Breast ultrasound image segmentation is vital in medical imaging, enabling precise delineation of tissues and lesions, which contributes to the diagnosis and treatment of breast diseases. This article reviews both tradition-al methods and recent advancements in deep learning techniques for breast ultrasound image segmentation. The discussion begins by highlighting the significance of image segmentation in breast disease diagnosis and its background within medical imaging. Traditional segmentation methods, such as thresholding, edge detection, and region growing, are examined, with an analysis of their applications and limitations in breast ultrasound segmen-tation. Subsequently, the focus shifts to deep learning approaches, including classic models like Convolutional Neural Networks, Fully Convolutional Networks, and U-Net, along with their improved algorithms. These methods learn hierarchical features directly from raw data, reducing reliance on manual preprocessing. U-Net, in particular, is highlighted as the benchmark for medical image segmentation due to its efficient data usage and ability to pre-serve fine-grained details. Comparative analysis demonstrates the advantages of deep learning in enhancing seg-mentation accuracy, reducing noise, and handling complex texture structures. The article concludes by summariz-ing current achievements and challenges in the field, while offering an outlook on the future developments aimed at advancing breast ultrasound image segmentation for improved diagnosis and treatment of breast diseases.