Computer-Aided Diagnosis for 3-D Power Doppler Breast Ultrasound

In recent studies, both tumor morphology and vascularity played an important role in differentiating breast tumors. In this article, a computer-aided diagnosis (CAD) system was proposed to quantify the tumor morphology of vascularity on three-dimensional (3-D) power Doppler breast ultrasound (PDUS) images. We segmented the tumor margin by the level set method and skeletonized vessels by the 3-D thinning algorithm from 3-D PDUS data to capture the B-mode and vascularity features. The B-mode features including texture, shape and ellipsoid fitting and the vascularity features containing volume, complexity, length, radius and tortuosity were used to differentiate breast tumors. In the experiment, 82 biopsy-verified lesions including 41 benign and 41 malignant lesions were used to test the performance of the proposed system. The proposed method performed well, achieving accuracy, sensitivity, specificity and Az values of 85.37% (70/82), 85.37% (35/41), 85.37% (35/41) and 0.9104, respectively.     

Multi-task learning for segmentation and classification of tumors in 3D automated breast ultrasound images

Tumor classification and segmentation are two important tasks for computer-aided diagnosis (CAD) using 3D automated breast ultrasound (ABUS) images. However, they are challenging due to the significant shape variation of breast tumors and the fuzzy nature of ultrasound images (e.g., low contrast and signal to noise ratio). Considering the correlation between tumor classification and segmentation, we argue that learning these two tasks jointly is able to improve the outcomes of both tasks. In this paper, we propose a novel multi-task learning framework for joint segmentation and classification of tumors in ABUS images. The proposed framework consists of two sub-networks: an encoder-decoder network for segmentation and a light-weight multi-scale network for classification. To account for the fuzzy boundaries of tumors in ABUS images, our framework uses an iterative training strategy to refine feature maps with the help of probability maps obtained from previous iterations. Experimental results based on a clinical dataset of 170 3D ABUS volumes collected from 107 patients indicate that the proposed multi-task framework improves tumor segmentation and classification over the single-task learning counterparts.     

Analysis of Elastographic and B-mode Features at Sonoelastography for Breast Tumor Classification

The purpose of this study was to evaluate the accuracy of neural network analysis of elastographic features at sonoelastography for the classification of biopsy-proved benign and malignant breast tumors. Sonoelastography of 181 solid breast masses (113 benign and 68 malignant tumors) was performed for 181 patients (mean age, 47 years; range, 24–75 years). After the manual segmentation of the tumors, five elastographic features (strain difference, strain ratio, mean, median and mode) and six B-mode features (orientation, undulation, angularity, average gradient, gradient variance and intensity variance) were computed. A neural network was used to classify tumors by the use of these features. The Student's t test and receiver operating characteristic (ROC) curve analysis were used for statistical analysis. Area under ROC curve (Az) values of the three elastographic features– mean (0.87), median (0.86) and mode (0.83)–were significantly higher than the Az values for the six B-mode features (0.54–0.69) (p < 0.01). Accuracy, sensitivity, specificity and Az of the neural network for the classification of solid breast tumors were 86.2% (156/181), 83.8% (57/68), 87.6% (99/113) and 0.84 for the elastographic features, respectively, and 82.3% (149/181), 70.6% (48/68), 89.4% (101/113) and 0.78 for the B-mode features, respectively, and 90.6% (164/181), 95.6% (65/68), 87.6% (99/113) and 0.92 for the combination of the elastographic and B-mode features, respectively. We conclude that sonoelastographic images and neural network analysis of features has the potential to increase the accuracy of the use of ultrasound for the classification of benign and malignant breast tumors. (E-mail: rfchang@csie.ntu.edu.tw)     

Effect of a novel segmentation algorithm on radiologists' diagnosis of breast masses using ultrasound imaging

We investigated the effect of using a novel segmentation algorithm on radiologists’ sensitivity and specificity for discriminating malignant masses from benign masses using ultrasound. Five-hundred ten conventional ultrasound images were processed by a novel segmentation algorithm. Five radiologists were invited to analyze the original and computerized images independently. Performances of radiologists with or without computer aid were evaluated by receiver operating characteristic (ROC) curve analysis. The masses became more obvious after being processed by the segmentation algorithm. Without using the algorithm, the areas under the ROC curve (Az) of the five radiologists ranged from 0.70∼0.84. Using the algorithm, the Az increased significantly (range, 0.79∼0.88; p < 0.001). The proposed segmentation algorithm could improve the radiologists’ diagnosis performance by reducing the image speckles and extracting the mass margin characteristics.     

