CN112244765B - Method, device and system for detecting brain temporary abnormal state - Google Patents

Abstract

The invention provides a method for detecting transient abnormal state of brainMethods, apparatus, and systems. By adopting short-time high-dimensional signals to form the observation space of the brain dynamic system, the requirement of long enough time for observing the signals can be avoided, namely, the signal acquisition time is shortened, the detection efficiency is improved, and the brain dynamic system is convenient to use in actual combat. In addition, the method, apparatus and system of the present invention identify key global features by evolving trajectories from the brain dynamics system. Because the global features are different from the local features, the global features are not easily influenced by external input or noise, so that key feature information of brain temporary abnormal states can be deduced from complicated non-steady dynamic noise responses. And uses the emergent light intensity signal and HbO respectively₂ The signal constructs two state spaces, and the high-dimensional feature set is formed by fusing the two types of features, so that the accuracy of judgment is improved. Therefore, the abnormal personnel checking operation can be realized conveniently, quickly and accurately.

Description

Method, device and system for detecting brain temporary abnormal state

The present application is a divisional application of the invention patent application with application number 201910955286.1, application date 2019, 10 and 9, and the invention name "a method, device and system for detecting transient abnormal state of brain".

Technical Field

The invention relates to a brain detection system and method, in particular to a detection method, device and system for temporary abnormal state of brain.

Background

At present, the main problems of the drug administration control and drug administration inspection and other drug administration work are the hysteresis of the discovery. Young people are usually not noticed by families after a plurality of drug addiction, and the best dry expectation before addiction is very easy to miss. The toxic driving is basically detected in the accident cause investigation process after the severe traffic accident is caused, and the tragedy can not be avoided in advance due to the toxin absorption of a driver. Therefore, the most effective measure for solving the risk hazard caused by the drug solution problem and guaranteeing the health and welfare of people is to realize the prepositive prevention of drug absorption induction cases or timely discovery and intervention of drug absorption behaviors. However, the existing drug inhalation detection means cannot meet the requirements of pre-preventive prevention or timely detection of intervention.

In the prior art, the hair inspection and blood inspection are the most accurate technical means, but precise detection equipment is needed, the equipment is expensive and certain operation expertise is needed, so that the public security base layer of the first line in China is difficult to be widely adopted, and the detection is usually completed by professionals in judicial identification centers of various places. "urine test plate" and "saliva plate" are used to detect drugs or metabolites thereof in body fluids (urine and saliva) to achieve a drug-taking test that is relatively simple to handle compared to hair test and blood test. However, since urine inspection requires privacy and requires female staff to check females, the basic unit cannot ensure that female staff is always on duty, so that the operability of the urine inspection on roadside or public places is not strong. Meanwhile, the collection of body fluid (urine and saliva) easily causes the spread of diseases, most people who inhale the drug have various infectious diseases (such as liver diseases, AIDS, syphilis and the like), and first-line policemen or traffic policemen have a good safety protection measure when on-site duty, so that the health and safety of the on-duty policemen are seriously threatened. In addition, the use of "urine test plates" and "saliva plates" results in a large amount of disposable consumption, and the test time is as long as about five minutes.

The Functional near infrared spectrum imaging technology (fNIRS) is used as a noninvasive optical brain imaging technology, has the characteristics of noninvasive, portability, low cost, long-term continuous monitoring, no limitation on applicable people and the like, and can be used for carrying out real-time monitoring on brain Functional activities in natural environment. The fNIRS technology mainly utilizes different optical characteristics of biological tissues in an optical window with the near infrared wavelength of 650 nm-950 nm to realize the detection of brain activity by measuring the incident light intensity and the emergent light intensity. After absorption and scattering of the scalp, skull, cerebrospinal fluid and cerebral cortex, the light emitted by the light source can be detected by a detector at a certain distance from the light source, and the track of the near infrared light transmitted in the tissue is in a banana shape.

The traditional near infrared light technology assumes that the blood oxygen state of the cerebral cortex is related to local neuron activity, so that the dynamic change of the blood flow of the local area of the brain can be observed by measuring the change of the hemoglobin concentration of brain tissue, and the neural activity of the brain can be indirectly obtained. During the scattering propagation of near infrared light in tissue, photons are absorbed to cause light attenuation, and fNIRS uses this property to calculate the change in blood oxygen.

However, the conventional fNIRS technique requires a relatively long observation signal time based on the change of blood oxygen state or concentration of the cerebral cortex to obtain the neuron activity connection (function connection amount) between the parts of the brain, and is easily affected by external input or noise due to the need of finely acquiring the local features of the brain for a long time, which is inefficient and has high detection cost. The method is not beneficial to quick and accurate investigation of scenes such as highways, station cards and the like.

Disclosure of Invention

In order to solve at least one of the problems, the invention provides a method, a device and a system for detecting temporary abnormal state of a brain.

