AD Neural Regulation System Based on Feature Extraction and Closed-Loop Ultrasound Stimulation

Abstract

The present invention discloses an AD neuromodulation system based on feature extraction and closed-loop ultrasound stimulation, belonging to the field of transcranial ultrasound stimulation. The system includes a programmable ultrasound signal generation module, an ultrasound stimulation module, an electroencephalogram (EEG) signal acquisition and processing module, a closed-loop control module, a signal transmission and storage module, a host computer, and EEG electrodes connected to a test subject. The system uses transcranial ultrasound stimulation technology to stimulate specific regions of the cerebral cortex and implants EEG electrodes in the target region to detect the EEG signals of the target region in real time. Then, multi-dimensional features such as time domain, non-linear dynamics, and spatial domain of signals in different frequency bands are extracted, and the obtained features are used as the input of the third-level processor for real-time detection. According to the category diagnosis results, the parameters of transcranial ultrasound stimulation are continuously adjusted to inhibit the aggravation of AD. The present invention can provide ideas for optimizing the treatment parameters of Alzheimer's disease and lay a foundation for the development of treatment devices for Alzheimer's disease.

Description

Technical Field

[0001] The present invention relates to the field of transcranial ultrasound stimulation, and particularly to an AD neural regulation system based on feature extraction and closed-loop ultrasound stimulation. Background Art

[0002] Alzheimer's disease (AD) is a neurodegenerative disease characterized by progressive memory impairment, loss of cognitive function accompanied by mental and behavioral abnormalities. Its main pathological features include deposition of β-amyloid protein (Aβ), neurofibrillary tangles (NFTs), neuron loss, etc., accompanied by behavioral disorders such as chronic decline in cognitive function, memory decline, language disorders, and decline in learning ability.

[0003] As of 2019, there were 10 million Alzheimer's disease (AD) patients in China. It is estimated that by 2050, the number of Alzheimer's disease patients in China will reach 30.03 million, and the proportion of patients over 80 years old will be close to 50%. Coupled with the fact that there is currently no effective treatment for AD, this will cause a heavy burden.

[0004] Currently, the main clinical treatments for such diseases include drugs, surgical resection of lesion areas, transcranial electromagnetic stimulation, deep brain stimulation, etc. However, these treatments have limitations such as drug resistance, invasiveness, and being limited by low resolution and superficial brain tissues. Transcranial ultrasound stimulation (TUS) is a non-invasive and non-invasive transcranial nerve therapy technology that can focus deeply in the brain and has high resolution, and has attracted extensive attention. However, the cause, pathogenesis of AD and the mechanism of action of ultrasound stimulation are not yet clear. Therefore, experimental research on Alzheimer's disease mouse models plays a crucial role in clinical research. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide an AD neural regulation system based on feature extraction and closed-loop ultrasound stimulation to realize the detection of AD diagnosis and improvement in a specific cerebral cortex of mice, so as to apply appropriate ultrasound stimulation to a specific area of the cerebral cortex in real time.

[0006] To solve the above technical problems, the technical solution adopted by the present invention is:

[0007] An AD neuroregulation system based on feature extraction and closed-loop ultrasound stimulation, comprising a programmable ultrasound signal generation module, an ultrasound stimulation module, a signal acquisition and processing module, a closed-loop control module, a signal transmission and storage module, a host computer, and EEG electrodes; the EEG electrodes are connected to an experimental subject; the programmable ultrasound signal generation module sends the generated stimulation signal to the ultrasound stimulation module, and the ultrasound stimulation module emits the stimulation signal to the experimental subject in the form of ultrasound stimulation; the signal acquisition and processing module is connected to the EEG electrodes, and is used for acquiring the EEG signals recorded by the EEG electrodes, preprocessing and frequency division processing the EEG signals, and sending the processed EEG signals to the closed-loop control module; the system uses transcranial ultrasound stimulation technology to stimulate a specific area of the cerebral cortex of the experimental subject and implants EEG electrodes in the target area, detects the EEG signals of the target area in real time, then extracts the time domain, non-linear dynamics and spatial multi-dimensional features of signals in different frequency bands, and uses the obtained multi-dimensional features as the input of the third-level processor in the closed-loop control module, so as to perform real-time detection, and continuously adjust the parameters of transcranial ultrasound stimulation according to the category diagnosis result to inhibit the aggravation of AD in the experimental subject.