Fuzzy-based classification of breast lesions using ultrasound echography and elastography

ABSTRACT: Common breast lesions have different elasticity properties. Segmentation of contours of breast lesions from elastography and B mode images by incorporating variational level set method is involved in the proposed work. After segmentation, strain and shape features, such as differences in area, perimeter, and contour and width to height difference and solidity, as well as texture features like contrast, entropy, standard deviation, dissimilarity, homogeneity and energy, are estimated. A nonlinear fuzzy inference system is applied for classifying the breast lesions as benign cyst, benign solid mass, or malignant solid mass. Detection of malignant solid masses is our primary objective. A classification accuracy of 83% is obtained. One hundred percent sensitivity is reported. It can be concluded that the proposed fuzzy-based classification technique can be used as an aid for the automated detection of breast lesions.     

Breast tumor classification using fuzzy clustering for breast elastography

Elastography is a new ultrasound imaging technique to provide the information about relative tissue stiffness. The elasticity information provided by this dynamic imaging method has proven to be helpful in distinguishing benign and malignant breast tumors. In previous studies for computer-aided diagnosis (CAD), the tumor contour was manually segmented and each pixel in the elastogram was classified into hard or soft tissue using the simple thresholding technique. In this paper, the tumor contour was automatically segmented by the level set method to provide more objective and reliable tumor contour for CAD. Moreover, the elasticity of each pixel inside each tumor was classified by the fuzzy c-means clustering technique to obtain a more precise diagnostic result. The test elastography database included 66 benign and 31 malignant biopsy-proven tumors. In the experiments, the accuracy, sensitivity, specificity and the area index Az under the receiver operating characteristic curve for the classification of solid breast masses were 83.5% (81/97), 83.9% (26/31), 83.3% (55/66) and 0.902 for the fuzzy c-means clustering method, respectively, and 59.8% (58/97), 96.8% (30/31), 42.4% (28/66) and 0.818 for the conventional thresholding method, respectively. The differences of accuracy, specificity and Az value were statistically significant (p < 0.05). We conclude that the proposed method has the potential to provide a CAD tool to help physicians to more reliably and objectively diagnose breast tumors using elastography.(E-mail: rfchang@csie.ntu.edu.tw)     

Deformable mapping using biomechanical models to relate corresponding lesions in digital breast tomosynthesis and automated breast ultrasound images

This work investigates the application of a deformable localization/mapping method to register lesions between the digital breast tomosynthesis (DBT) craniocaudal (CC) and mediolateral oblique (MLO) views and automated breast ultrasound (ABUS) images. This method was initially validated using compressible breast phantoms. This methodology was applied to 7 patient data sets containing 9 lesions. The automated deformable mapping algorithm uses finite element modeling and analysis to determine corresponding lesions based on the distance between their centers of mass (d_COM) in the deformed DBT model and the reference ABUS model. This technique shows that location information based on external fiducial markers is helpful in the improvement of registration results. However, use of external markers are not required for deformable registration results described by this methodology. For DBT (CC view) mapped to ABUS, the mean d_COM was 14.9 ± 6.8 mm based on 9 lesions using 6 markers in deformable analysis. For DBT (MLO view) mapped to ABUS, the mean d_COM was 13.7 ± 6.8 mm based on 8 lesions using 6 markers in analysis. Both DBT views registered to ABUS lesions showed statistically significant improvements (p ≤ 0.05) in registration using the deformable technique in comparison to a rigid registration. Application of this methodology could help improve a radiologist's characterization and accuracy in relating corresponding lesions between DBT and ABUS image datasets, especially for cases of high breast densities and multiple masses.     

