.
• Quantitative tCho levels are associated with lymph node status in breast cancer.
• Quantitative tCho levels are higher in hormonal receptor positive compared to hormonal receptor negative lesions.
Objectives
To evaluate the diagnostic performance in the assessment setting of three protocols: one-view wide-angle digital breast tomosynthesis (WA-DBT) with synthetic mammography (SM), two-view ...WA-DBT/SM, and two-view digital mammography (DM).
Methods
Included in this retrospective study were patients who underwent bilateral two-view DM and WA-DBT. SM were reconstructed from the WA-DBT data. The standard of reference was histology and/or 2 years follow-up. Included were 205 women with 179 lesions (89 malignant, 90 benign). Four blinded readers randomly evaluated images to assess density, lesion type, and level of suspicion according to BI-RADS. Three protocols were evaluated: two-view DM, one-view (mediolateral oblique) WA-DBT/SM, and two-view WA-DBT/SM. Detection rate, sensitivity, specificity, and accuracy were calculated and compared using multivariate analysis. Reading time was assessed.
Results
The detection rate was higher with two-view WA-DBT/SM (
p
= 0.063). Sensitivity was higher for two-view WA-DBT/SM compared to two-view DM (
p
= 0.001) and one-view WA-DBT/SM (
p
= 0.058). No significant differences in specificity were found. Accuracy was higher with both one-view WA-DBT/SM and two-view WA-DBT/SM compared to DM (
p
= 0.003 and > 0.001, respectively). Accuracy did not differ between one- and two-view WA-DBT/SM. Two-view WA-DBT/SM performed better for masses and asymmetries. Reading times were significantly longer when WA-DBT was evaluated.
Conclusions
One-view and two-view WA-DBT/SM can achieve a higher diagnostic performance compared to two-view DM. The detection rate and sensitivity were highest with two-view WA-DBT/SM. Two-view WA-DBT/SM appears to be the most appropriate tool for the assessment of breast lesions.
Key Points
• Detection rate with two-view wide-angle digital breast tomosynthesis (WA-DBT) is significantly higher than with two-view digital mammography in the assessment setting.
• Diagnostic accuracy of one-view and two-view WA-DBT with synthetic mammography (SM) in the assessment setting is higher than that of two-view digital mammography.
• Compared to one-view WA-DBT with SM, two-view WA-DBT with SM seems to be the most appropriate tool for the assessment of breast lesions.
Objectives
AI-based algorithms for medical image analysis showed comparable performance to human image readers. However, in practice, diagnoses are made using multiple imaging modalities alongside ...other data sources. We determined the importance of this multi-modal information and compared the diagnostic performance of routine breast cancer diagnosis to breast ultrasound interpretations by humans or AI-based algorithms.
Methods
Patients were recruited as part of a multicenter trial (NCT02638935). The trial enrolled 1288 women undergoing routine breast cancer diagnosis (multi-modal imaging, demographic, and clinical information). Three physicians specialized in ultrasound diagnosis performed a second read of all ultrasound images. We used data from 11 of 12 study sites to develop two machine learning (ML) algorithms using unimodal information (ultrasound features generated by the ultrasound experts) to classify breast masses which were validated on the remaining study site. The same ML algorithms were subsequently developed and validated on multi-modal information (clinical and demographic information plus ultrasound features). We assessed performance using area under the curve (AUC).
Results
Of 1288 breast masses, 368 (28.6%) were histopathologically malignant. In the external validation set (
n
= 373), the performance of the two unimodal ultrasound ML algorithms (AUC 0.83 and 0.82) was commensurate with performance of the human ultrasound experts (AUC 0.82 to 0.84;
p
for all comparisons > 0.05). The multi-modal ultrasound ML algorithms performed significantly better (AUC 0.90 and 0.89) but were statistically inferior to routine breast cancer diagnosis (AUC 0.95,
p
for all comparisons ≤ 0.05).
Conclusions
The performance of humans and AI-based algorithms improves with multi-modal information.
Key Points
• The performance of humans and AI-based algorithms improves with multi-modal information.
