Non-human primate experimental animal upper limb movement ability comprehensive evaluation, data acquisition and analysis method

Abstract

The invention relates to the technical field of non-human primate experimental animal upper limb movement ability evaluation and analysis, in particular to a non-human primate experimental animal upper limb movement ability comprehensive evaluation, data acquisition and analysis method. The comprehensive evaluation comprises the steps that upper limb comprehensive exercise ability evaluation is carried out by adopting an upper limb comprehensive exercise ability evaluation scale on the basis of upper limb feeding operation; and the upper limb coarse movement and the finger fine movement can be quantitatively evaluated. The data acquisition and analysis method comprises the following steps: S1, acquiring comprehensive movement time of upper limbs; s2, coordinate values of all key points of the upper limbs on a two-dimensional plane are obtained; s3, acquiring a 3D motion space trajectory; s4, analyzing the movement speed of the upper limb marked key points; s5, the joint angle of the elbow joint and the movement range of the palm, the elbow and the shoulder are obtained; and S6, analyzing to obtain finger key point distance change data. According to the invention, comprehensive objective evaluation and analysis of upper limb coarse movement and finger fine movement of non-human primate experimental animals can be realized.

Description

Technical Field

[0001] The present invention relates to the technical field of upper limb motor ability evaluation and analysis of non-human primate experimental animals, and in particular to a method for comprehensive evaluation, data collection and analysis of upper limb motor ability of non-human primate experimental animals. Background Art

[0002] When non-human primates are used in experimental animals for the study of aging and neurological diseases, such as ischemic stroke, traumatic brain injury, spinal cord injury, etc., it is often necessary to conduct behavioral evaluations based on clinical characteristics in animals in order to assess the severity of motor impairment in neurological diseases. Upper limb movement includes the comprehensive ability of fine movements of the shoulder, upper arm, forearm, hand, and fingers, which requires comprehensive information acquisition and evaluation. Existing evaluation methods cannot combine shoulder, upper arm, forearm, hand, and finger movements to conduct an accurate and true comprehensive evaluation of the upper limbs. Therefore, it is impossible to provide a simple and comprehensive automated analysis method for the comprehensive evaluation of upper limb movement in non-human primates, which can promote the study of neurological diseases such as stroke, traumatic brain injury, spinal cord injury, Parkinson's disease, and Alzheimer's disease. Summary of the invention

[0003] In view of the above technical problems, the present invention provides a reliable and accurate method for comprehensive evaluation, data collection and analysis of upper limb motor ability of non-human primate experimental animals, thereby realizing comprehensive and objective evaluation and analysis of gross motor skills of upper limbs and fine motor skills of fingers of non-human primates.

[0004] To achieve the above object, the present invention adopts the following technical solution:

[0005] Comprehensive evaluation of upper limb motor ability of non-human primate experimental animals, including:

[0006] Evaluation of upper limb comprehensive motor ability

[0007] The upper limb comprehensive movement evaluation scale was used to evaluate the feeding movements of non-human primates.

[0008] Quantifiable evaluation

[0009] Based on the feeding movements of the fingers, wrists, arms and shoulders during the feeding process, a quantifiable evaluation of the gross motor skills of the upper limbs and the fine motor skills of the fingers was conducted.

[0010] Optionally, the upper limb comprehensive motor ability evaluation includes:

[0011] (1) Setting up the behavioral test cage

[0012] The behavior test cage is provided with a plurality of transparent baffles, and the feeding ports of the plurality of transparent baffles are provided at different positions; by replacing different transparent baffles, the non-human primate experimental animal is controlled to use the left hand or the right hand and to raise or lower the upper limbs to take food; a plurality of cameras are installed in front of the behavior test cage, and the whole process of the feeding behavior of the non-human primate experimental animal is recorded from the front, side and top view at the same time, so as to obtain a multi-angle feeding test video;

[0013] (2) Combined feeding test device

[0014] Assembling a feeding test device onto the behavior test cage to conduct a feeding test, so as to obtain comprehensive upper limb motor behavior data of non-human primate experimental animals and multi-angle and multi-scene feeding test videos;

[0015] The quantifiable evaluation includes:

[0016] Based on the multi-angle and multi-scene feeding test videos obtained, the upper limb comprehensive movement scoring scale was used to conduct a systematic upper limb comprehensive movement scoring of non-human primate experimental animals.

