Neurodegenerative disease brain image generation prediction method based on depth generation model

Abstract

The invention provides a neurodegenerative disease brain image generation prediction method based on a depth generation model. The method comprises the following steps: acquiring brain structure image data of a patient, and performing data division and arrangement according to the subtype expression and natural disease course period of clinical symptoms of the patient to obtain a disease subtype tag and a disease course tag; constructing an image prediction generation model by adopting a deep neural network; performing training test on the constructed image prediction generation model by using K-fold cross validation and brain image data with a disease subtype label and a disease course label to obtain reconstruction loss, cross entropy loss and discrete uniformity loss of a test set, and storing an optimal model; inputting the brain image data of the patient in the current disease course period into the optimal model, and completing the generation of the brain image data of the next natural disease course of the patient, namely predicting the natural disease course development of the patient. The invention provides an effective scientific basis for early diagnosis and timely intervention treatment of neurodegenerative diseases.

Description

technical field [0001] The invention relates to the technical field of computer applications, in particular to a method for generating and predicting brain images of neurodegenerative diseases based on deep generation models. Background technique [0002] As the aging society is getting worse, the prevalence of some aging-related diseases is also increasing year by year. Neurodegenerative diseases are a kind of disease state based on early degeneration or atrophy of nerve cells, and this disease state is to destroy Sexual, irreversible, worsens over time, eventually leading to dysfunction. Alzheimer's disease (Alzheimer's disease, AD) and Parkinson's disease (Parkinson disease, PD) are two common neurodegenerative diseases, and they are the two diseases with the highest prevalence in the elderly population. According to the latest epidemiological statistics, there are more than 2.7 million Parkinson's disease patients in my country, and more than 100,000 new patients are di...

AI Technical Summary

Problems solved by technology

[0005]According to the above-mentioned technical problems, a method for generating and predicting brain images of neurodegenerative diseases based on deep generative models is provided. The present invention is based on brain imaging data and aims at The course development of each subtype of neurodegenerative diseases is predicted, and the brain imaging data of the next period of disease course are generated through the brain imaging data of the previous period, which solves the problem that the traditional method cannot predict the natural course of the disease. Early diagnosis of disease and timely intervention and treatment provide scientific basis for effectiveness

Images