Motion function evaluation method, device and equipment and computer readable storage medium

Abstract

The invention discloses a motion function evaluation method, device and equipment and a computer readable storage medium, the method is applied to a motion function evaluation system, the motion function evaluation system comprises an image acquisition module and a motion analysis module, and the motion analysis module carries a motion analysis model. The motion analysis model comprises a convolutional neural network model, a long-short term memory neural network model and a similarity evaluation model. The method can comprise the following steps: acquiring at least one frame of motion image of an evaluated person through an image acquisition module; the method comprises the following steps: extracting features of a motion image through a convolutional neural network carried in a motion analysis module, inputting the features of the motion image into a long-short term memory neural network model, identifying the motion of an evaluated person in the motion image, obtaining the similarity between the motion of the evaluated person and a standard motion through a similarity evaluation model, and calculating the motion of the evaluated person according to the similarity. And obtaining a motion function evaluation result of the evaluated person according to the similarity, so that the motion function of the evaluated person can be scientifically evaluated.

Description

Technical Field

[0001] The present application relates to the field of smart medical technology, and specifically to a motor function assessment method, device, equipment and computer-readable storage medium. Background Art

[0002] Normal human movement is the balance and coordination of movement completed with the participation of the motor area of ​​the cerebral cortex, the basal ganglia of the cortex, the vestibular labyrinth system, deep sensation, vision, etc., which is called ataxia. Lesions of these structures lead to coordination disorders, which is called ataxia. According to the cause of the lesion, it can be divided into sensory ataxia, cerebellar ataxia, vestibular ataxia, etc. It has important reference significance for the diagnosis of diseases in related parts. At present, the diagnosis of ataxia function mainly relies on physical examination, such as finger-to-nose test, finger-to-finger test, alternating movements, heel-knee-shin test, closed-eyes difficult standing test, etc. Objective examinations such as imaging and electrophysiology cannot accurately reflect the patient's ataxia function.

[0003] The traditional method of assessing the motor function is achieved through a motor function assessment scale. Taking the commonly used Fugl-Meyer upper limb motor function assessment scale as an example, the scale includes 33 assessment indicators. Each item is scored from 0 to 2 points, where 0 points means that it cannot be completed at all, 1 point means that it can be partially completed, and 2 points means that it can be completed completely. The specific assessment process is that the doctor gives the corresponding score by subjectively observing the mobility of each joint of the limbs of the person being assessed.

[0004] It is understandable that the above method relies on the subjective observation of doctors and lacks objective standards, so the accuracy of the evaluation results is not high. How to improve the accuracy of motor function evaluation has become a technical problem that needs to be solved urgently. Summary of the invention

[0005] The present application discloses a motor function assessment method, device, equipment and computer-readable storage medium, which can scientifically assess the motor function of a person being assessed with high accuracy.

[0006] In a first aspect, the present application discloses a method for evaluating motor function, which is applied to a motor function evaluation system, wherein the motor function evaluation system comprises an image acquisition module and a motion analysis module connected in sequence, wherein the motion analysis module is equipped with a motion analysis model, and the motion analysis model comprises a convolutional neural network model, a long short-term memory neural network model and a similarity evaluation model, wherein the output of the convolutional neural network is the input of the long short-term memory neural network, and the output of the long short-term memory neural network is the input of the similarity evaluation model, and the method may comprise the following steps:

[0007] Acquire at least one frame of motion image of the person being evaluated through the image acquisition module, wherein the motion image is an image of the person being evaluated when training according to a preset training item, and each motion image indicates the position, posture and motion process information of the person being evaluated;

[0008] The features of the motion image are extracted by the convolutional neural network carried by the motion analysis module, and the features of the motion image include at least one of orientation features, posture features and motion features. The features of the motion image are input into the long short-term memory neural network model, and the movements of the person being evaluated in the motion image are identified. The similarity between the movements of the person being evaluated and the standard movements is obtained by the similarity evaluation model, and the motor function evaluation result of the person being evaluated is obtained according to the similarity.