[0017] Optionally, the upper limb comprehensive motor score includes:

[0018] The upper limb movements of non-human primate experimental animals were divided into 7 levels to determine their upper limb muscle strength and finger fine motor ability;

[0019] The 7 levels include:

[0020] Level 1: When the feeding motivation exists, but the affected upper limb does not have any activity, the upper limb score is 0, and the corresponding upper limb muscle strength is also 0. Since the fingers do not have any activity, there is no fine motor ability of the fingers, and the fingers have no score;

[0021] Level 2: When the feeding motivation exists, but only the upper limb muscle tremors or contractions are observed, the upper limb score is 1, and the corresponding upper limb muscle strength is 1. Since only the upper limb muscles are trembling or contracting, there is no fine motor ability of the fingers, and the fingers have no score;

[0022] Level 3: When the affected upper limb shows feeding intention movements (shoulder, upper arm, forearm, etc. lifting or flexion and extension movements), but the upper limb fails to extend out of the test cage, the upper limb score is 2, and the corresponding upper limb muscle strength is 2. Since only the upper limb shows feeding intention movements, but the fingers do not move, there is no fine motor ability of the fingers, and the fingers have no score;

[0023] Level 4: When the upper limb on the affected side shows a feeding action and stretches out of the cage (the upper limb can resist gravity), and the finger does not touch the food (the finger does not move), the upper limb score is 3, and the corresponding upper limb muscle strength is 3. Since the finger does not touch the food, the finger score is 0;

[0024] Level 5: When the affected upper limb shows feeding action and stretches out of the cage, the finger touches the food but fails to complete the feeding action, the upper limb score is 4, and the corresponding upper limb muscle strength is 3+. Since the finger touches the food but fails to complete the feeding action, the finger score is 1;

[0025] Level 6: When the affected upper limb shows feeding action and stretches out of the cage, the finger touches the food and successfully retrieves the food, but it is not as good as normal, the upper limb score is 5, and the corresponding upper limb muscle strength is 4 or 4+. Since the finger touches the food and successfully retrieves the food, the finger score is 2;

[0026] Level 7: When it can continuously and quickly take food many times as in the normal state, its upper limb score is 6, and its upper limb muscle strength is 5. Since its fingers can continuously and quickly take food many times, its finger score is 3. The method for collecting and analyzing the upper limb motor ability data of a non-human primate experimental animal includes collecting and analyzing the upper limb motor ability of a non-human primate experimental animal as described above, and the specific steps are as follows:

[0027] S1: Using a feeding test device to obtain the time taken by the non-human primate's upper limbs and fine fingers to take food, the comprehensive movement time of the upper limbs is obtained;

[0028] S2: The DeepLabCut program is used to obtain the coordinate numerical table csv file of each key point of the upper limb of non-human primates in the feeding process on the two-dimensional plane;

[0029] S3: Then, the pose3d package in Matlab is used for 3D construction, and a three-dimensional motion situation is constructed through multiple 2D data to obtain a 3D motion space trajectory, so as to observe the motion space trajectory of non-human primates more intuitively;

[0030] S4: The JupyterNotebook (Anaconda3) program was used to analyze the movement speed of key upper limb markers (such as shoulder, upper arm, forearm, and hand), thereby achieving automated evaluation of the comprehensive movement of the upper limbs of non-human primates;

[0031] S5: Jupyter Notebook (Anaconda3) program was used to obtain the elbow joint angle and the range of motion of the palm, elbow and shoulder, thereby achieving automated evaluation of upper limb gross motor skills in nonhuman primates;

[0032] S6: Use Matlab program to analyze finger movement data to obtain finger key point distance change data, thereby realizing automated evaluation of non-human primate finger fine movement.

[0033] Optionally, the acquisition of the upper limb comprehensive movement time described in S1:

[0034] A feeding test device is provided, and the feeding test device includes a transparent front plate, a left plate and a right plate; a first feeding port is provided on each of the left and right sides of the transparent front plate; a left plate is provided on the left side of the transparent front plate, and a right plate is provided on the right side thereof; the left plate and the right plate are provided between the two first feeding ports; a second feeding port is provided on the left plate and the right plate, respectively; a food shelf is provided between the left plate and the right plate; a first timing sensor is provided at the first feeding port; a second timing sensor is provided at the second feeding port; the first timing sensor and the second timing sensor are connected to a circuit board, and the circuit board transmits data to a data acquisition panel to realize automatic data acquisition;

[0035] in:

[0036] (1) The timing starts when the non-human primate experimental animal's hand extends from the first feeding port, and the time when the hand reaches the second feeding port is the non-human primate experimental animal's upper limb gross motor extension time;

[0037] (2) When the hand of the non-human primate experimental animal obtains food and extends from the second feeding port, the time starts, and when the hand returns to the first feeding port, the time is the return time of the gross motor movement of the upper limb of the non-human primate experimental animal;

