Jan van Zelst

Computer-aided detection of breast cancers using Haar-like features in automated 3D breast ultrasound.

PURPOSE Automated 3D breast ultrasound (ABUS) has gained interest in breast imaging. Especially for screening women with dense breasts, ABUS appears to be beneficial. However, since the amount of data generated is large, the risk of oversight errors is substantial. Computer aided detection (CADe) may be used as a second reader to prevent oversight errors. When CADe is used in this fashion, it is essential that small cancers are detected, while the number of false positive findings should remain acceptable. In this work, the authors improve their previously developed CADe system in the initial candidate detection stage. METHODS The authors use a large number of 2D Haar-like features to differentiate lesion structures from false positives. Using a cascade of GentleBoost classifiers that combines these features, a likelihood score, highly specific for small cancers, can be efficiently computed. The likelihood scores are added to the previously developed voxel features to improve detection. RESULTS The method was tested in a dataset of 414 ABUS volumes with 211 cancers. Cancers had a mean size of 14.72 mm. Free-response receiver operating characteristic analysis was performed to evaluate the performance of the algorithm with and without using the aforementioned Haar-like feature likelihood scores. After the initial detection stage, the number of missed cancer was reduced by 18.8% after adding Haar-like feature likelihood scores. CONCLUSIONS The proposed technique significantly improves our previously developed CADe system in the initial candidate detection stage.

Multiplanar Reconstructions of 3D Automated Breast Ultrasound Improve Lesion Differentiation by Radiologists

RATIONALE AND OBJECTIVES To investigate the value of multiplanar reconstructions (MPRs) of automated three-dimensional (3D) breast ultrasound (ABUS) compared to transverse evaluation only, in differentiation of benign and malignant breast lesions. MATERIALS AND METHODS Five breast radiologists evaluated ABUS scans of 96 female patients with biopsy-proven abnormalities (36 malignant and 60 benign). They classified the most suspicious lesion based on the breast imaging reporting and data system (BI-RADS) lexicon using the transverse scans only. A likelihood-of-malignancy (LOM) score (0-100) and a BI-RADS final assessment were assigned. Thereafter, the MPR was provided and readers scored the cases again. In addition, they rated the presence of spiculation and retraction in the coronal plane on a five-point scale called Spiculation and Retraction Severity Index (SRSI). Reader performance was analyzed with receiver-operating characteristics analysis. RESULTS The area under the curve increased from 0.82 to 0.87 (P = .01) after readers were shown the reconstructed planes. The SRSI scores are highly correlated (Spearmans r) with the final LOM scores (range, r = 0.808-0.872) and ΔLOM scores (range, r = 0.525-0.836). Readers downgraded 3%-18% of the biopsied benign lesions to BI-RADS 2 after MPR evaluation. Inter-reader agreement for SRSI was substantial (intraclass correlation coefficient, 0.617). Inter-reader agreement of the BI-RADS final assessment improved from 0.367 to 0.536 after MPRs were read. CONCLUSIONS Full 3D evaluation of ABUS using MPR improves differentiation of breast lesions in comparison to evaluating only transverse planes. Results suggest that the added value of MPR might be related to visualization of spiculation and retraction patterns in the coronal reconstructions.

Consistency of breast density categories in serial screening mammograms: A comparison between automated and human assessment.

Reliable breast density measurement is needed to personalize screening by using density as a risk factor and offering supplemental screening to women with dense breasts. We investigated the categorization of pairs of subsequent screening mammograms into density classes by human readers and by an automated system. With software (VDG) and by four readers, including three specialized breast radiologists, 1000 mammograms belonging to 500 pairs of subsequent screening exams were categorized into either two or four density classes. We calculated percent agreement and the percentage of women that changed from dense to non-dense and vice versa. Inter-exam agreement (IEA) was calculated with kappa statistics. Results were computed for each reader individually and for the case that each mammogram was classified by one of the four readers by random assignment (group reading). Higher percent agreement was found with VDG (90.4%, CI 87.9-92.9%) than with readers (86.2-89.2%), while less plausible changes from non-dense to dense occur less often with VDG (2.8%, CI 1.4-4.2%) than with group reading (4.2%, CI 2.4-6.0%). We found an IEA of 0.68-0.77 for the readers using two classes and an IEA of 0.76-0.82 using four classes. IEA is significantly higher with VDG compared to group reading. The categorization of serial mammograms in density classes is more consistent with automated software than with a mixed group of human readers. When using breast density to personalize screening protocols, assessment with software may be preferred over assessment by radiologists.

Incorporating texture features in a computer-aided breast lesion diagnosis system for automated three-dimensional breast ultrasound.

