An olfactory function standardized evaluation device and method based on brain-computer interaction technology

Abstract

The application discloses a kind of olfactory function standardization evaluation device and method based on brain-computer interaction technology, specifically: the device includes front end module, data acquisition module, data processing module and user interaction module;For collecting the electrophysiological activity of olfactory mucosa, olfactory bulb and prefrontal cortex of examinee under resting state and respiratory signal, amplify and collect to multiple channel signal, generate olfactory function evaluation index by neural network model calculation, generate olfactory function examination report and display and store.The application records the olfactory physiological signal of examinee under resting state, obtains quantitative olfactory function evaluation index by data analysis, without the smell stimulation in traditional olfactory function detection method, can effectively exclude subjective factor influence, reduce individual difference.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of human and mammal olfactory detection and function evaluation, and in particular to an olfactory function standardized evaluation device and method based on brain-computer interaction technology. BACKGROUND

[0002] Olfactory disorder is an early symptom of various diseases such as neurodegenerative diseases, viral infections, tumors, etc. Timely, rapid, convenient and accurate olfactory function evaluation is of great significance for early intervention and disease delay. Current olfactory function detection methods include subjective detection methods and objective detection methods. Most of the methods currently used in clinical practice are subjective methods, i.e. olfactory function of subjects is evaluated by stimulating them with odors, including odor identification, odor threshold, odor discrimination, etc. Such methods are easily affected by subjective factors, detection procedures and background interference, and have low stability and comparability. Objective detection methods include electrophysiological detection and olfactory function imaging. Among them, olfactory function detection based on electrophysiology refers to recording olfactory-related electrophysiological signals of subjects using an electrophysiological acquisition system. Abnormal changes in electrophysiological activity of patients with olfactory disorders occur under spontaneous and odor stimulation, which reflect the olfactory function status of the subjects. This method is relatively objective, convenient and low-cost. However, current methods for obtaining and analyzing olfactory-related electrophysiological signals are difficult to establish standardized electrophysiological detection indicators that can be applied in clinical practice.

[0003] Many studies have shown that in the spontaneous state, i.e. without odor stimulation, there is a respiratory-synchronized neural rhythm in the olfactory system and part of the cerebral cortex of humans and mammals. Analogous to brain electrophysiology examination, the functional brain produces regular electrical activity, and when epilepsy or functional abnormalities occur, the electrical activity shows abnormalities. When olfactory disorders occur, this spontaneous neural rhythm also changes. Therefore, compared with odor-induced neural response signals, spontaneous neural signals of the olfactory system are less affected by individual differences and subjectivity, and are more conducive to establishing quantifiable and standardized olfactory function detection indicators. SUMMARY

[0004] The present application proposes a non-odor-induced quantitative olfactory function evaluation device and method based on brain-computer interaction technology to address the deficiencies of current olfactory function evaluation methods and devices.

[0005] According to the first aspect of the specification, an olfactory function standardized evaluation device based on brain-computer interaction technology is provided, which comprises a front-end module, a data acquisition module, a data processing module and a user interaction module.

[0006] The front-end module includes: a respiratory sensing module for detecting temperature and airflow changes caused by the subject's breathing; a sensing module for detecting olfactory nerve signals in the olfactory bulb and olfactory mucosa; and an electroencephalogram (EEG) detection module for detecting neural activity in the prefrontal cortex.

[0007] The data acquisition module is used to synchronously amplify the physiological signals transmitted from the front-end module through hardware circuitry, and perform analog-to-digital conversion to convert analog signals into digital signals, thereby completing the acquisition of multi-channel physiological signals.

[0008] The data processing module is used for preprocessing respiratory and electrophysiological data, completing feature extraction and fusion, establishing a multi-channel physiological signal database, and building a mapping model between multi-dimensional data and olfactory function scores. Real-time measured data is input into the model, and olfactory function scores are output.

[0009] The user interaction module is used to display, save, and read the data collected by the front-end module, display the olfactory function assessment results obtained by the data processing module, generate reports, save the results and subject information locally, and upload them to the cloud server.

[0010] Furthermore, in the front-end module, the breathing sensing module detects temperature and airflow changes caused by breathing through temperature and airflow sensors placed near the nasal cavity, and converts them into voltage signals.

