CN114664434A - Cerebral apoplexy rehabilitation training system for different medical institutions and training method thereof - Google Patents
Cerebral apoplexy rehabilitation training system for different medical institutions and training method thereof Download PDFInfo
- Publication number
- CN114664434A CN114664434A CN202210310987.1A CN202210310987A CN114664434A CN 114664434 A CN114664434 A CN 114664434A CN 202210310987 A CN202210310987 A CN 202210310987A CN 114664434 A CN114664434 A CN 114664434A
- Authority
- CN
- China
- Prior art keywords
- model
- data
- motor imagery
- server
- patient
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000012549 training Methods 0.000 title claims abstract description 110
- 238000000034 method Methods 0.000 title claims abstract description 17
- 208000006011 Stroke Diseases 0.000 title description 16
- 230000002490 cerebral effect Effects 0.000 title description 5
- 206010008190 Cerebrovascular accident Diseases 0.000 title description 4
- 238000002474 experimental method Methods 0.000 claims description 18
- 230000005540 biological transmission Effects 0.000 claims description 16
- 210000004556 brain Anatomy 0.000 claims description 15
- 238000007781 pre-processing Methods 0.000 claims description 12
- 230000001360 synchronised effect Effects 0.000 claims description 12
- 230000003993 interaction Effects 0.000 claims description 9
- 238000007726 management method Methods 0.000 claims description 9
- 230000002776 aggregation Effects 0.000 claims description 6
- 238000004220 aggregation Methods 0.000 claims description 6
- 238000007405 data analysis Methods 0.000 claims description 6
- 238000013523 data management Methods 0.000 claims description 6
- 230000033001 locomotion Effects 0.000 claims description 6
- 238000012360 testing method Methods 0.000 claims description 6
- 230000000694 effects Effects 0.000 claims description 5
- 230000008569 process Effects 0.000 claims description 4
- 230000003321 amplification Effects 0.000 claims description 3
- 238000013480 data collection Methods 0.000 claims description 3
- 238000003199 nucleic acid amplification method Methods 0.000 claims description 3
- 230000000717 retained effect Effects 0.000 claims description 3
- 230000000638 stimulation Effects 0.000 claims description 3
- 230000000007 visual effect Effects 0.000 claims description 3
- 238000004886 process control Methods 0.000 claims description 2
- 238000004891 communication Methods 0.000 description 3
- 239000013610 patient sample Substances 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 2
- 238000012880 independent component analysis Methods 0.000 description 2
- 239000000523 sample Substances 0.000 description 2
- 208000022306 Cerebral injury Diseases 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 230000006378 damage Effects 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 210000000337 motor cortex Anatomy 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
- G06F21/60—Protecting data
- G06F21/62—Protecting access to data via a platform, e.g. using keys or access control rules
- G06F21/6218—Protecting access to data via a platform, e.g. using keys or access control rules to a system of files or objects, e.g. local or distributed file system or database
- G06F21/6245—Protecting personal data, e.g. for financial or medical purposes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H20/00—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance
- G16H20/30—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance relating to physical therapies or activities, e.g. physiotherapy, acupressure or exercising
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Medical Informatics (AREA)
- Biomedical Technology (AREA)
- Public Health (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Data Mining & Analysis (AREA)
- Biophysics (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Artificial Intelligence (AREA)
- Bioethics (AREA)
- Mathematical Physics (AREA)
- Epidemiology (AREA)
- Primary Health Care (AREA)
- Pathology (AREA)
- Computer Hardware Design (AREA)
- Computer Security & Cryptography (AREA)
- Physical Education & Sports Medicine (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
- Rehabilitation Tools (AREA)
Abstract
The invention discloses a stroke rehabilitation training system and a training method thereof for different medical institutions, wherein the system comprises: the server distributes the global model identified by the motor imagery to the plurality of motor imagery brain-computer interface devices; the motor imagery brain-computer interface equipment is positioned in a medical institution, realizes independent model training according to the global model and local training data, and finally transmits the trained model to the server; the server aggregates the models trained by the motor imagery brain-computer interface devices to form a new global model, and distributes the new global model to the motor imagery brain-computer interface devices; the medical institution computer, the motor imagery brain-computer interface equipment is connected with the patient and the medical institution computer. The invention improves the accuracy of the model and saves the time and the training cost for the patient to carry out off-line training.