The invention provides a method for detecting transient abnormal state of brain, which comprises the following steps:

step A100: constructing a first type of phase space of the brain of the subject based on the emergent light intensity signal;

step A200: extracting a first type of phase space features of the brain of the subject to obtain a first feature quantity set;

step B100: based on the subject brain high-dimensional HbO₂ Constructing a second phase space of the brain dynamic system by signals;

step B200: extracting the features of the second phase space based on a common mode space mode to obtain a second feature quantity set;

step C: integrating the first characteristic quantity set and the second characteristic quantity set to form a characteristic set with higher dimension;

step D: determining an active mode of the brain dynamic system based at least in part on the feature set;

step E: determining whether the subject is abnormal based on the determination result.

Wherein, the step a100 includes:

step A110: acquiring a first observation space of a brain dynamic system of a subject based on the emergent light intensity signal;

step A120: reconstructing a first phase space of the brain dynamic system based on the first observation space, resulting in a first eigenvalue construction.

Wherein, the step a200 includes:

step a210: identifying a first evolution track of the brain dynamic system based on the first intrinsic component structure, obtaining a first fundamental component v1 (t) and a first residual force r1 (t);

step A220: a first feature quantity set is obtained based on the first basic component v1 (t) and a first residual force r1 (t).

The second phase space construction method comprises the following steps:

step B110, collecting light waves with two wavelengths radiated by each light source in the probe, and calculating to obtain HbO through the modified Lambert-Beer law₂ A signal;

step B120, hbO obtained by the low-pass filter pair₂ Filtering the signal to remove noise;

step B130, hbO after the filtering treatment₂ The signals generate a brain phase space matrix.

The proposal uses the emergent light intensity signal and HbO simultaneously and respectively₂ The signal constructs two state spaces, and by fusing the two types of features, a high-dimensional feature set can be formed, so that the accuracy of judgment is improved.

The invention also provides a device for detecting the temporary abnormal state of the brain, which comprises a host machine and a probe,

the host comprises a processor, a memory and a communication interface;

the memory is used for storing at least codes, data and results;

the communication interface is used for communicating with the probe;

the probe collects outgoing light intensity signals of the corresponding area of the brain of the subject and transmits the outgoing light intensity signals to the host;

wherein the processor is used for executing the detection method of the brain temporary abnormal state so as to identify whether the brain is under the influence of medicines or not, and feeding back the result.

Further, the host computer also comprises a display screen for displaying the detection result.

Further, the probe comprises a plurality of light sources and a plurality of detectors, wherein the plurality of light sources can radiate at least one light wavelength, and the detection spectral range of the detectors covers the radiation wavelength of the light sources.

Furthermore, the plurality of detectors can receive the emergent light intensity signals of the light rays emitted by the same light source through the brain so as to form a plurality of detection signal channels.

Further, the probe also comprises a gyroscope and/or an accelerometer sensor, which are used for detecting the posture change or the shake of the probe and removing the motion artifact in the emergent light intensity signal output by the detector.

The invention also provides a brain temporary abnormal state detection system, which comprises a remote data management platform and a plurality of brain temporary abnormal state detection devices as described above, wherein the detection devices comprise,

communication circuitry for wirelessly communicating with the data management platform.

Further, the detection device transmits the detection result to the data management platform in real time through the communication circuit, or,

the detection device carries out local storage on the detection result, transmits the detection result to a data management platform at fixed time or when the network is unobstructed,

the data management platform comprises a database, and the received detection result is stored in the database.

Further, the detection device further comprises an identification unit, and the identification unit at least partially comprises:

the identity card reading unit is used for supporting the identity card information reading and carrying out the networking inquiry and comparison function with the data management platform, and/or

Fingerprint acquisition unit for supporting fingerprint acquisition and input and performing networking query comparison function with data management platform, and/or

The camera unit is used for supporting the real-time face recognition comparison of field photographing or networking comparison function with the data management platform.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below. It will be apparent to those of ordinary skill in the art that the drawings in the following description are exemplary only and that other implementations can be obtained from the extensions of the drawings provided without inventive effort.

The structures, proportions, sizes, etc. shown in the present specification are shown only for the purposes of illustration and description, and are not intended to limit the scope of the invention, which is defined by the claims, so that any structural modifications, changes in proportions, or adjustments of sizes, which do not affect the efficacy or the achievement of the present invention, should fall within the ambit of the technical disclosure.

FIG. 1 is a schematic diagram of dopamine transmission paths;

FIG. 2 is a schematic diagram of a method for detecting transient abnormalities in the brain;

FIG. 3 is a schematic diagram of acquiring an observation space of a brain dynamic system;

FIG. 4 is a schematic diagram of a brain phase space matrix;

FIG. 5 is a schematic diagram of the normal living dynamic evolution track of the brain;

FIG. 6 is a schematic diagram of a transient abnormal state evolution trace of the brain;

FIG. 7 is a schematic diagram of determining an active mode of a brain dynamic system;

FIG. 8 is a schematic diagram of a method of constructing a classification model;

FIG. 9 is a schematic diagram of a constructed classification model;

FIG. 10 is a schematic diagram showing a method for detecting transient abnormal states of the brain;

FIG. 11 is a diagram showing an example of the result of applying the method shown in FIG. 10;

FIG. 12 is a schematic diagram showing a usage mode of the brain transient abnormality detection device;

FIG. 13 is a schematic diagram of a device for detecting transient abnormal states of the brain;

FIG. 14 is a schematic view of an arrangement of a light source and a photodetector in a probe;

FIG. 15 is a schematic view of another arrangement of a light source and a photodetector in a probe;

fig. 16 is a schematic diagram of a system for detecting transient abnormal states of the brain.