Diagnosis of breast tumors with sonographic texture analysis using wavelet transform and neural networks

To increase the ability of ultrasonographic technology for the differential diagnosis of solid breast tumors, we describe a novel computer-aided diagnosis (CADx) system using neural networks for classification of breast tumors. Tumor regions and surrounding tissues are segmented from the physician-located region-of-interest (ROI) images by applying our proposed segmentation algorithm. Cooperating with the segmentation algorithm, three feasible features, including variance contrast, autocorrelation contrast and distribution distortion of wavelet coefficients, were extracted from the ROI images for further classification. A multilayered perceptron (MLP) neural network trained using error back-propagation algorithm with momentum was then used for the differential diagnosis of breast tumors on sonograms. In the experiment, 242 cases (including benign breast tumors from 161 patients and carcinomas from 82 patients) were sampled with k-fold cross-validation (k = 10) to evaluate the performance. The receiver operating characteristic (ROC) area index for the proposed CADx system is 0.9396 +/- 0.0183, the sensitivity is 98.77%, the specificity is 81.37%, the positive predictive value is 72.73% and the negative predictive value is 99.24%. Experimental results showed that our diagnosis model performed very well for breast tumor diagnosis.     

Breast Cancer Classification in Automated Breast Ultrasound Using Multiview Convolutional Neural Network with Transfer Learning

To assist radiologists in breast cancer classification in automated breast ultrasound (ABUS) imaging, we propose a computer-aided diagnosis based on a convolutional neural network (CNN) that classifies breast lesions as benign and malignant. The proposed CNN adopts a modified Inception-v3 architecture to provide efficient feature extraction in ABUS imaging. Because the ABUS images can be visualized in transverse and coronal views, the proposed CNN provides an efficient way to extract multiview features from both views. The proposed CNN was trained and evaluated on 316 breast lesions (135 malignant and 181 benign). An observer performance test was conducted to compare five human reviewers' diagnostic performance before and after referring to the predicting outcomes of the proposed CNN. Our method achieved an area under the curve (AUC) value of 0.9468 with five-folder cross-validation, for which the sensitivity and specificity were 0.886 and 0.876, respectively. Compared with conventional machine learning-based feature extraction schemes, particularly principal component analysis (PCA) and histogram of oriented gradients (HOG), our method achieved a significant improvement in classification performance. The proposed CNN achieved a >10% increased AUC value compared with PCA and HOG. During the observer performance test, the diagnostic results of all human reviewers had increased AUC values and sensitivities after referring to the classification results of the proposed CNN, and four of the five human reviewers’ AUCs were significantly improved. The proposed CNN employing a multiview strategy showed promise for the diagnosis of breast cancer, and could be used as a second reviewer for increasing diagnostic reliability.     

Stepwise logistic regression analysis of tumor contour features for breast ultrasound diagnosis.

To assist the ultrasound (US) differential diagnosis of solid breast tumors by using stepwise logistic regression (SLR) analysis of tumor contour features, we retrospectively reviewed 111 medical records of digitized US images of breast pathologies. They were pathologically proved benign breast tumors from 40 patients (i.e., 40 fibroadenomas) and malignant breast tumors from 71 patients (i.e., 71 infiltrative ductal carcinomas). Radiologists, before analysis by the computer-aided diagnosis (CAD) system, segmented the tumors manually. The contour features were calculated by measuring the radial length of tumor boundaries. The features selection process was accomplished using a stepwise analysis procedure. Then, an SLR model with contour features was used to classify tumors as benign or malignant. In this experiment, cases were sampled with "leave-one-out" test methods to evaluate the SLR performance using a receiver operating characteristic (ROC) curve. The accuracy of our SLR model with contour features for classifying malignancies was 91.0% (101 of 111 tumors), the sensitivity was 97.2% (69 of 71), the specificity was 80.0% (32 of 40), the positive predictive value was 89.6% (69 of 77), and the negative predictive value was 94.1% (32 of 34). The CAD system using SLR can differentiate solid breast nodules with relatively high accuracy and its high negative predictive value could potentially help inexperienced operators to avoid misdiagnoses. Because the SLR model is trainable, it could be optimized if a larger set of tumor images were supplied.     

