CN102579040A - Acupuncture point selecting method for acquiring brain waves - Google Patents

Abstract

The invention discloses an acupoint selection method for collecting brainwave signals, and finally obtains the correct collection channel. Obtain multiple brain acupoint signals through EEG equipment, use the method of phase space reconstruction to analyze the signals of these acupoints, then compare the embedding dimension of each acupoint signal, and select channels containing more information and relatively independent channels as For follow-up research objects, the information of other channels can be reconstructed from channels with a larger embedding dimension, which simplifies the complexity of brain wave acupoint signal acquisition. The invention enhances the real-time analysis of brain wave signals; its mathematical model is simple, its physical meaning is clear, and it is easy to realize.

Description

A point selection method for brain wave acquisition

technical field

The invention relates to the field of electronic medicine, in particular to a method for sampling brain wave signals.

Background technique

EEG is the electrophysiological activity of brain nerve cells on the surface of the cerebral cortex and scalp. It can effectively reflect the physiological and pathological conditions of the brain and the functional state of the brain. Different states of thinking and emotional changes can reflect different EEG signals in different positions of the cerebral cortex, so EEG signals contain a wealth of useful information. In biomedicine, EEG is used as an effective means of medical diagnosis and disease treatment; in cognitive research, EEG is an important tool for the origin of human thinking. It was the German psychiatrist Hans Berger who first discovered that humans have the characteristics of brain electrical activity. He published the first paper on human brain electricity in 1929.

Traditional methods mainly study EEG from the perspectives of time domain, frequency spectrum and statistics. Time-domain analysis is mainly to study the characteristics of EEG waveform, amplitude, phase, etc., and frequency-domain analysis method is to analyze the distribution of EEG in different frequency bands. They can describe some features of EEG, but their sensitivity is poor, and they cannot analyze the essential features of brain activity. In recent years, methods such as wavelet transform, neural network and nonlinear dynamics have also been applied to the analysis of EEG signals, which represent a new direction of EEG signal analysis.

Using chaos theory to study EEG has become a very important field, and the brain is considered as a nonlinear dynamical system. Divided from the purpose of research, there are two main areas: (1) To reveal the working mechanism of the brain, and to study the nonlinear dynamic characteristics of the human body in different physiological states and different brain function states. (2) To study the changes of nonlinear dynamics in the pathological state of the human body, and provide the basis for clinical analysis and diagnosis. Whitney's phase space reconstruction theory and Takens' delay reconstruction theorem provide the theoretical basis for reconstructing the phase space of a system from an experimental time series. Based on this theory, the analysis and research of nonlinear time series can describe the target system very effectively, which also makes it possible to study unknown complex systems.

When performing phase space reconstruction on a nonlinear sequence, the two main parameters that affect the reconstruction effect are delay time and embedding dimension. The general principle is to ensure that the sampling time sequence can reproduce the nonlinear characteristics of the original system to the greatest extent. , keep as much information about the system as possible. Nonlinear dynamics method is a new way to study EEG signals, which will play a huge role in deepening human understanding of the brain and disease detection, diagnosis and treatment.

"Applied Mathematics and Computation" (2009, 207(1): 63-74) discussed the nonlinear dynamics of EEG signals, and "IEEE Transactions on Biomedical Engineering" (2011, 58(4): 1084-1093) used The theory of phase space reconstruction studies the changes of brain signals when people are in different physical states. They only processed the collected EEG signals, but did not mention the collection of EEG signals. We believe that the collection of EEG signals is the most important part of EEG signal processing. If the signal can contain the largest amount of information, it will certainly facilitate subsequent processing. This is mainly because the brain is not a single node, and the signals collected in different parts of the brain are very different. How to select the optimal signal so that it can represent the state of the brain in the largest amount and contain the largest amount of information . The traditional method generally selects the signal of the acupoints on the head as the candidate signal, but it is unable to make an optimal selection for the collected acupoint signals.

Contents of the invention

In order to overcome the complexity caused by incomplete brain wave information and multi-channel measurement when selecting the brain to observe acupuncture points in the prior art, the present invention proposes a new method for brain wave observation of acupuncture points, which can ensure the integrity of brain wave information collection At the same time, it simplifies the complexity of brain wave acquisition and saves a lot of time for subsequent real-time brain wave analysis.

The technical solution adopted by the present invention to solve its technical problems comprises the following steps:

First, obtain the brain wave signal, that is, collect the brain wave signal of the human brain; take the 32 acupoints of the brain as the acquisition objects, respectively construct the acupoint channels, and obtain the brain wave signals of the 32 acupoint channels by setting the time length; The data of the acupoint channels are stored separately to obtain 32 arrays of equal length;

Then, use the phase space reconstruction theory to study each array as a separate system component, and determine the embedding dimension m and delay time τ of the system reconstruction; first use the FNN method ("Phys.Rev.A", 1992 , 45:3403-3411) to determine the embedding dimension m, and then use the first extreme point of the AD method ("Physica D", 1994:73:82-98) as the value of the delay time τ;

Secondly, by analyzing each array one by one, m ₁ ...m _i ..m ₃₂ can be obtained, where m _i represents the embedding dimension of the data of the i-th degree channel; research is carried out by comparing the embedding dimension Screening of channels: if m _max > 32 (m _max represents the maximum embedding dimension in all channels), increase the acquisition object, that is, increase the number of acupoints collected to make it equal to m _max , if m _max ≤ 32, select a point with m The channel of _max is the research object, while retaining the channel of m _i =1; the channel containing m _max is the channel containing the most information, and the channel of m _i =1 is the relatively independent channel of brain wave information;

Finally, obtain the brain wave reconstruction model information: take m _max as the embedding dimension, and use its corresponding τ as the delay time to reconstruct the system model:

X(0)=[x(0), x(τ), x(2τ), L, x((m-1)τ)]

X(1)=[x(1), x(τ+1), x(2τ+1), L, x((m-1)τ+1)]

...................................

