A Classification Method of Brain Functional Networks Based on Variational Autoencoders

Abstract

A method for classifying brain functional networks based on variational autoencoders, comprising the following steps: collecting T1 weighted magnetic resonance images T1 weighted MRI and resting state functional magnetic resonance images rs‑fMRI of several normal people and patients with brain cognitive impairment ; carry out preprocessing; use the pretreated rs‑fMRI as the regression dependent variable, and the brain function network as the regression independent variable for double regression analysis to obtain the individual level brain function network; construct a deep variational autoencoder (VAE) model, and The obtained individual-level brain function network map is used as the input and output of VAE, and the encoder part is used as a feature extraction module to obtain the hidden code of the individual function network; construct a multi-layer perceptron network to classify the code obtained by VAE in step 4; use A plurality of classifiers trained for different brain function networks infer the samples in the test set, and fuse the inference results of the plurality of classifiers to obtain the final classification result; the invention improves the classification accuracy.

Description

Technical field [0001] The present invention relates to the field of medical image analysis, and in particular, to the method of treating classification of functional magnetic distribution images in the field of brain sciences, specifically a brain function network classification method based on variational self-encoder. Background technique [0002] The human brain is one of the most complicated systems in nature. The various neurophysiological activities of the brain are the results of interaction between different brain regions. The implementation of a complex cognitive task often requires the cooperation of multiple brain regions, and the study of mutual dependencies between different brain regions helps deepen human work. From the perspective of the network, the function of the brain has been proven to be an effective research method, and brain function network analysis has become one of the important research directions in the field of brain. [0003] Functional Magnenance ...

AI Technical Summary

Problems solved by technology

[0006] In order to overcome the problem that the input feature vector contains insufficient information in the traditional brain function network classification method, ignores the topological structure relationship between different brain regions in the brain function network, and the traditional machine learning algorithm has limited data modeling capabilities, the present invention The purpose is to propose a brain function network classification method based on variational autoencoders. Group independent component analysis (Group ICA) can effectively distinguish signals with physiological significance from noise signals, and can separate brain networks with co-activity patterns. Network, and then use the double regression method to extract the individual level brain function network; Group ICA as a data-driven method does not need to define the template network in advance, while retaining the topological structure information of different brain regions; variational autoencoder as a deep On the one hand, the neural network generation model can capture the activation degree of different brain regions in the brain network and the non-local relationship between different brain regions, and on the other hand, it can learn the distribution of hidden variables to obtain a set of hidden variables with good distribution properties.

Images