基于分形特征序列的乳腺X线图像分类方法

目的表征乳腺图像中肿块部分纹理特征,通过纹理分析实现乳腺图像中肿块部分与正常腺体部分的分类。方法应用分形特征值表征乳腺图像纹理特征,利用多级分形特征提取法将乳腺图像分解成一系列细节图像,提取出多个分形特征值;利用分类精度、ROC曲线及曲线下面积(AUC)进行特征选择构建分形特征序列,最后应用支持向量机(SVM)方法进行分类。结果对60幅图像的可疑病变区域进行分形特征序列提取分析,SVM交叉验证分类精度达84.50%。结论基于分形维数的乳腺图像分类方法不仅能对肿块与正常腺体进行图像分类,还可有效表征乳腺图像的纹理信息,有助于提高乳腺肿块诊断的准确率。     

3-D breast ultrasound segmentation using active contour model

In this study, we made use of the discrete active contour model to overcome the natural properties of ultrasound (US) images, speckle, noise and tissue-related textures, to segment the breast tumors precisely. Determination of the real tumor boundary with the snake-deformation process requires an initial contour estimate. However, the manual way to sketch an initial contour is very time-consuming. Thus, we propose an automatic initial contour-finding method that not only maintains the tumor shape, but also is close to the tumor boundary and inside the tumor. During the deformation process, to prevent the snake trapping into the false position caused by tissue-related texture or speckle, we added the edge information as an image feature to define the external force. In addition, because the 3-D volume of a tumor is essentially constructed by a sequence of 2-D images, our method for finding boundaries of a tumor can be extended to 3-D cases. By precisely counting the volume of the 3-D images, we can get the volume of tumor. Finally, we will show that the proposed techniques have rather good performance and lead to a satisfactory result in comparison with the estimated volume and physician's estimate.     

计算机辅助诊断技术可提高肿块最大径≤10 mm早期乳腺癌的超声诊断效能

目的探讨计算机辅助诊断在早期乳腺癌诊断中的价值。方法对120枚病理证实的最大径≤20 mm乳腺肿块(乳腺癌结节50枚，良性结节70枚)超声图像进行回顾性分析，根据肿块大小分为最大径≤10 mm组(56枚)、最大径11~20 mm组(64枚)两组，由2名超声医师参照BI-RADS-US分类予以诊断，结合计算机超声辅助诊断结果后再次诊断，以病理结果为金标准，对比分析计算机辅助诊断在超声诊断早期乳腺癌中的作用。结果乳腺肿块最大径≤10 mm组中，应用普通超声对早期乳腺癌的敏感性、特异性和准确性分别为62.5%、59.4%、60.7%，操作性曲线下面积(AUC)为0.61。结合计算机辅助诊断技术结果为79.2%、81.3%、80.4%；AUC为0.80。对于最大径11~20 mm组，常规超声的敏感性、特异性和准确性分别为69.2%、68.4%、68.8%，ROC曲线AUC为0.69。结合计算机辅助诊断结果为80.8%、81.6%、81.3%；AUC为0.81。计算机辅助诊断后两组不同大小乳腺肿块的敏感性、特异性、准确性及AUC均有提高，乳腺肿块最大径≤10 mm组的准确率及AUC提高尤为显著，差异有统计学意义(P < 0.05)。结论计算机辅助诊断技术有助于提高早期乳腺癌的超声诊断效能，尤其是辅助最大径≤10 mm的早期乳腺癌的诊断。      

自动乳腺超声诊断系统结合计算机辅助检测乳腺恶性肿瘤敏感度的影响因素分析

目的评估影响计算机辅助检测（CAD）识别自动乳腺超声诊断系统（ABUS）乳腺恶性肿瘤敏感度的因素。 方法收集自2016年1月至2017年2月于空军军医大学西京医院行ABUS检查并经外科手术或组织学活检病理证实的乳腺恶性肿瘤患者232例，共240个恶性病灶。所有病例均经CAD软件检测，统计CAD对病灶的总敏感度，并统计分析病灶组织学类型、最大径、距乳头距离、距皮肤距离及象限等因素与CAD敏感度之间的关系。以外科手术或组织学活检病理结果为诊断"金标准"，采用χ²检验分析病灶组织学类型、最大径、距乳头距离、距皮肤距离、象限、病灶边缘特征等因素与CAD敏感度的关系。 结果CAD对恶性病灶的总敏感度为85%（204/240），对不同病理学类型的敏感度分别为：浸润性导管癌89.0%（186/209）、导管原位癌53.9%（14/29）、黏液癌75.0%（3/4）、恶性叶状肿瘤100%（1/1），差异有统计学意义（χ²=18.836，P<0.001）。病灶最大径、距乳头距离、距皮肤距离及象限均与CAD敏感度之间比较，差异无统计学意义（P>0.05）。病灶距皮肤距离、病灶边缘特征与CAD对浸润性导管癌的敏感度之间比较，差异有统计学意义（P<0.05）。 结论CAD对恶性病灶的敏感度较高（85.0%），尤其是对浸润性导管癌的检出（89.0%），医师在借助CAD读图时，应注意是否有遗漏的导管原位癌、位置深或边缘模糊的浸润性导管癌。     