[0011] Furthermore, the olfactory nerve signal detection module measures the neural activity signals transmitted from the olfactory bulb to the surface of the skull through silver-silver chloride electrodes placed externally near the bridge of the nose and brow bone, and measures the olfactory mucosal nerve activity signals through neural microelectrodes placed at the olfactory mucosa of the nasal cavity.

[0012] Furthermore, the brain signals were detected using EEG cap electrodes to detect neural activity in the prefrontal cortex.

[0013] Furthermore, the preprocessing of the data processing module includes: filtering and removing noise from the multi-channel physiological signals; restoring the olfactory nerve signals of the olfactory bulb through a source reconstruction algorithm; calculating the time-domain and frequency-domain parameters of the time-series signals; and completing feature extraction, including the frequency of the respiratory signals, the variance, entropy, frequency band energy, cross-coupling coefficients between different frequency bands, and the cross-correlation coefficients between different physiological signals.

[0014] Further, the data processing module establishes a database by recording signals of animal models, patients with different degrees of olfactory dysfunction, and normal healthy people, takes the olfactory function degree diagnosed by an authoritative doctor as a label, and sets it from 100 to 0 according to the decreasing olfactory function; secondly, three one-dimensional convolutional neural networks are used to receive signals and features of each physiological signal channel, and data output by each convolutional neural network is input into a multi-layer perception machine for fusion, and finally a regression value from 100 to 0 is output by the model; the data set is divided into a training set, a validation set, and a test set, wherein the data of the training set is used to run a neural network learning algorithm, train the model, and obtain parameter values of each part of the network; the validation set is used to observe the performance of the model during training to prevent overfitting; and the test set is used to evaluate the performance of the final model.

[0015] According to another aspect of the specification, a standardized olfactory function evaluation method based on brain-computer interaction technology is provided, and the method steps are:

[0016] The first step is to connect the front-end module of signal detection with the subject, specifically: for breath recording, the temperature and airflow sensors of the breath front-end module are placed near the nostrils of the subject; for the olfactory mucosa of the subject, the neural microelectrode is inserted into the nasal cavity and closely attached to the olfactory mucosa; for the olfactory bulb, the silver-silver chloride electrode is closely attached to the nasal bridge and brow bone; for the prefrontal cortex recording, the subject wears an electroencephalogram cap electrode, and conductive paste is injected to increase conductivity;

[0017] The second step is to start the data acquisition module, connect to the computer or mobile device installed with the data processing module and user interaction module, guide the subject to enter a resting state, and adjust the detection and transmission of physiological signals to be stable;

[0018] The third step is to detect and record the breath waveform and electrophysiological signals synchronously and in real time in the resting state of the subject. The recording mode is started, the data is displayed in real time on the user interaction module interface, and is stored synchronously;

[0019] The fourth step is to transmit the collected signals to the data processing module for online preprocessing, time-frequency analysis, parameter calculation of time sequence, and extraction of olfactory function related information, and to complete feature extraction and fusion based on machine learning; in the early stage, a database is established by recording signals of animal models, patients with different degrees of olfactory dysfunction, and normal healthy people, and a multi-dimensional data and olfactory function score mapping model is built; the real-time measured data is input into the built olfactory function evaluation model, and the olfactory function score is output; the performance of the model is evaluated in debugging, and finally the objective olfactory function evaluation is realized;

[0020] The fifth step is to display the olfactory function evaluation result in the user interaction software module, generate a test report, save the subject's related information and result, and display the historical record score.

[0021] Further, the method for building the mapping model of the multi-dimensional data and the olfactory function score is as follows: first, multi-channel physiological signals of patients with different degrees of diagnosed olfactory dysfunction, animal models and healthy people are collected to build a data set, and the degree of olfactory function diagnosed by an authoritative doctor is taken as a label and set from 100 to 0 according to the decreasing olfactory function. Secondly, three one-dimensional convolutional neural networks are used to receive signals and features of each physiological signal channel respectively, and data output by each convolutional neural network is input into a multi-layer perception machine for fusion, and finally a regression value from 100 to 0 is output by the model. The data set is divided into a training set, a validation set and a test set, wherein the data of the training set is used to run a neural network learning algorithm, train the model and obtain parameter values of each part of the network. The validation set is used to observe the performance of the model during the training process to prevent overfitting. The test set is used to evaluate the performance of the final model.