Detailed Description

Other advantages and advantages of the present invention will become apparent to those skilled in the art from the following detailed description, which, by way of illustration, is to be read in connection with certain specific embodiments, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention. It should be noted that, without conflict, the embodiments of the present invention and features of the embodiments may be combined with each other. In order that the above objects, features and advantages of the invention will be readily understood, a more particular description of the invention will be rendered by reference to specific embodiments thereof which are illustrated in the appended drawings.

The dynamic brain system has dynamic behaviors such as chaos, bifurcation, singular attractors and the like, and can be described by a highly nonlinear dynamic system. I.e. a non-resting steady state (active stationary state) that the brain system can maintain by its own dynamics without receiving external input, two basic conditions are required for this:

1) The neurons are connected with each other in an excitatory mode, and the network can maintain stable activity by positive feedback among the neurons under the condition that no external input exists.

2) The system also has a suppressive connection to avoid "explosion" of system activity due to repeated positive feedback. The brain dynamic system is balanced in the proportion of excitation and inhibition under normal dynamic conditions, and the brain maintains a non-resting steady state by self dynamics without external input.

Under the action of some drugs affecting the brain, the above-mentioned stable state is broken, and after the brain is in abnormal excited state and lasts for about ten or more hours, the brain is restored to normal activity. Abnormal excited states of the brain under the influence of drugs are called "transient abnormal states of the brain". Such stimulants include, but are not limited to, amphetamines, mestrane, cocaine, methamphetamine, amphetamine-type stimulants, and the like. As shown in fig. 1, the human body produces an excessive amount of Dopamine (Dopamine) excitatory neurotransmitters in the ventral midbrain tegmental area (ventral tagmental area, VTA) after ingestion of the excitatory drug. The two main transmission pathways of dopamine are the mesenteric system channel, i.e. from the VTA to the nucleus accumbens (nucleus accumbens), and the mesocortical channel, from the VTA to the frontal lobe, respectively. Dopamine promotes the excitatory connection of neurons of the forehead leaves and nucleus accumbens to proliferate, and breaks the balance of the excitation and inhibition proportion, so that the dynamics system of the brain is separated from an unsteady stable state to form an abnormal high excitation activity state, and the average half-life of the effect is about ten or more hours.

Based on this, as shown in fig. 2, a method for detecting transient abnormal state of the brain according to the present invention is specifically as follows:

step S100: acquisition of the observation space of a brain dynamic system

The brain may be considered as a dynamic system, and the detected intensity signal of the emitted light at each moment in time or each spatial position reflects a specific state in the brain dynamic system, being a point in the phase space of the brain dynamic system. By acquiring the emergent light intensity signals of different space positions, the observation space of the brain dynamic system can be constructed.

As shown in fig. 3, acquiring an observation space of a brain dynamic system includes the steps of:

step S110 of collecting short-time high-dimensional signals

A multi-path detection channel is formed by theprobe 100 including a plurality of near infraredlight sources 110 andphotodetectors 120. After absorption and scattering of the light emitted from the near infraredlight source 110 through the scalp, skull, cerebrospinal fluid and cerebral cortex, the emitted light can be detected by thephotodetector 120 at a distance from thelight source 110. The light intensity variation of the outgoing light reflects the brain activity state of the corresponding spatial portion.

And in a shorter acquisition time (for example, within 10 seconds), the multi-channel detection channels acquire the emergent light intensity signals at the same time, namely, the multi-channel short-time emergent light intensity signals are obtained. The multiplexed short-time output intensity signal is referred to as a short-time high-dimensional signal.

Step S120 short-time high-dimensional signal preprocessing

The preprocessing may include normalization and filtering of the short-time high-dimensional signals. Will emit light intensity signal I_k Normalization can be performed by subtracting the mean (I_k ) Dividing variance sigma (I)_k ) Realizing the method. The filtering may employ a filter to filter out low and high frequency noise. For example, three-order Butterworth filters are used, and the low-pass and high-pass filter cut-off frequencies are set to 0.01Hz and 0.8Hz, respectively.

