Construction method and application of morphological fusion classification index of neural image marker

Abstract

The invention relates to a construction method and application of a morphological fusion classification index of a neural image marker, and the construction method comprises the following steps: obtaining the structural MRI data of M centers, and extracting the feature data of a brain structure image; independently training the data of each center, respectively establishing a classification model of each center, and obtaining classification models of M centers; for any sample, calculating a classification weight value of the sample in each feature of each model in classification models of all centers, namely an SHAP matrix; and then taking the sample size used during training of each model as a weight, and calculating according to a formula (1) to obtain a single morphological fusion classification index MICI value: Si represents the sample size of the model i, B represents the total number of the features, ai represents the SHAP value of the feature a in the model i, and i = 1-M. The MICI value can well realize identification between mental disease patients and normal people, and has good interpretability, evolutionary property and expandability.

Description

technical field [0001] The invention relates to the field of neuroimaging markers, and proposes a method for constructing neuroimaging markers based on machine learning and multi-center data—morphological fusion classification index (MICI value), which is used to assist individualized diagnosis and treatment of neuropsychiatric diseases. Background technique [0002] Neuropsychiatric diseases characterized by diffuse brain damage such as schizophrenia, severe depression, and Alzheimer's disease seriously affect human health and impose a huge burden on individuals and society. At present, mental illness mainly relies on doctors to make subjective diagnosis of clinical symptoms, and there are certain misdiagnosis and missed diagnoses. Magnetic resonance imaging (MRI) has attracted more and more attention due to its advantages of simplicity, non-invasiveness and comprehensiveness. A large number of studies have reported that there are significant differences between the brain ...

AI Technical Summary

Problems solved by technology

However, there are some unavoidable problems in multi-center data sharing. For example, the original MRI data analysis needs to consume massive storage, network and computing resources, which greatly increases the human and material costs of the analysis organization; in addition, the original MRI data contains personal identification. information, how to effectively protect the privacy of the subjects is also a difficult problem

The machine learning model is a "black box" with a large number of parameters. The interpretability of the prediction results is relatively poor, and it is difficult to establish human-understandable connections with the neurobiological characteristics and clinical symptoms of mental diseases, which makes it unable to be used in clinical work. Get fast promotion and conversion

Images