[0022] Further, the performance evaluation method of the mapping model of the multi-dimensional data and the olfactory function score is as follows: the mean square error, the root mean square error and the mean absolute error of the regression prediction of the model on the test set are calculated to judge the prediction error of the model, so as to evaluate the performance of the olfactory function score model. The smaller the mean square error value is, the better the performance of the model is.

[0023] Further, the test report includes: basic information of the user, judgment of whether the olfactory function is normal, olfactory function score, and part causing the olfactory dysfunction.

[0024] The present application has the following beneficial effects relative to the current olfactory detection technology:

[0025] The present application provides a brain-computer interaction-based olfactory function detection technology and evaluation device, and provides a thought for quantitatively evaluating olfactory function based on respiratory signals and electrophysiological data, without the need for large-scale instruments and complex detection processes, and is suitable for large-scale screening of olfactory dysfunction in communities.

[0026] The present application establishes a standardized process for olfactory neural signal detection and an olfactory function evaluation model, and proposes a quantitative index for measuring olfactory function, which is objective, accurate and repeatable. Relative to the current neurophysiological test for olfactory function test, the subject's active cooperation is not required, and the interference of subjective factors of the subject and the doctor can be prevented, and it is suitable for elderly people or children with communication difficulties.

[0027] The present application uses human spontaneous signal data for standardized and unified olfactory function test, without the need for odor stimulation, and can prevent individual differences from interfering with the diagnosis results. In addition, the olfactory function evaluation model established by the present application can avoid intentional concealment of the subject and can be used for judicial identification of olfactory function damage.

[0028] The device is expected to be popularly applied to the monitoring of olfactory electrophysiological activities of actual clinical human bodies and the standardized olfactory function evaluation. BRIEF DESCRIPTION OF DRAWINGS

[0029] In order to more clearly illustrate the specific embodiments of the present application, the drawings required to be used in the specific embodiments will be briefly introduced

[0030] Figure 1 is a flow chart of the olfactory function standardized evaluation device of the present application;

[0031] Figure 2 is a schematic diagram of the measurement signal position of the present application;

[0032] Figure 3 is an example of the neural electrical signal of the normal olfactory and the olfactory disorder sample of the present application;

[0033] Figure 4 is a schematic diagram of the data processing and interaction module of the present application;

[0034] Figure 5 is an example of the output olfactory function evaluation result of the present application. DETAILED DESCRIPTION

[0035] The technical solutions of the present application will be described clearly and completely in combination with the drawings.

[0036] The present application designs an olfactory function standardized evaluation device and method based on brain-computer interaction, according to the first aspect of the embodiment, the present application provides an olfactory function standardized evaluation device based on brain-computer interaction technology, the device comprises: a front-end module, a data acquisition module, a data processing module and a user interaction module;

[0037] The front-end module comprises: a respiration sensing module for detecting temperature changes and airflow changes caused by respiration of a subject, a sensing module for detecting olfactory bulb and olfactory mucosa olfactory nerve signals, and an electroencephalogram detection module for detecting prefrontal cortex neural activity;

[0038] The data acquisition module is used for synchronously amplifying the physiological signals transmitted by the front-end module through a hardware circuit, and performing analog-to-digital conversion to convert the analog signals into digital signals, thereby completing the acquisition of multi-channel physiological signals;

[0039] The data processing module is used for the preprocessing of respiratory data and electrophysiological data, the feature extraction and fusion, the establishment of a multi-channel physiological signal database and the building of a multi-dimensional data and olfactory function score mapping model, the input of real-time measured data into the model, and the output of the olfactory function score;

[0040] The user interaction module is used to display, save and read the collected data of the front-end module, display the olfactory function evaluation results obtained by the data processing module, generate a report, save the results and the information of the examinee to the local, and upload them to the cloud server.

[0041] Further, in the front-end module, the respiratory sensing module detects the temperature change and airflow change caused by respiration through temperature and airflow sensors placed near the nasal cavity and converts them into voltage signals.

