Matthias Benndorf

False-Positive Findings at Contrast-Enhanced Breast MRI: A BI-RADS Descriptor Study

OBJECTIVE Breast MRI has high sensitivity in breast cancer detection, and the BI-RADS MRI lexicon was a step toward standardized description of lesions. However, false-positive findings occur and lead to unnecessary biopsy. The purpose of this investigation was to identify criteria for false-positive findings in clinical practice. MATERIALS AND METHODS Eligible for investigation were all breast MRI examinations from a consecutive 16-month time period that had histopathologic verification and findings classified as BI-RADS category 4-6 in the initial MRI report. Accordingly, 132 patients with 120 malignant and 31 benign lesions were enrolled. Two blinded observers categorized lesions into mass or nonmass and used BI-RADS to identify descriptor distribution differences between the benign and malignant subgroups. RESULTS The ratio of mass to nonmass lesions differed significantly (p < 0.001) between benign (1.2:1) and malignant (7:1) findings. Seventeen mass and 14 nonmass lesions were false-positive, and 105 mass and 15 nonmass lesions were true-positive. Among mass lesions, it was possible to differentiate malignant and benign lesions on the basis of margin (smooth, irregular, or spiculated) and dynamic enhancement features (p < 0.05). Among nonmass lesions, only stippled enhancement had a significant difference between the subgroups (p < 0.05). Tumor diameter had no influence on the correct diagnosis of nonmass lesions (p = 0.301). Conversely, among mass lesions, false-positive lesions were smaller than true-positive lesions (p = 0.01). CONCLUSION Nonmass lesions were the major cause of false-positive breast MRI findings. BI-RADS descriptors are not sufficient for differentiating benign and malignant nonmass lesions.

Sensitivity and specificity of unenhanced MR mammography (DWI combined with T2-weighted TSE imaging, ueMRM) for the differentiation of mass lesions

ObjectiveThis study was performed to assess the sensitivity and specificity for malignant and benign mass lesions of a diagnostic approach combining DWI with T2-weighted images (unenhanced MR mammography, ueMRM) and compare the results with contrast-enhanced MR mammography (ceMRM).Materials and methodsConsecutive patients undergoing histopathological verification of mass lesions after MR mammography without prior breast interventions (contrast-enhanced T1-weighted, T2-weighted and DWI sequences) were eligible for this retrospective investigation. Two blinded observers first rated ueMRM and then ceMRM according to the BIRADS scale. Lesion size, ADC values and T2-weighted TSE descriptors were assessed.ResultsThis study examined 81 lesions (27 benign, 54 malignant). Sensitivity of ueMRM was 93% (observer 1) and 86% (observer 2), respectively. Sensitivity of ceMRM was 96.5% (observer 1) and 98.3% (observer 2). Specificity was 85.2% (ueMRM) and 92.6% (ceMRM) for both observers. The differences between both methods and observers were not significant (P ≥ 0.09). Lesion size measurements did not differ significantly among all sequences analyzed. Tumor visibility was worse using ueMRM for both benign (P < 0.001) and malignant lesions (P = 0.004).ConclusionSensitivity and specificity of ueMRM in mass lesions equal that of ceMRM. However, a reduced lesion visibility in ueMRM may lead to more false-negative findings.

Breast MRI as an adjunct to mammography: Does it really suffer from low specificity? A retrospective analysis stratified by mammographic BI-RADS classes.

Background: Reports on the specificity of breast MRI are heterogeneous, depending on the respective setting of the performed study. Purpose: To retrospectively estimate the sensitivity and especially the specificity of breast MRI in the non-screening setting as an adjunct to mammography sorted by breast density and to estimate the accuracy of breast MRI in cases rated BI-RADS 0 and 3 mammographically. Material and Methods: A total of 216 consecutive patients with referral to breast MRI and previously acquired mammography were enrolled in this analysis. Negative findings were followed up with a mean time of 26.7 months. The loss to follow-up was 10.8%. The single breast was regarded as the study subject (n=399, 364 cases were eligible for calculation of diagnostic accuracy). BI-RADS 1 and 2 were rated as benign, 4 and 5 as malignant. BI-RADS 0 and 3 were analyzed separately. The 95% confidence intervals (CIs) were calculated from the normally approximated binomial distribution and taken to represent significant differences for the two imaging modalities if they did not overlap. Results: Among the study population, 62 malignant neoplasms were detected. For cases rated BI-RADS 1, 2, 4, and 5 (n=251), the sensitivity of breast MRI was 95.7% (95% CI 89.9–100.0%) and 74.5% (95% CI 62.0–87.0%) for mammography, respectively. The specificity of breast MRI was 96.1% (95% CI 93.4–98.8%) and 92.2% (95% CI 88.5–95.9%) for mammography, respectively. The diagnostic accuracy of breast MRI did not depend on breast density. In cases rated BI-RADS 0, n=57 (3, n=56), breast MRI achieved a sensitivity of 100% (90.9%) and a specificity of 98.1% (88.9%). There was a significant (P< 0.01) accumulation of dense breast tissue (ACR IV) in breasts rated BI-RADS 0 in mammography. Breast MRI missed three malignant lesions, two of them being smaller than 3 mm. Conclusion: There is no rationale to criticize the low specificity of breast MRI when used as an adjunct to mammography. The independency of the diagnostic accuracy of breast MRI from breast density makes it a worthwhile choice in mammographic BI-RADS 0 cases.