Segmentation of breast tumor in three-dimensional ultrasound images using three-dimensional discrete active contour model

In this paper, we apply the three-dimensional (3-D) active contour model to a 3-D ultrasonic data file for segmenting of the breast tumor. The 3-D ultrasonic file is composed of a series of two-dimensional (2-D) images. Most of traditional techniques of 2-D image segmentation will not use the information between adjacent images. To suit the property of the 3-D data, we introduce the concept of the 3-D stick, the 3-D morphologic process and the 3-D active contour model. The 3-D stick can get over the problem that the ultrasonic image is full of speckle noise and highlight the edge information in images. The 3-D morphologic process helps to determine the contour of the tumor and the resulting contour can be regarded as the initial contour of the active contour model. Finally, the 3-D active contour model will make the initial contour approach to the real contour of the tumor. However, there is emphasis on these 3-D techniques that they do not consist of a series of 2-D techniques. When they work, they will consider the horizontal, vertical and depth directions at the same time. The use of these 3-D techniques not only segments the 3-D shape but also obtains the volume of the tumor. The volume of the tumor calculated by the proposed method will be compared with the volume calculated by the VOCAL software with the physician's manually drawn shape and it shows that the performance of our method is satisfactory.     

Texture feature analysis for breast ultrasound image enhancement

Texture analysis of breast ultrasound B-scans has been widely applied to the segmentation and classification of breast tumors. We present a parametric imaging method based on the texture features to preserve tumor edges and retain the texture information simultaneously. Four texture-feature parameters--homogeneity, contrast, energy and variance--were evaluated using the gray-level co-occurrence matrix. The local texture-feature parameter was assigned as the new pixel located at the center of the sliding window at each position. This process yielded the texture-feature parametric image as the map of texture-feature values. The signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) were estimated to show the quality improvement of the images. The contours outlined from 11 experienced physicians and the gradient vector flow (GVF) snake algorithm segmentations were adopted to verify the edge enhancement of texture-feature parametric images. In addition, the Fisher's linear discriminant analysis (FLDA) and receiver-operating-characteristic (ROC) curve were used to test the performance of breast tumor classifications between texture-feature parametric images and B-scan images. The results show that the variance images have higher CNR and SNR estimates than those in the B-scan images. There was a high agreement between the physician's manual contours and the GVF snake automatic segmentations in the variance images, and the mean area overlap was over 93%. The area under the ROC curve from the B-scan images had 0.81 and 95% confidence interval of 0.72-0.88, and the texture-feature parametric images had 0.90 and 95% confidence interval of 0.84-0.96. These findings indicate that the texture-feature parametric imaging method can be not only useful for determining the location of the lesion boundary but also as a tool to improve the accuracy of breast tumor classifications.     

Whole-Breast Ultrasound for Breast Screening and Archiving

The incidence of breast cancer is increasing worldwide, reinforcing the importance of breast screening. Conventional hand-held ultrasound (HHUS) for breast screening is efficient and relatively easy to perform; however, it lacks systematic recording and localization. This study investigated an electromagnetic tracking-based whole-breast ultrasound (WBUS) system to facilitate the use of HHUS for breast screening. One-hundred nine breast masses were collected, and the detection of suspicious breast lesions was compared between the WBUS system, HHUS and a commercial automated breast ultrasound (ABUS) system. The positioning error between WBUS and ABUS (1.39 ± 0.68 cm) was significantly smaller than that between HHUS and ABUS (1.62 ± 0.91 cm, p = 0.014) and HHUS and WBUS (1.63 ± 0.9 cm, p = 0.024). WBUS is a practical clinical tool for breast screening that can be used instead of the often unavailable and costly ABUS.     

