Breast MRI Data Research Articles

Abstract PO4-07-06: Development of MRI-based Radiomics Nomogram for the Prediction of Recurrence in Patients with Luminal-type Breast Cancer: A Nested Case-Control Study

Abstract Purpose: The aim of this study was to develop a radiomics nomogram using breast MRI data and preoperative pathologic information to predict oncotype DX recurrence score (RS) in patients with estrogen receptor-positive early-stage breast cancer (EBC). Materials and methods: In this retrospective study, a total of 493 patients with EBC, diagnosed through core needle biopsy and who underwent preoperative breast MRI at a single institution between 2011 and 2017, were included. The patients were categorized based on RS, age at surgery, and nodal status, as suggested by the TAILORX trial, and out of the included patients, 249 were identified as likely to benefit from chemotherapy, while 244 were deemed safe to forgo chemotherapy. Radiomic features were extracted from three-dimensional segmentations of each tumor, and computer-extracted image phenotypes (CEIP) were generated from early post-contrast T1-weighted images, percent enhancement (PE) maps, and signal enhancement ratio (SER) maps. The patient cohort was divided into a training set (n=329) and a validation set (n=164). Feature selection and radiomics score construction were performed using elastic net, and a prediction model was developed using multivariate logistic regression analysis. The radiomics score, along with independent pathologic risk factors, was incorporated into a radiomics nomogram. Internal validation was conducted using an independent validation set (n=164). Results: The radiomics score, composed of 24 selected CEIPs, demonstrated a significant association with recurrence prediction (C-index, 0.769 for training set and 0.745 for validation set). The independent pathologic predictors included in the nomogram were histology, estrogen receptor status, progesterone receptor status, nuclear grade, histologic grade, HER1, CK56, P53, and Ki67 status, with a C-index of 0.858 for the training set and 0.774 for the validation set. Adding the radiomics score to the pathologic nomogram resulted in incremental values of 0.054 and 0.092, respectively. The radiomics nomogram showed favorable prediction of low-risk RS, with a C-index of 0.912 for the training set and 0.866 for the validation set. Conclusion: This study demonstrated that an MRI-based radiomics nomogram incorporating the radiomics score and preoperatively obtained pathologic data can effectively predict oncotype DX RS and the benefit of adjuvant chemotherapy in EBC patients. The model improves risk assessment availability and timeliness for patients with EBC. Citation Format: Minji Song, Hee Jung Shin. Development of MRI-based Radiomics Nomogram for the Prediction of Recurrence in Patients with Luminal-type Breast Cancer: A Nested Case-Control Study [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO4-07-06.

Abstract PO2-03-01: Personalized 3D tumor models drive shared decision-making in early breast cancer