The path of the near-infrared light propagating in the tissue is "banana-shaped" with a penetration depth corresponding to 1/4 of the distance of thephotodetector 120 from thelight source 110. Thus, by different distances of thephotodetector 120 from thelight source 110, outgoing light intensity signals reflected at different depths can be obtained. When the distance between thephotodetector 120 and thelight source 110 is short, only the emergent light intensity signal of the brain tissue in a shallow layer can be obtained, and only when the distance between thephotodetector 120 and thelight source 110 is long, the emergent light intensity signal of the brain tissue in a deep layer can be obtained. However, photons must pass through the shallow layer and then reach the deep layer, so that shallow layer signals are superimposed in the deep layer signals, and whole body physiological activities such as respiration, heart activity, low-frequency oscillation and the like have influence on the shallow layer and the deep layer signals. Therefore, the shallow signal is used as a reference signal and the deep signal is used as a detection signal. For example, for the samelight source 110, the outgoing light signal detected by thephotodetector 120 closest to thislight source 110 is used as a reference signal, and the outgoing light signals detected by the remainingphotodetectors 120 that receive the outgoing light of thislight source 110 are used as detection signals.

Thus, in some embodiments, the preprocessing is performed by fitting noise reduction. Fitting noise reduction is a fitting signal obtained by fitting each detection signal (deep signal) by the reference signal (shallow signal) corresponding to the samelight source 110, and then subtracting the reference signal corresponding to the samelight source 110 from each detection signal. Thus, superficial disturbances and system level disturbances (e.g., respiratory, cardiac activity and low frequency oscillations, etc. whole body physiological activities) can be eliminated.

For example, the detection signal Is represented by Is, the reference signal Is represented by Ir, I_f Representing the fitted signal. The fitting noise reduction steps are as follows:

step S121: obtaining I by linear fitting_f ，I_f (t) =a×ir (t) +b. where t represents a discrete-time index. By minimizing the fitting signal I at each time point over the whole time interval_f And the sum of absolute differences between the detection signals Is to determine the parameters a and b, i.e

min_a，b (∑[abs(I_s (t)-I_f (t))])

Step S122: fitting the noise-reduced emergent light intensity signal to obtain a fitting signal I obtained by subtracting the reference signal corresponding to the samelight source 110 from the detection signal Is_f I.e.

I(t)＝I_s (t)-I_f (t)

In some embodiments, each of the short-time high-dimensional signals may be low-pass filtered to remove physiological noise. For example: respiratory (0.2-0.3 Hz), heartbeat signal (0.8-1.2 Hz) and physiological noise generated by Mayer wave (about 0.1 Hz) of blood pressure oscillation.

In some embodiments, a third order Chebyshev type II filter is used, with a cut-off frequency of 0.1Hz, a stop band frequency of 0.5Hz, a passband loss of no more than 6dB and a stop band attenuation of at least 50dB.

In some embodiments, a wavelet filter, such as Daubechies-12 wavelet, is used for ten-level decomposition.

In other embodiments, a third order Butterworth filter is used, with cut-off frequencies taken at 0.8Hz and 0.01Hz.

Step S130 utilizes the preprocessed short-time high-dimensional signals to form the observation space of the brain dynamic system

As shown in fig. 4, the observation space I composed of the short-time high-dimensional signal after the preprocessing can be expressed as:

wherein I is_j Representing the j-th path of emergent light intensity signals in the short-time high-dimensional signals, wherein j is more than or equal to 1 and less than or equal to k, and the short-time high-dimensional signals comprise k paths of emergent light intensity signals in total. m represents the total sampling point number of the emergent light intensity signal is m. Thereby, the normal phase space matrix I can be generated under the normal activity state of the brain, namely the state not influenced by drugs_N And brain transient abnormal state, i.e. transient abnormal state affected by drug, abnormal phase space matrix I_A 。

Step S200: reconstructing phase space of brain dynamic system

It is assumed that the brain dynamic phase space is created by a hidden nonlinear system with dynamic states. There is a dynamic phase space M of a brain dynamic system constructed from an observation space I_v (t) is capable of approximating the original nonlinear dynamics.

Singular value decomposition (singular value decomposition, SVD) is performed on the observation space I of the brain dynamic system, i.e. the following formula:

I＝U∑V^*

dynamic phase space of brain dynamic system

It can be constructed from the first r eigenvectors of the right singular matrix V. For example: r is taken as 3, the kinetic phase space +.>

Step S300: identification of evolution trajectories of brain dynamics systems

In order to identify patterns of overall evolution of the brain's dynamic system, linearization is applied to nonlinear dynamics. Using linear analytical analysis (linear resolvent analysis, LRA), i.e.

Where A and B are regression coefficients of the linear dynamics of the fundamental component v (t) and the residual force r (t), respectively.

Fig. 5 and 6 show an example of a normal activity state evolution trace of a brain without the action of a drug and an example of an abnormal activity state evolution trace of a brain with the action of a drug, respectively. It can be seen that there is a significant difference between the two, and the resulting evolution track can be divided into two regions, namely askirt loop region 31 and acentral disk region 32. The normal active trajectories are mainly concentrated in theskirt loop region 31, while the temporary abnormal trajectories are mainly concentrated in thecentral disk region 32.

Step S400: determining an active mode of a brain dynamic system

Active modes are classified into normal active states and temporary abnormal states. As shown in fig. 7, determining the active mode of the brain dynamic system includes the steps of:

step S410: obtaining characteristic quantities

Using the basic component v (t) and the residual force r (t), three feature quantities, for example, can be obtained: the first feature quantity is v (t) itself; the second feature quantity is r (t) itself; the third feature quantity is the Hilbert transform of the sum of r (t) and v (t), i.e., H [ v (t) +r (t) ].

