A Ventricular Image Segmentation Method Based on Feature Compression and Noise Suppression

Abstract

A ventricular image segmentation method based on feature compression and noise suppression can eliminate noise features in features, solve the impact of noise features on network learning, indirectly emphasize important features in the network, and make network training more stable. By calculating the similarity between different features, select several channels with high similarity for feature fusion, reduce the number of feature channels while retaining the original features, and greatly reduce the number of parameters of the entire network, making network training more efficient. Simple and efficient.

Description

technical field [0001] The invention relates to the technical field of medical image processing, in particular to a ventricular image segmentation method based on feature compression and noise suppression. Background technique [0002] "Global Burden of Cardiovascular Disease and Risk Factors 1990-2019" shows that the incidence of cardiovascular disease is increasing year by year, and the number of deaths due to cardiovascular disease accounts for one-third of the total global deaths, and has become the largest cause of death in the world. In the diagnosis and treatment of cardiovascular diseases, it is necessary to accurately obtain parameters such as the shape and volume of the ventricles based on cardiac images, so as to assist doctors in making accurate judgments on diseases. However, due to the large amount of cardiac image data and complex professional knowledge, it is very difficult to accurately interpret cardiac images. Moreover, the number of professional physicia...

AI Technical Summary

Problems solved by technology

However, this approach has limited improvement in the learning ability of the network, and will bring two serious problems: 1) a large number of intermediate features will lead to a sharp increase in the complexity of the model, and a large number of parameters make the training of the network extremely difficult; 2) huge There are a lot of noise features in the intermediate features, which seriously hinder the learning of the network

Images