[0009] In this way, at least one frame of motion image of the person being evaluated is obtained through the image acquisition module, the features of the motion image are extracted through the convolutional neural network carried by the motion analysis module, and the extracted features of the motion image are input into the long short-term memory neural network model to identify the work of the person being evaluated in the motion image, and the similarity between the actions of the person being evaluated and the standard actions is obtained through the similarity evaluation model, so that the motor function evaluation result of the person being evaluated can be obtained according to the similarity. The evaluation process can avoid human subjective judgment, making the evaluation process more objective, more refined, and with higher accuracy.

[0010] In a possible implementation, the motor function assessment system further includes a display module, and the display module is connected to the motion analysis module. The method further includes:

[0011] The motor function assessment result of the assessed person is displayed through the display module.

[0012] In this way, the motor function assessment results of the person being assessed are displayed through the display module so that the doctor can timely understand the motor function assessment results of the person being assessed, and the motor function assessment results can be used for rehabilitation guidance, rehabilitation big data collection and analysis, health monitoring, etc. of the person being assessed.

[0013] In a possible implementation, the motor function assessment system further includes a playback module, and the playback module is connected to the image acquisition module. The method further includes:

[0014] The video of the preset training item is played through the playing module.

[0015] In this way, it is convenient for the person being assessed to learn the preset training items.

[0016] In a possible implementation, the motion image includes an image of a training target action of the person being assessed, and the method further includes:

[0017] Acquire historical training images of the training target action of the person being evaluated;

[0018] Extracting features of the historical training images through the convolutional neural network carried by the motion analysis module;

[0019] The step of inputting the features of the motion image into the long short-term memory neural network model to identify the actions of the person being evaluated in the motion image includes:

[0020] The features of the historical training images and the features of the motion images are input into the long short-term memory neural network model to identify the actions of the person being evaluated.

[0021] In this way, historical training images as well as motion images acquired in real time are comprehensively considered in the evaluation, which can improve the evaluation accuracy.

[0022] In a possible implementation, the long short-term memory neural network model cascades T cells, and the value of T is equal to the number of features of the motion image.

[0023] In this way, the long short-term memory neural network model cascades T cells, and the value of T is equal to the number of features of the motion image, which can improve the evaluation accuracy.

[0024] In a possible implementation, the preset training items include one of finger-to-nose exercise, finger-to-finger exercise, alternating exercise, heel-knee-shin exercise, and eyes-closed difficult standing exercise.

[0025] In a second aspect, the present application provides a motor function assessment device, which is applied to a motor function assessment system, wherein the motor function assessment system includes an image acquisition module and a motion analysis module connected in sequence, wherein the motion analysis module is equipped with a motion analysis model, and the motion analysis model includes a convolutional neural network model, a long short-term memory neural network model and a similarity evaluation model, wherein the output of the convolutional neural network is the input of the long short-term memory neural network, and the output of the long short-term memory neural network is the input of the similarity evaluation model, wherein,

[0026] Acquire at least one frame of motion image of the person being evaluated through the image acquisition module, wherein the motion image is an image of the person being evaluated when training according to a preset training item, and each motion image indicates the position, posture and motion process information of the person being evaluated;

[0027] The features of the motion image are extracted by the convolutional neural network carried by the motion analysis module, and the features of the motion image include at least one of orientation features, posture features and motion features. The features of the motion image are input into the long short-term memory neural network model, and the movements of the person being evaluated in the motion image are identified. The similarity between the movements of the person being evaluated and the standard movements is obtained by the similarity evaluation model, and the motor function evaluation result of the person being evaluated is obtained according to the similarity.

[0028] In a third aspect, an embodiment of the present application provides a terminal device, comprising a processor, an input device, an output device and a memory, wherein the processor, input device, output device and memory are interconnected, wherein the memory is used to store a computer program that supports the device to execute the above method, the computer program includes program instructions, and the processor is configured to call the program instructions to execute the method of the above first aspect.

[0029] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, wherein the computer storage medium stores a computer program, wherein the computer program includes program instructions, and when the program instructions are executed by a processor, the processor executes the method of the first aspect above.

[0030] In a fifth aspect, an embodiment of the present application further provides a computer program, wherein the computer program includes program instructions, and when the program instructions are executed by a processor, the processor executes the method of the first aspect. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the background technology, the drawings required for use in the embodiments of the present application or the background technology will be described below.

