Ultrasonic breast tumor automatic segmentation method based on improved u-network based on attention enhancement

Abstract

The invention discloses a method for automatic segmentation of ultrasonic breast tumors based on attention-enhanced improved U-shaped network, step (A), constructing an attention-enhanced improved U-shaped network structure for extracting the target features of ultrasonic breast tumors; step (B) ), establish a mixed attention loss function that improves the U-shaped network structure with attention enhancement, and is used to obtain a more accurate network loss value; step (C), improves the U-shaped network structure and the mixed attention loss function according to the attention enhancement, by The network layered training method combining thickness and fineness fine-tunes the network parameters, improves the accuracy of the network parameters, and realizes the rapid positioning and segmentation of the lesion area in the breast ultrasound image. The invention can be used to extract the lesion area of ​​breast ultrasound images, which can effectively improve the accuracy of breast tumor segmentation, assist doctors to quickly and accurately locate the lesion area, reduce the workload of doctors, and alleviate the defects of young doctors' lack of clinical experience. Medicine has very important research value and application prospect.

Description

technical field [0001] The invention relates to the technical field of automatic breast tumor segmentation of ultrasonic images, in particular to an automatic segmentation method of ultrasonic breast tumors based on attention enhancement and improved U-shaped network. Background technique [0002] At present, breast cancer is a disease that seriously endangers women's health after skin cancer. With the development of modern medicine, early diagnosis and treatment can greatly improve the survival rate of breast cancer patients. At present, the diagnosis of breast tumors can be divided into invasive and non-invasive diagnosis. Invasive diagnosis mainly refers to biopsy, which causes physical damage to the breast tissue and brings pain to the patient; non-invasive diagnosis includes the examination of breast diseased areas using techniques such as X-rays, MRI (magnetic resonance imaging), and ultrasound imaging . Among various examination methods, due to the advantages of lo...

AI Technical Summary

Problems solved by technology

However, due to the lack of training data, the low contrast of ultrasound images, the mutual invasion between suspicious lesions and surrounding tissues, and the diversity of tumor shapes, the research on ultrasound breast tumor region segmentation based on deep neural networks is still in its infancy. , the segmentation effect still needs to be further improved

Images