Data acquisition and processing of breast cancer assisted diagnosis based on ultrasound imaging

Breast disease is a common disease in women. The analysis and judgment of B-mode ultrasound images by doctors depend heavily on the operation experience and technical level of doctors. Computer image processing technologies such as natural image classification, target detection and semantic segmentation, represented by deep learning, have been relatively mature, and have been widely used successfully in automatic driving, security, finance and other fields. In this paper, through consultation and cooperation with medical institutions, a large mammary ultrasound image data set is constructed, which basically meets the needs of deep neural network training and validation testing. It is used to develop and validate algorithms for subsequent subtasks of ultrasound image analysis.