Abstract Background: Decision-making for early (Stages 1-3) breast cancer (eBC) care is complex and highly personal, requiring discussions between patients and multiple physicians. Shared decision-making (SDM) is an accepted process that combines input from both patients and physicians, although its overall use is low due to a lack of time and incentives, complicated application techniques, and few communication tools. Reviews show that SDM results in “improved affective-cognitive outcomes” but evidence linking to patient outcomes is sparse. Consequently, patients often receive mixed messages and feel anxious when making treatment decisions. SimBioSys (SBS) developed the TumorSight(TM) platform for individualized treatment planning, coordinated care, and SDM. The first application on this platform, TumorSight Viz, rapidly builds 3D computational models from standard breast MRI data, and further integrates deep learning, genomic/genetic testing, metabolomics, and clinical and demographic data to tailor these models for each individual diagnosed with eBC. The platform can also transform acquired data (e.g., prone) to surgical position (e.g., supine) to create realistic images for precision surgical planning and tools for better SDM with eBC patients. We are assessing how surgeons and patient advocates (PA) perceive TumorSight, and how it may support better SDM and treatment implementation. Methods: SBS conducted an exploratory survey with 31 eBC surgeons to assess the utility of TumorSight Viz visualizations across a range of didactic and planning applications and to query preferences for future components like predicted response to chemotherapy and immunotherapy. Questions were developed with input from breast surgeon advisors. More exploration on TumorSight Viz utility is planned with semi-structured qualitative interviews to assess patient advocates’ perceptions of 3D visualizations, surgical planning, and ability to support decision making. PAs will represent geographic, experience, and race/ethnicity diversity. Questions to elicit patient needs, perception of SBS tools, and quality-of-life are being finalized with academic advisors. These will also be used to develop a future eBC patient survey. Results: The SBS surgeon survey revealed that 96% of breast surgeons find visualization important to surgical planning, and these depictions were considered to be critical by a third of breast surgeons surveyed. 96% also listed the conversion of prone to supine visualizations as important (45% critical, 35% valuable, 16% nice to have, 0% not important) for surgical planning value. Surgeons also assigned strong utility for: Graphical 3D rotation to determine tumor shape, size, location, positioning, and landmarks relative to other breast structures – 100% Ability to instantly provide tumor-to-breast volume determination with high accuracy – 96% Ability to instantly provide anticipated tumor extirpation volume to breast volume with high accuracy – 96% Automatic tumor-to-landmark structures measurements (nipple, chest, skin) – 96% Visualization of tumor in the context of other anatomical structures (e.g., skin, blood vessels, chest, heart) – 87% Initial PA discussions (n=3) listed the importance of individualized treatment based on evidence instead of hunches; reasons to avoid prophylactic mastectomies; and ways to lower patient anxiety. Conclusions: Survey-based feedback shows value to eBC surgeons and patients to coordinate care and personalize approaches for optimal outcomes. Computational tools can facilitate patient understanding and patient-tailored treatment. Additional surveys are planned with patients and eBC patient advocacy groups to assess the utility and potential expansion of patient-specific 3D tumor visualizations, both of which will be further discussed. Citation Format: Deborah Collyar, Matthew Biancalana, Anuja Antony, Nicole Liadis, Jyoti Palaniappan, Rebecca Seago-Coyle. Personalized 3D tumor models drive shared decision-making in early breast cancer [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO2-03-01.

A publicly available deep learning model and dataset for segmentation of breast, fibroglandular tissue, and vessels in breast MRI

Breast density, or the amount of fibroglandular tissue (FGT) relative to the overall breast volume, increases the risk of developing breast cancer. Although previous studies have utilized deep learning to assess breast density, the limited public availability of data and quantitative tools hinders the development of better assessment tools. Our objective was to (1) create and share a large dataset of pixel-wise annotations according to well-defined criteria, and (2) develop, evaluate, and share an automated segmentation method for breast, FGT, and blood vessels using convolutional neural networks. We used the Duke Breast Cancer MRI dataset to randomly select 100 MRI studies and manually annotated the breast, FGT, and blood vessels for each study. Model performance was evaluated using the Dice similarity coefficient (DSC). The model achieved DSC values of 0.92 for breast, 0.86 for FGT, and 0.65 for blood vessels on the test set. The correlation between our model’s predicted breast density and the manually generated masks was 0.95. The correlation between the predicted breast density and qualitative radiologist assessment was 0.75. Our automated models can accurately segment breast, FGT, and blood vessels using pre-contrast breast MRI data. The data and the models were made publicly available.

Fully Automatic Thoracic Cavity Segmentation in Dynamic Contrast Enhanced Breast MRI Using Deep Convolutional Neural Networks

Dynamic Contrast Enhanced Magnetic Resonance Imaging (DCE-MRI) is regarded as one of the main diagnostic tools for breast cancer. Several methodologies have been developed to automatically localize suspected malignant breast lesions. Changes in tissue appearance in response to the injection of the contrast agent (CA) are indicative of the presence of malignant breast lesions. However, these changes are extremely similar to the ones of internal organs, such as the heart. Thus, the task of chest cavity segmentation is necessary for the development of lesion detection. In this work, a data-efficient approach is proposed, to automatically segment breast MRI data. Specifically, a study on several UNet-like architectures (Dynamic UNet) based on ResNet is presented. Experiments quantify the impact of several additions to baseline models of varying depth, such as self-attention and the presence of a bottlenecked connection. The proposed methodology is demonstrated to outperform the current state of the art both in terms of data efficiency and in terms of similarity index when compared to manually segmented data.