[0008] A further improvement of the technical solution of the present invention lies in: the experimental subject is selected as a mouse, the ultrasound stimulation module is placed on the motor cortex of the mouse, and the EEG electrodes are implanted in the hippocampal CA1 area of the mouse.

[0009] A further improvement of the technical solution of the present invention lies in: a three-stage cascade processor is implanted in the closed-loop control module, which is used for performing three-stage processing on the EEG signals after frequency division processing by the signal acquisition and processing module. The first-stage processor is a strong classifier for quickly screening suspected AD EEG signals; the EEG signals that pass the screening enter the second-stage processor. In the second-stage processor, a multivariate variational mode decomposition algorithm is used to realize multi-channel input of the signals, and the time domain features and non-linear dynamics features of the signals are extracted from the decomposed signal components. At the same time, the signal components are combined to construct a new signal matrix, and CSP is used to extract spatial features from the signal matrix; in the third-stage processor, the three obtained features are combined to obtain the multi-modal features of the EEG signals, and finally classified by SVM; if the classification result is not an AD abnormal signal, the calculation is stopped, otherwise it is diagnosed as AD aggravation, and the diagnosis result is sent to the programmable ultrasound signal generation module to timely adjust the stimulation parameters to achieve the treatment purpose;

[0010] The signal transmission and storage module is used for receiving the working parameters of each module configured by the host computer and the EEG signals transmitted by the closed-loop control module, and storing the EEG signals as a data set;

[0011] The host computer is used to train the parameters of the three - stage cascade processor implanted in the closed - loop control module according to the data set, and communicate with the signal transmission and storage module in real time; continuously adjust the working parameters during the operation of each module, update various parameters in the three - stage cascade processor implanted in the closed - loop control module, and display the collected EEG signals in real time.

[0012] The programmable ultrasonic signal generation module is used to change the output of ultrasonic stimulation in real time according to the result obtained by the closed - loop control module or the host computer instruction.

[0013] A further improvement of the technical solution of the present invention is that: the first - stage processor is a strong classifier trained by the Ada Boost algorithm.

[0014] A further improvement of the technical solution of the present invention is that: multi - modal feature extraction is to be performed in the second - stage processor, and the multi - modal data comes from the EEG data of publicly available AD mice on the one hand, and the EEG data of the mice in the selected model group, sham - stimulation group and normal control group on the other hand.

[0015] A further improvement of the technical solution of the present invention is that: the components of multiple signals in the three - stage cascade processor are obtained by the MVMD method.

[0016] A further improvement of the technical solution of the present invention is that: the steps of pre - processing the EEG signal include filtering and noise reduction.

[0017] A further improvement of the technical solution of the present invention is that: after the SVM classifier trains the training set to obtain a classification model, it can test the test set.

[0018] An AD neuromodulation method based on feature extraction and closed - loop ultrasonic stimulation includes the following steps:

[0019] Step 1: implant the ultrasonic stimulation module and EEG electrodes into the preset sites of several mice respectively; the preset sites are located in the brain regions of the mice; the mice are divided into different stimulation groups.

[0020] Step 2: After all the experimental mice implanted with EEG electrodes have recovered for t time, use the programmable ultrasonic signal generation module to regulate the output parameters of the ultrasonic stimulation signal of the ultrasonic stimulation module.

[0021] Step 3: After the ultrasonic transducer of the ultrasonic stimulation module receives the stimulation signal, perform ultrasonic stimulation on the preset intracranial brain sites of the experimental mice.