In some embodiments, a common mode spatial mode (Common spatial pattern filtering, CSP) may be employed to extract features of brain phase space. The objective of CPS is to maximize the differentiation of brain normal activity dynamics (corresponding to normal phase space matrix I_N ) And temporary abnormal state (corresponding to abnormal empty phaseInter matrix I_A )。

Assuming Co v_N And Co v_A Respectively normal phase space matrix I_N And an anomaly phase space matrix I_A Is a mean covariance matrix of (b). CSP filters can be obtained by Rayleigh quotient, i.e

Where w represents the spatial filter and T represents the transpose.

Where NN represents the total number of data of the normal class and NA represents the total number of data of the abnormal class.

A series of spatial filters w= [ W ]₁ ,w₂ ,…,w_n ]Can be obtained by solving eigenvalue problems

Coν_N w＝λCoν_A w

Lambda represents eigenvalues and is arranged in descending order. The first three and the last three eigenvalues may be selected and the corresponding eigenvectors represent two classes of spatial filters.

The estimated source signal may be expressed as:

Y＝W^T I

the mean, slope and variance of the signal over the detection time window are calculated and characterized.

In other embodiments, principal component analysis (Principle component analysis, PCA), independent component analysis (Independent component analysis, ICA), or common average reference (Common average referencing, CAR) may also be used to extract features of brain phase space.

Step S420: and determining the active mode of the brain dynamic system according to the feature quantity set and the classification model.

The classification model is trained by a large amount of data in advance, and newly built measured data can be added into a sample library continuously, a training set is updated, and the classification model is optimized. The method for constructing the classification model is shown in fig. 8:

step a): and obtaining a sample of the normal living dynamic evolution track of the brain. By collecting the emergent light intensity signals of the persons who do not ingest drugs, the evolution track of the brain dynamic system is obtained for each person according to the steps S100 to S300, and is used as a sample of the normal living dynamic evolution track of the brain.

Step b): and obtaining a sample of the transient abnormal state evolution track of the brain. By collecting the emergent light intensity signals of the personnel taking the drugs, according to the steps S100 to S300, the evolution track of the brain dynamic system of each personnel involved in the drugs is obtained respectively and is used as a sample of the transient abnormal state evolution track of the brain.

Step c): a classification model of the active mode of the brain dynamic system is established.

By linear decision analysis (Linear discriminant analysis, LDA), decision hyperplane is determined, dividing the data points into different categories with maximum spacing. Where Linear Decision Analysis (LDA) is a generalization of fischer linear discrimination methods, using statistical, pattern recognition and machine learning methods, attempts to find a linear combination of features of two classes of objects or events to be able to characterize or distinguish them. The resulting combination may be used as a linear classifier or may be used to perform a dimension reduction process for subsequent classification. LDA works effectively when the measurement is continuous for each observation of the argument. The following describes a specific calculation method of LDA:

consider a set of observations of each object or event in a known class y

(also known as a feature, attribute, variable, or measurement). This set of samples is known as a training set. The problem of classification is that only one observation is given +.>

A good predictor is found for class y and any samples with the same distribution (not necessarily from the training set) are judged to be correct.

LDA is achieved by assuming a conditional probability density function

And->

Are all normal distributions with mean and covariance, respectively +.>

And->

Based on this assumption, the bayesian best solution is to consider a predicted point as belonging to the second class if its logarithm of the likelihood ratio is below a certain threshold T, calculated according to the following formula:

the LDA makes the additional simplifying variance Ji Xing assumption (i.e., the covariance between the different classes is the same, so Σ₀ ＝∑₁ = Σ, and covariance is full rank. In this case, some items may be eliminated:

because of sigma_i Is a Hermite matrix

Thus, the above criterion becomes a threshold for determining a dot product

For some threshold constant c, when

This means that one input

The criteria belonging to class y are purely a function of a linear combination of known observations.

From a geometric point of view: judging an input

Whether it is the class y standard is a point on the multidimensional space +.>

Projection to vector +.>

(we only consider its direction) function. In other words, if corresponding ++>

Is positioned at right angles to->

On one side of the hyperplane of (c), then the observations fall into class y. The position of the plane is determined by the threshold c. From this, a classification model diagram as shown in fig. 9 can be obtained.

In other embodiments, support vector machines (Support vector machine, SVM), deep learning (ANN), etc. may also be used.

Fig. 9 shows classification models obtained by the above method, in which abnormal type measured data are obtained from 18 persons and normal type measured data are obtained from 14 persons. Wherein the thick dashed line represents the interface.

Wherein the normal class data all fall into the normal class. The 15 abnormal measured data fall into abnormal classification, and 3 abnormal measured data fall out of two classes, so that the abnormal measured data become uncertain points which cannot be correctly classified.