Abstract. We investigated the benefits of incorporating texture features into an existing computer-aided diagnosis (CAD) system for classifying benign and malignant lesions in automated three-dimensional breast ultrasound images. The existing system takes into account 11 different features, describing different lesion properties; however, it does not include texture features. In this work, we expand the system by including texture features based on local binary patterns, gray level co-occurrence matrices, and Gabor filters computed from each lesion to be diagnosed. To deal with the resulting large number of features, we proposed a combination of feature-oriented classifiers combining each group of texture features into a single likelihood, resulting in three additional features used for the final classification. The classification was performed using support vector machine classifiers, and the evaluation was done with 10-fold cross validation on a dataset containing 424 lesions (239 benign and 185 malignant lesions). We compared the classification performance of the CAD system with and without texture features. The area under the receiver operating characteristic curve increased from 0.90 to 0.91 after adding texture features (p<0.001).

Dedicated computer-aided detection software for automated 3D breast ultrasound; an efficient tool for the radiologist in supplemental screening of women with dense breasts

ObjectivesTo determine the effect of computer-aided-detection (CAD) software for automated breast ultrasound (ABUS) on reading time (RT) and performance in screening for breast cancer.Material and methodsUnilateral ABUS examinations of 120 women with dense breasts were randomly selected from a multi-institutional archive of cases including 30 malignant (20/30 mammography-occult), 30 benign, and 60 normal cases with histopathological verification or ≥ 2 years of negative follow-up. Eight radiologists read once with (CAD-ABUS) and once without CAD (ABUS) with > 8 weeks between reading sessions. Readers provided a BI-RADS score and a level of suspiciousness (0-100). RT, sensitivity, specificity, PPV and area under the curve (AUC) were compared.ResultsAverage RT was significantly shorter using CAD-ABUS (133.4 s/case, 95% CI 129.2-137.6) compared with ABUS (158.3 s/case, 95% CI 153.0-163.3) (p < 0.001). Sensitivity was 0.84 for CAD-ABUS (95% CI 0.79-0.89) and ABUS (95% CI 0.78-0.88) (p = 0.90). Three out of eight readers showed significantly higher specificity using CAD. Pooled specificity (0.71, 95% CI 0.68-0.75 vs. 0.67, 95% CI 0.64-0.70, p = 0.08) and PPV (0.50, 95% CI 0.45-0.55 vs. 0.44, 95% CI 0.39-0.49, p = 0.07) were higher in CAD-ABUS vs. ABUS, respectively, albeit not significantly. Pooled AUC for CAD-ABUS was comparable with ABUS (0.82 vs. 0.83, p = 0.53, respectively).ConclusionCAD software for ABUS may decrease the time needed to screen for breast cancer without compromising the screening performance of radiologists.Key Points• ABUS with CAD software may speed up reading time without compromising radiologists’ accuracy.• CAD software for ABUS might prevent non-detection of malignant breast lesions by radiologists.• Radiologists reading ABUS with CAD software might improve their specificity without losing sensitivity.

Improvements of an objective model of compressed breasts undergoing mammography: Generation and characterization of breast shapes

Purpose To develop a set of accurate 2D models of compressed breasts undergoing mammography or breast tomosynthesis, based on objective analysis, to accurately characterize mammograms with few linearly independent parameters, and to generate novel clinically realistic paired cranio‐caudal (CC) and medio‐lateral oblique (MLO) views of the breast. Methods We seek to improve on an existing model of compressed breasts by overcoming detector size bias, removing the nipple and non‐mammary tissue, pairing the CC and MLO views from a single breast, and incorporating the pectoralis major muscle contour into the model. The outer breast shapes in 931 paired CC and MLO mammograms were automatically detected with an in‐house developed segmentation algorithm. From these shapes three generic models (CC‐only, MLO‐only, and joint CC/MLO) with linearly independent components were constructed via principal component analysis (PCA). The ability of the models to represent mammograms not used for PCA was tested via leave‐one‐out cross‐validation, by measuring the average distance error (ADE). Results The individual models based on six components were found to depict breast shapes with accuracy (mean ADE‐CC = 0.81 mm, ADE‐MLO = 1.64 mm, ADE‐Pectoralis = 1.61 mm), outperforming the joint CC/MLO model (P ≤ 0.001). The joint model based on 12 principal components contains 99.5% of the total variance of the data, and can be used to generate new clinically realistic paired CC and MLO breast shapes. This is achieved by generating random sets of 12 principal components, following the Gaussian distributions of the histograms of each component, which were obtained from the component values determined from the images in the mammography database used. Conclusion Our joint CC/MLO model can successfully generate paired CC and MLO view shapes of the same simulated breast, while the individual models can be used to represent with high accuracy clinical acquired mammograms with a small set of parameters. This is the first step toward objective 3D compressed breast models, useful for dosimetry and scatter correction research, among other applications.