[0022] Step 4, using the signal acquisition and processing module to collect the EEG signals recorded on the EEG electrodes, sending the processed EEG signals to the closed-loop control module, and after being processed by the three-stage series processor, judging whether the experimental mice need to adjust the ultrasonic stimulation parameters according to the classification results; the signal transmission and storage module receives the EEG signals transmitted by the closed-loop control module, and stores the EEG signals as a data set;

[0023] Step 5: The host computer will train the parameters of the three-stage series processor implanted in the closed-loop control module according to the existing data set, and communicate with the signal transmission and storage module in real time; continuously update the various parameters of the three-stage series processor implanted in the closed-loop control module, and display the collected EEG signals in real time;

[0024] Step 6: according to the judgment result of step 4, if the stimulation parameters do not need to be adjusted, proceed to step 7; otherwise, return to step 3;

[0025] Step 7, performing a Morris water maze test on the experimental mice every T period to obtain evaluation indicators of the Morris water maze test; the evaluation indicators of the Morris water maze test include escape latency and escape path length; judging the difference in the evaluation indicators of the Morris water maze test under the interaction of group × day number, if the difference between the time in the third quadrant and the time in other quadrants between the two groups is greater than a preset threshold, completing the regulation, otherwise, modifying the ultrasonic stimulation parameters and returning to step 3.

[0026] A further improvement of the technical solution of the present invention is that the steps of performing Morris water maze experiment on experimental mice include:

[0027] 701, dividing the water maze into four areas; the water maze is provided with a visible platform located above the water surface;

[0028] 702, allowing the experimental mice to enter the water, with the entry point being the pool wall at the midpoint of each area;

[0029] 703, using a CCD camera to collect the swimming trajectory of the organism and store it in a video capture card;

[0030] 704, the video capture card uploads the swimming trajectory of the experimental mouse to the computer; the computer performs image recognition on the swimming trajectory of the experimental mouse to obtain the evaluation index of the Morris water maze experiment; the escape latency is the time from the experimental mouse entering the water to climbing onto the visible platform.

[0031] Due to the adoption of the above technical solution, the technical progress achieved by the present invention is:

[0032] The AD neuromodulation system and method based on feature extraction and closed-loop ultrasound stimulation provided by the present invention can detect whether the EEG signals of a specific cerebral cortex are abnormal AD signals, and can adjust appropriate ultrasound stimulation parameters according to the detection results, so as to achieve precise monitoring and neuromodulation of a specific cerebral cortex.

[0033] The present invention selects the motor cortex of mice as the stimulation target, designs a transcranial ultrasound stimulation paradigm, conducts chronic stimulation experiments, designs an experimental control group, and evaluates the effect of transcranial ultrasound stimulation from the perspectives of animal behavior and EEG signals, so as to find the optimal stimulation parameters. In order to detect the neuromodulation effect, physiological signal acquisition and analysis are carried out by recording the hippocampal CA1 (AP: 2.06, ML: ±1.5, DV: 1.25) of mice implanted with EEG electrodes, to evaluate the differences in the regulation of AD neural activities by ultrasound stimulation with different parameters, and to provide ideas for optimizing the treatment parameters for Alzheimer's disease. Brief Description of the Drawings

[0034] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts;

[0035] Figure 1 It is the system block diagram in the embodiment of the present invention;

[0036] Figure 2 It is the structural schematic diagram of the closed-loop control module in the system of the embodiment of the present invention;

[0037] Figure 3 It is the schematic diagram of the multi-modal feature extraction and processing process in the embodiment of the present invention;

[0038] Among them, 1. Programmable ultrasound signal generation module; 2. Ultrasound stimulation module; 3. EEG electrode; 4. Signal acquisition and processing module; 5. Closed-loop control module; 51. Data reception sub-module; 52. Category diagnosis sub-module; 53. Parameter configuration sub-module; 54. Stimulation control sub-module; 6. Signal transmission and storage module; 7. Host computer. Detailed Embodiments

