Schizophrenia classification method and system based on multi-center model

Abstract

The invention relates to a schizophrenia classification method and system based on a multi-center model, and the method comprises the steps: 1, data preparation: collecting a brain MRI image of a sample, processing the brain MRI image, and extracting a plurality of brain structure features to obtain a feature matrix; carrying out covariant regression processing on the feature matrix, and then carrying out standardization processing; enabling each center to prepare a respective data set according to the step; 2, enabling each center to construct a respective single-center model by using a machine learning classifier, and training the respective single-center model by using the respective data set; 3, classifying to-be-classified test samples by using each single center model to obtain classification probability values of the to-be-classified test samples corresponding to each center; and performing weighted summation on the classification probability value and the weight of each center to obtain a classification probability value based on a multi-center model, and integrating all the single-center models into the multi-center model for classification. Data sharing of each center is realized, and each center does not need to share original data.

Description

technical field [0001] The invention relates to the technical field of neurological disease diagnosis, in particular to a method and system for schizophrenia classification based on a multicenter model. Background technique [0002] Schizophrenia is a serious mental illness, and objective auxiliary examination methods are the basis for early diagnosis of schizophrenia, and are helpful for the treatment and improvement of prognosis of schizophrenia. A large number of studies have shown that there are differences in the brain structure of patients with schizophrenia and normal people. Therefore, magnetic resonance imaging (MRI) is generally used to obtain comprehensive brain structure and function information to diagnose schizophrenia, but there are many differences in the diagnosis results. Large heterogeneity can easily lead to inaccurate diagnostic results. [0003] At present, although there are many methods for classification and analysis of neuroimaging combined with ma...

AI Technical Summary

Problems solved by technology

A large number of studies have shown that there are differences in the brain structure of patients with schizophrenia and normal people. Therefore, magnetic resonance imaging (MRI) is generally used to obtain comprehensive brain structure and function information to diagnose schizophrenia, but there are many differences in the diagnosis results. Large heterogeneity can easily lead to inaccurate diagnostic results

[0003] At present, although there are many methods for classification and analysis of neuroimaging combined with machine learning, most of them have the following limitations: 1) For single-center data, the sample size is small, and the limited sample size often leads to poor generalization performance of the classification model ; 2) Training on the original data of all centers cannot avoid the problem of large amount of original data and ethical issues

Images