Breast cancer remains a critical global health concern, necessitating early detection for improved patient outcomes. This study proposes an automated system for enhanced breast cancer diagnosis through intelligent ultrasound imaging. The system begins with the assembly of a diverse ultrasound breast cancer (USBC) image dataset, encompassing normal, benign, and malignant cases. Various preprocessing techniques, including data augmentation, cropping, and resizing, are applied to standardize and augment the dataset. Transfer learning is leveraged to extract features from USBC images using established deep learning models, namely VGG-16, Inception V3, and VGG-19. Extracted features are stored for efficient data handling during model training and evaluation. The dataset is partitioned using 10-fold cross-validation, ensuring robust model validation. Two machine learning models, Multi-Layer Perceptron (MLP) and Support Vector Machines (SVM), are employed for breast cancer classification. MLP architectures with varying layer sizes and SVM models with polynomial and radial basis function kernels are fine-tuned for optimal performance. Ultimately, MLP serves as the final classifier, offering a promising approach for accurate and efficient breast cancer identification.

KEYWORDS: Breast cancer, ultrasound imaging, transfer learning, deep learning models, data augmentation, preprocessing techniques, machine learning classification, MLP architecture, SVM kernels, model validation.