[0032] Figure 1 It is a structural diagram of the internal processing logic of an LSTM neural network model provided in an embodiment of the present application;

[0033] Figure 2 It is a structural diagram of a cascaded LSTM neural network model provided in an embodiment of the present application;

[0034] Figure 3 It is a flowchart of a motor function assessment method provided in an embodiment of the present application;

[0035] Figure 4 is a structural schematic diagram of a motor function assessment system provided in an embodiment of the present application;

[0036] Figure 5 It is an application scenario diagram provided by an embodiment of the present application;

[0037] Figure 6 is a flow chart of another motor function assessment method provided in an embodiment of the present application;

[0038] Figure 7 It is a structural diagram of a terminal device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0039] The technical solutions in the embodiments of the present application will be described below in conjunction with the drawings in the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0040] First, the main inventive concept of the present application is introduced.

[0041] The traditional method of assessing the motor function is achieved through a motor function assessment scale. Taking the commonly used Fugl-Meyer upper limb motor function assessment scale as an example, the scale includes 33 assessment indicators. Each item is scored from 0 to 2 points, where 0 points means that it cannot be completed at all, 1 point means that it can be partially completed, and 2 points means that it can be completed completely. The specific assessment process is that the physician gives a corresponding score by subjectively observing the mobility of each joint of the limb of the person being assessed. This method relies on the physician's subjective observation and lacks objective standards, so the accuracy of the assessment results is not high. Based on this, the present application proposes a motor function assessment method, which is applied to a motor function assessment system, wherein the motor function assessment system includes an image acquisition module and a motion analysis module connected in sequence, wherein the motion analysis module is equipped with a motion analysis model, and the motion analysis model includes a convolutional neural network model, a long short-term memory neural network model and a similarity evaluation model, the output of the convolutional neural network is the input of the long short-term memory neural network, and the output of the long short-term memory neural network is the input of the similarity evaluation model, at least one frame of motion image of the person being assessed is obtained through the image acquisition module, the features of the motion image are extracted through the convolutional neural network carried by the motion analysis module, and the extracted features of the motion image are input into the long short-term memory neural network model, the work of the person being assessed in the motion image is identified, and the similarity between the actions of the person being assessed and the standard actions is obtained through the similarity evaluation model, so that the motor function assessment result of the person being assessed can be obtained according to the similarity, and the assessment process can avoid human subjective judgment, making the assessment process more objective, more refined, and with higher accuracy.

[0042] Next, some terms in this application are explained to facilitate understanding by those skilled in the art.

[0043] (1) Long Short-Term Memory Neural Network

[0044] In order to solve the problems of traditional recurrent neural network (RNN), the Long-Short Term Memory (LSTM) model came into being. The LSTM model uses input gate, output gate, forget gate and cell structure to control the learning and forgetting of historical information, making the model suitable for processing long sequence problems. Figure 1 , is a schematic diagram of the structure of an LSTM provided in an embodiment of the present application. Figure 1 As shown, let time t, the memory unit of the LSTM model is represented by C t , the output of the forget gate is represented by f t , the output of the input gate is represented by i t , the output of the output gate is represented by O (t) , the element values ​​of the three gates are all in the interval [0,1].

[0045] Specifically, the forget gate controls whether to forget, that is, to control whether to forget the hidden cell state of the previous layer with a certain probability. At time t, for the forget gate, its input is the hidden state h(t-1) of the previous sequence and the data x(t) of the current sequence. Under the action of the activation function, the output of the forget gate is obtained. Specifically, the activation function here can be sigmoid.

[0046] In practical applications, the processing logic of the forget gate can be expressed as the following mathematical expression (1):

[0047] f t =σ(W f h (t-1) +U f x (t) +b f

[0048] Among them, W f , U f 、b f are the coefficients and bias of the linear relationship, and σ represents the activation function sigmoid.

[0049] Specifically, the input gate is responsible for processing the input of the current sequence position and deciding what new information to put into the "cell state". Figure 1 As can be seen, the input gate consists of two parts. The first part, under the action of the activation function sigmoid, outputs i t , the second part is activated by the tanh function, and the output is a t , these two parts are multiplied together to update the cell state. In general, the role of the input gate is to prepare for state updates.