Referring to fig. 7 again, the active mode of the brain dynamic system is determined according to the feature quantity set and the classification model, that is, the obtained feature quantity set is input into the classification model, that is, whether the current test is normal active state or temporary abnormal state can be determined according to the projection space position.

S500: determining whether the subject is abnormal based on the determination result and outputting a result.

Outputting an indication that the subject is currently normal when it is determined that the subject is in normal activity; outputting an indication of its current abnormality when it is determined that the subject is in temporary abnormal state; when the subject is determined to be in an uncertain state, a risk cue is output so that it can be further tested to exclude the possibility of misuse of the drug to cause mental state abnormalities.

The method of the invention adopts short-time high-dimensional signals to form the observation space of the brain dynamic system, thereby avoiding the requirement of long enough time for observing signals, namely shortening the signal acquisition time, improving the detection efficiency and being convenient for practical use. In addition, the method of the present invention identifies key global features by evolving trajectories from the brain dynamics system. Because the global features are different from the local features, the global features are not easily influenced by external input or noise, so that key feature information of brain temporary abnormal states can be deduced from complicated non-steady dynamic noise responses.

Fig. 10 shows another example of a method for detecting transient abnormal states of the brain.

As shown in fig. 10, the detection method is as follows:

step A100: and constructing a first type of phase space of the brain of the subject based on the emergent light intensity signals. The method specifically comprises the following steps:

step A110: acquiring a first observation space of a brain dynamic system of a subject based on the emergent light intensity signal;

step A120: reconstructing a first phase space of the brain dynamic system based on the first observation space, resulting in a first eigenvalue construction;

step A200: and extracting the first type of phase space characteristics of the brain of the subject. The method specifically comprises the following steps:

step a210: identifying a first evolutionary trajectory of the brain dynamic system based on the first eigenvector construct to obtain a first principal component v₁ (t) and a first residual force r₁ (t)；

Step A220: based on the first basic component v₁ (t) and a first residual force r₁ (t) obtaining a first feature quantity set;

step B100: based on the subject brain high-dimensional HbO₂ The signals construct a second phase space of the brain dynamic system. Wherein Hb is hemoglobin, hbO₂ Refers to hemoglobin carrying oxygen, i.e., oxyhemoglobin.

The second phase space construction method comprises the following steps:

step B110, collecting the light waves of two wavelengths radiated by eachlight source 110 in theprobe 100, and calculating to obtain HbO through the modified Lambert-Beer law₂ A signal.

Step B120, hbO obtained by the low-pass filter pair₂ Filtering the signal to remove noise; for example: a 20-order FIR hamming window filter may be employed with a cut-off frequency of 0.1Hz.

Step B130, hbO after the filtering treatment₂ The signals generate a brain phase space matrix.

Step B200: and extracting the features of the second phase space based on the common mode space mode to obtain a second feature quantity set. This step is similar to step S410 and will not be described here.

Step C: and integrating the first characteristic quantity set and the second characteristic quantity set to form a characteristic set with higher dimension.

Step D: determining an active mode of the brain dynamic system based at least in part on the feature set; the brain is classified into temporary abnormal state and normal state. Similar to step S420, a description is not repeated here.

Step E: determining whether the subject is abnormal based on the determination result and outputting a result.

Outputting an indication that the subject is currently normal when it is determined that the subject is in normal activity; outputting an indication of its current abnormality when it is determined that the subject is in temporary abnormal state; when the subject is determined to be in an uncertain state, a risk cue is output so that it can be further tested to exclude the possibility of misuse of the drug to cause mental state abnormalities.

Fig. 11 shows classification models obtained by obtaining abnormal actual measurement data from 40 persons and normal actual measurement data from 50 persons, obtaining a first feature quantity set containing three features from a first phase space and obtaining a second feature quantity set containing three features from a second phase space, integrating the feature quantity sets into a six-dimensional feature set according to the above method. To display the classification result, fig. 11 selects three of the features for display, and gray shading represents the interface. Wherein the normal class data all fall into the normal class. The 39 abnormal measured data fall into abnormal classification, and 1 abnormal measured data fall into normal data.

It can be seen that the method shown in FIG. 10 is different from the method shown in FIG. 2 in that the outgoing light intensity signal and HbO are used simultaneously₂ The signal constructs two state spaces, and by fusing the two types of features, a high-dimensional feature set can be formed, thereby being beneficial to improving the accuracy of judgment.

The invention also provides a brain transient abnormal state detection device capable of implementing the transient abnormal state detection method, and the use mode of the brain transient abnormal state detection device is shown in fig. 12. The detection device can be divided into ahost 200 and aprobe 100, and in the testing process, theprobe 100 can be attached to the forehead or temple of the tested person. As can be seen from the dopamine transmission path in fig. 1, the frontal lobe brain region corresponding to the forehead or the nucleus accumbens brain region corresponding to the temple are both located in the dopamine transmission path.

