M Dietzel

Automatic multimodal 2D/3D breast image registration using biomechanical FEM models and intensity-based optimization.

Due to their different physical origin, X-ray mammography and Magnetic Resonance Imaging (MRI) provide complementary diagnostic information. However, the correlation of their images is challenging due to differences in dimensionality, patient positioning and compression state of the breast. Our automated registration takes over part of the correlation task. The registration method is based on a biomechanical finite element model, which is used to simulate mammographic compression. The deformed MRI volume can be compared directly with the corresponding mammogram. The registration accuracy is determined by a number of patient-specific parameters. We optimize these parameters--e.g. breast rotation--using image similarity measures. The method was evaluated on 79 datasets from clinical routine. The mean target registration error was 13.2mm in a fully automated setting. On basis of our results, we conclude that a completely automated registration of volume images with 2D mammograms is feasible. The registration accuracy is within the clinically relevant range and thus beneficial for multimodal diagnosis.

Clinical MR Mammography: Impact of Hormonal Status on Background Enhancement and Diagnostic Accuracy

PURPOSEnHormonal stimulation can induce background enhancement (BE) in MR mammography (MRM). This fact has been assumed to decrease the accuracy of MRM. Consequently, this report investigates: 1. The prevalence of BE in postmenopausal vs. premenopausal women in correlation to hormonal cycle phase (CP). 2. The impact of hormonal status (HS) and BE on diagnostic accuracy.nnnMATERIALS AND METHODSnConsecutive patients over 22 months with complete HS information (week of CP or postmenopausal) were included in this prospective investigation. Exclusion criteria were any hormonal therapy, hysterectomy as well as cancer proven by biopsy. The standard of reference was histopathology. All MRM scans were acquired using the same protocol (1.5 T, dynamic T 1w GRE after 0.1 mmol/kg bw Gd-DTPA i. v.). Two radiologists rated all examinations in consensus according to BI-RADS. BE was defined as: 0 = missing, 1 = moderate, 2 = distinct.nnnRESULTSn224 patients (150 postmenopausal, 74 premenopausal, 45 in the second week of CP) were included in this study (83 benign and 141 malignant findings). BE was more frequent in premenopausal women (p = 0.006), but did not differ between CP (p = 0.460). Neither HS nor BE had a significant impact on the diagnostic parameters of MRM (p ≥ 0.375). However, regarding BE, the relative number of false positive (FP) findings was highest (5 / 10; 50 %) in the distinct BE group. Regarding HS, 17 % more FP findings were observed in premenopausal women examined outside the second week of CP.nnnCONCLUSIONnIn premenopausal women, HS leads to increased BE of breast tissue, independent of CP. Distinct BE and less pronounced, non-optimal CP may lead to an increased number of false positive findings.

Association between survival in patients with primary invasive breast cancer and computer aided MRI.

To identify the potential of semi‐quantitative enhancement‐analysis in breast MRI to predict disease‐related death in primary breast cancer patients.

Potential of MR mammography to predict tumor grading of invasive breast cancer.

PURPOSEnTumor grading (TG) is one of the most widely used prognostic factors in the case of breast cancer. This study aims to identify the potential of magnetic resonance mammography (MRM) to non-invasively assess TG.nnnMATERIALS AND METHODSn399 invasive breast cancers were included (IRB approval; standardized clinical MRM protocols). All breast cancers were prospectively evaluated by two experienced (> 500 MRM) and blinded radiologists in consensus. In every cancer a set of 18 previously published MRM descriptors was assessed. These were assessed by univariate and multivariate analysis to identify the potential of MRM to predict TG (X2 statistics; binary logistic regression; area under the ROC curve [AUC]).nnnRESULTSn8 of 18 MRM descriptors were associated with TG, e. g. internal structure, edema (p < 0.001), as well as skin thickening and destruction of the nipple line (p < 0.05). MRM was feasible to predict TG by multivariate analysis (p < 0.001). The highest potential could be identified to predict well differentiated breast cancers with good prognosis (AUC = 0.930).nnnCONCLUSIONnMR mammography was able to non-invasively assess tumor grading in a standard protocol. Since tumor grading is a surrogate for overall survival, these results provide further evidence to the clinical application of MR mammography as a noninvasive prognostic tool.

