Brain state prediction model training device, prediction device and electronic equipment

Abstract

The invention provides a brain state prediction model training device, a prediction device and electronic equipment. The training device comprises a construction module which is used for constructing a training data set, wherein the training data set comprises multiple pieces of sample data, and each piece of sample data comprises a sample brain image, a real brain age and a real brain gender; a calculation module, which is used for extracting abstract features of the sample brain image through a convolutional neural network, calculating to obtain an output brain gender probability and an output brain age probability in a plurality of preset brain age groups, calculating the similarity between the output brain age probability and a real brain age probability and the similarity between the output brain gender probability and a real brain gender, and outputting the similarity between the output brain gender probability and the real brain age group, obtaining a first calculation result and a second calculation result; a training module, which is used for optimizing network parameters of the convolutional neural network according to the first calculation result and the second calculation result to obtain a brain state prediction model; a brain state prediction model, which predicts brain age prediction and brain gender at the same time, with a higher precision of the brain age prediction precision.

Description

technical field [0001] The present application relates to the field of brain state prediction, and in particular, relates to a brain state prediction model training device, prediction device and electronic equipment. Background technique [0002] With the development of machine learning, more and more recognition models for the attributes shown in images have emerged. After learning, these image recognition models can process, analyze and understand images to identify targets and objects in various patterns. Aiming at the attribute of the image, a relatively objective prediction result for the attribute shown in the image is obtained. The human brain will undergo some subtle changes in structure with age. The image recognition model can predict the brain age of the brain based on these changes in the brain images of the brain, thereby quantifying the aging degree of the patient's brain , describes a normal human brain aging trajectory, combined with its real age, it can ref...

AI Technical Summary

Problems solved by technology

[0003] At this stage, there is a technology that regards age recognition as a regression problem, that is, input the age label of a real person into the neural network, and return the true value through iterative training. This method calculates two points For brain imaging, due to the small number of sample data and the large age span, the number of sample data specific to each age group is insufficient. Therefore, the accuracy of the trained image recognition model poor rate

Images