Description
Technical Field
The invention relates to a rehabilitation training system and a training method thereof, in particular to a stroke rehabilitation training system and a training method thereof for different medical institutions.
Background
China is one of the high-incidence countries of cerebral apoplexy, and the cerebral apoplexy is used as a cerebral injury disease, and about 70 to 80 percent of patients lose the control capability of limbs due to the damage of cerebral motor cortex. The motor imagery brain-computer interface technology can judge the motor intention of a patient by recording and decoding the electroencephalogram signals, and the motor intention triggers a rehabilitation peripheral to perform treatment work, so that closed-loop training and active rehabilitation are realized. Compared with the traditional rehabilitation training technology, the motor imagery brain-computer interface has a better effect on the rehabilitation training of the stroke patient.
However, motor imagery brain-computer interfaces require extensive training by the patient in advance before use, or by building deep learning models with the help of large data from others and small data from new patients. The method can obtain higher modeling accuracy by acquiring the motor imagery electroencephalogram data of a plurality of mechanisms, can save the modeling training time and cost, and is a better choice. However, the method needs a large number of other patient samples, and due to the fact that privacy of medical data is extremely high, data transmission and sharing are very sensitive, information sharing of data sets stored in different organizations is difficult to achieve, and full utilization of medical big data is limited. How to solve the islanding problem of the medical data of the stroke patient under the condition of protecting privacy and safety becomes a technical problem to be solved urgently by realizing more accurate modeling based on data from different medical institutions.
Disclosure of Invention
In order to overcome the defects of the prior art, the invention provides a stroke rehabilitation training system and a stroke rehabilitation training method for different medical institutions.
The invention solves the technical problems through the following technical scheme: a stroke rehabilitation training system for different medical institutions is characterized by comprising:
the server distributes the global model identified by the motor imagery to the plurality of motor imagery brain-computer interface devices;
the motor imagery brain-computer interface equipment is positioned in a medical institution, realizes independent model training according to the global model and local training data, and finally transmits the trained model to the server; the server aggregates the models trained by the motor imagery brain-computer interface devices to form a new global model, and distributes the new global model to the motor imagery brain-computer interface devices;
the medical institution computer is connected with the patient and the medical institution computer, the medical institution computer is connected with the server, the patient only needs to perform a small amount of local training to generate local training data before performing rehabilitation training, and independent model training is realized according to the global model and the local training data.
Preferably, the motor imagery brain-computer interface device is brain electrical signal acquisition equipment for acquiring motor imagery brain electrical signals when a patient performs related operations according to guidance of a user interaction interface, medical institution professionals use medical institution computers to help the patient to acquire motor imagery sample data, motor imagery brain electrical characteristics and a motor instruction label are used as input samples, and supervised learning is performed on the global model; after training and learning are completed, the computer of the medical institution encrypts the trained model parameters and transmits the trained model parameters to the server through the router.
Preferably, the medical institution computer comprises a data acquisition module, an experiment management module, a model training module and a model transmission module, wherein the data acquisition module realizes acquisition, amplification and transmission of electroencephalogram signals; the experiment management module adopts a training mode of intention expression, namely, the patient is required to carry out corresponding motor imagery according to instructions to obtain corresponding experiment data; the model training module trains according to the global model and the experimental data obtained from the server to obtain a motion imagery identification model with a better effect, the model transmission module needs to obtain and decrypt a new global model from the server, and after each experiment is finished, the model parameters updated locally are encrypted and sent to the server.
Preferably, the model transmission module includes:
the first encryption unit encrypts the trained model parameters;
the first sending unit sends the encrypted trained model parameters to the server;
the server comprises a first receiving unit and a first decryption unit, wherein the first receiving unit and the first decryption unit respectively receive and decrypt the global model parameters sent by the server.
Preferably, the server includes:
the second receiving unit and the second decryption unit respectively receive and decrypt the trained model parameters encrypted by the first encryption unit;
the model aggregation unit aggregates the models trained by the motor imagery brain-computer interface devices to form a new global model;
a second encryption unit that encrypts the global model parameter;
and the second sending unit is used for sending the encrypted global model parameters to the motor imagery brain-computer interface equipment.