[0039] It should be noted that the terms "including" and "having" and any variations thereof in the specification, claims and above-mentioned drawings of the present invention are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units does not necessarily limit to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0040] The present invention will be further described in detail below with reference to the drawings and embodiments:

[0041] As Figure 1 shown, this embodiment provides an AD neural regulation system based on feature extraction and closed-loop ultrasound stimulation, including a programmable ultrasound signal generation module 1, an ultrasound stimulation module 2, electroencephalogram electrodes 3, a signal acquisition and processing module 4, a closed-loop control module 5, a signal transmission and storage module 6, and a host computer 7;

[0042] The programmable ultrasound signal generation module 1 can send the generated stimulation signal to the ultrasound stimulation module 2 according to the processing result obtained by the closed-loop control module 5 or the instruction of the host computer 7, and change the output of the ultrasound stimulation in real time; the electroencephalogram electrodes 3 are electrically connected to the experimental object (mouse) for collecting the electroencephalogram signal of the mouse;

[0043] The signal acquisition and processing module 4 is electrically connected to the electroencephalogram electrodes 3, and is used for collecting the electroencephalogram signal recorded by the electroencephalogram electrodes 3, and performing a series of processing such as analog-to-digital conversion and filtering and noise reduction on the collected electroencephalogram signal, and sending the processed electroencephalogram signal to the closed-loop control module 5.

[0044] The signal transmission and storage module 6 can transmit the received electroencephalogram signal to the host computer 7 in a wired or wireless manner for real-time display and analysis of the electroencephalogram signal. In the low-power operation mode, the signal transmission and storage module 6 does not establish a physical connection with the host computer 7, and directly stores the received electroencephalogram signal in the on-board SD memory for subsequent offline analysis and processing.

[0045] After receiving the preprocessed electroencephalogram signal, the closed-loop control module 5 performs multi-dimensional (time domain, non-linear dynamics, spatial domain, etc.) feature extraction and classification on the electroencephalogram signal, and judges in real time whether the aggravation of AD occurs in a specific cerebral cortex. After this classification result is transmitted to the stimulation control sub-module 54, the stimulation control sub-module 54 configures different ultrasound stimulation modes and parameters according to the configured stimulation parameters and the classification result and feeds them back to the programmable ultrasound signal generation module 1. The programmable ultrasound signal generation module 1 applies corresponding ultrasound stimulation pulses according to the received stimulation parameters to intervene in the neuron situation in the corresponding area of the intracranial cavity, and completes the closed-loop intervention process of the system for AD diagnosis.

[0046] Further, as Figure 2 shown, the closed-loop control module 5 includes a data receiving sub-module 51, a category diagnosis sub-module 52, a parameter configuration sub-module 53, and a stimulation control sub-module 54; the data receiving sub-module 51 serves as the interface between the signal acquisition and processing module 4, the signal transmission and storage module 6, and the closed-loop control module 5. Through the SPI communication method, it is responsible for receiving and caching the EEG signals of the signal acquisition and processing module 4, as well as the host computer configuration parameters transmitted through the signal transmission and storage module 6. When the data buffer in the data receiving sub-module 51 obtains a neural signal time series sequence of a predetermined length, this neural signal sequence will enter the category diagnosis sub-module 52. The essence of this module is a three-stage cascade processor, which determines whether the current EEG signal segment is an AD signal exacerbation segment. The classification result will be transmitted to the stimulation control sub-module 54, and the stimulation control sub-module 54 will convey different stimulation mode parameters to the programmable ultrasonic signal generation module 1 according to the corresponding configuration parameters such as stimulation time, stimulation intensity, duty cycle, etc. transmitted by the host computer in the parameter configuration sub-module 53 and the classification result. If the diagnosis result of a specific cerebral cortex neural signal is an abnormal AD signal, the corresponding area will receive corresponding ultrasonic stimulation.