Diffusion weighted imaging of liver lesions suspect for metastases: Apparent diffusion coefficient (ADC) values and lesion contrast are independent from Gd-EOB-DTPA administration

PURPOSE Gadolinium ethoxybenzyl diethylenetriamine pentaacetic acid (Gd-EOB-DTPA) enhanced liver MRI is widely used for detection and differentiation of focal liver lesions. Diffusion weighted imaging (DWI) including apparent diffusion coefficient (ADC) measurements is increasingly utilised as a fast and, with limitations, quantitative method for liver lesion detection and characterisation. Herein we investigate whether the administration of Gd-EOB-DTPA affects DWI. MATERIALS AND METHODS 31 consecutive patients referred to standardised liver MRI (1.5T, Gd-EOB-DTPA, 0.025mmol/kg) were retrospectively reviewed. All underwent a breathhold DWI sequence before and after contrast agent administration (EPI-DWI, TR/TE (effective): 2100/62ms, b-values: 0 and 800s/mm(2)). Patients with previously treated liver lesions were excluded. Signal intensity of lesion, parenchyma and noise on DWI images as well as the ADC value were measured after identification by two observers in consensus using manually placed regions of interest. The reference standard was imaging follow-up determined separately by two radiologists. Data analysis included signal-to-noise (SNR) ratio and contrast-to-noise ratio (CNR) calculations, comparisons were drawn by employing multiple Bonferroni corrected Wilcoxon signed-rank tests. RESULTS 50 malignant and 39 benign lesions were identified. Neither SNR, CNR nor ADC values showed significant differences between pre- and postcontrast DWI. Both pre- and postcontrast ADC values differed significantly between benign and malignant lesions (P<0.001). CONCLUSION We did not identify a significant influence of Gd-EOB-DTPA on DWI of liver lesions. This allows for individual tailoring of imaging protocols according to clinical needs.

Diagnosis of focal liver lesions suspected of metastases by diffusion-weighted imaging (DWI): systematic comparison favors free-breathing technique

Two echo planar imaging diffusion-weighted imaging (DWI) techniques [one breath hold (DWI(bh)), repetition time/echo time (TR/TE) 2100/62 ms; one at free breathing (DWI(fb)), TR/TE 2000/65 ms] were compared regarding diagnosis of focal liver lesions (FLLs) in 45 patients with suspected liver metastasis without prior treatment. Apparent diffusion coefficient values of 46 benign and 67 malignant FLLs were analyzed by receiver operating characteristics (ROC) analysis. DWI(fb) detected more malignant lesions than DWI(bh) (P=.002). Lesion size ≤10 mm was associated with FLLs missed by DWI(bh) (P=.018). Area under the ROC curve of DWI(fb) (0.801) was higher compared to that of DWI(bh) (0.669, P<.0113), demonstrating the diagnostic superiority of DWI(fb).

Kinetic characteristics of ductal carcinoma in situ (DCIS) in dynamic breast MRI using computer-assisted analysis

Background: Enhancement characteristics of breast lesions are regarded as a major criterion for their differential diagnosis in dynamic breast MRI (bMRI). However, ductal carcinoma in situ (DCIS) exhibits a highly heterogeneous enhancement pattern when kinetic analysis is performed conventionally by manual placement of region of interest (ROI) and therefore its diagnosis remains challenging. Purpose: To compare enhancement characteristics of DCIS lesions on dynamic bMRI using manual ROI placement with computer-aided analysis and to evaluate whether the latter might increase the detection rate of kinetic features suspicious for malignancy. Material and Methods: The enhancement patterns of 47 histopathologically verified pure DCIS lesions were evaluated on bMRI images using manual ROI placement as well as a commercially available computer analysis software. The latter is able to automatically assess enhancement characteristics of a whole lesion pixelwise. Kinetic features evaluated included classification of lesion enhancement pattern into washout, plateau or persistent curve type. A washout and plateau enhancement pattern are regarded as suggestive for malignancy. Results: Morphological classification revealed focus-like enhancement in 2 lesions, mass enhancement in 11, and non-mass enhancement in 34. Manual placement of ROI demonstrated a suspicious enhancement pattern in 51.1% of the DCIS lesions, which could not be significantly increased using computer-aided analysis. Of the mass and non-mass-enhancing DCIS lesions, 90.9% and 38.3%, respectively, demonstrated suspicious kinetic curves. After application of the automated analysis software, the detection rate of suspicious enhancement patterns was unchanged in mass DCIS lesions and increased to 52.9% in non-mass DCIS lesions (P=0.33). However, the increase in the detection of washout curves alone was significant (P=0.02). In all, 40% of G1, 41.1% of G2, and 60% of G3 lesions demonstrated a suspicious curve type with manual evaluation. Computer analysis increased the detection of suspicious enhancement patterns in a non-significant manner to 50%, 58.8%, and 70%, respectively. Conclusion: The detection of suspicious enhancement curves could not be significantly increased in DCIS lesions when using computer-aided analysis despite a significantly higher detection rate of washout curves alone.

