Identification and Localization of Breast Tumor Components via a Convolutional Neural Network Based on High-Frequency Ultrasound Combined With Histopathologic Registration: Prospective Study.

Key Points

• The aim is to establish an effective method for identifying and localizing breast tumor components using imaging techniques.
• Developed a technique for spatial registration of whole slide images (WSIs) and ultrasound images.
• Utilized an advanced convolutional neural network for identifying cancer regions.
• Targeted identification was conducted at the pixel level in high-frequency ultrasound images.
• Breast cancer regions were accurately localized using high-frequency ultrasound images.
• The method achieved pixel-level identification of tumor components based on histopathologic standards.