[0047] Further, the classifier used to form the three-stage cascade processor in the closed-loop control module 5 is obtained by the Real AdaBoost algorithm based on the criterion of minimizing the loss function in the positive and negative sample sets in the training set. Among them, the classifier c i consists of a threshold and a segmented output function. When the corresponding eigenvalue f of the signal is greater than the threshold θ, a numerical value is output, otherwise another numerical value is output. The segmented function and threshold output by the classifier are obtained by training the collected EEG signals.

[0048]

[0049] The first-stage processor is a strong classifier, and H(x) is trained by the Real AdaBoost algorithm. The classifier c used in the first stage i corresponds to computationally small features such as the amplitudes and spectra of different rhythms after frequency band division, which is conducive to the rapid screening of suspected abnormal AD signals.

[0050] H(x) = ∑a = 1,...na (2)

[0051] The second-stage processor first uses the MVMD method to further decompose the signals quickly screened, extracts the time-domain features and non-linear dynamic features of the signals from the decomposed signal components, combines the signal components at the same time, constructs a new signal matrix, and uses CSP to extract the spatial features of this signal matrix.

[0052] Signal decomposition: The MVMD method realizes the transformation from a single-channel input signal to a multi-channel input signal, and can keep the frequencies of each IMF component the same when decomposing data. The components obtained by decomposition are jointly used as the input of the iterator, and the central frequency and bandwidth are used as the update targets of this iterator, and its output is the required k components. Assuming that the signals of C sampling channels are X(t), it can be expressed in mathematical form as [x1(t), x2(t), … x C (t).

[0053] (1) First, assume that the signal contains k components and satisfies:

[0054]

[0055] (2) In the vector u k (t), through the Hilbert-Huang transform (HHT), the data is analytically expressed as and the single-sided spectrum is obtained with this as a reference. Then, by multiplying the exponential term the adjustment of the central frequency is realized. Then calculate for all the signals to be decomposed. The optimization objective of the objective function is to minimize the bandwidth of the components while keeping each of the obtained components as much as possible to continue to form the original signal. The following is the optimization problem to be solved:

[0056]

[0057] where, is the analytical expression form of the data.

[0058] (3) To solve this variational problem, the Lagrangian in the following form is constructed:

[0059]

[0060] (4) Update and According to the updated values obtained, the value of u k (t) and the magnitude of the central frequency can be calculated, and thus the decomposed signal components can be obtained. The further update mode is:

[0061]

[0062] The update frequency is:

[0063]

[0064] Time-domain feature extraction: After adopting the HHT method, by analyzing the characteristics of components, the variation characteristics of the EEG signal in the time direction can be obtained. According to the instantaneous amplitude of the IMF, the instantaneous energy H[u k (t)] can be obtained, and the information in the frequency domain and the amplitude variation can be obtained:

[0065] U k (t) = u k (t) + jH[u k (t)] (8)

[0066]

[0067] Calculate the energy amplitude of the sampled signal:

[0068]

[0069] where n is the number of sampling points, is the amplitude of the discrete signal i. The average instantaneous energy value reflects the variation of the signal in the time domain, and it is denoted as F1.

[0070] Nonlinear dynamics features: Introduce multi-scale entropy to observe the information differences of the signal in multiple modes and analyze the signal complexity. After sampling the decomposed IMF functions to obtain discrete signals of corresponding different modes, perform a series of analyses, averaging, and dimensionality transformations, and finally obtain the sample entropy value when the length of the time series is M:

[0071] SampEn(m, r, M) = -ln[C m+1 (r) / C m (r)] (10)

[0072] Repeat the above calculations to obtain the entropy features at multiple scales, and combine these features to obtain the multi-scale entropy features of the EEG signal, denoted as F2.