Intermodal correlation of quantitative CT-data and MRI-biomarkers derived from synchronous spectral CT-maps and breast MRI-examinations with molecular biomarkers in invasive ductal breast carcinomas

ObjectiveTo asses the correlation of data derived from dual-layer (DL)-CT material-maps and breast MRI data with molecular biomarkers in invasive breast carcinomas. MethodsAll patients at the University Breast Cancer Center who underwent a clinically indicated DLCT-scan and a breast MRI for staging of invasive ductal breast cancer from 2016 to 2020 were prospectively included. Iodine concentration-maps, and Zeffective-maps were reconstructed from the CT-datasets. T1w- and T2w-signal intensities, ADC and the clustered shapes of the dynamic-curves (washout, plateau, persistent) were derived from the MRI-datasets. ROI-based evaluations of the cancers and the reference “musculature” were performed semi-automatically in identical anatomical positions using dedicated evaluation software. Statistical analysis was essentially descriptive using Spearmańs rank correlation and (multivariable) partial correlation. ResultsThe signal intensities measured in the 3rd phase of the contrast dynamics correlated at an intermediate level of significance with the iodine content and the Zeffective-values derived from the breast target lesions (Spearmańs rank correlation-coefficient r = 0.237/0.236, p = 0.002/0.003). The bivariate and the multivariate analyses displayed correlations of an intermediate significance level of the iodine content and the Zeff-values measured in the breast target lesions with immunhistochemical subtyping (r = 0.211–0.243, p = 0.002–0.009, respectively). The Zeff-values showed the strongest correlations when normalized to the values measured in the musculature and in the aorta (r = -0.237 to −0.305, r=<0.001–0.003). The MRI-assessments showed correlations of intermediate to high significance and low to intermediate significance between the ratios of the T2w-signal intensities and the trends of the dynamic curves measured in the breast target lesions and in musculature and immunohistochemical cancer subtyping, respectively (T2w: r = 0.232–0.249, p = 0.003/0.002; dynamics: r = -0.322/-0.245, p=<0.001/0.002). The ratios of the clustered trends of the dynamic curves measured in the breast target lesions and in musculature correlated with tumor grading on intermediate significance level (r = -0.213 and −0.194, p = 0.007/0.016) and with Ki-67 on a low significance level (bivariate analysis: r = -0.160, p = 0.040). There was only a weak correlation between the ADC-values measured in the breast target lesions and HER2-expression (bivariate ansalysis: r = 0.191, p = 0.030). ConclusionsOur preliminary results indicate that evaluation of perfusion based DLCT-data and MRI-biomarkers show correlations with the immunhistochemical subtyping of invasive ductal breast carcinomas. Further clinical research is warranted in order to validate the value of the results and define clinical situations in which the use of the described DLCT-biomaker and MRI biomarkers may be helpful in clinical patient care.

Convolutional Neural Network of Multiparametric MRI Accurately Detects Axillary Lymph Node Metastasis in Breast Cancer Patients With Pre Neoadjuvant Chemotherapy

BackgroundAccurate assessment of the axillary lymph nodes (aLNs) in breast cancer patients is essential for prognosis and treatment planning. Current radiological staging of nodal metastasis has poor accuracy. This study aimed to investigate the machine learning convolutional neural networks (CNNs) on multiparametric MRI to detect nodal metastasis with 18FDG-PET as ground truths. Materials and MethodsData were obtained via a retrospective search. Inclusion criteria were patients with bilateral breast MRI and 18FDG-PETand/or CT scans obtained before neoadjuvant chemotherapy. In total, 238 aLNs were obtained from 56 breast cancer patients with 18FDG-PET and/or CT and breast MRI data. Radiologists scored each node based on all MRI as diseased and non–diseased nodes. Five models were built using T1-W MRI, T2-W MRI, DCE MRI, T1-W + T2-W MRI, and DCE + T2-W MRI model. Performance was evaluated using receiver operating curve (ROC) analysis, including area under the curve (AUC). ResultsAll CNN models yielded similar performance with an accuracy ranging from 86.08% to 88.50% and AUC ranging from 0.804 to 0.882. The CNN model using T1-W MRI performed better than that using T2-W MRI in detecting nodal metastasis. CNN model using combined T1- and T2-W MRI performed the best compared to all other models (accuracy = 88.50%, AUC = 0.882), but similar in AUC to the DCE + T2-W MRI model (accuracy = 88.02%, AUC = 0.880). All CNN models performed better than radiologists in detecting nodal metastasis (accuracy = 65.8%). Conclusionxxxxxx

