A Brain Temporal Signal Processing Method Based on Hidden Markov Model
A hidden Markov, time signal technology, applied in the field of medical signal processing, can solve problems such as sparse connections in brain regions that cannot reflect, multiple false connections in the brain, and multiple classification features.
Active Publication Date: 2022-02-22
XI AN JIAOTONG UNIV
View PDF7 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0004] 1. The fully connected brain network based on Pearson correlation or partial correlation will lead to more false connections in the brain, which cannot reflect the objective fact of sparse connections in brain regions in brain neural activity
[0005] 2. Based on l 1 Norm regularization to build a sparse network of brain connections improves the correlation-based brain network structure to a certain extent, but due to l 1 The coefficient of the norm regularization term is calculated from the data of each individual according to mathematical criteria (cross-validation, Bayesian information criterion, etc.), which leads to inconsistent sparsity of individual brain networks and unstable brain network structures
[0006] 3. Existing analysis methods only use time series to solve the relationship between the two brain regions, resulting in a waste of information on the time scale
[0007] When using the features obtained by existing analysis methods to classify fMRI images, there are often too many classification features, resulting in serious over-fitting of the classifier.
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment
[0070] In this embodiment, functional magnetic resonance data is taken as an example, and the processing method of other brain temporal signals such as EEG is the same.
[0071] Step 1: Follow as in figure 1 fMRI data were preprocessed in the manner shown;
[0072] Step 2: Use the AAL90 template to extract the time series of 90 brain regions;
[0073] Step 3: The signal length of each subject is T, and the time series of each brain region of each subject is centered and standardized. x(t) represents the time series of any brain region of any subject;
[0074] Step 4: Take the time series of the unified brain regions of all healthy subjects as a set, and train the hidden Markov model λ=(A, B, π) of the corresponding brain regions. After using the EM algorithm to solve the problem, the specific calculation method of the model parameters is as follows :
[0075] A=[a ij ]10x10 represents the hidden activation state transition matrix of the brain, assuming that the activatio...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The invention discloses a brain time signal processing method based on a hidden Markov model, comprising the steps of: 1) preprocessing the collected functional magnetic resonance images, and requiring the collected functional magnetic resonance images to have the same echo time ; 2) Centralize and standardize the time series of each brain region after preprocessing, and use healthy subjects to train the hidden Markov model of each brain region separately; 3) According to the hidden Markov model sought Solve the likelihood value of each brain region sequence of each subject with parameters, and scale according to the length of time series of different subjects to obtain a feature of each brain region of each subject; 4) Use step 3) to get each After testing the characteristics of a brain region, the SVM-RFE method was used to classify the two groups of subjects. Based on the analysis method proposed by the present invention, the accuracy rate of classifying functional magnetic resonance data of autism spectrum disorder has reached 80.37%. Compared with the construction of brain network, deep learning and other methods have greatly improved.
Description
technical field [0001] The invention belongs to the technical field of medical signal processing, in particular to a brain time signal processing method based on a hidden Markov model. Background technique [0002] Functional magnetic resonance technology measures the hemodynamic changes caused by neuron activity according to the difference between paramagnetism and diamagnetism exhibited by deoxygenated hemoglobin and oxygenated hemoglobin under high-intensity magnetic field. Functional magnetic resonance technology has become the mainstream technology in brain science research due to its advantages of non-intrusion, non-injury, high spatial resolution and high time resolution, among which resting state functional magnetic resonance images are the most important. Become the basic signal of brain science research. Resting-state functional magnetic resonance (rest-fMRI) can reflect the low-frequency fluctuating signals generated by the brain's spontaneous neural activity in ...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More Patent Type & Authority Patents(China)
IPC IPC(8): G06K9/00G06K9/62G06T7/30
CPCG06T7/30G06T2207/30016G06V10/267G06F2218/12G06F18/2411G06F18/2415
Inventor 刘天范庚杨明陈宇豪
Owner XI AN JIAOTONG UNIV