Fig. 13 shows a device for detecting transient abnormal state of the brain.Host 200 includes aprocessor 210, amemory 220, and acommunication interface 230. Theprocessor 210 is configured to perform a method for detecting transient abnormal states of the brain. Theprocessor 210 may be a single-chip microcomputer, a CPU, or the like. Thememory 220 is used to store codes, data, results, etc. Thecommunication interface 230 is used to communicate with thecommunication interface 130 of theprobe 100. Thememory 220 may be a flash memory, a cloud disk, a hard disk, etc. Thehost 200 and probe 100 may be connected by a wired connection (e.g., USB, serial, etc.) or a wireless connection (e.g., bluetooth, WIFI, etc.). The communication between thehost 200 and theprobe 100 includes control instructions (e.g., detection instructions) transmitted by thehost 200 to theprobe 100, data (e.g., collected outgoing light intensity signals) transmitted by theprobe 100 to thehost 200, and so on. Theprobe 100 collects the outgoing light intensity signal and transmits the outgoing light intensity signal to thehost 200, and thehost 200 executes a method for detecting the brain temporary abnormal state to identify whether the brain is in the brain temporary abnormal state under the influence of drugs. Further, thehost 200 may further include a display device for displaying the detection result, where the display device may be a touch-sensitive display screen, a handheld intelligent terminal, a road station card warning screen, an interactive headset device, and so on.

Theprobe 100 includes a plurality oflight sources 110 and a plurality ofphotodetectors 120. The plurality oflight sources 110 may radiate at least one or more wavelengths of light, the wavelengths of light radiated being between 700-900 nm. In some embodiments, eachlight source 110 may radiate a single wavelength of light, such as 840nm. In some embodiments, eachlight source 110 may radiate light waves of two wavelengths, one above the point of isosbestic (i.e., a wavelength greater than the point of isosbestic) and one below the point of isosbestic (i.e., a wavelength less than the point of isosbestic), such as 770nm and 840nm. The detection spectral range of thephotodetector 120 encompasses the radiation wavelength of thelight source 110.

Thelight source 110 may be multiplexed (i.e., themultiple photodetectors 120 may receive the outgoing light intensity signals of the light beam emitted by the samelight source 110 through the brain to form multiple detection signal channels), so that the size of theprobe 100 is reduced while the high-dimensional signal is obtained, which is helpful for improving the portability of the device.

Fig. 14 is an exemplary arrangement of a plurality oflight sources 110 and a plurality ofphotodetectors 120 inprobe 100. Theprobe 100 includes four light sources 110 (indicated by letter S) and four photodetectors 120 (indicated by letter D). The distance between thelight source 110 and thephotodetector 120 may range from 5mm to 30mm, and the distance between thelight source 110 may range from 5mm to 10 mm. Each line between thelight source 110 and thephotodetector 120 represents a detection signal path, and each line represents a reference signal path. Alight source 110 forms a plurality of detection signal channels with surroundingphotodetectors 120, and a corresponding reference signal channel is formed with anearest photodetector 120. For example, the light source S2 forms S2-D1, S2-D2, S2-D4, S2-D5 and S2-D6 five detection signal channels with D1, D2, D4, D5 and D6, respectively, and forms a corresponding reference signal channel S2-D3 with D3. One light source forms a plurality of detection signal channels with a plurality of surroundingphotodetectors 120, whereby 16 spatial dimension detection channels can be formed with four light sources and fourphotodetectors 120.

It should be noted that fig. 14 is only exemplary, and the number and arrangement of thelight sources 110 and thephotodetectors 120 in theprobe 100 are not limited thereto, as long as a high-dimensional detection signal channel can be formed. For example, threelight sources 110 and sixphotodetectors 120 may be employed, i.e., any one of the columns oflight sources 110 andphotodetectors 120 in FIG. 14 is eliminated. For another example, fourlight sources 110 and fourphotodetectors 120 may be employed, i.e., the first row ofphotodetectors 120 in FIG. 14 is eliminated.

In some embodiments, the arrangement of the plurality oflight sources 110 and the plurality ofdetectors 120 in theprobe 100 may be as shown in fig. 15, where theprobe 100 includes four light sources 110 (indicated by letter S) and four detectors 120 (indicated by letter D). The spacing between thelight source 110 and thedetector 120 may range from 5mm to 30mm, and the spacing between thelight source 110 may range from 5mm to 10 mm. Each connection between thelight source 110 and thedetector 120 represents a detection signal path. Onelight source 110 forms a plurality of detection signal channels with surrounding plurality ofdetectors 120, whereby 10 spatial dimension detection channels can be formed with fourlight sources 110 and fourdetectors 120.

In some embodiments, theprobe 100 may further include a sensor such as a gyroscope or an accelerometer for detecting changes in the attitude or jitter of theprobe 100 and for removing motion artifacts in the signal output by the nearinfrared photodetector 120.