• Multimodal AI-based algorithms do not necessarily outperform expert humans.
• Unimodal AI-based algorithms do not represent optimal performance to classify breast masses.
To analyze the rate of potentially avoidable needle biopsies in mammographically suspicious calcifications if supplementary Contrast-Enhanced MRI (CE-MRI) is negative.
Using predefined criteria, a ...systematic review was performed. Studies investigating the use of supplemental CE-MRI in the setting of mammographically suspicious calcifications undergoing stereotactic biopsy and published between 2000 and 2020 were eligible. Two reviewers extracted study characteristics and true positives (TP), false positives, true negatives and false negatives (FN). Specificity, in this setting equaling the number of avoidable biopsies and FN rates were calculated. The maximum pre-test probability at which post-test probabilities of a negative CE-MRI met with BI-RADS benchmarks was determined by a Fagan nomogram. Random-effects models, I2-statistics, Deek’s funnel plot testing and meta-regression were employed. P-values <0.05 were considered significant.
Thirteen studies investigating 1414 lesions with a cancer prevalence of 43.6% (range: 22.7–66.9%) were included. No publication bias was found (P = 0.91). CE-MRI performed better in pure microcalcification studies compared to those also including associate findings (P < 0.001). In the first group, the pooled rate of avoidable biopsies was 80.6% (95%-CI: 64.6–90.5%) while the overall and invasive cancer FN rates were 3.7% (95%-CI: 1.2–6.2%) and 1.6% (95%-CI 0–3.6%), respectively. Up to a pre-test probability of 22%, the post-test probability did not exceed 2%.
A negative supplementary CE-MRI could potentially avoid 80.6% of unnecessary stereotactic biopsies in BI-RADS 4 microcalcifications at a cost of 3.7% missed breast cancers, 1.6% invasive. BI-RADS benchmarks for downgrading mammographic calcifications would be met up to a pretest probability of 22%.
•A negative breast MRI can downgrade up to 80.6% of suspicious microcalcifications, potentially avoiding vacuum-assisted breast biopsies.•Up to a pretest probability of 22% , a negative breast MRI result would not exceed the 2% cancer rate required for a BI-RADS 3 category assignment.
Diffusion‐weighted MRI (DWI) provides insights into tissue microstructure by visualization and quantification of water diffusivity.
Quantitative evaluation of the apparent diffusion coefficient (ADC) ...obtained from DWI has been proven helpful for differentiating between malignant and benign breast lesions, for cancer subtyping in breast cancer patients, and for prediction of response to neoadjuvant chemotherapy. However, to further establish DWI of breast lesions it is important to evaluate the quantitative imaging biomarker (QIB) characteristics of reproducibility, repeatability, and diagnostic accuracy.
In this intra‐individual prospective clinical study 40 consecutive patients with suspicious findings, scheduled for biopsy, underwent an identical 3T breast MRI protocol of the breast on two consecutive days (>24 h). Mean ADC of target lesions was assessed (two independent readers) in four separate sessions. Reproducibility, repeatability, and diagnostic accuracy between examinations (E1, E2), readers (R1, R2), and measurements (M1, M2) were assessed with intraclass correlation coefficients (ICCs), coefficients of variation (CVs), Bland–Altman plots, and receiver operating characteristic (ROC) analysis with calculation of the area under the ROC curve (AUC). The standard of reference was either histopathology (n = 38) or imaging follow‐up of up to 24 months (n = 2).
Eighty breast MRI examinations (median E1–E2, 2 ± 1.7 days, 95% confidence interval (CI) 1–2 days, range 1–11 days) in 40 patients (mean age 56, standard deviation (SD) ±14) were evaluated. In 55 target lesions (mean size 25.2 ± 20.8 (SD) mm, range 6–106 mm), mean ADC values were significantly (P < 0.0001) higher in benign (1.38, 95% CI 1.27–1.49 × 10−3 mm2/s) compared with malignant (0.86, 95% CI 0.81–0.91 × 10−3 mm2/s) lesions. Reproducibility and repeatability showed high agreement for repeated examinations, readers, and measurements (all ICCs >0.9, CVs 3.2–8%), indicating little variation. Bland–Altman plots demonstrated no systematic differences, and diagnostic accuracy was not significantly different in the two repeated examinations (all ROC curves >0.91, P > 0.05).
