FMRI visual reconstruction method based on discrete representation and conditional autoregression

Abstract

The invention provides an fMRI visual reconstruction method based on discrete representation and conditional autoregression. The method comprises the following steps: step 1, constructing an image discrete representation model and an image reconstruction model; step 2, creating an fMRI visual reconstruction data set, wherein the data set comprises a training set and a test set, and the training set comprises a first set of images and a first set of fMRI brain signals after a subject sees stimulation of the first set of images; wherein the test set comprises a second set of images and a secondset of fMRI brain signals after a subject sees stimulation of the second set of images; 3, constructing a conditional autoregression model from the fMRI brain signal to image discrete representation according to the training set and the image discrete representation model; and 4, obtaining a visual reconstruction result of the second set of fMRI brain signals according to the conditional autoregression model and the image reconstruction model. The method is mainly used for reconstruction of complex natural images.

Description

technical field [0001] The invention relates to the technical field of visual reconstruction, in particular to an fMRI visual reconstruction method based on discrete representation and conditional autoregressive. Background technique [0002] fMRI-based visual reconstruction is the key technology of fMRI-based visual information decoding, which aims to reconstruct image stimuli according to the neural information of the cerebral cortex, which is the most difficult problem in the visual decoding of the brain. The reasons for the difficulty of visual reconstruction mainly include noise, high dimensionality, and lack of information. In the fMRI data acquisition process, various noises such as machines and subjects are inevitably introduced; the number of voxels in the cerebral cortex is large; due to the invariance of the visual information processing process in the brain, the visual information contained in the voxels is missing, resulting in brain One-to-many problem of sign...

AI Technical Summary

Problems solved by technology

[0017] Aiming at the problem that the matching degree between the image features extracted by the existing method and the voxel response is still not high, the present invention provides a fMRI visual reconstruction method based on discrete representation and conditional autoregressive, mainly for the reconstruction of complex natural images, through discrete Characterize the compressed image content, realize the correlation between fMRI brain signal and discrete features through conditional autoregression, and then realize image reconstruction

Images