Development of an online, publicly accessible naive Bayesian decision support tool for mammographic mass lesions based on the American College of Radiology (ACR) BI-RADS lexicon

AbstractPurposeTo develop and validate a decision support tool for mammographic mass lesions based on a standardized descriptor terminology (BI-RADS lexicon) to reduce variability of practice.Materials and methodsWe used separate training data (1,276 lesions, 138 malignant) and validation data (1,177 lesions, 175 malignant). We created naïve Bayes (NB) classifiers from the training data with tenfold cross-validation. Our “inclusive model” comprised BI-RADS categories, BI-RADS descriptors, and age as predictive variables; our “descriptor model” comprised BI-RADS descriptors and age. The resulting NB classifiers were applied to the validation data. We evaluated and compared classifier performance with ROC-analysis.ResultsIn the training data, the inclusive model yields an AUC of 0.959; the descriptor model yields an AUC of 0.910 (P < 0.001). The inclusive model is superior to the clinical performance (BI-RADS categories alone, P < 0.001); the descriptor model performs similarly. When applied to the validation data, the inclusive model yields an AUC of 0.935; the descriptor model yields an AUC of 0.876 (P < 0.001). Again, the inclusive model is superior to the clinical performance (P < 0.001); the descriptor model performs similarly.ConclusionWe consider our classifier a step towards a more uniform interpretation of combinations of BI-RADS descriptors. We provide our classifier at www.ebm-radiology.com/nbmm/index.html.Key Points• We provide a decision support tool for mammographic masses atwww.ebm-radiology.com/nbmm/index.html. • Our tool may reduce variability of practice in BI-RADS category assignment. • A formal analysis of BI-RADS descriptors may enhance radiologists’ diagnostic performance.

An exception to tumour neoangiogenesis in a malignant breast-lesion.

Abstract:  Magnetic resonance‐mammography is regarded as the most sensitive diagnostic modality in the detection of breast cancer. It uses the tumour neoangiogenesis to depict lesions after intravenous contrast agent injection. It is said, that for tumours exceeding a diameter of three millimetres contrast agent enhancement is mandatory. In our case report we describe a rare tumour growth condition. We observed a large invasive carcinoma (18 millimetres diameter) without contrast enhancement in breast MRI due to an almost missing tumour neoangiogenesis. The cancer had a low cellularity and a strong desmoplastic reaction.

A history of breast cancer and older age allow risk stratification of mammographic BI-RADS 3 ratings in the diagnostic setting.

OBJECTIVE The objective was to investigate whether risk stratification of mammographic Breast Imaging: Reporting and Data System (BI-RADS) 3 can be accomplished in the diagnostic setting. METHODS We analyzed 4941 BI-RADS-3-rated patients (23 malignant outcomes) and built logistic-regression models with age, personal and family history of breast cancer, fibroglandular density, and additional mammographic findings as predictive variables. RESULTS A personal history of breast cancer (odds ratio: 5.53) and older age (odds ratio: 12.44/10.93 for age 50-64/>64) are independent risk factors. Patients with both risk factors have a risk >2%. CONCLUSION Biopsy may be warranted in older patients with a history of breast cancer who would be otherwise assigned BI-RADS 3.

Diagnostic Accuracy of Robot-Guided, Software Based Transperineal MRI/TRUS Fusion Biopsy of the Prostate in a High Risk Population of Previously Biopsy Negative Men

Objective. In this study, we compared prostate cancer detection rates between MRI-TRUS fusion targeted and systematic biopsies using a robot-guided, software based transperineal approach. Methods and Patients. 52 patients received a MRIT/TRUS fusion followed by a systematic volume adapted biopsy using the same robot-guided transperineal approach. The primary outcome was the detection rate of clinically significant disease (Gleason grade ≥ 4). Secondary outcomes were detection rate of all cancers, sampling efficiency and utility, and serious adverse event rate. Patients received no antibiotic prophylaxis. Results. From 52 patients, 519 targeted biopsies from 135 lesions and 1561 random biopsies were generated (total n = 2080). Overall detection rate of clinically significant PCa was 44.2% (23/52) and 50.0% (26/52) for target and random biopsy, respectively. Sampling efficiency as the median number of cores needed to detect clinically significant prostate cancer was 9 for target (IQR: 6–14.0) and 32 (IQR: 24–32) for random biopsy. The utility as the number of additionally detected clinically significant PCa cases by either strategy was 0% (0/52) for target and 3.9% (2/52) for random biopsy. Conclusions. MRI/TRUS fusion based target biopsy did not show an advantage in the overall detection rate of clinically significant prostate cancer.