[0073] CSP spatial domain features: Combine the obtained IMF components and the sampled signals of the components to obtain a signal matrix composed of the total number k of components and the number n of sampling points, that is, k×n as the object of CSP processing. Taking the components of channels C3 and C4 as an example, denote their spatial domain features as F3, and its matrix can be expressed as:

[0074]

[0075] The third-level processor then combines the multiple feature information extracted to obtain the multi-modal feature denoted as F = {F1, F2, F3}, and the entire processing process is as Figure 3 shown. To avoid the numerical differences of different features, perform a normalization operation on the extracted features:

[0076] Fe = (F e - μ e ) / σ e , e = 1, 2, 3 (12)

[0077] where μ e and σ e respectively represent the mean value and standard deviation when the feature is e. Classify F with the normalized feature. On the basis of this common representation, introduce an SVM classifier to obtain the final diagnostic result.

[0078] An AD neuromodulation method based on feature extraction and closed-loop ultrasound stimulation includes the following:

[0079] To evaluate the therapeutic effect of transcranial ultrasound stimulation through a control experiment, use an AD mouse model and group the mice as follows: The AD mice are randomly divided into 2 groups, including a stimulation group (ADT group) and a sham stimulation group (ADS group), while healthy mice are used as a normal control group (WT group). The sham stimulation group is to reserve a stimulation area in the motor cortex of the AD mice and implant electroencephalogram electrodes 3 in the hippocampal region, but no ultrasonic stimulation is applied.

[0080] Step 1, implant the ultrasonic stimulation module 2 and the electroencephalogram electrodes 3 into the preset sites of the mice in the stimulation group (ADT group), the sham stimulation group (ADS group), and the control group (WT group) respectively; the preset sites are located in the brain region of the mice;

[0081] Specifically, place the mice participating in the experiment in a gas anesthesia induction box, adjust the anesthesia to 2.5 L / min, and let it stand for about 2 minutes until the mice do not have a leg retraction reaction when pinching their toes. Use 1% chloral hydrate and perform intraperitoneal injection according to the body weight ratio to achieve surgical anesthesia.

[0082] Perform a craniotomy in the motor cortex (AP: -1.54, ML: ±1.5) to form an observation window for applying ultrasonic stimulation and implant a glass slide. The electroencephalogram signal acquisition electrode is implanted in the hippocampal CA1 (AP: 2.06, ML: ±1.5, DV: 1.25), and the electroencephalogram electrode for acquisition / recording is implanted in the CA1 region of the hippocampus of the mice. Two cranial nails are placed at the nasal bone position for grounding and reference.

[0083] Step 2, after all the experimental mice implanted with the electroencephalogram electrodes 3 have recovered for t time, use the programmable ultrasonic signal generation module 1 to regulate the output parameters of the ultrasonic stimulation signal of the ultrasonic stimulation module 2;

[0084] Specifically, the mice diagnosed with AD were treated 1 week after recovery, and all the mice were 5 months old at this time. The mice in the stimulation group received the signal transmitted by the stimulation control sub-module 54 to the programmable ultrasonic signal generation module 1. The programmable ultrasonic signal generation module 1 sent the generated stimulation signal to the ultrasonic stimulation module 2, and the ultrasonic stimulation module 2 emitted the stimulation signal in the form of ultrasonic stimulation to the observation window area where the glass slide was implanted.

[0085] Step 3, after the ultrasonic transducer of the ultrasonic stimulation module 2 received the stimulation signal, ultrasonic stimulation was performed on the preset site in the intracranial brain of the experimental mice;

[0086] Step 4, the electroencephalogram (EEG) signals recorded on the EEG electrodes 3 were collected by the signal acquisition and processing module 4, and the processed EEG signals were sent to the closed-loop control module 5. After being processed by the three-stage cascade processor, it was judged whether the experimental mice needed to adjust the ultrasonic stimulation parameters according to the classification results; the signal transmission and storage module 6 received the EEG signals transmitted by the closed-loop control module 5 and stored the EEG signals as a data set;

[0087] Step 5, the host computer 7 would train the parameters of the three-stage cascade processor implanted in the closed-loop control module 5 based on the existing data set and communicate with the signal transmission and storage module 6 in real time; continuously update various parameters in the three-stage cascade processor implanted in the closed-loop control module 5 and display the collected EEG signals in real time;

