Michiel Kallenberg

Volumetric breast density estimation from Full-Field Digital Mammograms: A validation study

A method is presented for estimation of dense breast tissue volume from mammograms obtained with full-field digital mammography (FFDM). The thickness of dense tissue mapping to a pixel is determined by using a physical model of image acquisition. This model is based on the assumption that the breast is composed of two types of tissue, fat and parenchyma. Effective linear attenuation coefficients of these tissues are derived from empirical data as a function of tube voltage (kVp), anode material, filtration, and compressed breast thickness. By employing these, tissue composition at a given pixel is computed after performing breast thickness compensation, using a reference value for fatty tissue determined by the maximum pixel value in the breast tissue projection. Validation has been performed using 22 FFDM cases acquired with a GE Senographe 2000D by comparing the volume estimates with volumes obtained by semi-automatic segmentation of breast magnetic resonance imaging (MRI) data. The correlation between MRI and mammography volumes was 0.94 on a per image basis and 0.97 on a per patient basis. Using the dense tissue volumes from MRI data as the gold standard, the average relative error of the volume estimates was 13.6%.

Unsupervised Deep Learning Applied to Breast Density Segmentation and Mammographic Risk Scoring

Mammographic risk scoring has commonly been automated by extracting a set of handcrafted features from mammograms, and relating the responses directly or indirectly to breast cancer risk. We present a method that learns a feature hierarchy from unlabeled data. When the learned features are used as the input to a simple classifier, two different tasks can be addressed: i) breast density segmentation, and ii) scoring of mammographic texture. The proposed model learns features at multiple scales. To control the models capacity a novel sparsity regularizer is introduced that incorporates both lifetime and population sparsity. We evaluated our method on three different clinical datasets. Our state-of-the-art results show that the learned breast density scores have a very strong positive relationship with manual ones, and that the learned texture scores are predictive of breast cancer. The model is easy to apply and generalizes to many other segmentation and scoring problems.

Volumetric Breast Density from Full-Field Digital Mammograms and Its Association with Breast Cancer Risk Factors: A Comparison with a Threshold Method

Introduction: Breast density, a strong breast cancer risk factor, is usually measured on the projected breast area from film screen mammograms. This is far from ideal, as breast thickness and technical characteristics are not taken into account. We investigated whether volumetric density measurements on full-field digital mammography (FFDM) are more strongly related to breast cancer risk factors than measurements with a computer-assisted threshold method. Methods: Breast density was measured on FFDMs from 370 breast cancer screening participants, using a computer-assisted threshold method and a volumetric method. The distribution of breast cancer risk factors among quintiles of density was compared between both methods. We adjusted for age and body mass index (BMI) with linear regression analysis. Results: High percent density was strongly related to younger age, lower BMI, nulliparity, late age at first delivery and pre/perimenopausal status, to the same extent with both methods (all P < 0.05). Similarly strong relationships were seen for the absolute dense area but to a lesser extent for absolute dense volume. A larger dense volume was only significantly associated with late age at menopause, use of menopausal hormone therapy, and, in contrast to the other methods, high BMI. Conclusion: Both methods related equally well to known breast cancer risk factors. Impact: Despite its alleged higher precision, the volumetric method was not more strongly related to breast cancer risk factors. This is in agreement with other studies. The definitive relationship with breast cancer risk still needs to be investigated. Cancer Epidemiol Biomarkers Prev; 19(12); 3096–105. ©2010 AACR.

Automatic breast density segmentation: an integration of different approaches

Mammographic breast density has been found to be a strong risk factor for breast cancer. In most studies, it is assessed with a user-assisted threshold method, which is time consuming and subjective. In this study, we develop a breast density segmentation method that is fully automatic. The method is based on pixel classification in which different approaches known in the literature to segment breast density are integrated and extended. In addition, the method incorporates the knowledge of a trained observer, by using segmentations obtained by the user-assisted threshold method as training data. The method is trained and tested using 1300 digitized film mammographic images acquired with a variety of systems. Results show a high correspondence between the automated method and the user-assisted threshold method. Pearsons correlation coefficient between our method and the user-assisted method is R = 0.911 for percent density and R = 0.895 for dense area, which is substantially higher than the best correlation found in the literature (R = 0.70, R = 0.68). The area under the receiver operating characteristic curve obtained when discriminating between fatty and dense pixels is 0.987. A combination of segmentation strategies outperforms the application of single segmentation techniques.