Preferably, the data acquisition module is connected with a storage module, and the storage module matches and stores the acquired amplified electroencephalogram signal and the tag to form an electroencephalogram database; the electroencephalogram database stores the electroencephalogram data of the patient which is retained after the patient agrees.
Preferably, a signal amplifier is connected between the motor imagery brain-computer interface device and the medical institution computer and is used for amplifying off-line training brain electrical data and on-line testing brain electrical data to generate corresponding amplified brain electrical signals.
Preferably, a multi-parameter synchronizer is connected between the motor imagery brain-computer interface device and the medical institution computer and is used for marking off-line training synchronous label information and on-line testing synchronous label information to ensure parameter synchronization.
Preferably, the medical institution computer is connected with a manipulator auxiliary device, and the medical institution computer drives the corresponding hand of the patient to move through the manipulator auxiliary device according to the recognized movement intention.
Preferably, the experiment management module comprises a user interaction interface for performing experiment process control, and the user interaction interface comprises at least one of the following three sub-interfaces: the system comprises a starting interface, a data management interface and a rehabilitation training interface, wherein the starting interface is used for collecting personnel registration information and selecting the operation to be performed, the data management interface is used for managing an experimental process and checking, selecting and storing experimental data, and the rehabilitation training interface is used for playing two experimental paradigms for a patient and guiding the patient, and comprises a left-hand grasping imagination interface and a right-hand grasping imagination interface.
Preferably, the model training module comprises:
the EEG data preprocessing unit is used for preprocessing the acquired original electroencephalogram signals to obtain electroencephalogram data which can be analyzed and modeled;
the CNN-based offline decoding modeling unit is used for classifying electroencephalogram data and known labels in a database by adopting a CNN algorithm, and establishing and updating two classification data analysis models of left-hand grasping and right-hand grasping;
and the online identification unit based on the CNN is used for carrying out online classification identification on the data acquired in real time by adopting the established two-classification data analysis model and outputting a prediction label.
The invention also provides a training method of the stroke rehabilitation training system for different medical institutions, which is characterized in that the method adopts the stroke rehabilitation training system for different medical institutions, and comprises the following steps:
step one, a professional helps a patient to wear a motor imagery brain-computer interface device and connects the motor imagery brain-computer interface device, a signal amplifier, a medical institution computer and a server;
step two, starting a medical institution computer, and acquiring a global model identified by the motor imagery from a server by the medical institution computer;
thirdly, connecting a medical institution computer with a multi-parameter synchronizer;
step four, the medical institution computer collects the signals through the motor imagery brain-computer interface device, the patient performs motor imagery according to the prompt, meanwhile, the motor imagery brain-computer interface device collects electroencephalogram signals, and after the signals are amplified by a signal amplifier, the synchronous label signals transmit the motor imagery electroencephalogram signals to a data collection module of the medical institution computer in a wired or wireless transmission mode;
step five, after the patient agrees, the electroencephalogram data and the synchronous label information of the patient are added to an electroencephalogram database;
step six, preprocessing the original data by an EEG data preprocessing unit;
step seven, classifying the electroencephalogram data and the known labels in the electroencephalogram database by adopting the CNN model obtained from the server in the step two, and improving the accuracy and precision of the model in use;
step eight, judging whether the training reaches the required accuracy, if not, turning to the step four, otherwise, turning to the step nine;
step nine, obtaining a personalized model of the motor imagery of the patient, and performing auxiliary rehabilitation training on the patient by using mechanical arm auxiliary equipment in combination with electroencephalogram information and visual stimulation;
step ten, after the consent of the patient, encrypting the trained model parameters and sending the encrypted model parameters to a server;
and step eleven, the server decrypts the received model parameters and carries out aggregation to obtain the global model parameters of the server side.
The positive progress effects of the invention are as follows: the invention uses the local training data of a plurality of motor imagery brain-computer interface devices of a plurality of medical institutions to obtain a global model of federal learning training, and finally obtains a more accurate motor imagery model according with the characteristics of a user through a small amount of local training, thereby not only improving the precision of the model, but also saving the time and the training cost of off-line training of patients. The invention does not directly obtain the original privacy data of the patient when carrying out model training, and only exchanges model parameters with a computer of a medical institution, thereby ensuring the privacy security of the patient.
Drawings
Fig. 1 is a diagram of a stroke rehabilitation training system for different medical institutions according to the present invention.