[0038] (3) The time for the non-human primate experimental animal to extend its upper limb gross motor and the time for the non-human primate experimental animal to return its upper limb gross motor are summed up as the time used for the non-human primate experimental animal's upper limb gross motor; (4) When the non-human primate experimental animal's hand passes through the second feeding port, the time taken to leave the second feeding port after taking the food is recorded as the time used for the fine motor movement of the fingers. Optionally, the acquisition of the coordinate value table csv file in S2 mainly includes the following steps:

[0039] (1) Extract and annotate frames from the video: First, extract representative frames from the video data; these frames should cover the greatest diversity, including different upper limb feeding movements, different individuals, and various brightness conditions; then, annotate these frames consistently to ensure the accuracy and consistency of the annotations;

[0040] (2) Generate training data set: Based on the annotated images, generate an annotated data set for training; this step includes integrating the annotation information with the image data to construct the input set of the deep learning model;

[0041] (3) Training a deep neural network: Using the generated training dataset, a deep neural network model is trained; the goal of the model is to predict the key point positions of the upper limb joints based on the input image; the training process includes multiple iterations to optimize the model parameters to minimize the prediction error;

[0042] (4) Automatic extraction: After training is completed, the model is used to predict the unlabeled new video data using transfer learning to extract the positions of the key points of the upper limbs; this includes the 2D marked points, generating a csv file that records the coordinate values ​​of each marked point on the two-dimensional plane and a coordinate diagram of each marked point.

[0043] Optionally, in S3, a plurality of 2D data are obtained by recording the front, side, and top views of the behavior test cage to construct a 3D data, that is, a first camera is installed on the front of the behavior test cage to collect and obtain the motion trajectory of the y-axis and z-axis during the animal's feeding process from the front perspective; a second camera is installed on the side of the behavior test cage to collect and obtain the motion trajectory of the x-axis and z-axis during the animal's feeding process from the side perspective; a third camera is installed above the behavior test cage to collect and obtain the motion trajectory of the x-axis and y-axis during the animal's feeding process from a top-down perspective, and then 3D construction is performed. The construction process of the 3D motion space trajectory mainly includes the following operations:

[0044] (1) Data import: Import the 2D key point coordinates extracted by DeepLabCut into Matlab;

[0045] (2) Camera calibration: In order to perform 3D reconstruction, the camera needs to be calibrated to obtain the camera's intrinsic and extrinsic parameters; open 'checkerboardPattern.pdf' in Matlab and print the checkerboard, and fix the checkerboard on a stable surface for easy calibration;

[0046] (3) 3D pose estimation: Use the functions in the pose3d package to combine the camera calibration results and the 2D key point coordinates to perform 3D pose estimation; the pose3d package provides a variety of methods to achieve this, such as direct linear transformation (DLT) and perspective n-point problem (PnP).

[0047] (4) 3D trajectory generation: Connect the 3D postures in consecutive frames to form the 3D spatial motion trajectory of the upper limb key points.

[0048] Optionally, the movement speed analysis step of the upper limb marking key points in S4 is as follows:

[0049] (1) Read the csv file and skip the first three rows: Call the read_csv() function to read the csv file in the specified path, skiprows=3. The first three rows in the file do not have valid data information, so they are skipped and the data is stored in the DataFrame variable df.

[0050] (2) Extracting time series and coordinate columns: The time series and x and y coordinate columns of two points are extracted from the DataFrame for subsequent analysis.

[0051] (3) Calculate the position change and velocity: Call the function of the NumPy library to calculate the velocity between two consecutive points;

[0052] (4) Display the speed graph: Call the plot function to display the drawn speed graph.

[0053] Optionally, S5 includes the following steps:

[0054] (1) Calculate joint angles (JointAngles): Use the dlc2kinematics.compute_joint_angles function to calculate the joint angles based on the DataFrame (df) output from DeepLabCut and the joint dictionary (joints_dict);

[0055] (2) Plot joint angles: Use the dlc2kinematics.plot_joint_angles function to plot joint angles and show how the elbow angle changes over time.

[0056] (3) Calculate the activity range: Use the dlc2kinematics.compute_correlation function to calculate the point distance matrix and obtain the correlation distance of each point to other points, and then draw the correlation matrix;

[0057] (4) Draw the activity optical flow map and output the activity range map: Draw the calculated correlation matrix and obtain the activity direction and range of the palm, elbow, and shoulder.