MRI Background Parenchymal Enhancement Is Not Associated with Breast Cancer

Background Previously, a strong positive association between background parenchymal enhancement (BPE) at magnetic resonance imaging (MRI) and breast cancer was reported in high-risk populations. We sought to determine, whether this was also true for non-high-risk patients. Methods 540 consecutive patients underwent breast MRI for assessment of breast findings (BI-RADS 0–5, non-high-risk screening (no familial history of breast cancer, no known genetic mutation, no prior chest irradiation, or previous breast cancer diagnosis)) and subsequent histological work-up. For this IRB-approved study, BPE and fibroglandular tissue FGT were retrospectively assessed by two experienced radiologists according to the BI-RADS lexicon. Pearson correlation coefficients were calculated to explore associations between BPE, FGT, age and final diagnosis of breast cancer. Subsequently, multivariate logistic regression analysis, considering covariate colinearities, was performed, using final diagnosis as the target variable and BPE, FGT and age as covariates. Results Age showed a moderate negative correlation with FGT (r = -0.43, p<0.001) and a weak negative correlation with BPE (r = -0.28, p<0.001). FGT and BPE correlated moderately (r = 0.35, p<0.001). Final diagnosis of breast cancer displayed very weak negative correlations with FGT (r = -0.09, p = 0.046) and BPE (r = -0.156, p<0.001) and weak positive correlation with age (r = 0.353, p<0.001). On multivariate logistic regression analysis, the only independent covariate for prediction of breast cancer was age (OR 1.032, p<0.001). Conclusions Based on our data, neither BPE nor FGT independently correlate with breast cancer risk in non-high-risk patients at MRI. Our model retained only age as an independent risk factor for breast cancer in this setting.

DCE-MRI of the breast in a stand-alone setting outside a complementary strategy - results of the TK-study

AbstractObjectivesTo evaluate the accuracy of MRI of the breast (DCE-MRI) in a stand-alone setting with extended indications.Materials and methodsAccording to the inclusion criteria, breast specialists were invited to refer patients to our institution for DCE-MRI. Depending on the MR findings, patients received either a follow-up or biopsy. Between 04/2006 and 12/2011 a consecutive total of 1,488 women were prospectively examined.ResultsOf 1,488 included patients, 393 patients were lost to follow-up, 1,095 patients were evaluated. 124 patients were diagnosed with malignancy by DCE-MRI (76 TP, 48 FP, 971 TN, 0 FN cases). Positive cases were confirmed by histology, negative cases by MR follow-ups or patient questionnaires over the next 5xa0years in 1,737 cases (sensitivity 100xa0%; specificity 95.2xa0%; PPV 61.3xa0%; NPV 100xa0%; accuracy 95.5xa0%). For invasive cancers only (DCIS excluded), the results were 63 TP; 27 FP; 971 TP and 0 FN (sensitivity 100xa0%; specificity 97.2xa0%; PPV 70xa0%; NPV 100xa0%; accuracy 97.5xa0%).ConclusionThe DCE-MRI indications tested imply that negative results in DCE-MRI reliably exclude cancer. The results were achieved in a stand-alone setting (single modality diagnosis). However, these results are strongly dependent on reader experience and adequate technical standards as prerequisites for optimal diagnoses.Key Points• DCE-MRI of the breast has a high accuracy in finding breast cancer.n • The set of indications for DCE-MRI of the breast is still very limited.n • DCE-MRI can achieve a high accuracy in a ‘screening-like’ setting.n • Accuracy of breast DCE-MRI is strongly dependent on technique and reader experience.n • A negative DCE-MRI effectively excludes cancer.

MR-spectroscopy at 1.5 tesla and 3 tesla. Useful? A systematic review and meta-analysis.

Dynamic contrast enhanced magnetic resonance imaging is the most sensitive method for detection of breast cancer [1,2]. These T1-weighted imaging studies measure the extracellular distribution of paramagnetic contrast agents which are mainly affected by vascular properties of the tissue investigated. Although cancers show a characteristically early and strong enhancement, substantial overlaps of enhancement characteristics between benign and malignant breast lesions have been described. Consequently, for lesion classification in clinical practice, a combination of morphologic criteria and dynamic enhancement pattern analysis is applied [3]. Such an approach is mainly subjective and thus prone to experience related mistakes and interobserver variation. Proton magnetic resonance spectroscopy (H MR spectroscopy, MRS) is a noninvasive examination technique for the assessment of biochemical tissue properties. Presence of a compound resonance around 3.23 ppm is attributed to choline metabolites such as choline, phosphocholine and glycerophosphocholine and simply referred to as total choline (tCho). Increased levels of tCho have been detected in malignant cancers and are ascribed to an increased cellular membrane turnover [4–6]. In vivo qualitative and quantitative tCho measurements have been used as a diagnostic test in the workup of neoplastic breast lesions. The present systematic review and meta-analysis aims to investigate the diagnostic capability of tCho measurements for the differentiation of breast lesions focusing on state of the art field strengths for clinical breast imaging of 1.5 and 3 tesla.