Detailed Description
The following provides a detailed description of the preferred embodiments of the present invention with reference to the accompanying drawings.
As shown in fig. 1, the stroke rehabilitation training system for different medical institutions of the present invention includes:
the server distributes the global model identified by the motor imagery to the plurality of motor imagery brain-computer interface devices;
the motor imagery brain-computer interface equipment is positioned in a medical institution, realizes independent model training according to the global model and local training data, and finally transmits the trained model to the server; the server aggregates the models trained by the motor imagery brain-computer interface devices to form a new global model, and distributes the new global model to the motor imagery brain-computer interface devices;
the medical institution computer, the motor imagery brain-computer interface equipment is connected with the patient and the medical institution computer, the patient only needs to carry out a small amount of local training to generate local training data before carrying out rehabilitation training, independent model training is realized according to the global model and the local training data, and a personalized motor imagery intention recognition model with higher precision can be obtained for subsequent rehabilitation training.
Only the model and the hyper-parameters of the model are exchanged between the server and a plurality of medical institution computers, and the data of the patient sample is not exchanged, thereby ensuring the privacy and the safety of the patient.
The medical institution is a hospital or a rehabilitation institution, and has wide application range.
The motor imagery brain-computer interface equipment adopts Quick30 series lead dry electrode cap of CGX company, and is connected with medical institution computer by wire serial communication or wireless Bluetooth mode, and the structure is simple and convenient for connection.
The motor imagery brain-computer interface device is brain electrical acquisition equipment and is used for acquiring motor imagery brain electrical signals when a patient performs related operations according to guidance of a user interaction interface, medical institution professionals use medical institution computers to help the patient to acquire motor imagery sample data, motor imagery brain electrical characteristics and a motor instruction label are used as input samples, and supervised learning is performed on a global model; after training and learning are completed, the computer of the medical institution encrypts the trained model parameters and transmits the trained model parameters to the server through the router, so that secret leakage is prevented, the safety of communication data is guaranteed, and the patient obtains a personalized motor imagery intention recognition model for subsequent rehabilitation treatment.
The medical institution computer comprises a data acquisition module, an experiment management module, a model training module and a model transmission module, wherein the data acquisition module realizes acquisition, amplification and transmission of electroencephalogram signals; the experiment management module adopts a training mode of intention expression, namely, the patient is required to carry out corresponding motor imagery according to instructions to obtain corresponding experiment data; the model training module trains according to the global model and the experimental data obtained from the server to obtain a motion imagery identification model with a better effect, the model transmission module needs to obtain and decrypt a new global model from the server, and after each experiment is finished, the model parameters updated locally are encrypted and sent to the server.
The model transmission module includes:
the first encryption unit encrypts the trained model parameters;
the first sending unit sends the encrypted trained model parameters to the server;
the server comprises a first receiving unit and a first decryption unit, wherein the first receiving unit and the first decryption unit respectively receive and decrypt the global model parameters sent by the server.
The server includes:
the second receiving unit and the second decryption unit respectively receive and decrypt the trained model parameters encrypted by the first encryption unit;
the model aggregation unit aggregates the models trained by the plurality of motor imagery brain-computer interface devices to form a new global model;
a second encryption unit that encrypts the global model parameter;
and the second sending unit is used for sending the encrypted global model parameters to the motor imagery brain-computer interface equipment.
The data acquisition module is connected with a storage module, and the storage module matches and stores the acquired amplified electroencephalogram signal and the label to form an electroencephalogram database; the electroencephalogram database stores the electroencephalogram data of the patient which is retained after the patient agrees.
A signal amplifier is connected between the motor imagery brain-computer interface device and the medical institution computer and is used for amplifying off-line training brain-computer data and on-line testing brain-computer data to generate corresponding amplified brain-computer signals.
A multi-parameter synchronizer is connected between the motor imagery brain-computer interface equipment and a medical institution computer, and specifically can be connected through a USB serial port and used for marking off-line training synchronous label information and on-line testing synchronous label information and ensuring parameter synchronization.
The medical institution computer is connected with a manipulator auxiliary device, and can be specifically connected through a USB serial port, and the medical institution computer drives the corresponding hand of the patient to move through the manipulator auxiliary device according to the recognized movement intention.
