Breast cancer remains a primary cause of death among women globally, making early detection essential for boosting survival rates. In this study, we propose a dual-approach methodology for breast cancer detection using both structured (tabular) and unstructured (image) data. The first approach utilizes the Wisconsin Breast Cancer Dataset (WBCD), to train various machine learning (ML) models such as Logistic Regression (LR), Random Forest (RF), Support Vector Machine (SVM), K-Nearest Neighbors (KNN), and XGBoost (XGB), alongside deep learning (DL) models including Convolutional Neural Networks (CNN), Artificial Neural Networks (ANN), and Recurrent Neural Networks (RNN). Out of all models tested, Logistic Regression demonstrated the highest accuracy at 99.12%. The second approach focuses on the Mammographic Image Analysis Society (MIAS) dataset comprising mammographic images. Three models were developed for this dataset: a standard CNN model, a pre-trained CNN-DNN hybrid model, and a pre-trained CNN-ML ensemble hybrid model. The ensemble hybrid model utilizes pre-trained CNN architectures (InceptionV3, ResNet50, and VGG16) for feature extraction, followed by a majority voting-based ensemble of traditional ML classifiers (SVM and RF) for final classification. This configuration outperformed all other models with an accuracy of 98.48%. These findings demonstrate the effectiveness of integrating traditional and deep learning techniques across heterogeneous data sources, providing a promising direction for the development of real-time AI-assisted diagnostic systems in healthcare.