Breast Segmentation and Density Estimation in Breast MRI: A Fully Automatic Framework

Breast density measurement is an important aspect in breast cancer diagnosis as dense tissue has been related to the risk of breast cancer development. The purpose of this study is to develop a method to automatically compute breast density in breast MRI. The framework is a combination of image processing techniques to segment breast and fibroglandular tissue. Intra- and interpatient signal intensity variability is initially corrected. The breast is segmented by automatically detecting body-breast and air-breast surfaces. Subsequently, fibroglandular tissue is segmented in the breast area using expectation-maximization. A dataset of 50 cases with manual segmentations was used for evaluation. Dice similarity coefficient (DSC), total overlap, false negative fraction (FNF), and false positive fraction (FPF) are used to report similarity between automatic and manual segmentations. For breast segmentation, the proposed approach obtained DSC, total overlap, FNF, and FPF values of 0.94, 0.96, 0.04, and 0.07, respectively. For fibroglandular tissue segmentation, we obtained DSC, total overlap, FNF, and FPF values of 0.80, 0.85, 0.15, and 0.22, respectively. The method is relevant for researchers investigating breast density as a risk factor for breast cancer and all the described steps can be also applied in computer aided diagnosis systems.

Segmentation of the pectoral muscle in breast MRI using atlas-based approaches

Pectoral muscle segmentation is an important step in automatic breast image analysis methods and crucial for multi-modal image registration. In breast MRI, accurate delineation of the pectoral is important for volumetric breast density estimation and for pharmacokinetic analysis of dynamic contrast enhancement. In this paper we propose and study the performance of atlas-based segmentation methods evaluating two fully automatic breast MRI dedicated strategies on a set of 27 manually segmented MR volumes. One uses a probabilistic model and the other is a multi-atlas registration based approach. The multi-atlas approach performed slightly better, with an average Dice coefficient (DSC) of 0.74, while with the much faster probabilistic method a DSC of 0.72 was obtained.

Effect of compression paddle tilt correction on volumetric breast density estimation.

For the acquisition of a mammogram, a breast is compressed between a compression paddle and a support table. When compression is applied with a flexible compression paddle, the upper plate may be tilted, which results in variation in breast thickness from the chest wall to the breast margin. Paddle tilt has been recognized as a major problem in volumetric breast density estimation methods. In previous work, we developed a fully automatic method to correct the image for the effect of compression paddle tilt. In this study, we investigated in three experiments the effect of paddle tilt and its correction on volumetric breast density estimation. Results showed that paddle tilt considerably affected accuracy of volumetric breast density estimation, but that effect could be reduced by tilt correction. By applying tilt correction, a significant increase in correspondence between mammographic density estimates and measurements on MRI was established. We argue that in volumetric breast density estimation, tilt correction is both feasible and essential when mammographic images are acquired with a flexible compression paddle.

Computer-aided detection of masses in full-field digital mammography using screen-film mammograms for training

It would be of great value when available databases of screen-film mammography (SFM) images can be used to train full-field digital mammography (FFDM) computer-aided detection (CAD) systems, as compilation of new databases is costly. In this paper, we investigate this possibility. Firstly, we develop a method that converts an FFDM image into an SFM-like representation. In this conversion method, we establish a relation between exposure and optical density by simulation of an automatic exposure control unit. Secondly, we investigate the effects of using the SFM images as training samples compared to training with FFDM images. Our FFDM database consisted of 266 cases, of which 102 were biopsy-proven malignant masses and 164 normals. The images were acquired with systems of two different manufacturers. We found that, when we trained our FFDM CAD system with a small number of images, training with FFDM images, using a five-fold crossvalidation procedure, outperformed training with SFM images. However, when the full SFM database, consisting of 348 abnormal cases (including 204 priors) and 810 normal cases, was used for training, SFM training outperformed FFDMA training. These results show that an existing CAD system for detection of masses in SFM can be used for FFDM images without retraining.

Compression paddle tilt correction in full-field digital mammograms.

During the acquisition of a mammogram the breast is compressed between the compression paddle and the support table. When compression is applied, the upper plate is tilted which results in variation in breast thickness from the chest wall to the breast margin. Variation in breast thickness influences the grey-level values of the image and hampers image analysis, such as volumetric breast density estimation. In this paper, we present and compare two methods that estimate and correct image tilt. The first method estimates tilt from fatty tissue regions. The second method is based on the entropy of the grey-level distribution of the image. Both methods use a classifier that distinguishes fatty areas from dense tissue based on texture features independent of tilt. The tilt correction methods are evaluated by assessing their accuracies in estimating artificial tilts that are added to images that are known to have only a small tilt. On average, both methods are able to estimate the artificial tilt. To the best of our knowledge, this is the first paper that presents and validates tilt correction methods on individual mammograms.

Multi-class probabilistic atlas-based segmentation method in breast MRI

Organ localization is an important topic in medical imaging in aid of cancer treatment and diagnosis. An example are the pharmacokinetic model calibration methods based on a reference tissue, where a pectoral muscle delineation in breast MRI is needed to detect malignancy signs. Atlas-based segmentation has been proven to be powerful in brain MRI. This is the first attempt to apply an atlas-based approach to segment breast in T1 weighted MR images. The atlas consists of 5 structures (fatty and dense tissues, heart, lungs and pectoral muscle). It has been used in a Bayesian segmentation framework to delineate the mentioned structures. Global and local registration have been compared, where global registration showed the best results in terms of accuracy and speed. Overall, a Dice Similarity Coefficient value of 0.8 has been obtained which shows the validity of our approach to Breast MRI segmentation.