The experiment management module comprises a user interaction interface for controlling the experiment process, wherein the user interaction interface comprises at least one of the following three sub-interfaces: the system comprises a starting interface, a data management interface and a rehabilitation training interface, wherein the starting interface is used for collecting personnel registration information and selecting the operation to be performed, the data management interface is used for managing an experimental process and checking, selecting and storing experimental data, and the rehabilitation training interface is used for playing two experimental paradigms for a patient and guiding the patient, and comprises a left-hand grasping imagination interface and a right-hand grasping imagination interface.
The model training module comprises:
the EEG data preprocessing unit is used for preprocessing the acquired original EEG signals to obtain EEG data which can be analyzed and modeled;
the CNN-based offline decoding modeling unit is used for classifying electroencephalogram data and known labels in a database by adopting a CNN algorithm, and establishing and updating two classification data analysis models of left-hand grasping and right-hand grasping;
and the CNN-based online identification unit is used for carrying out online classification identification on the data acquired in real time by adopting the established two-classification data analysis model and outputting a prediction label.
The invention relates to a training method of a stroke rehabilitation training system for different medical institutions, which adopts the stroke rehabilitation training system for different medical institutions, and comprises the following steps:
step one, a professional helps a patient to wear a motor imagery brain-computer interface device and connects the motor imagery brain-computer interface device, a signal amplifier, a medical institution computer and a server;
step two, starting a medical institution computer, and acquiring a global model identified by the motor imagery from a server by the medical institution computer;
thirdly, connecting a medical institution computer with a multi-parameter synchronizer;
step four, the medical institution computer collects the signals through the motor imagery brain-computer interface device, the patient performs motor imagery according to the prompt, meanwhile, the motor imagery brain-computer interface device collects electroencephalogram signals, and after the signals are amplified by a signal amplifier, the synchronous label signals transmit the motor imagery electroencephalogram signals to a data collection module of the medical institution computer in a wired or wireless transmission mode;
step five, after the patient agrees, the electroencephalogram data and the synchronous label information of the patient are added to an electroencephalogram database;
preprocessing the original data by an EEG data preprocessing unit;
step seven, classifying the electroencephalogram data and the known labels in the electroencephalogram database by adopting the CNN model obtained from the server in the step two, and improving the accuracy and precision of the model in use;
step eight, judging whether the training reaches the required accuracy, if not, turning to the step four, otherwise, turning to the step nine;
step nine, obtaining a personalized model of the motor imagery of the patient, and performing auxiliary rehabilitation training on the patient by using mechanical arm auxiliary equipment in combination with electroencephalogram information and visual stimulation;
step ten, after the consent of the patient, encrypting the trained model parameters and sending the encrypted model parameters to a server;
and step eleven, the server decrypts the received model parameters and carries out aggregation to obtain the global model parameters of the server side.
In the second step and the tenth step, the motor imagery brain-computer interface device communicates with the server by adopting a TCP/IP protocol to transmit model data. The client and the server use homomorphic encryption to ensure the safety of communication data.
The sixth step comprises the following substeps:
sixty one, channel screening is carried out on the original EEG data, and unused channels A1 and A2 are removed;
sixty-two, loading a channel spatial position configuration file, and carrying out spatial position distribution on EEG data;
sixty-three, re-referencing by using the average potential, and updating an EEG data reference point;
sixty-four steps, carrying out band-pass filtering of 0.1-30 Hz on the EEG data by using an eight-order Butterworth filter;
sixty-five steps, segmenting the data according to the label information, taking the label position as an initial position, and taking the data length of [ -200ms, 800ms ] for segmentation to obtain a data segment with the time length of 1 s;
sixteenth, performing base line correction of [ -200ms, 0] on all data segments;
sixty-seventeen, distinguishing each independent component in the data by using an Independent Component Analysis (ICA) method, and removing the electro-oculogram and electro-myoelectric components.
In the eleventh step, the server uses a federal average algorithm to aggregate the trained models, so that the privacy and safety of user data can be protected.
The invention adopts a centralized framework, a server is arranged, each medical institution computer is taken as a client for federal learning, the server distributes a global model for motor imagery identification to a plurality of clients, a patient only needs to carry out a small amount of local training before carrying out rehabilitation training, the client realizes independent model training according to the model and local training data, and a personalized motor imagery intention identification model with higher precision can be obtained for subsequent rehabilitation training. And simultaneously, the client transmits the trained model to the server. And the server aggregates the client models to form a new global model and distributes the new global model to the clients. Only the model and the hyper-parameters of the model are exchanged between the server and the client, and the data of the patient sample is not exchanged.