[0042] Further, the olfactory nerve signal detection module measures the neural activity signals conducted from the olfactory bulb to the surface of the skull through silver-silver chloride electrodes placed near the nasal bridge and the brow bone, and measures the olfactory mucosa neural activity signals through the neural microelectrode placed at the olfactory mucosa position in the nasal cavity.

[0043] Further, the electroencephalogram signals are detected through the electroencephalogram cap electrodes to detect the neural activity of the prefrontal cortex.

[0044] Further, the preprocessing of the data processing module is to filter the multi-channel physiological signals to remove noise, restore the olfactory nerve signals of the olfactory bulb through the source reconstruction algorithm, calculate the parameters in the time domain and frequency domain of the time series signals, complete feature extraction, including the frequency of the respiratory signal, the variance, entropy, frequency band energy of the olfactory nerve signal and the electroencephalogram signal, the cross-coupling coefficient between different frequency bands, and the cross-correlation coefficient between different physiological signals.

[0045] Further, the data processing module establishes a database by recording the signals of animal models, patients with different degrees of olfactory dysfunction and normal healthy people, uses the olfactory function degree diagnosed by authoritative doctors as the label, sets it from 100 to 0 according to the decreasing olfactory function; secondly, a three-layer one-dimensional convolutional neural network is used to receive the signals and features of each physiological signal channel, and the data output by each convolutional neural network is input into a multi-layer perceptron for fusion, and the final model outputs a regression value from 100 to 0; the data set is divided into a training set, a validation set and a test set, wherein the data of the training set is used to run the neural network learning algorithm, train the model, and obtain the parameter values of each part of the network; the validation set is used to observe the performance of the model during the training process to prevent overfitting; the test set is used to evaluate the performance of the final model.

[0046] According to another aspect of the embodiment, a standardized olfactory function evaluation method based on brain-computer interaction technology is provided, and the steps of the method are:

[0047] First step: connect the front-end module of signal detection with the subject, specifically: for respiratory recording, place the temperature and airflow sensors of the respiratory front-end module near the subject's nostrils; for the subject's olfactory mucosa, extend the neural microelectrode into the nasal cavity and attach it closely to the olfactory mucosa; for the olfactory bulb, attach the silver-silver chloride electrode to the nasal bridge and brow bone; for prefrontal cortex recording, the subject wears an electroencephalogram cap electrode and injects conductive paste to increase conductivity;

[0048] Second step: start the data acquisition module and connect it to the computer or mobile device installed with the data processing module and user interaction module, guide the subject to enter a resting state, and adjust the detection and transmission of physiological signals to be stable;

[0049] Third step: in the resting state of the subject, conduct synchronous and real-time detection and recording of respiratory waveforms and electrophysiological signals. Start the recording mode, and the data is displayed in real time on the user interaction module interface and stored synchronously;

[0050] Fourth step: transmit the collected signals to the data processing module for online preprocessing, time-frequency analysis, parameter calculation of time series, and extraction of olfactory function-related information. Based on machine learning, complete feature extraction and fusion. In the early stage, through signal recording of animal models, patients with different degrees of olfactory dysfunction, and normal healthy people, establish a database, and build a multi-dimensional data and olfactory function score mapping model. Use the built olfactory function evaluation model to input real-time measurement data into the model, output the olfactory function score, evaluate the performance of the model in debugging, and finally realize objective olfactory function evaluation;

[0051] Fifth step: display the olfactory function evaluation results in the user interaction software module, generate a test report, save the subject's relevant information and results, and display the historical record scores.

[0052] Further, the method for building the multi-dimensional data and olfactory function score mapping model is as follows: first, collect multi-channel physiological signals of patients with different degrees of diagnosed olfactory dysfunction, animal models, and healthy people, build a data set, and use the authoritative doctor's diagnosis of olfactory function degree as the label, set from 100 to 0 according to the decreasing olfactory function. Second, use a three-layer one-dimensional convolutional neural network to receive each physiological signal channel signal and feature, and input the data output by each convolutional neural network into a multi-layer perceptron for fusion, and finally the model outputs a regression value from 100 to 0. Divide the data set into a training set, a validation set, and a test set, where the data in the training set is used to run the neural network learning algorithm, train the model, and obtain the parameter values of each part of the network. The validation set is used to observe the performance of the model during training to prevent overfitting. The test set is used to evaluate the performance of the final model.

