Comparative Heatmap Explanations for Deep Learning-Based Breast Density Classification

Deep Learning-Based Automated Reports for Breast Density Assessment in Mammography Images.

Intelligent Application and Model for Enhancing Breast Cancer Classification

Breast cancer diagnosis using mammography remains a challenging task, particularly in the presence of dense and heterogeneous tissue structures. While deep learning (DL) methods have shown promise, their interpretability and performance are often constrained by suboptimal configurations and limited contextual learning. This paper introduces a Vision Transformer (ViT)-based model, enhanced by the Reptile Search Algorithm (RSA), to achieve both accurate classification and explainability. The ViT framework captures global dependencies in mammographic image patches, while RSA optimizes critical hyperparameters, including patch size, learning rate, and attention depth, to enhance model convergence and generalization. Applied on the CBIS-DDSM: Breast Cancer Image Dataset, the ViT-RSA model attains a classification accuracy of 96.2% and an Area Under the Curve (AUC) of 0.97. Interpretability is ensured through attention visualization maps from ViT layers and Grad-Cam-based explanations. The experimental results confirm that ViT-RSA outperforms conventional CNN models, offering a robust and interpretable tool for breast cancer diagnosis that supports clinical decision-making.

Explainable Lightweight Deep Learning with Grad-CAM for Breast Cancer USG Classification

DeepScan: A Deep Learning-Based Clinical Decision Support System for Breast Cancer Diagnosis Using DenseNet121-CBAM and LLM Integration

Breast cancer diagnosis using ultrasound imaging remains challenging due to image noise, class imbalance, and limited interpretability of automated systems. This study presents DeepScan, a deep learning–based Clinical Decision Support System (CDSS) designed for research and educational use, which integrates a DenseNet121 backbone with a Convolutional Block Attention Module (CBAM) to enhance feature discrimination in breast ultrasound images. The system performs three-class classification (Normal, Benign, Malignant) using a structured preprocessing pipeline that includes resizing, ImageNet-based normalization, data augmentation, and controlled oversampling to address class imbalance. To improve transparency, Grad-CAM–based visual explanations are incorporated to highlight diagnostically relevant regions influencing model predictions. Beyond image-level classification, DeepScan integrates a Large Language Model (LLM)–based reasoning engine to generate BI-RADS–aligned, structured clinical reports and provide interactive explanations for users. Experimental evaluation on the BUSI dataset demonstrates strong discriminative performance, achieving AUC values of 0.998 for Normal tissue and 0.957 for both Benign and Malignant classes, with near real-time inference latency. The results indicate that combining attention-enhanced convolutional models with explainable AI and LLM-based reporting can improve both performance and interpretability, positioning DeepScan as a supportive CDSS framework for breast ultrasound analysis in pre-clinical and educational settings.

Prediction Modeling for Breast Cancer Treatment Planning Using Contrast-Enhanced Mammography

Objective The goal of this study was to develop a deep learning model to support clinicians in the diagnosis and treatment planning of breast cancer. Specifically, we aim to leverage Grad-CAM (Gradient-weighted Class Activation Mapping)-based classification techniques for the weakly supervised localization of breast lesions in contrast-enhanced mammography (CEM) images. Furthermore, by integrating radiomics features and histopathological data, the model seeks to predict treatment outcomes more accurately. This approach helps in clinical decision-making, improve personalized care, and ultimately contribute to better patient survival rates. Materials and Methods The proposed model extracts radiomics features from 2,289 CEM images, sourced from both a public data set CDD (categorized digital database)-CEM and a private data set, employing pretrained convolutional neural network (CNN) architectures such as Xception, Inception V3, DenseNet 201, ResNet 152, and VGG16. Results The model's performance was evaluated using three key metrics—accuracy, precision, and area under the curve (AUC)—for the classification task, and HD (Hausdorff distance) and Dice coefficient for the segmentation task. Experimental results demonstrate that the support vector machine model is particularly effective for breast cancer diagnosis, achieving an overall accuracy of 83%, precision of 80%, and an AUC of 0.83, based on a threefold cross-validation method applied specifically to the prediction modeling phase. Five different models were trained and tested on features selected differently from Least Absolute Shrinkage and Selection Operator (LASSO) and Minimum Redundancy Maximum Relevance; LASSO selected features gave better result with every model used. Conclusion These findings suggest that integrating advanced CNN architectures with radiomics features can significantly enhance the accuracy and reliability of both lesion segmentation and treatment prediction in breast cancer, potentially leading to better clinical outcomes.

An attention-based transfer learning framework for breast cancer classification in mammography under limited data conditions

Breast cancer is a leading cause of death for women worldwide. There is a critical need for an early and accurate diagnosis to improve survival rates. This study evaluates the performance of transfer learning-based MobileNetV2 and ResNet50 architectures for binary breast cancer classification using mammography images. To enhance channel-wise feature learning, the Squeeze-and-Excitation (SE) attention mechanism was integrated into the network and partial fine tuning was performed by only freezing the early layers of the model. Furthermore, gradient-weighted class activation mapping was applied to visualize the important regions of mammogram images responsible for the model’s predictions. The performance evaluation was carried out on the Contrast Limited Adaptive Histogram Equalization (CLAHE) binary, Digital Database for Screening Mammography (DDSM), INbreast, and Mammographic Image Analysis Society (MIAS) datasets using stratified k-fold cross-validation. The experimental results demonstrate that one of the proposed frameworks, MobileNetV2-SE, achieved mean classification accuracies of 99.14%, 93.53%, 92.35%, and 96.33% on the DDSM, INbreast, MIAS, and CLAHE-binary datasets, respectively. In comparison, the ResNet50-SE model attained mean accuracies of 100%, 88.89%, 71.60%, and 97.98% on the same datasets. The experimental results demonstrate that the frameworks achieve reliable and competitive performance for breast cancer detection under realistic evaluation conditions and improve the diagnostic accuracy of mammogram classification. This work contributes to the field of medical image analysis and computer-aided diagnosis. Future work should focus on applying advanced class balancing methods to improve diagnostic accuracy and model robustness in a variety of medical imaging applications.