While there have been shown and described what are at present considered the fundamental principles and essential features of the invention and its advantages, it will be apparent to those skilled in the art that the invention is not limited to the details of the foregoing exemplary embodiments, but is capable of other specific forms without departing from the spirit or essential characteristics thereof. The present embodiments are therefore to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. Any reference sign in a claim should not be construed as limiting the claim concerned.
Furthermore, it should be understood that although the present description refers to embodiments, not every embodiment includes only a single embodiment, and such description is for clarity only, and those skilled in the art should make the description as a whole, and the embodiments may be appropriately combined to form other embodiments understood by those skilled in the art.
Claims (12)
1. A stroke rehabilitation training system for different medical institutions is characterized by comprising:
the server distributes the global model identified by the motor imagery to the plurality of motor imagery brain-computer interface devices;
the motor imagery brain-computer interface equipment is positioned in a medical institution, realizes independent model training according to the global model and local training data, and finally transmits the trained model to the server; the server aggregates the models trained by the motor imagery brain-computer interface devices to form a new global model, and distributes the new global model to the motor imagery brain-computer interface devices;
the medical institution computer is connected with the patient and the medical institution computer, the medical institution computer is connected with the server, the patient only needs to perform a small amount of local training to generate local training data before performing rehabilitation training, and independent model training is realized according to the global model and the local training data.
2. The stroke rehabilitation training system for different medical institutions according to claim 1, wherein the motor imagery brain-computer interface device is an electroencephalogram acquisition device, and is used for acquiring motor imagery electroencephalogram signals when a patient performs related operations according to guidance of a user interaction interface, and medical institution professionals use medical institution computers to help the patient acquire motor imagery sample data, and perform supervised learning on a global model by using motor imagery electroencephalogram characteristics and a motor instruction label as input samples; after training and learning are completed, the computer of the medical institution encrypts the trained model parameters and transmits the trained model parameters to the server through the router.
3. The stroke rehabilitation training system for different medical institutions according to claim 1, wherein the medical institution computer comprises a data acquisition module, an experiment management module, a model training module and a model transmission module, and the data acquisition module realizes acquisition, amplification and transmission of electroencephalogram signals; the experiment management module adopts a training mode of intention expression, namely, the patient is required to carry out corresponding motor imagery according to instructions to obtain corresponding experiment data; the model training module trains according to the global model and the experimental data obtained from the server to obtain a motion imagery identification model with a better effect, the model transmission module needs to obtain and decrypt a new global model from the server, and after each experiment is finished, the model parameters updated locally are encrypted and sent to the server.
4. The stroke rehabilitation training system for different medical institutions according to claim 3, wherein the model transmission module comprises:
the first encryption unit encrypts the trained model parameters;
the first sending unit sends the encrypted trained model parameters to the server;
the server comprises a first receiving unit and a first decryption unit, wherein the first receiving unit and the first decryption unit respectively receive and decrypt the global model parameters sent by the server.
5. The stroke rehabilitation training system for different medical institutions according to claim 4, wherein the server comprises:
the second receiving unit and the second decryption unit respectively receive and decrypt the trained model parameters encrypted by the first encryption unit;
the model aggregation unit aggregates the models trained by the motor imagery brain-computer interface devices to form a new global model;
a second encryption unit that encrypts the global model parameter;
and the second sending unit is used for sending the encrypted global model parameters to the motor imagery brain-computer interface equipment.
6. The stroke rehabilitation training system for different medical institutions according to claim 5, wherein the data acquisition module is connected with a storage module, and the storage module is used for matching and storing the acquired amplified brain electrical signals and the tags to form a brain electrical database; the electroencephalogram database stores the electroencephalogram data of the patient which is retained after the patient agrees.
7. The stroke rehabilitation training system for different medical institutions according to claim 5, wherein a signal amplifier is connected between the motor imagery brain-computer interface device and the medical institution computer and used for amplifying off-line training brain electrical data and on-line testing brain electrical data to generate corresponding amplified brain electrical signals.