Breast tissue characterization using FARMA modeling of ultrasonic RF echo

A number of empirical and analytical studies demonstrated that the ultrasound RF echo reflected from tissue exhibits 1/f characteristics. In this paper, we propose to model 1/f characteristics of the ultrasonic RF echo by a novel parsimonious model, namely the fractional differencing auto regressive moving average (FARMA) process, and evaluated diagnostic value of model parameters for breast cancer malignancy differentiation. FARMA model captures the fractal and long term correlated nature of the backscattered speckle texture and facilitates robust efficient estimation of fractal parameters. In our study, in addition to the computer generated FARMA model parameters, we included patient age and radiologist's prebiopsy level of suspicion (LOS) as potential indicators of malignant and benign masses. We evaluated the performance of the proposed set of features using various classifiers and training methods using 120 in vivo breast images. Our study shows that the area under the receiver operating characteristics (ROC) curve of FARMA model parameters alone is superior to the area under the ROC curve of the radiologist's prebiopsy LOS. The area under the ROC curve of the three sets of features yields a value of 0.87, with a confidence interval of [0.85, 0.89], at a significance level of 0.05. Our results suggest that the proposed method of ultrasound RF echo model leads to parameters that can differentiate breast tumors with a relatively high precision. This set of RF echo features can be incorporated into a comprehensive computer-aided diagnostic system to aid physicians in breast cancer diagnosis.     

Use of the K-distribution for classification of breast masses

The K-distribution had been introduced as a valid model to represent the statistics of the envelope of the backscattered echo from phantom and tissue. This paper investigates the efficacy of the parameters of this statistical model; namely, the effective number and the effective cross-section, to characterize breast lesions as benign or malignant. Based on the normalized values of the effective number and the effective scattering cross-section, images containing benign and malignant masses were classified for a data set from 52 patients having breast masses/lesions. The receiver operating characteristic (ROC) curves were then obtained to test the classification based on these two parameters. The results indicate that the parameters of the K-distribution may be useful in classification of the breast lesions as benign and malignant.     

Integrated local binary pattern texture features for classification of breast tissue imaged by optical coherence microscopy

This paper proposes a texture analysis technique that can effectively classify different types of human breast tissue imaged by Optical Coherence Microscopy (OCM). OCM is an emerging imaging modality for rapid tissue screening and has the potential to provide high resolution microscopic images that approach those of histology. OCM images, acquired without tissue staining, however, pose unique challenges to image analysis and pattern classification. We examined multiple types of texture features and found Local Binary Pattern (LBP) features to perform better in classifying tissues imaged by OCM. In order to improve classification accuracy, we propose novel variants of LBP features, namely average LBP (ALBP) and block based LBP (BLBP). Compared with the classic LBP feature, ALBP and BLBP features provide an enhanced encoding of the texture structure in a local neighborhood by looking at intensity differences among neighboring pixels and among certain blocks of pixels in the neighborhood. Fourty-six freshly excised human breast tissue samples, including 27 benign (e.g. fibroadenoma, fibrocystic disease and usual ductal hyperplasia) and 19 breast carcinoma (e.g. invasive ductal carcinoma, ductal carcinoma in situ and lobular carcinoma in situ) were imaged with large field OCM with an imaging area of 10 × 10 mm² (10, 000 × 10, 000 pixels) for each sample. Corresponding H&E histology was obtained for each sample and used to provide ground truth diagnosis. 4310 small OCM image blocks (500 × 500 pixels) each paired with corresponding H&E histology was extracted from large-field OCM images and labeled with one of the five different classes: adipose tissue (n = 347), fibrous stroma (n = 2,065), breast lobules (n = 199), carcinomas (pooled from all sub-types, n = 1,127), and background (regions outside of the specimens, n = 572). Our experiments show that by integrating a selected set of LBP and the two new variant (ALBP and BLBP) features at multiple scales, the classification accuracy increased from 81.7% (using LBP features alone) to 93.8% using a neural network classifier. The integrated feature was also used to classify large-field OCM images for tumor detection. A receiver operating characteristic (ROC) curve was obtained with an area under the curve value of 0.959. A sensitivity level of 100% and specificity level of 85.2% was achieved to differentiate benign from malignant samples. Several other experiments also demonstrate the complementary nature of LBP and the two variants (ALBP and BLBP features) and the significance of integrating these texture features for classification. Using features from multiple scales and performing feature selection are also effective mechanisms to improve accuracy while maintaining computational efficiency.