The invention also discloses a system for detecting the temporary abnormal state of the brain. As shown in fig. 16, the detection system includes a remote data management platform and a plurality of detection devices for temporary abnormal states of the brain. The detection device further includes communication circuitry for wirelessly communicating with the data management platform. The detection device can transmit the detection result to the data management platform in real time through the communication circuit. The data can also be stored locally, and the detection result is transmitted to the data management platform when the network is timed or detected. The data management platform comprises a database, and the received detection result is stored in the database. In addition, the detection device can further comprise an identity card reading unit which is used for supporting the reading of the identity card information and carrying out the networking inquiry and comparison function with the data management platform. The detection device can further comprise a fingerprint acquisition unit for supporting fingerprint acquisition and input and performing a function of networking query comparison with the data management platform. The detection device can further comprise a camera unit for supporting the real-time face recognition comparison of field photographing or the networking comparison function with the data management platform.

While the invention has been described in detail in the foregoing general description and specific examples, it will be apparent to those skilled in the art that modifications and improvements can be made thereto. Accordingly, such modifications or improvements may be made without departing from the spirit of the invention and are intended to be within the scope of the invention as claimed.

Claims (12)

1. A method for detecting transient abnormal states of the brain, the method comprising the steps of:

step A100: constructing a first type of phase space of the brain of the subject based on the emergent light intensity signal;

step A200: extracting a first type of phase space features of the brain of the subject to obtain a first feature quantity set;

step B100: based on the subject brain high-dimensional HbO₂ Constructing a second phase space of the brain dynamic system by the signals;

step B200: extracting the features of the second phase space based on a common mode space mode to obtain a second feature quantity set;

step C: integrating the first characteristic quantity set and the second characteristic quantity set to form a characteristic set with higher dimension;

step D: determining an active mode of the brain dynamic system based at least in part on the feature set;

step E: determining whether the subject is abnormal based on the determination result.

2. The method for detecting transient abnormal state of the brain according to claim 1, wherein the step a100 comprises:

step A110: acquiring a first observation space of a brain dynamic system of a subject based on the emergent light intensity signal;

step A120: reconstructing a first phase space of the brain dynamic system based on the first observation space, resulting in a first eigenvalue construction.

3. The method for detecting transient abnormal state of the brain according to claim 2, wherein the step a200 comprises:

step a210: identifying a first evolutionary trajectory of the brain dynamic system based on the first eigenvector construct to obtain a first principal component v₁ (t) and a first residual force r₁ (t)；

Step A220: based on the first basic component v₁ (t) and a first residual force r₁ (t) obtaining a first feature quantity set.

4. The method for detecting transient abnormal state of the brain according to claim 1, wherein the second phase space constructing method comprises:

step B110, collecting light waves with two wavelengths radiated by each light source in the probe, and calculating to obtain HbO through the modified Lambert-Beer law₂ A signal;

step B120, hbO obtained by the low-pass filter pair₂ Filtering the signal to remove noise;

step B130, hbO after the filtering treatment₂ The signals generate a brain phase space matrix.

5. A brain transient abnormal state detection device, which is characterized by comprising a host and a probe, wherein the host comprises a processor, a memory and a communication interface;

the memory is used for storing at least codes, data and results;

the communication interface is used for communicating with the probe;

the probe collects outgoing light intensity signals of the corresponding area of the brain of the subject and transmits the outgoing light intensity signals to the host;

wherein the processor is configured to perform the method for detecting a transient abnormal state of the brain according to any one of claims 1 to 4 to identify whether the brain is under the influence of a drug, and to feed back the result.

6. The brain transient abnormality detection device according to claim 5, wherein said host computer further includes a display screen for displaying the detection result.

7. The brain transient abnormality detection device according to claim 5, wherein said probe includes a plurality of light sources and a plurality of detectors, wherein said plurality of light sources can radiate at least one wavelength of light, and wherein the detection spectral range of said detectors covers the radiation wavelength of said light sources.

8. The device for detecting transient abnormal brain states according to claim 7, wherein the plurality of detectors can receive the outgoing light intensity signals of the light rays emitted from the same light source through the brain to form a plurality of detection signal channels.

9. The brain transient abnormal state detection apparatus according to claim 7, wherein the probe further comprises a gyroscope and/or an accelerometer sensor for detecting a change in posture or shake of the probe for removing motion artifacts in the outgoing light intensity signal output from the detector.

10. A brain transient abnormal state detection system, characterized by comprising a remote data management platform and a plurality of brain transient abnormal state detection devices according to any one of claims 5-9, said detection devices comprising,

communication circuitry for wirelessly communicating with the data management platform.

11. The system for detecting transient abnormal brain conditions according to claim 10, wherein the detection means transmits the detection result to a data management platform in real time through the communication circuit, or,

the detection device carries out local storage on the detection result, transmits the detection result to a data management platform at fixed time or when the network is unobstructed,

the data management platform comprises a database, and the received detection result is stored in the database.

12. The brain transient abnormality detection system according to claim 10, wherein said detection means further includes an identification unit, said identification unit including, at least in part:

the identity card reading unit is used for supporting the identity card information reading and carrying out the networking inquiry and comparison function with the data management platform, and/or

Fingerprint acquisition unit for supporting fingerprint acquisition and input and performing networking query comparison function with data management platform, and/or

The camera unit is used for supporting the real-time face recognition comparison of field photographing or networking comparison function with the data management platform.

Images