[0053] Further, the performance evaluation method of the mapping model of the multi-dimensional data and the olfactory function score is: calculating the mean square error, root mean square error and mean absolute error of the regression prediction of the model on the test set to judge the prediction error of the model, so as to evaluate the performance of the olfactory function score model, and the smaller the mean square error value, the better the model performance.

[0054] Further, the test report includes: basic information of the user, judgment of whether the olfactory function is normal, olfactory function score, and part (olfactory mucosa, olfactory bulb, prefrontal cortex) causing the olfactory function disorder.

[0055] The specific embodiments of the present application are as follows:

[0056] As shown in Figure 1 , the respiratory signal, olfactory nerve signal and electroencephalogram of the prefrontal cortex of the subject are collected, data processing, analysis and feature extraction are performed, a machine learning model is input, an olfactory function score is output, and an evaluation report is generated. The present application mainly includes the following steps:

[0057] (1) The subject correctly wears the signal collection front-end module with the help of medical care, including a respiratory sensing module, an olfactory signal detection module and an electroencephalogram detection module, and the placement position of the front-end module sensor is as shown in Figure 2 the measurement site schematic diagram. First, the electroencephalogram cap electrodes for measuring the prefrontal cortex electroencephalogram data are correctly placed on the forehead of the subject. The electrodes are injected with conductive paste to increase the conductivity; secondly, four silver-silver chloride electrodes for recording the olfactory bulb nerve activity are placed on the nose bridge and the brow bone to detect the electro-physiological signal conducted to the surface of the olfactory bulb, and the subsequent source reconstruction algorithm can be used to reconstruct the electro-physiological signal of the olfactory bulb. Thirdly, the neural microelectrode is inserted into the nasal cavity, and the electrode is attached to the olfactory mucosa to record the activity of the neurons in the olfactory mucosa. Finally, the respiratory detection front-end module is placed at the outlet of the nasal cavity, and the temperature change or airflow change caused by respiration can be detected. The above-mentioned electrodes for measuring the olfactory nerve signal and the respiratory signal can be fixed by means of a mirror frame, a nose clip and the like to prevent shaking or falling off.

[0058] (2) The front-end module is connected with the data collection module and the host. The front-end module and the data collection module are connected in a wired manner, and the recorded signals are amplified and conditioned in the data collection module. The data collection module is connected with the host installed with the data processing module and the user interaction module program through wired or wireless connection, including a computer or a smart mobile device.

[0059] The data acquisition module in the host computer is controlled to sample data, and the data acquisition module transmits the collected signals to the host computer in real time and displays them through a user interactive interface. The data processing module detects the connectivity and stability of the signals and gives corresponding indications. Medical personnel visually check the signal quality and give prompts to adjust and process the corresponding front-end module.

[0060] (3) After confirming the signal quality, formal signal synchronization acquisition and recording are started. Respiratory signals, electro-physiological signals of the olfactory epithelium and olfactory bulb, and electroencephalogram signals of the prefrontal cortex are recorded. Example signals are shown in Figure 3 , in which abnormal high-frequency oscillations appear in the signals of olfactory abnormalities. These signals are collected and input into the data processing module for automatic analysis.