X(n)=[x(n), x(n+τ), x(n+2τ), L, x(n+(m-1)τ)]

................................................... .........

X(m-1)=[x(m-1), x(m-1+τ), x(m-1+2τ), L, x((m-1)(τ+1))]

At this time, all the data of X(0)....X(m-1) comes from the channel with the largest embedding dimension m _max , but it can reflect the information of other channels. For example, X(0) represents 1 channel. information, X(1) indicates 2-channel information, and X(n) indicates n+1 channel information. Therefore, the channel containing m _max contains relevant information of all channels except the channel m _i =1, and the subsequent analysis only needs to collect the data of m _max and m=1 channel, and it is not necessary to collect data for all acupoints.

The beneficial effect of the present invention is that: the present invention starts from the collection of brain wave signals, utilizes the theory of phase space reconstruction, and compares the embedded dimensions of each channel signal reconstruction, and selects the channel containing the most information and the information Relatively independent channels, according to the phase space reconstruction theory, can restore the information of other channels from the channel containing the most information. At the same time, the present invention proposes a joint algorithm for phase space reconstruction, which greatly improves the speed of reconstruction while ensuring the calculation accuracy, and provides a guarantee for the real-time analysis of brain wave signals; the mathematical model of the present invention is simple, and the physical meaning is clear. It is easy to implement and the calculation effect is accurate.

The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

Description of drawings

Figure 1. The initially selected acupoints for brain wave observation;

Fig. 2 is a flow chart of the phase space reconstruction brain wave analysis operation according to the present invention;

Detailed ways

In order to find out the quantitative basis, the present invention firstly collects a large amount of acupuncture point information, then proposes a standard to measure the size of each acupuncture point information, and finally selects a single channel signal for processing.

This embodiment is based on the operation flow shown in FIG. 2 and a simple implementation scheme based on the acquisition of human brain electric wave signals shown in FIG. 1 .

Firstly, the signals of the initially selected 32 acupoints can be collected by the EEG signal acquisition equipment, and the obtained data can be transmitted to the PC through the USB data cable, and the type of the transmitted data can be converted in the PC, and stored according to the acquisition channel {x (t _i )}.

Then compare the embedding dimension m value of each channel to obtain the maximum embedding dimension value m _max , if m _max > 32, increase the collection object, that is, increase the number of acupoints collected to make it equal to m _max , if m _max ≤ 32, Then the channel corresponding to m _max contains the largest amount of information, and select this channel as the final preferred channel, while retaining the channel of m=1; m _max is the channel containing the most information, and m=1 is the relatively independent channel of brain wave information ;

Finally, select the channel with the largest embedding dimension m _max as the research object, while retaining the channel with embedding dimension m=1; because m _max is the channel that contains the most information, and m=1 is a relatively independent channel of brain wave information

The present invention finally obtains suitable acupoints for human brain signal acquisition, simplifies the complexity of signal acquisition through phase space reconstruction theory, enhances the effect of real-time analysis of brain wave signals, and can accurately reconstruct the output of the entire brain signal , the mathematical model of system reconstruction is simple, the physical meaning is clear, easy to implement, and the effect is accurate.

Claims (1)

1. an acupuncture point selection method for brain wave collection, is characterized in that comprising the steps: First, take 32 acupoints of the brain as the acquisition objects, construct acupoint channels respectively, and obtain the brain wave signals of the 32 acupoint channels by setting the time length; store the data of the 32 acupoint channels separately, and obtain 32 equal-length array; Then, use the phase space reconstruction theory to study each array as a separate system component, use the FNN method to determine the embedding dimension m, and use the first extreme point of the AD method as the value of the delay time τ; Secondly, analyze each array one by one to get m ₁ ...m _i ..m ₃₂ , where m _i represents the embedding dimension of the data of the i-th degree channel; filter the research channel by comparing the embedding dimension : If m _max > 32, m _max represents the maximum embedding dimension in all channels, increase the number of collected acupuncture points to make it equal to m _max , if m _max ≤ 32, select the channel with m _max as the research object, and keep The channel of m _i =1; the channel containing m _max is the channel that contains the most information, and the channel of mi ₌ 1 is a relatively independent channel of brain wave information; Finally, obtain the brain wave reconstruction model information: take m _max as the embedding dimension, and use its corresponding τ as the delay time to reconstruct the system model: X(0)=[x(0), x(τ), x(2τ), L, x((m-1)τ)] X(1)=[x(1), x(τ+1), x(2τ+1), L, x((m-1)τ+1)] ................................... X(n)=[x(n), x(n+τ), x(n+2τ), L, x(n+(m-1)τ)] ................................................... ......... X(m-1)=[x(m-1), x(m-1+τ), x(m-1+2τ), L, x((m-1)(τ+1))] Subsequent analysis only collected m _max and m = 1 channel data.

Images