Multi-modal brain image depression identification method and system based on graph node embedding

Abstract

The invention provides a multi-modal brain image depression identification method and system based on graph node embedding. Deep learning is applied to the depression recognition of a multi-modal brain image, and a bridge is built between a multi-modal brain network and a convolutional neural network (CNN) through graph node embedding, so that the CNN can be used for the depression recognition ofthe multi-modal brain image, and therefore, depression recognition accuracy is improved. The method comprises the following steps of: 1) acquiring the resting state fMRI and DTI image data of a depressive patient and a normal control group; 2) preprocessing the acquired fMRI and DTI image data; 3) respectively constructing a brain function network and a brain structure network according to the preprocessed fMRI and DTI image data to obtain a brain network adjacency matrix, and 4) adopting graph node embedding to express the adjacency matrix as an image, inputting the image into a convolutionalneural network for classifying the image, and establishing a classification model for identifying depressive patients and normal subjects.

Description

technical field [0001] The present invention relates to the technical fields of brain neuroscience, medical imaging, and deep learning, in particular to a multimodal brain image depression recognition system based on graph node embedding. Background technique [0002] Depression (MDD) is the fourth leading disease in the world, and patients will have clinical manifestations such as low mood, slow thinking and cognitive impairment. The diagnosis of MDD is usually based on Diagnostic and Statistical Manual of Mental Disorders (DSM) criteria and clinical response. Due to overlapping phenotypes among various psychiatric disorders, as well as heterogeneity within disorders such as MDD, clinical diagnoses often have a high rate of underdiagnosis and misdiagnosis. Neuroimaging provides noninvasive measurements of brain function and structure and can serve as a powerful tool to study discriminative biomarkers, thereby reducing missed and misdiagnosed diagnoses. Common to other psy...

AI Technical Summary

Problems solved by technology

However, the existing classification methods are mainly based on machine learning, especially support vector machines (SVM), and the classification accuracy has always been low.

In addition, due to the excellent performance of convolutional neural network (CNN) in many fields such as image processing, speech recognition, computer-aided diagnosis, and natural language processing, some researchers directly use brain imaging data as the input of CNN, but have not achieved good results. the result of

Images