[0061] (4) The collected signals are input into the data processing module. A method for evaluating olfactory function based on physiological signals is constructed: in the early stage, signals of animal models and patients with different stages of olfactory disorders and healthy people are used for feature screening, classification / regression model establishment, and evaluation index, and model parameters are trained. In the test stage, the neural signals recorded by the neural electrode are preprocessed, including effective frequency band filtering, baseline drift and power frequency interference removal, standardization, and frame processing. Then the features of neural activity are extracted, and the feature indexes reflecting the olfactory function are screened. Common features of time series signals are calculated, such as variance, frequency band energy, Shannon entropy, kurtosis, and skewness, to obtain a set of feature vectors. All the calculated features are fused and reduced. A classification / regression model is built by signal feature data to map the physiological signals and the olfactory function, and the output data is converted to 0-100 points as a quantitative index for evaluating the olfactory function. In the clinical diagnosis of olfactory function, the data of the person with impaired olfactory function are input into the pre-trained model to fine-tune the weight parameters of the last multilayer perceptron to obtain the highest performance. In the actual application, the real-time data of the test subject are transmitted to the host computer, and the real-time waveform is displayed on the user interactive interface. Through the same data processing module, the real-time data are input into the evaluation model for pattern recognition to determine the olfactory function score of the sample individual. The test subject information and the result are displayed and saved, as shown in Figure 4 . The historical information and data can be recalled in the user interactive software to display the personal information and olfactory function scores of different test subjects, as shown in Figure 5 . Finally, a standardized olfactory function evaluation based on brain-computer interaction is realized.

[0062] The above examples are used to explain and illustrate the present application, but not to limit the present application. Any modifications and changes made to the present application within the spirit and protection scope of the claims fall within the protection scope of the present application.

Claims

1. A standardized assessment device for olfactory function based on brain-computer interface technology, characterized in that, The device includes: a front-end module, a data acquisition module, a data processing module, and a user interaction module; The front-end module includes: a respiratory sensing module for detecting temperature and airflow changes caused by the subject's breathing; a sensing module for detecting olfactory nerve signals in the olfactory bulb and olfactory mucosa; and an electroencephalogram (EEG) detection module for detecting neural activity in the prefrontal cortex. The data acquisition module is used to synchronously amplify the physiological signals transmitted from the front-end module through hardware circuitry, and perform analog-to-digital conversion to convert analog signals into digital signals, thereby completing the acquisition of multi-channel physiological signals. The data processing module is used for preprocessing respiratory data and electrophysiological data, and completes feature extraction and fusion. The preprocessing of the data processing module includes: filtering multi-channel physiological signals to remove noise, restoring the olfactory nerve signals of the olfactory bulb through source reconstruction algorithm, calculating the time domain and frequency domain parameters of the time-series signals, and completing feature extraction, including the frequency of respiratory signals, variance, entropy, frequency band energy, cross-coupling coefficient between different frequency bands, and cross-correlation coefficient between different physiological signals; A multi-channel physiological signal database was established and a mapping model between multi-dimensional data and olfactory function scores was constructed. Real-time measured data were input into the model, and olfactory function scores were output. The data processing module establishes a database by recording signals from animal models, patients with varying degrees of olfactory dysfunction, and healthy individuals. The database uses the degree of olfactory function diagnosed by authoritative physicians as a label, ranging from 100 to 0, decreasing in olfactory function. Next, a three-layer one-dimensional convolutional neural network receives the signals and features of each physiological signal channel, and the data produced by each convolutional neural network is input into a multilayer perceptron for fusion. The final model outputs regression values ​​from 100 to 0. The dataset is divided into training, validation, and test sets. The training set is used to run the neural network learning algorithm, train the model, and obtain parameter values ​​for each part of the network. The validation set is used to observe model performance during training to prevent overfitting. The test set is used to evaluate the final model performance. The user interaction module is used to display, save, and read the data collected by the front-end module, display the olfactory function assessment results obtained by the data processing module, generate reports, save the results and subject information locally, and upload them to the cloud server.

2. The standardized assessment device for olfactory function based on brain-computer interface technology according to claim 1, characterized in that, In the front-end module, the breathing sensing module detects temperature and airflow changes caused by breathing through temperature and airflow sensors placed near the nasal cavity, and converts them into voltage signals.

3. The standardized assessment device for olfactory function based on brain-computer interface technology according to claim 1, characterized in that, The olfactory nerve signal detection module measures the neural activity signals transmitted from the olfactory bulb to the surface of the skull through silver-silver chloride electrodes placed externally near the bridge of the nose and brow bone, and measures the olfactory nerve signals of the olfactory mucosa through neural microelectrodes placed at the olfactory mucosa of the nasal cavity.

4. The standardized assessment device for olfactory function based on brain-computer interface technology according to claim 1, characterized in that, Electroencephalogram (EEG) signals are detected using electrodes on an EEG cap to detect neural activity in the prefrontal cortex.

Images