U-Net-based blood vessel image segmentation method, device and equipment

Abstract

The invention discloses a blood vessel image segmentation method, device and equipment based on U-Net. The method comprises the following steps: acquiring a blood vessel segmentation data set; preprocessing the blood vessel segmentation data set; performing image block cutting operation on the preprocessed blood vessel segmentation image to obtain sample data; according to the sample data, building a blood vessel image segmentation network through a Pytorch deep learning framework; and performing blood vessel image segmentation according to the blood vessel image segmentation network, and evaluating a blood vessel image segmentation result. A convolution block in a tube image segmentation network is replaced by a multi-scale feature aggregation block; the first input of the multi-scale feature aggregation block is a multi-scale high-level feature, and the second input of the multi-scale feature aggregation block is a multi-scale low-level feature; according to the blood vessel image segmentation network, the multi-scale high-level features and the multi-scale low-level features in the multi-scale feature aggregation block are fused through the MS-CAM module, the segmentation performance can be improved, and the blood vessel image segmentation network can be widely applied to the technical field of artificial intelligence.

Description

technical field [0001] The invention relates to the technical field of artificial intelligence, in particular to a U-Net-based blood vessel image segmentation method, device and equipment. Background technique [0002] The visible structure of retinal blood vessels can be indicative of many diseases. Accurate segmentation helps capture visible changes in the retinal vascular structure, which helps doctors diagnose eye-related diseases. Therefore, it is particularly important in current retinal image analysis tasks. For example, hypertensive retinopathy is a disease of the retina caused by high blood pressure, and increased curvature or narrowing of blood vessels can be found in patients with high blood pressure. Traditionally, manual segmentation is performed by experts, which is laborious, time-consuming, and suffers from inter-expert subjectivity. In clinical practice, there is a high demand for automatic segmentation methods in order to improve efficiency and reliabili...

AI Technical Summary

Problems solved by technology

The main deficiencies of these two types of encoders include: First, there are huge differences in features and imaging principles between medical images and natural images, resulting in poor portability of feature extraction models on these two datasets

[0005] Second, ordinary skip connections in each stage usually directly incorporate local information, which introduces too much irrelevant background noise, making it difficult to distinguish retinal vessels from surrounding mimics and noise, especially tiny vessels

In principle, the high-level features obtained from the deep stage have rich semantic information but lack sufficient resolution, while the low-level features obtained from the shallow stage have rich spatial details but lack global semantic information

Images