An A.I. classifier derived from 4D radiomics of dynamic contrast-enhanced breast MRI data: potential to avoid unnecessary breast biopsies

ObjectivesDue 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.MethodsThis 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 (C1, 100%, and C2, ≥ 95% sensitivity).ResultsFour 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% (C1) and 36.2% (C2).ConclusionThe 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.

Three-Dimensional Whole Breast Segmentation in Sagittal and Axial Breast MRI With Dense Depth Field Modeling and Localized Self-Adaptation for Chest-Wall Line Detection.

Whole breast segmentation is an essential task in quantitative analysis of breast MRI for cancer risk assessment. It is challenging, mainly, because the chest-wall line (CWL) can be very difficult to locate due to its spatially varying appearance-caused by both nature and imaging artifacts-and neighboring distracting structures. This paper proposes an automatic three-dimensional (3-D) segmentation method, termed DeepSeA, of whole breast for breast MRI. DeepSeA distinguishes itself from previous methods in three aspects. First, it reformulates the challenging problem of CWL localization as an equivalent problem that optimizes a smooth depth field and so fully utilizes the CWL's 3-D continuity. Second, it employs a localized self-adapting algorithm to adjust to the CWL's spatial variation. Third, it applies to breast MRI data in both sagittal and axial orientations equally well without training. A representative set of 99 breast MRI scans with varying imaging protocols is used for evaluation. Experimental results with expert-outlined reference standard show that DeepSeA can segment breasts accurately: the average Dice similarity coefficients, sensitivity, specificity, and CWL deviation error are 96.04%, 97.27%, 98.77%, and 1.63mm, respectively. In addition, the configuration of DeepSeA is generalized based on experimental findings, for application to broad prospective data. A fully automatic method-DeepSeA-for whole breast segmentation in sagittal and axial breast MRI is reported. DeepSeA can facilitate cancer risk assessment with breast MRI.

Patient-based 4D digital breast phantom for perfusion contrast-enhanced breast CT imaging.

PurposeThe purpose of this study was to develop a realistic patient‐based 4D digital breast phantom including time‐varying contrast enhancement for simulation of dedicated breast CT perfusion imaging.MethodsA 3D static phantom is first created by segmenting a breast CT image from a healthy patient into skin, fibroglandular tissue, adipose tissue, and vasculature. For the creation of abnormal cases, a breast lesion model was developed and can be added to the phantom. After defining the necessary perfusion parameters for each tissue (e.g., arterial input function for vasculature, blood volume and blood flow for the other normal tissues) based on contrast‐enhanced dynamic breast MRI data, the corresponding time‐enhancement curves are computed for each voxel in the phantom, according to tissue type. These curves are calculated by convolution between the arterial input function and a shifted exponential function. This exponential depends on the perfusion parameters associated with each tissue voxel, and, to incorporate normal biological variability, a uniform random distribution is used to vary the perfusion parameters on a voxel‐basis. Finally, a 4D array is produced by sampling the continuous time‐enhancement curves at the desired sampling rate. Beside modeling different enhancement dynamics according to the given input perfusion parameters, the phantom also includes the possibility to realistically simulate different spatial enhancement patterns for the breast parenchyma, taking into account the arterial sources supplying the breast. Finally, different patterns of contrast medium uptake can also be simulated for the tumor models (homogeneous and rim enhancement).ResultsAs an example, a typical 4D phantom has dimensions of 426 × 421 × 260 × 559 (x, y, z, t), with a voxel size of 273 μm and a sampling time of 1 s. The characteristics of the tumor model can be modified at will to evaluate perfusion in different types of breast lesions. Results show the expected enhancement of tissues, consistent with the given input parameters. Moreover, the tumor models evaluated in this work show different enhancement dynamics according to the tumor type (defined by different input perfusion parameters), and also present a higher enhancement compared to the other healthy tissues, as expected.ConclusionsThe proposed digital phantom can model the breast tissue perfusion during 4D breast CT image acquisition, displaying the different enhancement dynamics that could be found in a real patient breast. This phantom can be used during the development of dynamic contrast‐enhanced dedicated breast CT imaging, for optimization of image acquisition, image reconstruction, and image analysis. This modality could provide functional information of the breast, resulting in detection, diagnosis, and treatment improvements of breast cancer with breast CT.