Segmentation of malignant lesions in 3D breast ultrasound using a depth-dependent model

PURPOSE Automated 3D breast ultrasound (ABUS) has been proposed as a complementary screening modality to mammography for early detection of breast cancers. To facilitate the interpretation of ABUS images, automated diagnosis and detection techniques are being developed, in which malignant lesion segmentation plays an important role. However, automated segmentation of cancer in ABUS is challenging since lesion edges might not be well defined. In this study, the authors aim at developing an automated segmentation method for malignant lesions in ABUS that is robust to ill-defined cancer edges and posterior shadowing. METHODS A segmentation method using depth-guided dynamic programming based on spiral scanning is proposed. The method automatically adjusts aggressiveness of the segmentation according to the position of the voxels relative to the lesion center. Segmentation is more aggressive in the upper part of the lesion (close to the transducer) than at the bottom (far away from the transducer), where posterior shadowing is usually visible. The authors used Dice similarity coefficient (Dice) for evaluation. The proposed method is compared to existing state of the art approaches such as graph cut, level set, and smart opening and an existing dynamic programming method without depth dependence. RESULTS In a dataset of 78 cancers, our proposed segmentation method achieved a mean Dice of 0.73 ± 0.14. The method outperforms an existing dynamic programming method (0.70 ± 0.16) on this task (p = 0.03) and it is also significantly (p < 0.001) better than graph cut (0.66 ± 0.18), level set based approach (0.63 ± 0.20) and smart opening (0.65 ± 0.12). CONCLUSIONS The proposed depth-guided dynamic programming method achieves accurate breast malignant lesion segmentation results in automated breast ultrasound.

Automated quality assessment in three-dimensional breast ultrasound images

Abstract. Automated three-dimensional breast ultrasound (ABUS) is a valuable adjunct to x-ray mammography for breast cancer screening of women with dense breasts. High image quality is essential for proper diagnostics and computer-aided detection. We propose an automated image quality assessment system for ABUS images that detects artifacts at the time of acquisition. Therefore, we study three aspects that can corrupt ABUS images: the nipple position relative to the rest of the breast, the shadow caused by the nipple, and the shape of the breast contour on the image. Image processing and machine learning algorithms are combined to detect these artifacts based on 368 clinical ABUS images that have been rated manually by two experienced clinicians. At a specificity of 0.99, 55% of the images that were rated as low quality are detected by the proposed algorithms. The areas under the ROC curves of the single classifiers are 0.99 for the nipple position, 0.84 for the nipple shadow, and 0.89 for the breast contour shape. The proposed algorithms work fast and reliably, which makes them adequate for online evaluation of image quality during acquisition. The presented concept may be extended to further image modalities and quality aspects.

Automated Three-dimensional Breast US for Screening: Technique, Artifacts, and Lesion Characterization

Automated breast (AB) ultrasonography (US) scanners have recently been brought to market for breast imaging. AB US devices use mechanically driven wide linear-array transducers that can image whole-breast US volumes in three dimensions. AB US is proposed for screening as a supplemental modality to mammography in women with dense breasts and overcomes important limitations of whole-breast US using handheld devices, such as operator dependence and limited reproducibility. A literature review of supplemental whole-breast US for screening was performed, which showed that both AB US and handheld US allow detection of mammographically negative early-stage invasive breast cancers but also increase the false-positive recall rate. Technicians with limited training can perform AB US; nevertheless, there is a learning curve for acquiring optimal images. Proper acquisition technique may allow avoidance of common artifacts that could impair interpretation of AB US results. Regardless, interpretation of AB US results can be challenging. This article reviews the US appearance of common benign and malignant lesions and presents examples of false-positive and false-negative AB US results. In situ breast cancers are rarely detected with supplemental whole-breast US. The most discriminating feature that separates AB US from handheld US is the retraction phenomenon on coronal reformatted images. The retraction phenomenon is rarely seen with benign findings but accompanies almost all breast cancers. In conclusion, women with dense breasts may benefit from supplemental AB US examinations. Understanding the pitfalls in acquisition technique and lesion interpretation, both of which can lead to false-positive recalls, might reduce the potential harm of performing supplemental AB US. Online supplemental material is available for this article. ©RSNA, 2018.

Predicting treatment response in invasive ductal breast carcinoma using three-dimensional quantitative ultrasound analysis

About 1 in 8 women will develop breast cancer. Histological tumor grade and biological markers such as oestrogen receptors (ER), progesterone receptors (PR), and human epidermal growth factor receptor 2 (HER2) are prognostic indicators for the clinical response to medical treatment and patient prognosis. Breast ultrasound (US) is mainly used to diagnose breast cancer or guide breast biopsies. However, quantitative analysis of US could also be of clinical value by giving additional information on tissue type. The automated breast volume scanner (ABUS) allows an automated assessment of a complete 3D US breast volume. By analyzing the texture of the B-mode images of the ABUS we have already shown that 3D quantitative breast ultrasound (3DQBUS) provides good discrimination between fibroadenomas (benign lesions) and invasive ductal carcinomas (IDC)(AUC=0.89). In this study we investigated whether 3DQBUS can be used to predict grade and receptor status.