[0088] Step 6, according to the judgment result of Step 4, if the stimulation parameters do not need to be adjusted, go to Step 7; otherwise, return to Step 3;

[0089] Step 7, the Morris water maze experiment was performed on the experimental mice every T period to obtain the evaluation indexes of the Morris water maze experiment; the evaluation indexes of the Morris water maze experiment included escape latency and escape path length; judge the differences in the evaluation indexes of the Morris water maze experiment under the interaction of group × days. If the difference between the time in the third quadrant and the time in other quadrants among groups was greater than the preset threshold, the regulation was completed; otherwise, the ultrasonic stimulation parameters were modified and returned to Step 3.

[0090] Specifically, in order to evaluate the safety of ultrasonic stimulation and ensure that it does not cause anxiety-related side effects, a Morris water maze experiment was designed. The Morris water maze experiment was performed on experimental mice every T period to obtain the evaluation indicators of the Morris water maze experiment: escape latency and escape path length. The group and day were analyzed to see whether there were significant main effects on these two indicators, and there were significant differences in the interaction of group × day; if there was a significant difference between the time spent in the third quadrant and the time spent in other quadrants between the groups, then these results showed that the ADT group could effectively distinguish the quadrant where the platform was located from other quadrants, while the remaining groups were second in terms of their ability to distinguish. Otherwise, it means that there is a problem with the stimulation scheme, and the stimulation parameters can be adjusted.

[0091] Morris water maze (MWM) consists of a water maze, a computer, a video capture card, a CCD camera and other equipment. Two virtual vertical lines are set in the pool to divide the pool into four quadrants, I, II, III and IV. The mouse enters the water at the midpoint of each quadrant. A cylindrical visible platform with a diameter of 10 cm is placed in the middle of quadrant III. An appropriate amount of compound colorant is added to the water to make it white, so that the behavior analysis system can track the swimming trajectory of the mouse during the experiment.

[0092] The steps for conducting Morris water maze test on experimental mice include:

[0093] 1) Divide the water maze into 4 areas; the water maze is provided with a visible platform above the water surface;

[0094] 2) Let the experimental mice enter the water, and the entry point is the pool wall at the midpoint of each area;

[0095] 3) Use a CCD camera to capture the swimming trajectory of the experimental mice and store it in a video capture card;

[0096] 4) The video capture card uploads the swimming trajectory of the experimental mouse to the computer; the computer performs image recognition on the swimming trajectory of the experimental mouse to obtain the evaluation index of the Morris water maze experiment; the escape latency is the time from the experimental mouse entering the water to climbing onto the platform.

[0097] Specifically, the visible platform extends 1 cm above the water surface. The mice are placed in the pool at different quadrant walls. The time it takes for the mice to find and climb onto the visible platform within 60 s is recorded as the escape latency. If the mice do not find the visible platform within 60 s, they are guided to the visible platform and placed on the visible platform for 15 - 20 s, and the escape latency is recorded as 60 s. Meanwhile, the path length before the mice escape to the visible platform is recorded. If there are no significant differences in the escape latency and path length among the groups, it is considered that the mice in each group have similar motor and visual abilities.

