Search

Segmentation of the pectoral muscle is one of the most fundamental steps in developing a Computer-Aided Diagnosis (CAD) System. Considering the presence of various artifacts and homogeneity among the breast region, the pectoral muscle segmentation is a relatively tricky task. This study proposes a modified U-Net structure for automatic pectoral muscle segmentation from mammogram images. The modified U-Net architecture is different from the traditional U-Net architecture in having densely connected convolution layers. The proposed architecture was trained, validated, and tested by utilizing 322 and 200 mammogram images from two publicly available datasets: Mammographic Image Analysis Society (mini-MIAS) and the INBreast. Mini-MIAS contains scanned filmed mammograms, whereas INBreast contains full-field digital (FFDM) mammograms. Unlike other segmentation techniques, the modified U-Net architecture proposed does not require any pre-processing method and performs equally well on both scanned filmed and FFDM mammograms. We tested the system by fusing both the above datasets to verify our claim. Our proposed architecture has obtained satisfactory results on the mixed dataset. The experimental results showed that the pectoral muscle segmentation approach received a dice coefficient of 96.38 ± 1.52 and 96.82 ± 2.23 on mini-MIAS and the INBreast databases, respectively. Moreover on mixing both the datasets and randomly testing 20% of the samples, the algorithm achieved a Dice coefficient of 95.82 ± 1.23. [ABSTRACT FROM AUTHOR]

Abstract Background The indulgence of artificial intelligence (AI) has been considered recently in the work up for the detection and diagnosis of breast cancer through algorithms that could supply diagnosis as the radiologist do. The algorithm learns from a supervised and continuous input of large and new data sets unlike the standard programming, which requires clear step-by-step instructions. The aim of this study is to assess the ability of AI scanned mammograms to aid the ultrasound in the discrimination between phyllodes tumors and fibro-adenomas. Results This is a retrospective analysis included 374 proven phyllodes tumors (PT) and fibro-adenomas (FA). Digital mammogram and breast ultrasound was performed for all the cases and each breast was given a “Breast Imaging Reporting and Data System” (BI-RADS) score. Included mammograms were scanned by AI with resultant a qualitative heatmap and a quantitative abnormality scoring of suspicion percentage. The study included 164 PT (43.9%) and 210 FA (56.1%). BI-RADS category 2 was assigned in 40.1%, category 3 in 38.2%, category 4 in 18.5% and category 5 in 3.2% with median value of the AI abnormality scoring of 23%, 44%, 65% and 90% respectively. Sensitivity and specificity of the conventional imaging were 59.2% and 75.8% respectively. The AI abnormality scoring of 49.5% upgraded the sensitivity to 89.6% and specificity to 94.8% in the ability to discriminate PT from FA masses. Conclusion Artificial intelligence-aided mammograms could be used as method of distinction between PT from FA detected on sono-mammogram. The color hue and the quantification of the abnormality scoring percentage could be used as a one setting method for specification and so guide clinicians in their decision of conservative management or the choice of the surgical procedure.

Breast cancer is the most common cancer with highest mortality risk among the female worldwide and breast mass is the most effective sign for cancer identification. Thus, accurate segmentation of breast mass is regarded as a key step to reduce the death rate. Traditional segmentation methods require prior knowledge and manually set parameters, while recent studies prefer to construct neural networks based on feature reuse. However, breast mass can display in different orientations and the spatial context is complex, which makes the segmentation remain a challenging task. For these concerns, we propose a Spatial Enhanced Rotation Aware Network (SERAN) for automatic breast mass segmentation. SERAN consists of two critical components: 1) a residual attention encoder with spatial enhancement mechanism for effective feature extraction, and 2) a decoder constructed by multi-stream rotation aware blocks for feature fusion and prediction refinement. To optimize SERAN better and avoid misclassification in background area, a regulation item named Inside-outside Loss (IOL) is used in training procedure. The experimental results tested on a representative subset of Digital Database for Screening Mammography (DDSM) dataset show that SERAN outperforms state-of-the-art methods among most adopted evaluation metrics.

