Breast cancer stands as a formidable global health challenge, substantially impacting cancer-related mortality rates.Ultrasound (US) imaging has gained prominence in breast cancerdiagnosis, particularly for individuals with dense breast tissue.However, the efficacy of US imaging is reliant on operatorproficiency and is susceptible to noise, posing a substantialdiagnostic hurdle. In this study, we have devised an automatedultrasound-based Computer-Aided Diagnosis (CAD) systemdesigned to detect and classify breast cancer lesions. A datasetcomprising 6,319 images from 2889 patients was employed. Toensure the generalizability of our AI algorithm, images wereacquired using various US machines with different transducers (1-14 MHz). Deep learning methodologies were harnessed,encompassing the utilization of the EfficientNetV2-B0 architecturefor image classification (benign/malignant) and theimplementation of the Attention U-Net coupled with the Cosh logDice loss function for breast lesion segmentation. Our CAD systemdemonstrated an impressive sensitivity of 89.0% and specificity of92.0% for classification, along with a segmentation Dice score of86.0%. The integration of such CAD systems into breast imagingworkflows holds promise for diminishing the influence of humanerrors, consequently reducing diagnostic costs, and expediting thebreast US imaging process.