High reproducibility, repeatability, and diagnostic accuracy of DWI provide reliable characteristics for its use as a potential QIB, to further improve breast lesion detection, characterization, and treatment monitoring of breast lesions.
In an intra‐individual clinical study the quantitative imaging biomarker characteristics for DWI of breast lesions were evaluated. Both reproducibility and repeatability demonstrated an almost perfect agreement (all intraclass correlation coefficient values >0.9) with little variation (all coefficients of variation 3.2–8%). Diagnostic accuracy showed no significant difference in two repeated identical examinations (all ROC curves >0.91, P > 0.05). Consequently, DWI of breast lesions provides reliable characteristics for use as a potential quantitative imaging biomarker, to further improve breast lesion detection, characterization, and treatment monitoring.
In this prospective study, 117 female patients (mean age = 53 years) with 127 histologically proven breast cancer lesions (lymph node (LN) positive = 85, LN negative = 42) underwent simultaneous ...18F-FDG PET/MRI of the breast. Quantitative parameters were calculated from dynamic contrast-enhanced (DCE) imaging (tumor Mean Transit Time, Volume Distribution, Plasma Flow), diffusion-weighted imaging (DWI) (tumor ADCmean), and PET (tumor SUVmax, mean and minimum, SUVmean of ipsilateral breast parenchyma). Manual whole-lesion segmentation was also performed on DCE, T2-weighted, DWI, and PET images, and radiomic features were extracted. The dataset was divided into a training (70%) and a test set (30%). Multi-step feature selection was performed, and a support vector machine classifier was trained and tested for predicting axillary LN status. 13 radiomic features from DCE, DWI, T2-weighted, and PET images were selected for model building. The classifier obtained an accuracy of 79.8 (AUC = 0.798) in the training set and 78.6% (AUC = 0.839), with sensitivity and specificity of 67.9% and 100%, respectively, in the test set. A machine learning-based radiomics model comprising 18F-FDG PET/MRI radiomic features extracted from the primary breast cancer lesions allows high accuracy in non-invasive identification of axillary LN metastasis.
Purpose
To assess 18F-Fluoroethylcholine (18F-FEC) as a PET/MRI tracer in the evaluation of breast lesions, breast cancer aggressiveness, and prediction of lymph node status.
Materials and methods
...This prospective, monocentric study was approved by the ethics committee and patients gave written, informed consent. This clinical trial was registered in the EudraCT database (Number 2017-003089-29). Women who presented with suspicious breast lesions were included. Histopathology was used as reference standard. Simultaneous 18F-FEC PET/MRI of the breast was performed in a prone position with a dedicated breast coil. MRI was performed using a standard protocol before and after contrast agent administration. A simultaneous read by nuclear medicine physicians and radiologists collected the imaging data of MRI-detected lesions, including the maximum standardized 18F-FEC-uptake value of breast lesions (SUV
maxT
) and axillary lymph nodes (SUV
maxLN
). Differences in SUV
max
were evaluated with the Mann–Whitney U test. To calculate diagnostic performance, the area under the receiver operating characteristics curve (ROC) was used.
Results
There were 101 patients (mean age 52.3 years, standard deviation 12.0) with 117 breast lesions included (30 benign, 7 ductal carcinomas in situ, 80 invasive carcinomas). 18F-FEC was well tolerated by all patients. The ROC to distinguish benign from malignant breast lesions was 0.846. SUV
maxT
was higher if lesions were malignant (
p
< 0.001), had a higher proliferation rate (
p
= 0.011), and were HER2-positive (
p
= 0.041). SUV
maxLN
was higher in metastatic lymph nodes, with an ROC of 0.761 for SUV
maxT
and of 0.793 for SUV
maxLN.
Conclusion
Simultaneous 18F-FEC PET/MRI is safe and has the potential to be used for the evaluation of breast cancer aggressiveness, and prediction of lymph node status.