Cross-modal medical image domain adaptive classification method based on graph neural network

Abstract

The invention discloses a cross-modal medical image domain adaptive classification method based on a graph neural network. The collection and marking of cross-modal medical image data are affected because a cross-modal medical image presents a larger intra-class difference and a smaller inter-class difference. The graph neural network module can well distinguish medical images in different fields,model oscillation can be greatly reduced, and classification precision is improved.

Description

technical field [0001] The invention relates to the technical field of medical image classification, in particular to a cross-modal medical image domain adaptive classification method based on a graph neural network. Background technique [0002] Cancer is still one of the leading causes of human death, and timely and accurate disease grading plays a vital role in later treatment. Although medical imaging technology has developed rapidly in recent years, computer-aided diagnosis systems (CAD) usually rely on training machine learning models (such as deep convolutional neural networks) and a large number of correctly labeled samples, and the lack of large training data sets presents significant challenges to computer-aided diagnosis (CAD) systems. First, to gather data, experienced medical imaging experts need to scrutinize hundreds to thousands of gigabytes of images, an annotation process that is time-consuming and labor-intensive. Secondly, it is difficult to label medic...

AI Technical Summary

Problems solved by technology

[0004] The above introduction to existing methods shows that the current domain adaptation methods designed for medical image analysis are mainly based on the framework of adversarial learning, which is inefficient and often leads to model oscillations that are difficult to converge.

Images