Magnetic resonance mammography in small vs. advanced breast lesions - systematic comparison reveals significant impact of lesion size on diagnostic accuracy in 936 histologically verified breast lesions.

PURPOSEnThis study was conducted to investigate the appearance of breast lesions in MR mammography (MRM) as a function of size and to identify the potential impact on diagnostic accuracy.nnnMATERIALS AND METHODSn936 histologically verified breast lesions (standardized MRM protocol; consecutive 12-year period at our institution, diameter 5 - 50 mm) were prospectively evaluated in consensus by two radiologists with significant MRM experience. For this purpose previously published descriptors (n = 17) were used. These were summarized as the basic catalog and extended catalog of descriptors (BC vs. EC). According to a cut-off of 20 mm, the database was divided into the subgroups small (n = 669) and advanced (n = 267). The diagnostic accuracy of MRM in these two subgroups was then determined using BC and EC, separately (binary logistic regression analysis; AUC analysis).nnnRESULTSnThe majority of descriptors (n = 11) showed a significantly different prevalence in correlation with size (p < 0.05). The diagnostic accuracy of MRM for advanced lesions (AUC = 0.969) was significantly higher (p < 0.001). This difference was significantly decreased (p < 0.001), if instead of BC (AUC = 0.865) EC was applied for the assessment of small lesions (AUC: 0.908 vs. 0.865).nnnCONCLUSIONnThe typical appearance of breast lesions in MRM depends on lesion size. This resulted in lower diagnostic accuracy in small lesions compared to advanced findings. This difference was able to be significantly decreased by applying the catalog of extended descriptors.

Can Color-Coded Parametric Maps Improve Dynamic Enhancement Pattern Analysis in MR Mammography?

PURPOSEnPost-contrast enhancement characteristics (PEC) are a major criterion for differential diagnosis in MR mammography (MRM). Manual placement of regions of interest (ROIs) to obtain time/signal intensity curves (TSIC) is the standard approach to assess dynamic enhancement data. Computers can automatically calculate the TSIC in every lesion voxel and combine this data to form one color-coded parametric map (CCPM). Thus, the TSIC of the whole lesion can be assessed. This investigation was conducted to compare the diagnostic accuracy (DA) of CCPM with TSIC for the assessment of PEC.nnnMATERIALS AND METHODSn329 consecutive patients with 469 histologically verified lesions were examined. MRM was performed according to a standard protocol (1.5 T, 0.1 mmol/kgbw Gd-DTPA). ROIs were drawn manually within any lesion to calculate the TSIC. CCPMs were created in all patients using dedicated software (CAD Sciences). Both methods were rated by 2 observers in consensus on an ordinal scale. Receiver operating characteristics (ROC) analysis was used to compare both methods.nnnRESULTSnThe area under the curve (AUC) was significantly (p=0.026) higher for CCPM (0.829) than TSIC (0.749). The sensitivity was 88.5% (CCPM) vs. 82.8% (TSIC), whereas equal specificity levels were found (CCPM: 63.7%, TSIC: 63.0%).nnnCONCLUSIONnThe color-coded parametric maps (CCPMs) showed a significantly higher DA compared to TSIC, in particular the sensitivity could be increased. Therefore, the CCPM method is a feasible approach to assessing dynamic data in MRM and condenses several imaging series into one parametric map.

MR-mammography: high sensitivity but low specificity? New thoughts and fresh data on an old mantra

1. Background Despite significant advances in the therapeutic options, breast cancer remains the most frequent and most deadly malignant neoplasm in women [1]. Probably more than in any other oncologic setting, medical imaging plays a central part in the work up of every single patient diagnosed with this disease. This covers not only pretherapeutic staging, therapy monitoring and screening for cancer recurrence, but – most vital – initial detection of the disease. The latter is so important as there are virtually no other possibilities to detect breast cancers at a preclinical stage [2]. MR-mammography (MRM) is frequently used as an adjunct to conventional mammography. It provides excellent soft tissue contrast and allows – besides subtle morphologic analyses – to assess tissue vascularisation and neoangiogenesis [3]. As the latter is a hallmark of invasive cancer growth, MRM shows an unsurpassed sensitivity >>95%. Different from conventional mammography, standardization of MRM is poor. This concerns both technical issues and imaging protocols but also interpretation criteria. Accordingly a wide range of imaging protocols, personal preferences, reading routines and evaluation criteria are used worldwide. Due to the excellent sensitivity of MRM only few cancers are likely to be missed. Yet, the definite classification of a breast lesion as “positive” or “negative” can be challenging. Particularly the number of “false positives” and