Background: The indulgence of artificial intelligence (AI) has been considered recently in the work up for the detection and diagnosis of breast cancer through algorithms that could supply diagnosis as the radiologist do. The algorithm learns from a supervised and continuous input of large and new data sets unlike the standard programming, which requires clear step-by-step instructions. The aim of this study is to assess the ability of AI scanned mammograms to aid the ultrasound in the discrimination between phyllodes tumors and fibro-adenomas. Results: This is a retrospective analysis included 374 proven phyllodes tumors (PT) and fibro-adenomas (FA). Digital mammogram and breast ultrasound was performed for all the cases and each breast was given a "Breast Imaging Reporting and Data System" (BI-RADS) score. Included mammograms were scanned by AI with resultant a qualitative heatmap and a quantitative abnormality scoring of suspicion percentage. The study included 164 PT (43.9%) and 210 FA (56.1%). BI-RADS category 2 was assigned in 40.1%, category 3 in 38.2%, category 4 in 18.5% and category 5 in 3.2% with median value of the AI abnormality scoring of 23%, 44%, 65% and 90% respectively. Sensitivity and specificity of the conventional imaging were 59.2% and 75.8% respectively. The AI abnormality scoring of 49.5% upgraded the sensitivity to 89.6% and specificity to 94.8% in the ability to discriminate PT from FA masses. Conclusion: Artificial intelligence-aided mammograms could be used as method of distinction between PT from FA detected on sono-mammogram. The color hue and the quantification of the abnormality scoring percentage could be used as a one setting method for specification and so guide clinicians in their decision of conservative management or the choice of the surgical procedure. [ABSTRACT FROM AUTHOR]

Segmentation of the breast region and pectoral muscle are fundamental subsequent steps in the process of Computer-Aided Diagnosis (CAD) systems. Segmenting the breast region and pectoral muscle are considered a difficult task, particularly in mammogram images because of artefacts, homogeneity among the region of the breast and pectoral muscle, and low contrast along the region of breast boundary, the similarity between the texture of the Region of Interest (ROI), and the unwanted region and irregular ROI. This study aims to propose an improved threshold-based and trainable segmentation model to derive ROI. A hybrid segmentation approach for the boundary of the breast region and pectoral muscle in mammogram images was established based on thresholding and Machine Learning (ML) techniques. For breast boundary estimation, the region of the breast was highlighted by eliminating bands of the wavelet transform. The initial breast boundary was determined through a new thresholding technique. Morphological operations and masking were employed to correct the overestimated boundary by deleting small objects. In the medical imaging field, significant progress to develop effective and accurate ML methods for the segmentation process. In the literature, the imperative role of ML methods in enabling effective and more accurate segmentation method has been highlighted. In this study, an ML technique was built based on the Histogram of Oriented Gradient (HOG) feature with neural network classifiers to determine the region of pectoral muscle and ROI. The proposed segmentation approach was tested by utilizing 322, 200, 100 mammogram images from mammographic image analysis society (mini-MIAS), INbreast, Breast Cancer Digital Repository (BCDR) databases, respectively. The experimental results were compared with manual segmentation based on different texture features. Moreover, evaluation and comparison for the boundary of the breast region and pectoral muscle segmentation have been done separately. The experimental results showed that the boundary of the breast region and the pectoral muscle segmentation approach obtained an accuracy of 98.13% and 98.41% (mini-MIAS), 100%, and 98.01% (INbreast), and 99.8% and 99.5% (BCDR), respectively. On average, the proposed study achieved 99.31% accuracy for the boundary of breast region segmentation and 98.64% accuracy for pectoral muscle segmentation. The overall ROI performance of the proposed method showed improving accuracy after improving the threshold technique for background segmentation and building an ML technique for pectoral muscle segmentation. More so, this article also included the ground-truth as an evaluation of comprehensive similarity. In the clinic, this analysis may be provided as a valuable support for breast cancer identification.

Abstract Background The percentage of mammographic dense tissue (PD) defined by pixel value threshold is a well-established risk factor for breast cancer. Recently there has been some evidence to suggest that an increased threshold based on visual assessment could improve risk prediction. It is unknown, however, whether this also applies to volumetric density using digital raw mammograms. Method Two case-control studies nested within a screening cohort (ages of participants 46–73 years) from Manchester UK were used. In the first study (317 cases and 947 controls) cases were detected at the first screen; whereas in the second study (318 cases and 935 controls), cases were diagnosed after the initial mammogram. Volpara software was used to estimate dense tissue height at each pixel point, and from these, volumetric and area-based PD were computed at a range of thresholds. Volumetric and area-based PDs were evaluated using conditional logistic regression, and their predictive ability was assessed using the Akaike information criterion (AIC) and matched concordance index (mC). Results The best performing volumetric PD was based on a threshold of 5 mm of dense tissue height (which we refer to as VPD5), and the best areal PD was at a threshold level of 6 mm (which we refer to as APD6), using pooled data and in both studies separately. VPD5 showed a modest improvement in prediction performance compared to the original volumetric PD by Volpara with ΔAIC = 5.90 for the pooled data. APD6, on the other hand, shows much stronger evidence for better prediction performance, with ΔAIC = 14.52 for the pooled data, and mC increased slightly from 0.567 to 0.577. Conclusion These results suggest that imposing a 5 mm threshold on dense tissue height for volumetric PD could result in better prediction of cancer risk. There is stronger evidence that area-based density with a 6 mm threshold gives better prediction than the original volumetric density metric.

Abstract Background The percentage of mammographic dense tissue (PD) is an important risk factor for breast cancer, and there is some evidence that texture features may further improve predictive ability. However, relatively little work has assessed or validated textural feature algorithms using raw full field digital mammograms (FFDM). Method A case-control study nested within a screening cohort (age 46–73 years) from Manchester UK was used to develop a texture feature risk score (264 cases diagnosed at the same time as mammogram of the contralateral breast, 787 controls) using the least absolute shrinkage and selection operator (LASSO) method for 112 features, and validated in a second case-control study from the same cohort but with cases diagnosed after the index mammogram (317 cases, 931 controls). Predictive ability was assessed using deviance and matched concordance index (mC). The ability to improve risk estimation beyond percent volumetric density (Volpara) was evaluated using conditional logistic regression. Results The strongest features identified in the training set were “sum average” based on the grey-level co-occurrence matrix at low image resolutions (original resolution 10.628 pixels per mm; downsized by factors of 16, 32 and 64), which had a better deviance and mC than volumetric PD. In the validation study, the risk score combining the three sum average features achieved a better deviance than volumetric PD (Δχ2 = 10.55 or 6.95 if logarithm PD) and a similar mC to volumetric PD (0.58 and 0.57, respectively). The risk score added independent information to volumetric PD (Δχ2 = 14.38, p = 0.0008). Conclusion Textural features based on digital mammograms improve risk assessment beyond volumetric percentage density. The features and risk score developed need further investigation in other settings.

This paper presents a machine learning based approach for the discrimination of malignant and benign microcalcification (MC) clusters in digital mammograms. A series of morphological operations was carried out to facilitate the feature extraction from segmented microcalcification. A combination of morphological, texture, and distribution features from individual MC components and MC clusters were extracted and a correlation-based feature selection technique was used. The clinical relevance of the selected features is discussed. The proposed method was evaluated using three different databases: Optimam Mammography Image Database (OMI-DB), Digital Database for Screening Mammography (DDSM), and Mammographic Image Analysis Society (MIAS) database. The best classification accuracy (95.00 = 0.57%) was achieved for OPTIMAM using a stack generalization classifier with 10-fold cross validation obtaining an Az value equal to 0.97 = 0.01. [ABSTRACT FROM AUTHOR]

Featured Application: The proposed method can be used for detecting suspected incipient breast lesions that may cause breast cancer. The image processing of mammograms is very important for the early detection of breast pathologies, including cancer. This paper proposes a new technique based on directional fractal filtering for detecting microcalcification clusters or irregularly shaped microcalcifications. The proposed algorithm has two parts: a preprocessing step for detecting and locating microcalcification; and a second zooming, enhancement, and segmentation step. Detection is performed by image convolution using a set of masks with interesting fractal properties. Combined with other simple mathematical operations, remarkable contrast enhancement and segmentation are produced. The final result permits the clear delineation of the shape of individual microcalcifications. A comparison is made with other microcalcification enhancement techniques described in the literature. [ABSTRACT FROM AUTHOR]

Early detection is a key step for effective treatment of breast cancer and computer-aided diagnosis (CAD) is the most common tool used by the medical research community to detect early breast cancer development. Automated and accurate classification of mammogram images is an important criterion for the analysis and interpretation of these images and many methods have been proposed in this direction. In this paper, an improved CAD model is developed to classify the digital mammograms into normal and abnormal, and further, benign and malignant. The proposed model constitutes four different phases, namely, region of interest (ROI) generation, feature extraction, feature reduction, and classification. The proposed model first employs discrete Tchebichef transform (DTT) to extract the features from the ROIs. Subsequently, a technique based on a combination of principal component analysis (PCA) and linear discriminant analysis (LDA) is employed to reduce the dimensions of the feature vector. Next, the reduced features are sent to an extreme learning machine (ELM) for the classification. Here, to obtain a better generalization performance, the hidden node parameters of ELM are optimized through an improved grey wolf optimization-based ELM (IGWO-ELM). To validate the proposed CAD system, different performance metrics such as accuracy, sensitivity, specificity, and area under curve (AUC) are measured using k-fold stratified cross-validation (SCV). Moreover, to eliminate the issue of randomness, 10 independent runs are carried out on SCV. From a detailed analysis of the results, it is observed that the proposed model yields an average accuracy of 100% for MIAS dataset in both normal vs. abnormal, and benign vs. malignant cases. Further, the accuracy achieved for DDSM dataset is 99.50%, and 98.50% for normal vs. abnormal, and benign vs. malignant cases, respectively. The computation time taken by the proposed CAD model for MIAS and DDSM are 1.131 secs and 3.063 secs, respectively. The experimental analysis justifies the effectiveness of the proposed CAD model and as a result, this model can be considered as an effective tool to help the radiologists for better diagnosis. [ABSTRACT FROM AUTHOR]

Breast cancer is one of the most common malignant diseases among women. Mammography is at present one of the available methods for early detection of abnormalities, which is related to breast cancer different lesions to show characteristics such as masses, which can be detected through this technique. The images are divided according to the Mini-MIAS database. In order to classify the cluster as malignant or benign, the 2nd order co-occurrence statistical properties of the image such as contrast, correlation, homogeneity, and energy were adopted. The gray-scale convergent matrices (GLCM) are used with a suggested feature of gray level density matrices (GLDM) to identify abnormal tissues (malignant) and natural tissue (benign). The proposed method of analysis is tested on several images (taken from a UK organization interested in understanding breast X-ray images) from the database of digital mammograms and outstanding results were obtained for breast cancer detection. The proposed multilayered design performance significantly improved the diagnosis of breast cancer by more than 95% and 95.8% sensitivity, 95.5% specificity for all datasets, 91.1% with 100% sensitivity and 84% specificity for 70% training data and 30% testing data. [ABSTRACT FROM AUTHOR]

Breast cancer is the most life-threatening disease among women. The best way to decrease the mortality is early detection of cancer from digital mammogram. The diagnosis can be successful if the pre-processing and segmentation of the digital mammograms identifies the suspicious area correctly. In this paper, the Butterworth bandpass filter along with fusion of FCM and K-means clustering followed by morphological operations is used for the segmentation of calcification areas from the mammogram images. [ABSTRACT FROM AUTHOR]

Breast cancer is one of the life-threatening cancers occurring in women. In recent years, from the surveys provided by various medical organizations, it has become clear that the mortality rate of females is increasing owing to the late detection of breast cancer. Therefore, an automated algorithm is needed to identify the early occurrence of microcalcification, which would assist radiologists and physicians in reducing the false predictions via image processing techniques. In this work, we propose a new algorithm to detect the pattern of a microcalcification by calculating its physical characteristics. The considered physical characteristics are the reflection coefficient and mass density of the binned digital mammogram image. The calculation of physical characteristics doubly confirms the presence of malignant microcalcification. Subsequently, by interpolating the physical characteristics via thresholding and mapping techniques, a three-dimensional (3D) projection of the region of interest (RoI) is obtained in terms of the distance in millimeter. The size of a microcalcification is determined using this 3D-projected view. This algorithm is verified with 100 abnormal mammogram images showing microcalcification and 10 normal mammogram images. In addition to the size calculation, the proposed algorithm acts as a good classifier that is used to classify the considered input image as normal or abnormal with the help of only two physical characteristics. This proposed algorithm exhibits a classification accuracy of 99%. [ABSTRACT FROM AUTHOR]

Background and objective Early detection is the important key to reduce breast cancer mortality rate. Detecting the mammographic abnormality as a subtle sign of breast cancer is essential for the proper diagnosis and treatment. The aim of this preliminary study is to develop algorithms which detect suspicious lesions and characterize them to reduce the diagnostic errors regarding false positives and false negatives. Methods The proposed hybrid mechanism detects suspicious lesions automatically using connected component labeling and adaptive fuzzy region growing algorithm. A novel neighboring pixel selection algorithm reduces the computational complexity of the seeded region growing algorithm used to finalize lesion contours. These lesions are characterized using radiomic features and then classified as benign mass or malignant tumor using k -NN and SVM classifiers. Two datasets of 460 full field digital mammograms (FFDM) utilized in this clinical study consists of 210 images with malignant tumors, 30 with benign masses and 220 normal breast images that are validated by radiologists expert in mammography. Results The qualitative assessment of segmentation results by the expert radiologists shows 91.67% sensitivity and 58.33% specificity. The effects of seven geometric and 48 textural features on classification accuracy, false positives per image (FPsI), sensitivity and specificity are studied separately and together. The features together achieved the sensitivity of 84.44% and 85.56%, specificity of 91.11% and 91.67% with FPsI of 0.54 and 0.55 using k -NN and SVM classifiers respectively on local dataset. Conclusions The overall breast cancer detection performance of proposed scheme after combining geometric and textural features with both classifiers is improved in terms of sensitivity, specificity, and FPsI. [ABSTRACT FROM AUTHOR]