8. The stroke rehabilitation training system for different medical institutions according to claim 5, wherein a multi-parameter synchronizer is connected between the motor imagery brain interface device and the medical institution computer and used for marking off-line training synchronous label information and on-line testing synchronous label information to ensure parameter synchronization.
9. The stroke rehabilitation training system for different medical institutions according to claim 5, wherein the medical institution computer is connected with a manipulator auxiliary device, and the medical institution computer drives the corresponding hand of the patient to move through the manipulator auxiliary device according to the recognized movement intention.
10. The stroke rehabilitation training system for different medical institutions according to claim 5, wherein the experiment management module comprises a user interaction interface for performing experiment process control, and the user interaction interface comprises at least one of the following three sub-interfaces: the system comprises a starting interface, a data management interface and a rehabilitation training interface, wherein the starting interface is used for collecting personnel registration information and selecting the operation to be performed, the data management interface is used for managing an experimental process and checking, selecting and storing experimental data, and the rehabilitation training interface is used for playing two experimental paradigms for a patient and guiding the patient, and comprises a left-hand grasping imagination interface and a right-hand grasping imagination interface.
11. The stroke rehabilitation training system for different medical institutions according to claim 5, wherein the model training module comprises:
the EEG data preprocessing unit is used for preprocessing the acquired original EEG signals to obtain EEG data which can be analyzed and modeled;
the CNN-based offline decoding modeling unit is used for classifying electroencephalogram data and known labels in a database by adopting a CNN algorithm, and establishing and updating two classification data analysis models of left-hand grasping and right-hand grasping;
and the online identification unit based on the CNN is used for carrying out online classification identification on the data acquired in real time by adopting the established two-classification data analysis model and outputting a prediction label.
12. A training method of a stroke rehabilitation training system for different medical institutions, which is characterized by adopting the stroke rehabilitation training system for different medical institutions according to claim 1, and comprises the following steps:
step one, a professional helps a patient to wear a motor imagery brain-computer interface device, and the motor imagery brain-computer interface device, a signal amplifier, a medical institution computer and a server are connected;
step two, starting a medical institution computer, and acquiring a global model identified by the motor imagery from a server by the medical institution computer;
thirdly, connecting a medical institution computer with a multi-parameter synchronizer;
step four, the medical institution computer collects the signals through the motor imagery brain-computer interface device, the patient performs motor imagery according to the prompt, meanwhile, the motor imagery brain-computer interface device collects electroencephalogram signals, and after the signals are amplified by a signal amplifier, the synchronous label signals transmit the motor imagery electroencephalogram signals to a data collection module of the medical institution computer in a wired or wireless transmission mode;
step five, after the consent of the patient, the electroencephalogram data and the synchronous label information of the patient are added to an electroencephalogram database;
preprocessing the original data by an EEG data preprocessing unit;
step seven, classifying the electroencephalogram data and the known labels in the electroencephalogram database by adopting the CNN model obtained from the server in the step two, and improving the accuracy and precision of the model in use;
step eight, judging whether the training reaches the required accuracy, if not, turning to the step four, otherwise, turning to the step nine;
step nine, obtaining a personalized model of the motor imagery of the patient, and performing auxiliary rehabilitation training on the patient by using mechanical arm auxiliary equipment in combination with electroencephalogram information and visual stimulation;
step ten, after the consent of the patient, encrypting the trained model parameters and sending the encrypted model parameters to a server;
and step eleven, the server decrypts the received model parameters and carries out aggregation to obtain the global model parameters of the server side.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210310987.1A CN114664434A (en) | 2022-03-28 | 2022-03-28 | Cerebral apoplexy rehabilitation training system for different medical institutions and training method thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210310987.1A CN114664434A (en) | 2022-03-28 | 2022-03-28 | Cerebral apoplexy rehabilitation training system for different medical institutions and training method thereof |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114664434A true CN114664434A (en) | 2022-06-24 |
