In this paper, we evaluate bagging ensembles for breast cancer diagnosis using a multimodal dataset “PathoEMR” which combines pathology images and electronic medical records. We implement the bagging ensemble method to two deep learning models (Densenet201 and VGG16) for pathological images and seven machine learning classifiers for electronic medical records. The best bagging ensemble for electronic medical records achieved an accuracy mean value of 92.31

Contrast-enhanced spectral mammography (CESM) has emerged as a valuable tool for detecting breast tumors, offering enhanced sensitivity in specific clinical scenarios. However, accurate segmentation of tumors based on CESM imaging modality remains challenging. In this work, we propose a novel approach to utilizing generative modeling in the context of CESM for breast cancer diagnosis. We utilize paired imageto-image translation to generate high-fidelity dual-energy subtracted (DES) representations that closely align with real data. Our proposed model integrates two inputs: real DES images and synthesized ones, aiming to enhance segmentation accuracy and generate masks pinpointing tumor locations. Following segmentation, these masks are classified to distinguish between benign and malignant tumors. A Multi Scale Subtraction Network is utilized to improve tumor detection with the integration of DenseLossNet, a novel loss function based to assist in refining segmentation results, thereby achieving precise delineation of tumor boundaries. Through rigorous evaluation, our proposed framework demonstrates promising results in tumor detection and classification, achieving high accuracy in distinguishing between benign and malignant tumors while precisely localizing tumor positions. These findings underscore the potential of our approach in improving tumor detection in CESM, offering clinicians valuable insights for more informed decision making.

This paper introduces a novel bagged ensemble approach for a binary classification using convolutional neural networks (CNNs) and transfer learning strategies. The CNN is trained independently on different bootstrapped samples (bags) of the training data, and the predictions are aggregated using majority voting to obtain the final classification result of the deep bagging ensemble. The BreakHis dataset, encompassing four magnification factors (40×, 100×, 200×, and 400×), is used for evaluation, and in addition to the four-metrics (accuracy, precision, sensitivity, and F1-score), the Scott-knot statistical test and Borda count voting method are employed to comprehensively assess the performance of the different models and rank them. Experimental results demonstrated that the proposed approach achieved high values over the four metrics for breast cancer classification. The comparative analysis highlights the ability of the deep bagging ensembles to capture diverse and complementary features from different bags, and it showcases their superiority compared to the single CNNs and the hybrid bagging ensembles, where the pre-trained deep learning models are only used for feature extraction. The findings suggested that the simplicity and effectiveness of combining fine-tuned CNNs with bagging ensembles, make the proposed approach an attractive choice for practical implementation, as it holds a promise for accurate and reliable classification of histopathological breast cancer images. Furthermore, this approach can potentially be applied to other medical image classification tasks, providing a more efficient and accurate diagnosis for a range of diseases.

Accurate identification of breast cancer molecular subtypes significantly impacts patient prognosis and treatment decisions. Multimodal fusion techniques have shown promise in improving the performance of deep learning models by leveraging the complementary strengths of various modalities. However, the integration of multi-dimensional data poses a challenge due to the heterogeneity and complexity of the data sources. Random subspace can improve the robustness and generalization abilities of multimodal classification techniques by reducing variance and making it easier to identify the most informative features. The present study suggests a new approach referred to as Multimodal Random Subspace Support Vector Machine (MRSVM) ensemble that effectively combines multidimensional data encompassing copy number variation, clinical information, gene expression data and histopathological whole slide images to improve the prediction of breast cancer molecular subtypes. To highlight the prowess and distinctive qualities of the newly devised approach, we conducted an extensive comparative examination across all possible combinations of these four modalities, where we were able to address six research questions, which mainly involve identifying the optimal combination of the four available modalities, the most effective fusion strategy, and the most suitable classification technique. Results indicated that early fusion models outperformed late fusion models, notably, the MRSVM ensemble based on early fusion achieved the highest accuracy of 88.07

In breast cancer research, diverse data types and formats, such as radiological images, clinical records, histological data, and expression analysis, are employed. Given the intricate nature of natural phenomena, relying on the features of a single modality is seldom sufficient for comprehensive analysis. Therefore, it is possible to guarantee medical relevance and achieve improved clinical outcomes by combining several modalities. The presen study carefully maps and reviews 47 primary articles from six well-known digital libraries that were published between 2018 and 2023 for breast cancer classification based on multimodal deep learning fusion (MDLF) techniques. This systematic literature review encompasses various aspects, including the medical modalities combined, the datasets utilized in these studies, the techniques, models, and architectures used in MDLF and it also discusses the advantages and limitations of each approach. The analysis of selected papers has revealed a compelling trend: the emergence of new modalities and combinations that were previously unexplored in the context of breast cancer classification. This exploration has not only expanded the scope of predictive models but also introduced fresh perspectives for addressing diverse targets, ranging from screening to diagnosis and prognosis. The practical advantages of MDLF are evident in its ability to enhance the predictive capabilities of machine learning models, resulting in improved accuracy across diverse applications. The prevalence of deep learning models underscores their success in autonomously discerning complex patterns, offering a substantial departure from traditional machine learning approaches. Furthermore, the paper explores the challenges and future directions in this field, including the need for larger datasets, the use of ensemble learning methods, and the interpretation of multimodal models.

Purpose Histopathology biopsy imaging is currently the gold standard for the diagnosis of breast cancer in clinical practice. Pathologists examine the images at various magnifications to identify the type of tumor because if only one magnification is taken into account, the decision may not be accurate. This study explores the performance of transfer learning and late fusion to construct multi-scale ensembles that fuse different magnification-specific deep learning models for the binary classification of breast tumor slides. Design/methodology/approach Three pretrained deep learning techniques (DenseNet 201, MobileNet v2 and Inception v3) were used to classify breast tumor images over the four magnification factors of the Breast Cancer Histopathological Image Classification dataset (40×, 100×, 200× and 400×). To fuse the predictions of the models trained on different magnification factors, different aggregators were used, including weighted voting and seven meta-classifiers trained on slide predictions using class labels and the probabilities assigned to each class. The best cluster of the outperforming models was chosen using the Scott–Knott statistical test, and the top models were ranked using the Borda count voting system. Findings This study recommends the use of transfer learning and late fusion for histopathological breast cancer image classification by constructing multi-magnification ensembles because they perform better than models trained on each magnification separately. Originality/value The best multi-scale ensembles outperformed state-of-the-art integrated models and achieved an accuracy mean value of 98.82 per cent, precision of 98.46 per cent, recall of 100 per cent and F 1-score of 99.20 per cent.

: This paper proposes the use of transfer learning and ensemble learning for binary classification of breast cancer histological images over the four magnification factors of the BreakHis dataset: 40×, 100×, 200× and 400×. The proposed bagging ensembles are implemented using a set of hybrid architectures that combine pre-trained deep learning techniques for feature extraction with machine learning classifiers as base learners (MLP, SVM and KNN). The study evaluated and compared: (1) bagging ensembles with their base learners, (2) bagging ensembles with a different number of base learners (3, 5, 7 and 9), (3) single classifiers with the best bagging ensembles, and (4) best bagging ensembles of each feature extractor and magnification factor. The best cluster of the outperforming models was chosen using the Scott Knott (SK) statistical test, and the top models were ranked using the Borda Count voting system. The best bagging ensemble achieved a mean accuracy value of 93.98%, and was constructed using 3 base learners, 200× as a magnification factor, MLP as a classifier, and DenseNet201 as a feature extractor. The results demonstrated that bagging hybrid deep learning is an effective and a promising approach for the automatic classification of histopathological breast cancer images.