Relationships Between Human-Extracted MRI Tumor Phenotypes of Breast Cancer and Clinical Prognostic Indicators Including Receptor Status and Molecular Subtype

The purpose of this study was to investigate if human-extracted MRI tumor phenotypes of breast cancer could predict receptor status and tumor molecular subtype using MRIs from The Cancer Genome Atlas project. Our retrospective interpretation study utilized the analysis of HIPAA-compliant breast MRI data from The Cancer Imaging Archive. One hundred and seven preoperative breast MRIs of biopsy proven invasive breast cancers were analyzed by 3 fellowship-trained breast-imaging radiologists. Each study was scored according to the Breast Imaging Reporting and Data System lexicon for mass and nonmass features. The Spearman rank correlation was used for association analysis of continuous variables; the Kruskal-Wallis test was used for associating continuous outcomes with categorical variables. The Fisher-exact test was used to assess correlations between categorical image-derived features and receptor status. Prediction of estrogen receptor (ER), progesterone receptor, human epidermal growth factor receptor, and molecular subtype were performed using random forest classifiers. ER+ tumors were associated with the absence of rim enhancement (P = 0.019, odds ratio [OR] 5.5), heterogeneous internal enhancement (P = 0.02, OR 6.5), peritumoral edema (P = 0.0001, OR 10.0), and axillary adenopathy (P = 0.04, OR 4.4). ER+ tumors were smaller than ER- tumors (23.7 mm vs 29.2 mm, P = 0.02, OR 8.2). All of these variables except the lack of axillary adenopathy were also associated with progesterone receptor+ status. Luminal A tumors (n = 57) were smaller compared to nonLuminal A (21.8 mm vs 27.5 mm, P = 0.035, OR 7.3) and lacked peritumoral edema (P = 0.001, OR 6.8). Basal like tumors were associated with heterogeneous internal enhancement (P = 0.05, OR 10.1), rim enhancement (P = 0.05, OR6.9), and perituomral edema (P = 0.0001, OR 13.8). Human extracted MRI tumor phenotypes may be able to differentiate those tumors with a more favorable clinical prognosis from their more aggressive counterparts.

Associations between gene expression profiles of invasive breast cancer and Breast Imaging Reporting and Data System MRI lexicon.

PurposeTo evaluate whether the Breast Imaging Reporting and Data System (BI-RADS) MRI lexicon could reflect the genomic information of breast cancers and to suggest intuitive imaging features as biomarkers.MethodsMatched breast MRI data from The Cancer Imaging Archive and gene expression profile from The Cancer Genome Atlas of 70 invasive breast cancers were analyzed. Magnetic resonance images were reviewed according to the BI-RADS MRI lexicon of mass morphology. The cancers were divided into 2 groups of gene clustering by gene set enrichment an alysis. Clinicopathologic and imaging characteristics were compared between the 2 groups.ResultsThe luminal subtype was predominant in the group 1 gene set and the triple-negative subtype was predominant in the group 2 gene set (55 of 56, 98.2% vs. 9 of 14, 64.3%). Internal enhancement descriptors were different between the 2 groups; heterogeneity was most frequent in group 1 (27 of 56, 48.2%) and rim enhancement was dominant in group 2 (10 of 14, 71.4%). In group 1, the gene sets related to mammary gland development were overexpressed whereas the gene sets related to mitotic cell division were overexpressed in group 2.ConclusionWe identified intuitive imaging features of breast MRI associated with distinct gene expression profiles using the standard imaging variables of BI-RADS. The internal enhancement pattern on MRI might reflect specific gene expression profiles of breast cancers, which can be recognized by visual distinction.

A new user friendly visual environment for breast MRI data analysis

Introduction In recent years, there has been a great effort for the exploitation of powerful computer systems in order to support clinical decisions, the so-called CDSS (Clinical Desicion Support Systems). After processing imaging data in an automatic way they can provide useful diagnostic information facilitating like this clinical decisions. Purpose This study demonstrates the advantages of a newly developed friendly visual CDSS environment for Breast MRI Data Analysis, the co-called BreDAn. This software process all dynamic MRI data prior and post contrast medium injection generating kinematic graphs, color maps of signal increase and decrease and finally detecting high risk breast areas. The great advantage of BreDAn is the automation of the radiodiagnostic process in a reliable manner. Materials and methods BreDAn involves a series of steps of MRI Breast data processing, such as data initialization, windowing, intensity filter, intensity curve slope filter, color fragment filter and data display. It has been tested and evaluated using 534 MRI data sets (522 F, age group 31–79 y and 12 M, 55–68 y). A GE Signa HDxt 1.5 Tesla system was used throughout all MRI scans. A description of the MRI protocol is given. Results Out of the 534 cases studied, 9.55% of them found to be pathological and 90.45% were non-malignant. All automatically generated results by BreDAn were compared by radiologist’s diagnosis and found to be in agreement. BreDAn in practice simulates the radiological diagnostic methodology. Conclusion This study illustrates the advantages of BreDAn, a user friendly visual environment for MRI Breast data analysis, which have been tested thoroughly and validated successfully.

Contribution of FDG PET/CT for the Optimization of the Management of Additional Lesions Detected on Local Staging Breast MRI.

The purpose of this study is to evaluate the added value of FDG PET/CT for the management of additional lesions detected by breast MRI during initial staging of known breast cancer. We retrospectively queried our database for all consecutive patients with biopsy-confirmed breast cancer who underwent breast MRI and FDG PET/CT before neoadjuvant chemotherapy between November 2011 and November 2012. The final population comprised 80 patients. Initially, two readers retrospectively analyzed the breast MRI data for the presence of lesions in addition to the previously confirmed index neoplasm. Analysis of FDG PET/CT data followed; two different readers evaluated the presence or absence of FDG uptake in both breasts. The reference standard for additional lesions was percutaneous biopsy with subsequent 2-year follow-up for benign results. Statistical analysis was conducted with nonparametric analysis and the Fisher exact test. The readers detected 61 additional breast lesions at MRI in 45 patients (56.2%); 37 of 61 (61%) additional lesions were malignant and 24 of 61 (39%) were benign. Among the 61 additional breast lesions detected by MRI, only 32 were positive at FDG PET/CT, with a positive predictive value of 90.6% and negative predictive value of 72%. The sensitivity, specificity, and accuracy of FDG PET/CT were 78.3%, 87.5%, and 81.9%, respectively. In eight cases, FDG PET/CT missed additional malignant lesions. All missed malignant additional lesions were smaller than 1 cm. In three cases, additional lesions also detected at FDG PET/CT were benign. In the case of additional lesions detected at MRI, a negative FDG PET/CT finding could replace a breast biopsy for lesions larger than 1 cm.

Longitudinal, intermodality registration of quantitative breast PET and MRI data acquired before and during neoadjuvant chemotherapy: preliminary results.

The authors propose a method whereby serially acquired DCE-MRI, DW-MRI, and FDG-PET breast data sets can be spatially and temporally coregistered to enable the comparison of changes in parameter maps at the voxel level. First, the authors aligned the PET and MR images at each time point rigidly and nonrigidly. To register the MR images longitudinally, the authors extended a nonrigid registration algorithm by including a tumor volume-preserving constraint in the cost function. After the PET images were aligned to the MR images at each time point, the authors then used the transformation obtained from the longitudinal registration of the MRI volumes to register the PET images longitudinally. The authors tested this approach on ten breast cancer patients by calculating a modified Dice similarity of tumor size between the PET and MR images as well as the bending energy and changes in the tumor volume after the application of the registration algorithm. The median of the modified Dice in the registered PET and DCE-MRI data was 0.92. For the longitudinal registration, the median tumor volume change was -0.03% for the constrained algorithm, compared to -32.16% for the unconstrained registration algorithms (p = 8 × 10(-6)). The medians of the bending energy were 0.0092 and 0.0001 for the unconstrained and constrained algorithms, respectively (p = 2.84 × 10(-7)). The results indicate that the proposed method can accurately spatially align DCE-MRI, DW-MRI, and FDG-PET breast images acquired at different time points during therapy while preventing the tumor from being substantially distorted or compressed.