[0050] In practical applications, the processing logic of the input gate can be expressed as the following mathematical expression (2):

[0051] i t =σ(W i h (t-1) +U i x (t) +b i

[0052] a t =tanh(W a h (t-1) +U a x (t) +b a

[0053] Among them, W i , U i 、b i , W a , U a 、b a are the coefficients and bias of the linear relationship, and σ represents the activation function sigmoid.

[0054] After passing through the forget gate and the input gate, the deletion and addition of the transmitted information can be determined, that is, the "cell state" can be updated. Figure 1 It can be seen that the cell state C (t) It consists of two parts. The first part is C (t-1) And the forget gate output f t The second part is the product of the input gate i t and a t The product of can be expressed as the following mathematical expression (3):

[0055] C (t) =C (t-1) *f (t) +i (t) *a (t)

[0056] Wherein, * represents the Hadamard product.

[0057] Specifically, here, the Hadamard product is multiplied bit by bit.

[0058] from Figure 1 It can be seen that the hidden state h (t) The update consists of two parts. The first part is O (t) , which is composed of the hidden state h of the previous sequence (t-1) and the sequence data x (t) , and the activation function sigmoid, the second part is obtained by the hidden state C (t)and the activation function tanh, its processing logic can be expressed as the following mathematical expression (4):

[0059] O (t) =σ(W O h (t-1) +U O x (t) +b O

[0060] h (t) =O (t) *tanh(C (t) )

[0061] In the embodiment of the present application, the preset machine learning model may include but is not limited to the LSTM neural network model. Specifically, the specific architecture of the model can be as follows: Figure 2 As shown, in this LSTM neural network model, multiple cells are cascaded, for example, Figure 2 T cells are shown.

[0062] In the present application, the value of T is equal to the number of features of the motion image.

[0063] Figure 3 This is a motor function assessment method provided by the embodiment of the present application, which can be applied to Figure 4 The motor function evaluation system shown in the figure may include a playback module 40, an image acquisition module 41, a motion analysis module 42 and a display module 43 which are connected in sequence. Figure 5 is an actual scene graph), the method may include but is not limited to the following steps:

[0064] Step S301, obtaining at least one frame of motion image of the person being evaluated through an image acquisition module, wherein the motion image is an image of the person being evaluated when training according to a preset training item, and each motion image indicates the position, posture and motion process information of the person being evaluated;

[0065] Before the assessment, the person being assessed can open an account and password, and enter his or her user information into the motor function assessment system, such as name, age, assessment time, assessment location, assessment registration, etc. The motor function assessment system can be an application installed on a computer, mobile phone, or PAD. The person being assessed can enter the above information into the above motor function assessment system by himself or herself, or the medical center can arrange for the entry of the above information. After entering the above information, the motor function assessment system can automatically generate a code or number that uniquely identifies each person being assessed.

[0066] In practical applications, the motion image of the person being evaluated can be obtained through automated equipment (eg, a camera, etc.).

[0067] In one embodiment, before the assessment, a video of a preset training item may be played through the playback module so that the person being assessed can learn and better master the essentials of the exercise.

[0068] In one embodiment, before the assessment, the video and precautions of the preset training items can be played through the playback module.

[0069] In a specific implementation, the preset training items include one of finger-to-nose exercise, finger-to-finger exercise, alternating exercise, heel-knee-shin exercise, and eyes-closed difficult standing exercise.

[0070] Step S302: extract the features of the motion image through the convolutional neural network carried by the motion analysis module, the features of the motion image include at least one of the orientation features, posture features and motion features, input the features of the motion image into the long short-term memory neural network model, identify the movements of the person being evaluated in the motion image, and obtain the similarity between the movements of the person being evaluated and the standard movements through the similarity evaluation model, and obtain the motor function evaluation result of the person being evaluated based on the similarity.

[0071] In a specific implementation, the long short-term memory neural network model cascades T cells, and the value of T is equal to the number of features of the motion image. In one embodiment, the features of the motion image include orientation features, posture features, and motion features, and the long short-term memory neural network model includes 3 T cells. In one embodiment, the features of the motion image include orientation features and posture features, and the long short-term memory neural network model includes 2 T cells. Since the value of T is equal to the number of features of the motion image, in this way, the generalization ability of the model can be improved, thereby improving the accuracy of the evaluation.

[0072] In a specific implementation, the motor function assessment results may include but are not limited to scores, ratings, etc.

[0073] In this way, at least one frame of motion image of the person being evaluated is obtained through the image acquisition module, the features of the motion image are extracted through the convolutional neural network carried by the motion analysis module, and the extracted features of the motion image are input into the long short-term memory neural network model to identify the work of the person being evaluated in the motion image, and the similarity between the actions of the person being evaluated and the standard actions is obtained through the similarity evaluation model, so that the motor function evaluation result of the person being evaluated can be obtained according to the similarity. The evaluation process can avoid human subjective judgment, making the evaluation process more objective, more refined, and with higher accuracy.

[0074] It can be understood that after executing the above step S302, step S303 may also be executed, which is specifically described below:

[0075] Step S303: displaying the motor function assessment result of the person being assessed via a display module.

[0076] In actual application, after completing the motor function assessment of one item, the person being assessed can proceed to the motor function assessment of the next item. After obtaining the motor function assessment of multiple items, the motor function assessment results of multiple items can be combined to conduct an overall assessment of the person being assessed.

[0077] Figure 6 Another function evaluation method provided by the embodiment of the present application comprehensively considers historical training images and real-time acquired motion images, and may include but is not limited to the following steps:

[0078] Step S601, obtaining historical training images of the training target action of the person being evaluated;

[0079] Step S602: acquiring at least one frame of motion image of the training target action of the person being evaluated through an image acquisition module, wherein each motion image indicates the position, posture and motion process information of the person being evaluated;

[0080] Step S603: extract the features of the motion image and the features of the historical training image respectively through the convolutional neural network carried by the motion analysis module, wherein the features of the motion image include at least one of orientation features, posture features and motion features, input the features of the historical training image and the features of the motion image into the long short-term memory neural network model, identify the movements of the person being evaluated, and obtain the similarity between the movements of the person being evaluated and the standard movements through the similarity evaluation model, and obtain the motor function evaluation result of the person being evaluated according to the similarity.

[0081] In this way, historical training images as well as motion images acquired in real time are comprehensively considered in the evaluation, which can improve the evaluation accuracy.

[0082] It should be noted that, for the above-mentioned method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should know that the present disclosure is not limited by the order of the actions described, because according to the present disclosure, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily required by the present disclosure.

[0083] It should be further explained that Figure 3 For example, although Figure 3The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, Figure 3 At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.

[0084] One or more exemplary embodiments of the present specification also provide a terminal device, the terminal device 70 may include a processor 701, a memory 704 and a communication module 705, and the processor 701, the memory 704 and the communication module 705 may be interconnected via a bus 706. The memory 704 may be a high-speed random access memory (Random Access Memory, RAM) memory, or a non-volatile memory (non-volatile memory), such as at least one disk memory. The memory 704 may optionally be at least one storage system located away from the aforementioned processor 701. The memory 704 is used to store application code, and may include an operating system, a network communication module, a user interface module and a data processing program. The communication module 705 is used to interact with external devices for information; the processor 701 is configured to call the program code, and when executed, the motor function assessment method according to various embodiments of the present specification described in the above-mentioned "Exemplary Method" of the present specification is implemented.

[0085] Exemplary computer program products and storage media

[0086] In addition to the above-mentioned methods and devices, an embodiment of the present application may also be a computer program product, which includes computer program instructions, which, when executed by a processor, enable the processor to execute the steps of the overlap detection method for blind and buried holes according to various embodiments of the present application described in the above-mentioned "Exemplary Method" section of this specification.

[0087] The computer program product may be written in any combination of one or more programming languages ​​to write program codes for performing the operations of the embodiments of the present application, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" language or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0088] In addition, an embodiment of the present application may also be a storage medium on which a computer program is stored, and the computer program is executed by a processor to execute the steps of the blind buried hole overlap detection method according to various embodiments of the present application described in the above "Exemplary Method" section of this specification.