Family
ID=82033032
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210310987.1A Pending CN114664434A (en) | 2022-03-28 | 2022-03-28 | Cerebral apoplexy rehabilitation training system for different medical institutions and training method thereof |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114664434A (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110765920A (en) * | 2019-10-18 | 2020-02-07 | 西安电子科技大学 | Motor imagery classification method based on convolutional neural network |
| CN111110982A (en) * | 2019-12-02 | 2020-05-08 | 燕山大学 | Hand rehabilitation training method based on motor imagery |
| CN111638803A (en) * | 2020-06-19 | 2020-09-08 | 深圳前海微众银行股份有限公司 | Federal migration-based electroencephalogram data processing method, device, equipment and medium |
| WO2020229684A1 (en) * | 2019-05-16 | 2020-11-19 | Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. | Concepts for federated learning, client classification and training data similarity measurement |
| CN113128384A (en) * | 2021-04-01 | 2021-07-16 | 北京工业大学 | Brain-computer interface software key technical method of stroke rehabilitation system based on deep learning |
| CN114048780A (en) * | 2021-11-15 | 2022-02-15 | 中国科学院深圳先进技术研究院 | Electroencephalogram classification model training method and device based on federal learning |
-
2022
- 2022-03-28 CN CN202210310987.1A patent/CN114664434A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020229684A1 (en) * | 2019-05-16 | 2020-11-19 | Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. | Concepts for federated learning, client classification and training data similarity measurement |
| CN110765920A (en) * | 2019-10-18 | 2020-02-07 | 西安电子科技大学 | Motor imagery classification method based on convolutional neural network |
| CN111110982A (en) * | 2019-12-02 | 2020-05-08 | 燕山大学 | Hand rehabilitation training method based on motor imagery |
| CN111638803A (en) * | 2020-06-19 | 2020-09-08 | 深圳前海微众银行股份有限公司 | Federal migration-based electroencephalogram data processing method, device, equipment and medium |
| CN113128384A (en) * | 2021-04-01 | 2021-07-16 | 北京工业大学 | Brain-computer interface software key technical method of stroke rehabilitation system based on deep learning |
| CN114048780A (en) * | 2021-11-15 | 2022-02-15 | 中国科学院深圳先进技术研究院 | Electroencephalogram classification model training method and device based on federal learning |
Non-Patent Citations (1)
| Title |
|---|
| 魏立斐;陈聪聪;张蕾;李梦思;陈玉娇;王勤;: "机器学习的安全问题及隐私保护", 计算机研究与发展, no. 10, 9 October 2020 (2020-10-09), pages 60 - 79 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Amin et al. | Cognitive smart healthcare for pathology detection and monitoring | |
| Wang et al. | Real-time mental arithmetic task recognition from EEG signals | |
| CN109872803B (en) | Artificial intelligence pathology marking system | |
| CN106777873A (en) | Community's smart cloud health services system | |
| CN107595302B (en) | Device and method for monitoring mental state of user by electroencephalogram signals | |
| KR102162683B1 (en) | Reading aid using atypical skin disease image data | |
| US11217345B2 (en) | Anonymization of audio-visual medical data | |
| CN105574559B (en) | A kind of mobile high-voltage oxygen workstation system based on Internet of Things | |
| EP3679831A1 (en) | Method for determining movement patterns in dental treatment | |
| CN103294192A (en) | LED lamp switch control device and control method thereof based on motor imagery | |
| CN109222931A (en) | Intelligent guarding system | |
| KR102031958B1 (en) | Prefrontal based cognitive brain-machine interfacing apparatus and method thereof | |
| Mahajan et al. | A real time set up for retrieval of emotional states from human neural responses | |
| Song et al. | Mobile robot control by BCI based on motor imagery | |
| CN114664434A (en) | Cerebral apoplexy rehabilitation training system for different medical institutions and training method thereof | |
| CN116665281B (en) | Key emotion extraction method based on doctor-patient interaction | |
| CN108538385A (en) | A kind of critical value warning system of medical test | |
| Martisius et al. | Design of a neural interface based system for control of robotic devices | |
| US20210076942A1 (en) | Infrared thermography for intraoperative functional mapping | |
| CN114566275A (en) | Pre-hospital emergency auxiliary system based on mixed reality | |
| WO2018075754A1 (en) | System and method for determining amount of volition in a subject | |
| CN106095924A (en) | Personal data collecting system and personal data collection generate and application process | |
| Healy et al. | Experiences and insights from the collection of a novel multimedia EEG dataset | |
| Nour | Artificial Intelligence (AI) for Improving Performance at the Cutting Edge of Medical Imaging | |
| Kim et al. | Identifying error features in a MI-BCI system using microstates |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |