INbreast Database Research Articles

Residual shallow convolutional neural network to classify microcalcifications clusters in digital mammograms

We introduce a residual shallow Convolutional Neural Network (CNN) designed for classifying the presence or absence of Microcalcification Clusters (MCCs) in digital mammograms. MCCs are the most indirect signs of breast cancer in the early stages, and when they are detected on time, the women have a 5-year survival rate of 99%. However, MCCs are difficult to detect. The proposed architecture comprises three Convolutional Layers (CLs) with 6, 16, and 16 filters, each of size 9 × 9 at full scale, including a residual block. Subsequently, a Global Max Pooling operator transforms the three-dimensional input tensor into a 16-valued vector, avoiding flattening and dense layers. A sigmoid function at the output layer allows binary classification. The CNN was evaluated using the public INbreast database. We augmented the data through four geometric transformations to enhance the CNN accuracy. We show that the residual shallow CNN achieves an accuracy of 99.71% with 29,053 parameters, over 2,333 times smaller than the MobileNetV2, which attains 99.8% accuracy but with 67,797,505 parameters. Additionally, our proposed CNN was subjected to a systematic ablation study, based on the CLs energy, to observe the importance of the filters and reduce the number of parameters after training.

Multi-class Breast Cancer Classification Using CNN Features Hybridization

Breast cancer has become the leading cause of cancer mortality among women worldwide. The timely diagnosis of such cancer is always in demand among researchers. This research pours light on improving the design of computer-aided detection (CAD) for earlier breast cancer classification. Meanwhile, the design of CAD tools using deep learning is becoming popular and robust in biomedical classification systems. However, deep learning gives inadequate performance when used for multilabel classification problems, especially if the dataset has an uneven distribution of output targets. And this problem is prevalent in publicly available breast cancer datasets. To overcome this, the paper integrates the learning and discrimination ability of multiple convolution neural networks such as VGG16, VGG19, ResNet50, and DenseNet121 architectures for breast cancer classification. Accordingly, the approach of fusion of hybrid deep features (FHDF) is proposed to capture more potential information and attain improved classification performance. This way, the research utilizes digital mammogram images for earlier breast tumor detection. The proposed approach is evaluated on three public breast cancer datasets: mammographic image analysis society (MIAS), curated breast imaging subset of digital database for screening mammography (CBIS-DDSM), and INbreast databases. The attained results are then compared with base convolutional neural networks (CNN) architectures and the late fusion approach. For MIAS, CBIS-DDSM, and INbreast datasets, the proposed FHDF approach provides maximum performance of 98.706%, 97.734%, and 98.834% of accuracy in classifying three classes of breast cancer severities.

Real time mass classification for mammographic images: a Driven CADx scheme

Computer-Aided Diagnosis (CADx) schemes have been proposed to serve as a supplementary image analysis tool in mammography. Experienced radiologists tend to be more assertive to such schemes in assisting their interpretation rather than solely relying on their ability to detect suspicious signals. This study focuses on a simplified version of a previously developed mammography CADx scheme, which was initially designed for digitized film, but is now specifically aimed at classifying breast nodules marked as regions of interest on digital images. This “driven” CADx scheme provides prompt indications regarding whether the selected nodule is deemed normal or suspicious. Its performance was evaluated through tests conducted on different mammograms sets – one with large number of images selected from DDSM database for training, testing and validation of classification parameters, and other comprising direct digital images from InBreast database. Remarkably, similar rates were observed for sensitivity, specificity and accuracy across these two sets (83%, 67% and 72%, respectively). The classification attributes were associated to contour, density and texture. Furthermore, a third test was conducted involving radiologists analyzing digital mammograms obtained from a specific full field digital mammography (FFDM) unit. Results showed that the Driven CADx scheme positively influenced the final diagnoses made by 3 radiologists, consistently increasing accuracy rates. This promising result allows establishing this software as a valuable tool for radiologists in the analysis of masses in digital mammography. The scheme can be implemented on any operating system, or even accessed online.

Improvement of Segmentation Efficiency in Mammogram Images Using Dual-ROI Method

Mammogram segmentation utilizing multi-region of intrigue is a standout amongst the most rising exploration territory in the medical image analysis. The steps engaged with the research are grouped into two kinds: 1) segmentation of mammogram images and 2) extraction of texture features from mammogram images. To overcome these difficulties, a compelling technique is proposed in this paper that comprises of three phases. In the principal arrangement, mammogram images from INbreast database are selected and improved utilizing Laplacian filtering. At that point, the pre-processed mammogram images are utilized for segmentation utilizing modified adaptively regularized kernel-based fuzzy C means (M-ARKFCM). After segmentation, statistical texture FE is connected for recognizing the patterns of cancer and non-cancer regions in mammogram images. Finally, the experimental outcome demonstrated that the proposed approach enhanced the segmentation efficiency by methods of statistical parameters contrasted with the existing operating procedures.

Classifying presence or absence of calcifications on mammography using generative contribution mapping.

The purpose of this study was to verify the efficacy of generative contribution mapping (GCM), an explainable deep learning model for images, in classifying the presence or absence of calcifications on mammography. The learning dataset consisted of 303 full-field digital mammography (FFDM) images labeled with microcalcifications obtained from the public INbreast database without extremely dense images. FFDM images were divided into calcification and non-calcification patch images using a sliding window method with 25% overlap. The patch images of the mediolateral oblique (MLO) and craniocaudal (CC) views were divided into a training set of 70%, a validation set of 10%, and a testing set of 20%. The classification performance of GCM classifiers was evaluated and compared with that of EfficientNet classifiers. Visualization maps of GCM highlighted regions of interest more clearly than EfficientNet's gradient-weighted class activation maps. The results showed that GCM classifiers yielded an accuracy of 0.92 (CC), 0.91 (MLO), and an area under the receiver operating characteristic curve of 0.92 (CC), 0.94 (MLO). In conclusion, GCM could accurately classify the presence or absence of calcifications on mammograms and explain intuitively reasonable grounds for their classification with visualization maps highlighting regions of interest.

Classification of Mammographic ROI for Microcalcification Detection Using Multifractal Approach

Microcalcifications (MCs) are the main signs of precancerous cells. The development of aided-system for their detection has become a challenge for researchers in this field. In this paper, we propose a system for MCs detection based on the multifractal approach that classifies mammographic ROIs into normal (healthy) or abnormal ROIs containing MCs. The proposed method is divided into four main steps: a mammogram pre-processing step based on breast selection, breast density reduction using haze removal algorithm and contrast enhancement using multifractal measures. The second step consists of extracting the normal and abnormal ROIs and calculating the multifractal spectrum of each ROI. The next step represents the extraction of the multifractal features from the multifractal spectrum and the GLCM characteristics of each ROI. The last step is the classification of ROIs where three classifiers are tested (KNN, DT, and SVM). The system is evaluated on images from the INbreast database (308 images) with a total of 2688 extracted ROIs (1344 normal, 1344 with MC) from different BI-RADS classes. In this study, the SVM classifier gave the best classification results with a sensitivity, specificity, and precision of 98.66%, 97.77%, and 98.20% respectively. These results are very satisfactory and remarkable compared to the literature.

Multidirectional Gabor Filter-Based Approach for Pectoral Muscle Boundary Detection

Accurate pectoral muscle boundary (PMB) detection is a crucial stage in any computer-aided detection (CAD) system, used for automatic detection of various possible signs of malignancy in a mammographic image. The presence of high-density glandular tissues overlapping PMB, pectoral muscle of small size and low density, presence of pectoralis minors, etc., poses a significant challenge for researchers working on automatic PMB detection problem. In mammograms with pectoral muscle superimposed by high-density glandular tissues, especially near the lower portion of the pectoral muscle, PMB forms fuzzy textural edges with surrounding mammary tissues regions. The performance of any intensity-based approach in detecting such fuzzy textural edges is poor. Here, we present a multidirectional Gabor filter (MDGF)-based approach for PMB detection. A set of three high-frequency bandpass Gabor filters is designed, which covers all the possible orientations of PMB present in a preprocessed mediolateral oblique (MLO) view mammogram. These filters are used to extract high- and mid-frequency range edge information corresponding to strong as well as weak fuzzy textural edges of PMB. The information obtained from magnitude and phase response along with a boundary search and merge algorithm (designed to connect broken PMB edges) is used to detect complete PMB with high accuracy. With PMB detection accuracy of 95.28% for MIAS, 96.50% for CBIS-DDSM, and 96.95% for INbreast database mammograms, the proposed method outperforms many state-of-the-art methods, thus show its suitability for a CAD system.

ASU-Net: U-shape adaptive scale network for mass segmentation in mammograms

U-Net is a commonly used deep learning model for mammogram segmentation. Despite outstanding overall performance in segmenting, U-Net still faces from two aspects of challenges: (1) the skip-connections in U-Net have limitations, which may not be able to effectively extract multi-scale features for breast masses with diverse shapes and sizes. (2) U-Net only merges low-level spatial information and high-level semantic information through concatenating, which neglects interdependencies between channels. To address these two problems, we propose the U-shape adaptive scale network (ASU-Net), which contains two modules: adaptive scale module (ASM) and feature refinement module (FRM). In each level of skip-connections, ASM is used to adaptively adjust the receptive fields according to the different scales of the mass, which makes the network adaptively capture multi-scale features. Besides, FRM is employed to allows the decoder to capture channel-wise dependencies, which make the network can selectively emphasize the feature representation of useful channels. Two commonly used mammogram databases including the DDSM-BCRP database and the INbreast database are used to evaluate the segmentation performance of ASU-Net. Finally, ASU-Net obtains the Dice Index (DI) of 91.41% and 93.55% in the DDSM-BCRP database and the INbreast database, respectively.

Classification of Breast Cancer in Mammograms with Deep Learning Adding a Fifth Class

Breast cancer is one of the diseases of most profound concern, with the most prevalence worldwide, where early detections and diagnoses play the leading role against this disease achieved through imaging techniques such as mammography. Radiologists tend to have a high false positive rate for mammography diagnoses and an accuracy of around 82%. Currently, deep learning (DL) techniques have shown promising results in the early detection of breast cancer by generating computer-aided diagnosis (CAD) systems implementing convolutional neural networks (CNNs). This work focuses on applying, evaluating, and comparing the architectures: AlexNet, GoogLeNet, Resnet50, and Vgg19 to classify breast lesions after using transfer learning with fine-tuning and training the CNN with regions extracted from the MIAS and INbreast databases. We analyzed 14 classifiers, involving 4 classes as several researches have done it before, corresponding to benign and malignant microcalcifications and masses, and as our main contribution, we also added a 5th class for the normal tissue of the mammary parenchyma increasing the correct detection; in order to evaluate the architectures with a statistical analysis based on the received operational characteristics (ROC), the area under the curve (AUC), F1 Score, accuracy, precision, sensitivity, and specificity. We generate the best results with the CNN GoogLeNet trained with five classes on a balanced database with an AUC of 99.29%, F1 Score of 91.92%, the accuracy of 91.92%, precision of 92.15%, sensitivity of 91.70%, and specificity of 97.66%, concluding that GoogLeNet is optimal as a classifier in a CAD system to deal with breast cancer.

Automatic breast mass detection in mammograms using density of wavelet coefficients and a patch-based CNN.

For the purpose of accurate and efficient mass detection in full-field digital mammograms, weproposeamethodfor automated mass detection that consists oftwostages: suspicious region localization and false-positive (FP) reduction, by classifying these regions into mass and non-mass regions (normal tissues). In the first stage, the density of the wavelet coefficients based on Quincunx Lifting Scheme (DWC-QLS) is used to find suspicious regions (regions of interest, ROIs) in full mammograms. In the second stage, a patch-based CNN classifier is developed as an FP reduction to classify the suspicious regions. The main aim of this stage is to reduce the false-positive suspicious regions while keeping the true-positive suspicious regions. To further improve the performance of the FP reduction, the effectiveness of different transfer learning strategies is further explored and the best fine-tuning strategy in training InceptionV3 model is determined experimentally. The experimental results show that the proposed method can achieve an overall performance of 0.98 TPR @1.43 FPI on the INbreast database. In addition, the suggested segmentation method detects the mass location with 100% sensitivity and average of 5.4 false positives per image. Based on the obtained results,the introduced method was able to successfully detectand classify suspicious regions in digital mammograms and provide better TPR and FPI results in comparison with other state-of-the-art method.

RETRACTED ARTICLE: Cancer detection using deep learning techniques

Breast cancer has become the most common form of cancer in world recently having overtaken cervical cancer in urban cities. Immense research has been carried out on breast cancer and several automated machines for detection have been formed, however, they are far from perfection and medical assessments need more reliable services. Computer Assisted Diagnostics programs have been developed over the past 2 decades to help radiologists interpret mammogram screening. Deep convolutional neural networks (CNN), which have surpassed human output since 2012, have been an immense success in image recognition. Deep CNNs will revolutionize the analysis of medical images. We propose a method for breast cancer detection based on Faster R-CNN, the most common frameworks for object detection. In a non-human interference mammogram, the device detects and categorizes malignant or benign lesions. The method proposed sets the current status of the INbreast database public classification scheme, AUC = 0.95. In the digital mammography challenge DREAM with AUC = 0.85, the method mentioned here was second. When the device is used as a sensor, the accuracy of the INbreast data set is extremely low with very false positive image points.

Multi-scale attention-based convolutional neural network for classification of breast masses in mammograms.

Breast cancer is the cancer with the highest incidence in women, and early detection can effectively improve the survival rate of patients. Mammography is an important method for physicians to screening breast cancer, but the diagnosis of mammograms by physicians depends largely on clinical practice experience. Studies have shown that using computer-aided diagnosis techniques can help doctors diagnose breast cancer. In this paper, the method of convolutional neural network is mainly used to classify benign and malignant breast masses in the mammograms. First, we use multi-scale residual networks and densely connected networks as backbone networks to extract the features of global image patches and local image patches. Second, we use the attention module named convolutional block attention module (CBAM) to improve the two feature extraction networks to enhance the network's feature expression ability. Finally, we fuse the features of multi-scale image patches to achieve the classification of benign and malignant breast masses. In the digital database for screening mammography (DDSM) database, the accuracy, sensitivity, AUC value and corresponding standard deviation of our method are 0.9626±0.0110, 0.9719±0.0126, and 0.9576±0.0064, respectively. Compared with the commonly used ResNet (AUC=0.8823±0.0112) and DenseNet (AUC=0.9141±0.0085), the performance of our method has improved. In addition, we also used the INbreast database to train and validate the proposed method. The accuracy, sensitivity, AUC and corresponding standard deviations are 0.9554±0.0296, 0.9605±0.0228, and 0.9468±0.0085, respectively. Compared with the previous work, our proposed method uses multi-scale image features, has better classification performance in breast mass patches classification tasks, and can effectively assist physicians in breast cancer diagnosis.

MGBN: Convolutional neural networks for automated benign and malignant breast masses classification

Automated benign and malignant breast masses classification is a crucial yet challenging topic. Recently, many studies based on convolutional neural network (CNN) are presented to address this task, but most of these CNN-based methods neglect the effective global contextual information. Moreover, their methods do not further analyze the reliability and interpretability of CNN models, which does not correspond to the clinical diagnosis. In this work, we firstly propose a novel multi-level global-guided branch-attention network (MGBN) for mass classification, which aims to fully leverage the multi-level global contextual information to refine the feature representation. Specifically, the MGBN includes a stem module and a branch module. The former extracts the local information through standard local convolutional operations of ResNet-50. The latter embeds the global contextual information and establishes the relationships of different feature levels via global pooling and Multi-layer Perceptron (MLP). The final prediction is computed by local information and global information together. Then, we discuss the reliability and interpretability of our mass classification network by visualizing the coarse localization map through Gradient-weighted Class Activation Mapping (Grad-CAM), which is important in clinical diagnosis. Finally, our proposed MGBN is greatly demonstrated on two public mammographic mass classification databases including the DDSM and INbreast databases, resulting in AUC of 0.8375 and 0.9311, respectively.

Implementing Multilabeling, ADASYN, and ReliefF Techniques for Classification of Breast Cancer Diagnostic through Machine Learning: Efficient Computer-Aided Diagnostic System.

Multilabel recognition of morphological images and detection of cancerous areas are difficult to locate in the scenario of the image redundancy and less resolution. Cancerous tissues are incredibly tiny in various scenarios. Therefore, for automatic classification, the characteristics of cancer patches in the X-ray image are of critical importance. Due to the slight variation between the textures, using just one feature or using a few features contributes to inaccurate classification outcomes. The present study focuses on five different algorithms for extracting features that can extract further different features. The algorithms are GLCM, LBGLCM, LBP, GLRLM, and SFTA from 8 image groups, and then, the extracted feature spaces are combined. The dataset used for classification is most probably imbalanced. Additionally, another focal point is to eradicate the unbalanced data problem by creating more samples using the ADASYN algorithm so that the error rate is minimized and the accuracy is increased. By using the ReliefF algorithm, it skips less contributing features that relieve the burden on the process. Finally, the feedforward neural network is used for the classification of data. The proposed method showed 99.5% micro, 99.5% macro, 0.5% misclassification, 99.5% recall rats, specificity 99.4%, precision 99.5%, and accuracy 99.5%, showing its robustness in these results. To assess the feasibility of the new system, the INbreast database was used.

Breast Cancer Lesion Detection and Classification in mammograms using Deep Neural

A method to automatically detect and classify a lesion into either malignant or non-malignant is presented in this work. The dataset used is obtained from INbreast database and are in format of full-field digital mammography (FFDM). Some of the key challenges in detecting cancerous lesion in mammography are the low contrast between cancerous lesion and its surrounding tissues, apparent contrast similarities between lesion and pectoral muscle, presence of calcifications that may disrupt the detection process, and some level of morphological similarities between the lesion and some normal tissues. The work here consists of two main parts. The first part is the image processing section that aims to sample the lesion with intended lesion-to-surrounding ratio (0.4-0.6) and to avoid sampling from unintended regions such as pectoral muscle. Another key challenge is that the database is relatively small while machine learning requires a relatively large dataset. To improve size of samples, eighty fixed-size images (250 pixels x 250 pixels) are randomly cropped out of each of the previously processed image. The second part is to build the machine learning application based on deep neural network framework to classify samples into two classes, malignant and non-malignant. In present work we apply two different frameworks known as Plain and Residual Net (ResNet). Our calculations show that both models can detect a single lesion with more than 90% accuracy and area under ROC curve >0.94.

Breast Lesions Detection and Classification via YOLO-Based Fusion Models

With recent breakthroughs in artificial intelligence, the use of deep learning models achieved remarkable advances in computer vision, ecommerce, cybersecurity, and healthcare. Particularly, numerous applications provided efficient solutions to assist radiologists for medical imaging analysis. For instance, automatic lesion detection and classification in mammograms is still considered a crucial task that requires more accurate diagnosis and precise analysis of abnormal lesions. In this paper, we propose an end-to-end system, which is based on You-Only-Look-Once (YOLO) model, to simultaneously localize and classify suspicious breast lesions from entire mammograms. The proposed system first preprocesses the raw images, then recognizes abnormal regions as breast lesions and determines their pathology classification as either mass or calcification. We evaluated the model on two publicly available datasets, with 2907 mammograms from the Curated Breast Imaging Subset of Digital Database for Screening Mammography (CBIS-DDSM) and 235 mammograms from INbreast database. We also used a privately collected dataset with 487 mammograms. Furthermore, we suggested a fusion models approach to report more precise detection and accurate classification. Our best results reached a detection accuracy rate of 95.7%, 98.1% and 98% for mass lesions and 74.4%, 71.8% and 73.2% for calcification lesions, respectively on CBIS-DDSM, INbreast and the private dataset.

Multi-class classification of mammograms with hesitancy based Hanman transform classifier on pervasive information set texture features

With an aim to enhance the survival rate of women from breast cancer, this paper extends the Gabor, wavelet, and structure-based information set features into the pervasive information set features for more accurate breast cancer diagnosis from mammograms. The notion of a pervasive information set arises from expanding the information set to incorporate the intuitionistic fuzzy set. The effectiveness of the proposed features is evaluated on an annotated private dataset acquired from Superspeciality Cancer Hospital, New Delhi, and the public datasets comprising Digital Database for Screening Mammography, INbreast database, and the Mammographic Image Analysis Society database to validate their efficacy for multi-class categorization of mammograms employing the Hesitancy-based Hanman transform classifier. This classifier not only embodies the uncertainties in the errors between the training and test features but also typifies the deficiencies in the modelling of the membership function. The results of the Analysis of variance test confirm that the proposed features are statistically relevant and the experimental outcomes verified by expert radiologists validate the clinical significance of the proposed work. The proposed methods inhibiting the “Non-data hungry” aspect give 100% accuracy for multi-class classification on both public and private datasets which are higher than those obtained by the state of art methods available in the literature. These results will aid the radiologists in the early diagnosis of breast cancer.

Density map and fuzzy classification for breast density by using BI-RADS

Mammographic density (MD) is conformed by a different percentage of stromal, epithelial, and adipose tissue within the breast. One of the most critical findings in mammographic patterns for establishing a diagnosis of breast cancer is high breast tissue density. There is a wide variety of works focused on the study and automatic calculation of general breast density; however, they do not provide more detailed information about the changes that may occur within the breast tissue. This work proposes to generate a breast density map based on a texture analysis to identify the internal composition and distribution of the breast tissue through the diffuse division technique of the different densities inside the breast. Therefore, it is possible to obtain a density map associated with the breast that allows us to distinguish and quantify the different types of breast densities and their distribution according to the Breast Imaging Reporting and Data System (BI-RADS Breast Density Category). The proposed methodology was tested with mammograms from the BCDR and InBreast databases, demonstrating consistency in results and reaching an accuracy of 84.2% and 81.3%, respectively. Finally, the information obtained from the density map and its analysis could be a support tool for the specialist physician to monitor changes in breast density over time, since the fuzzy classification carried out allows quantifying the degree of membership in the BI-RADS breast density classes.

Breast cancer masses classification using deep convolutional neural networks and transfer learning

With the recent advances in the deep learning field, the use of deep convolutional neural networks (DCNNs) in biomedical image processing becomes very encouraging. This paper presents a new classification model for breast cancer masses based on DCNNs. We investigated the use of transfer learning from AlexNet and GoogleNet pre-trained models to suit this task. We experimentally determined the best DCNN model for accurate classification by comparing different models, which vary according to the design and hyper-parameters. The effectiveness of these models were demonstrated using four mammogram databases. All models were trained and tested using a mammographic dataset from CBIS-DDSM and INbreast databases to select the best AlexNet and GoogleNet models. The performance of the two proposed models was further verified using images from Egyptian National Cancer Institute (NCI) and MIAS database. When tested on CBIS-DDSM and INbreast databases, the proposed AlexNet model achieved an accuracy of 100% for both databases. While, the proposed GoogleNet model achieved accuracy of 98.46% and 92.5%, respectively. When tested on NCI images and MIAS databases, AlexNet achieved an accuracy of 97.89% with AUC of 98.32%, and accuracy of 98.53% with AUC of 98.95%, respectively. GoogleNet achieved an accuracy of 91.58% with AUC of 96.5%, and accuracy of 88.24% with AUC of 94.65%, respectively. These results suggest that AlexNet has better performance and more robustness than GoogleNet. To the best of our knowledge, the proposed AlexNet model outperformed the latest methods. It achieved the highest accuracy and AUC score and the lowest testing time reported on CBIS-DDSM, INbreast and MIAS databases.

Dataset of breast mammography images with masses.

Among many cancers, breast cancer is the second most common cause of death in women. Early detection and early treatment reduce breast cancer mortality. Mammography plays an important role in breast cancer screening because it can detect early breast masses or calcification region. One of the drawbacks in breast mammography is breast cancer masses are more difficult to be found in extremely dense breast tissue. We select 106 breast mammography images with masses from INbreast database. Through data augmentation, the number of breast mammography images was increased to 7632. We utilize data augmentation on breast mammography images, and then apply the Convolutional Neural Networks (CNN) models including AlexNet, DenseNet, and ShuffleNet to classify these breast mammography images.