[0098] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. An AD neural regulation system based on feature extraction and closed-loop ultrasound stimulation, characterized in that, It includes a programmable ultrasonic signal generation module (1), an ultrasonic stimulation module (2), a signal acquisition and processing module (4), a closed-loop control module (5), a signal transmission and storage module (6), a host computer (7) and electroencephalogram electrodes (3); the electroencephalogram electrodes (3) are connected to the experimental subject; the programmable ultrasonic signal generation module (1) sends the generated stimulation signal to the ultrasonic stimulation module (2), and the ultrasonic stimulation module (2) emits the stimulation signal to the experimental subject in the form of ultrasonic stimulation; the signal acquisition and processing module (4) is connected to the electroencephalogram electrodes (3) and is used to collect the electroencephalogram signals recorded by the electroencephalogram electrodes (3), preprocess and frequency-divide the electroencephalogram signals, and send the processed electroencephalogram signals to the closed-loop control module (5); the system uses transcranial ultrasonic stimulation technology to stimulate a specific area of the cerebral cortex of the experimental subject and implants electroencephalogram electrodes (3) in the target area, detects the electroencephalogram signals of the target area in real time, then extracts the time-domain, non-linear dynamics and spatial multi-dimensional features of signals in different frequency bands, and uses the obtained multi-dimensional features as the input of the third-level processor in the closed-loop control module (5), so as to perform real-time detection, and continuously adjust the parameters of transcranial ultrasonic stimulation according to the category diagnosis result to inhibit the aggravation of AD in the experimental subject; A three-stage cascade processor is implanted in the closed-loop control module (5) and is used to perform three-stage processing on the electroencephalogram signals frequency-divided by the signal acquisition and processing module (4). The first-stage processor is a strong classifier for quickly screening electroencephalogram signals suspected of AD; the electroencephalogram signals that pass the screening enter the second-stage processor. In the second-stage processor, a multivariate variational mode decomposition algorithm is used to realize multi-channel input of the signals, and the time-domain features and non-linear dynamics features of the signals are extracted from the decomposed signal components. At the same time, the signal components are merged to construct a new signal matrix, and CSP is used to extract spatial features from this signal matrix; In the third-stage processor, the three obtained features are combined to obtain the multi-modal features of the EEG signals, and finally classified by SVM; if the classification result is not an AD abnormal signal, the calculation is stopped, otherwise it is diagnosed as AD aggravation, and the diagnosis result is sent to the programmable ultrasonic signal generation module (1) to timely adjust the stimulation parameters to achieve the treatment purpose; The signal transmission and storage module (6) is used to receive the working parameters of each module configured by the host computer (7) and the electroencephalogram signals transmitted by the closed-loop control module (5), and store the electroencephalogram signals as a data set; The host computer (7) is used to train the parameters of the three-stage cascade processor implanted in the closed-loop control module (5) according to the data set, and perform real-time communication with the signal transmission and storage module (6); continuously adjust the working parameters during the operation of each module, update various parameters in the three-stage cascade processor implanted in the closed-loop control module (5), and display the collected electroencephalogram signals in real time; The programmable ultrasonic signal generation module (1) is used to change the output of ultrasonic stimulation in real time according to the result obtained by the closed-loop control module (5) or the instruction of the host computer (7).

2. The AD neural regulation system based on feature extraction and closed-loop ultrasonic stimulation according to claim 1, wherein The experimental subjects are mice. The ultrasonic stimulation module (2) is placed on the motor cortex of the mice, and the electroencephalogram electrodes (3) are implanted into the CA1 region of the hippocampus of the mice.

3. The AD nerve regulation system based on feature extraction and closed-loop ultrasonic stimulation according to claim 1, wherein The first-level processor is a strong classifier trained using the Ada Boost algorithm.

4. The AD neural regulation system based on feature extraction and closed-loop ultrasonic stimulation according to claim 1, characterized in that, Multimodal feature extraction is to be performed in the second-level processor. The multimodal data comes from the electroencephalogram data of AD mice that has been made public on the one hand, and the electroencephalogram data of the mice in the set model group, sham stimulation group, and normal control group on the other hand.

5. The AD neural regulation system based on feature extraction and closed-loop ultrasound stimulation according to claim 1, wherein, The components of multiple signals in the three-level cascaded processor are obtained using the MVMD method.

6. The AD neural regulation system based on feature extraction and closed-loop ultrasound stimulation according to claim 1, wherein, The steps for preprocessing the EEG signals include filtering and noise reduction.

7. The AD nerve regulation system based on feature extraction and closed-loop ultrasonic stimulation according to claim 1, wherein After the SVM classifier trains the training set to obtain a classification model, it can then test the test set.

Images