Joint estimation of shape and deformation for the detection of lesions in dynamic contrast-enhanced breast MRI

We propose a mathematical framework for simultaneously delineating the boundary of object and estimating its temporal motion in the application of lesion detection in a dynamic contrast-enhanced (DCE) breast MRI sequence where both the appearance and the shape of region of interest is assumed to change in time. A unified energy functional for a joint segmentation and registration is proposed based on the assumption that the statistical properties of dynamic intensity curves within a region of interest are homogeneous. Our algorithm is designed to provide the morphological properties of the enhanced region and its dynamic intensity profiles, called kinetic signatures, in the analysis of DCE imagery since these features are considered as significant cues in understanding images. The proposed energy comprises a combination of a segmentation energy and a registration energy. The segmentation energy is developed based on a convex formulation being insensitive to the initialization. The registration energy is designed to compensate motion artifacts that are usually involved in the temporal imaging procedure. The major objective of this work is to provide a mathematical framework for a joint segmentation and registration on a dynamic sequence of images, and we demonstrate the mutual benefit of the estimation of temporal deformations for the registration step and the localization of regions of interest for the segmentation step. The effectiveness of the developed algorithm has been demonstrated on a number of clinical DCE breast MRI data in the application of breast lesion detection and the results show its potential to improve the accuracy and the efficiency in the diagnosis of breast cancer.

A new user-friendly visual environment for breast MRI data analysis

In this paper a novel, user friendly visual environment for Breast MRI Data Analysis is presented (BreDAn). Given planar MRI images before and after IV contrast medium injection, BreDAn generates kinematic graphs, color maps of signal increase and decrease and finally detects high risk breast areas. The advantage of BreDAn, which has been validated and tested successfully, is the automation of the radiodiagnostic process in an accurate and reliable manner. It can potentially facilitate radiologists’ workload.

Model-free visualization of suspicious lesions in breast MRI based on supervised and unsupervised learning

Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) has become an important tool in breast cancer diagnosis, but evaluation of multitemporal 3D image data holds new challenges for human observers. To aid the image analysis process, we apply supervised and unsupervised pattern recognition techniques for computing enhanced visualizations of suspicious lesions in breast MRI data. These techniques represent an important component of future sophisticated computer-aided diagnosis (CAD) systems and support the visual exploration of spatial and temporal features of DCE-MRI data stemming from patients with confirmed lesion diagnosis. By taking into account the heterogeneity of cancerous tissue, these techniques reveal signals with malignant, benign and normal kinetics. They also provide a regional subclassification of pathological breast tissue, which is the basis for pseudo-color presentations of the image data. Intelligent medical systems are expected to have substantial implications in healthcare politics by contributing to the diagnosis of indeterminate breast lesions by non-invasive imaging.

Bilateral symmetry analysis of breast MRI

Mammographic interpretation often uses symmetry between left and right breasts to indicate the site of potential tumour masses. This approach has not been applied to breast images obtained from MRI. We present an automatic technique for breast symmetry detection based on feature extraction techniques which does not require any efforts to co-register breast MRI data. The approach applies computer-vision techniques to detect natural biological symmetries in breast MR scans based on three objective measures of similarity: multiresolution non-orthogonal wavelet representation, three-dimensional intensity distributions and co-occurrence matrices. Statistical distributions that are invariant to feature localization are computed for each of the extracted image features. These distributions are later compared against each other to account for perceptual similarity. Studies based on 51 normal MRI scans of randomly selected patients showed that the sensitivity of symmetry detection rate approached 94%. The symmetry analysis procedure presented in this paper can be applied as an aid in detecting breast tissue changes arising from disease.