[0089] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in the present application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0090] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0091] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims

1. A method for evaluating motor function, characterized in that: The method is applied to a motor function evaluation system, which includes an image acquisition module and a motion analysis module connected in sequence, wherein the motion analysis module is equipped with a motion analysis model, and the motion analysis model includes a convolutional neural network model, a long short-term memory neural network model and a similarity evaluation model, the output of the convolutional neural network is the input of the long short-term memory neural network, and the output of the long short-term memory neural network is the input of the similarity evaluation model, and the method includes: Acquire at least one frame of motion image of the person being evaluated through the image acquisition module, wherein the motion image is an image of the person being evaluated when training according to a preset training item, and each motion image indicates the position, posture and motion process information of the person being evaluated; The features of the motion image are extracted by the convolutional neural network carried by the motion analysis module, and the features of the motion image include at least one of orientation features, posture features and motion features. The features of the motion image are input into the long short-term memory neural network model, and the movements of the person being evaluated in the motion image are identified. The similarity between the movements of the person being evaluated and the standard movements is obtained by the similarity evaluation model, and the motor function evaluation result of the person being evaluated is obtained according to the similarity.

2. The method according to claim 1, characterized in that: The motor function assessment system further includes a display module, which is connected to the motion analysis module. The method further includes: The motor function assessment result of the assessed person is displayed through the display module.

3. The method according to claim 1, characterized in that The motor function assessment system further includes a playback module, which is connected to the image acquisition module. The method further includes: The video of the preset training item is played through the playing module.

4. The method according to claim 1, characterized in that The motion image includes an image of a training target action of the person being assessed, and the method further includes: Acquire historical training images of the training target action of the person being evaluated; Extracting features of the historical training images through the convolutional neural network carried by the motion analysis module; The step of inputting the features of the motion image into the long short-term memory neural network model to identify the actions of the person being evaluated in the motion image includes: The features of the historical training images and the features of the motion images are input into the long short-term memory neural network model to identify the actions of the person being evaluated.

5. The method according to any one of claims 1 to 4, characterized in that: The long short-term memory neural network model cascades T cells, and the value of T is equal to the number of features of the motion image.

6. The method according to any one of claims 1 to 4, characterized in that: The preset training items include one of finger-to-nose exercise, finger-to-finger exercise, alternating exercise, heel-knee-shin exercise, and eyes-closed difficult standing exercise.

7. A motor function assessment device, characterized in that: The device is applied to a motor function evaluation system, which includes an image acquisition module and a motion analysis module connected in sequence, wherein the motion analysis module is equipped with a motion analysis model, and the motion analysis model includes a convolutional neural network model, a long short-term memory neural network model and a similarity evaluation model, wherein the output of the convolutional neural network is the input of the long short-term memory neural network, and the output of the long short-term memory neural network is the input of the similarity evaluation model, wherein: Acquire at least one frame of motion image of the person being evaluated through the image acquisition module, wherein the motion image is an image of the person being evaluated when training according to a preset training item, and each motion image indicates the position, posture and motion process information of the person being evaluated; The features of the motion image are extracted by the convolutional neural network carried by the motion analysis module, and the features of the motion image include at least one of orientation features, posture features and motion features. The features of the motion image are input into the long short-term memory neural network model, and the movements of the person being evaluated in the motion image are identified. The similarity between the movements of the person being evaluated and the standard movements is obtained by the similarity evaluation model, and the motor function evaluation result of the person being evaluated is obtained according to the similarity.

8. The device according to claim 7, characterized in that The motor function evaluation system further comprises a display module, which is connected to the motion analysis module, wherein: The motor function assessment result of the assessed person is displayed through the display module.

9. A terminal device, characterized in that: The method comprises a processor and a memory, wherein the processor and the memory are connected to each other, wherein the memory is used to store a computer program, the computer program comprises program instructions, and the processor is configured to call the program instructions to execute the method according to any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, wherein the computer program includes program instructions, and when the program instructions are executed by a processor, the processor is caused to perform the method according to any one of claims 1 to 6.

Images