Purpose Breast cancer (BC) is the most common diagnosed cancer type and one of the top leading causes of death in women worldwide. This paper aims to investigate ensemble learning and transfer learning for binary classification of BC histological images over the four-magnification factor (MF) values of the BreakHis dataset: 40X, 100X, 200X, and 400X. Methods The proposed homogeneous ensembles are implemented using a hybrid architecture that combines: (1) three of the most recent deep learning (DL) techniques for feature extraction: DenseNet_201, MobileNet_V2, and Inception_V3, and (2) four of the most popular boosting methods for classification: AdaBoost (ADB), Gradient Boosting Machine (GBM), LightGBM (LGBM) and XGBoost (XGB) with Decision Tree (DT) as a base learner. The study evaluated and compared: (1) a set of boosting ensembles designed with the same hybrid architecture and different number of trees (50, 100, 150 and 200); (2) different boosting methods, and (3) the single DT classifier with the best boosting ensembles. The empirical evaluations used: four classification performance criteria (accuracy, recall, precision and F1-score), the fivefold cross-validation, Scott Knott statistical test to select the best cluster of the outperforming models, and Borda Count voting system to rank the best performing ones. Results The best boosting ensemble achieved an accuracy value of 92.52% and it was constructed using XGB with 200 trees and Inception_V3 as feature extractor (FE). Conclusions The results showed the potential of combining DL techniques for feature extraction and boosting ensembles to classify BC in malignant and benign tumors.

Breast cancer is the most common cancer in women worldwide. While the early diagnosis and treatment can significantly reduce the mortality rate, it is a challenging task for pathologists to accurately estimate the cancerous cells and tissues. Therefore, machine learning techniques are playing a significant role in assisting pathologists and improving the diagnosis results. This paper proposes a hybrid architecture that combines: three of the most recent deep learning techniques for feature extraction (DenseNet_201, Inception_V3, and MobileNet_V2) and random forest to classify breast cancer histological images over the BreakHis dataset with its four magnification factors: 40X, 100X, 200X and 400X. The study evaluated and compared: (1) the developed random forest models with their base learners, (2) the designed random forest models with the same architecture but with a different number of trees, (3) the decision tree classifiers with the best random forest models and (4) the best random forest models of each feature extractor. The empirical evaluations used: four classification performance criteria (accuracy, sensitivity, precision and F1-score), 5-fold cross-validation, Scott Knott statistical test, and Borda Count voting method. The best random forest model achieved an accuracy mean value of 85.88%, and was constructed using 9 trees, 200X as a magnification factor, and Inception_V3 as a feature extractor. The experimental results demonstrated that combining random forest with deep learning models is effective for the automatic classification of malignant and benign tumors using histopathological images of breast cancer.

Diagnosis of breast cancer in the early stages allows to significantly decrease the mortality rate by allowing to choose the adequate treatment. This paper develops and evaluates twenty-eight hybrid architectures combining seven recent deep learning techniques for feature extraction (DenseNet 201, Inception V3, Inception ReseNet V2, MobileNet V2, ResNet 50, VGG16 and VGG19), and four classifiers (MLP, SVM, DT and KNN) for binary classification of breast cytological images over the FNAC dataset. To evaluate the designed architectures, we used: (1) four classification performance criterias (accuracy, precision, recall and F1-score), (1) Scott Knott (SK) statistical test to cluster the developed architectures and identify the best cluster of the outperforming architectures, and (2) Borda Count voting method to rank the best performing architectures. Results showed the potential of combining deep learning techniques for feature extraction and classical classifiers to classify breast cancer in malignant and benign tumors. The hybrid architectures using MLP classifier and DenseNet 201 for feature extraction were the top performing architectures with a higher accuracy value reaching 99% over the FNAC dataset. As results, the findings of this study recommend the use of the hybrid architectures using DenseNet 201 for the feature extraction of the breast cancer cytological images since it gave the best results for the FNAC data images, especially if combined with the MLP classifier.