[0058] Optionally, the S6 includes:

[0059] (1) Data import: read data from the csv file in the specified path through the csvread function;

[0060] (2) Data extraction: The X-coordinate, Y-coordinate and corresponding likelihood information of the thumb and index finger tips were extracted;

[0061] (3) Data screening: By screening the confidence level, only the data points with the higher confidence level of both being greater than or equal to 0.9 are retained;

[0062] (4) Distance calculation: Calculate the Euclidean distance between the tip of the thumb and the tip of the index finger when the confidence level is greater than or equal to 0.9;

[0063] Euclidean distance:

[0064] Where d is the Euclidean distance, n is the dimension of the space, and x and y are the horizontal and vertical coordinate values ​​of two points in the vector space.

[0065] (5) Display results: Output the calculated distance and draw two scatter plots, which respectively show the coordinates of the thumb tip and the index finger tip, and a line graph, which shows the change of the distance between the thumb tip and the index finger tip over time, that is, the change of data points.

[0066] The beneficial effects of the present invention compared with the prior art are as follows:

[0067] 1. The present invention uses a feeding test device to automatically collect the time taken by non-human primates for the gross upper limbs and fine fingers when feeding; collects videos from the front, side and top views, constructs three-dimensional motion, marks key points of the shoulder, upper arm, forearm and hand, and analyzes their movement speed during operation, thereby realizing automatic evaluation of the comprehensive movement of the non-human primates' upper limbs; at the same time, collects and analyzes the joint angle of the elbow joint and the movement range of the palm, elbow and shoulder, thereby realizing automatic evaluation of the gross movement of the non-human primates' upper limbs, and automatically evaluates the fine movement of the non-human primates' fingers by using the distance change data of the key points of the fingers; finally, it is possible to perform a simple, comprehensive and objective automatic analysis of the comprehensive movement of the upper limbs of non-human primates, thereby promoting the research of neurological diseases such as stroke, traumatic brain injury, spinal cord injury, Parkinson's disease, and Alzheimer's disease.

[0068] 2. The present invention can realize the coherent and accurate automatic monitoring and collection of the time of gross motor movements of the upper limbs and the time of fine motor movements of the fingers of non-human primate experimental animals, provide accurate and interference-free raw data for subsequent related analysis, and significantly improve the accuracy and objectivity of subsequent evaluation and analysis.

[0069] 3. A simple and comprehensive upper limb comprehensive movement evaluation scale was adopted. It only required obtaining a simple feeding video of the experimental animals, without the need for high cooperation and training of the experimental animals, which reduced the training link before the experiment and reduced the relevant costs. The upper limb comprehensive movement evaluation scale combined the shoulder, upper arm, forearm, hand movement, and finger fine movement to conduct a comprehensive and objective evaluation of the upper limb gross movement and finger fine movement of non-human primates.

[0070] 4. By using the pose3d package in Matlab for 3D construction and constructing a three-dimensional motion situation through multiple 2D data, the spatial trajectory of the motion of non-human primates can be observed more intuitively; the JupyterNotebook (Anaconda3) program is used to analyze the movement speed of the key points of the upper limbs of non-human primates (such as shoulders, upper arms, forearms and hands) in the behavioral test cage, the joint angles of the elbow joints, and the movement ranges of the palms, elbows and shoulders, so as to achieve objective and reliable automated evaluation and analysis of the gross movements of the upper limbs of non-human primates; the Matlab program is used to analyze the finger movement data to obtain the distance change data of the key points of the fingers, so as to achieve objective and reliable automated evaluation and analysis of the fine movements of the fingers of non-human primates. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] Figure 1 This is a multi-scenario diagram of a feeding test for non-human primates;

[0072] Figure 2 It is a diagram of the upper limb comprehensive motor assessment scale;

[0073] Figure 3 This is a graph showing the scoring results of using the upper limb comprehensive motor scale to evaluate the non-human primate's eating 20 times before and after stroke in this embodiment;

[0074] Figure 4 It is a flow chart of the automated analysis of the comprehensive movement of upper limbs of the present invention;

[0075] Figure 5 Diagram of the feeding process of non-human primates;

[0076] Figure 6 This is the DeepLabCut program marking the key points of the hand;

[0077] Figure 7 It is a coordinate chart of each marker point on the hand;

[0078] Figure 8 It is the coordinate diagram of each marker point on the hand;

[0079] Fig. 9 Construct 3D motion space trajectory;

[0080] Fig.10 It is the joint point motion velocity diagram;

[0081] Fig.11 The palm, elbow, shoulder markings and elbow joint angles by DeepLabCut program (a);

[0082] Fig.12 is the result diagram of angle analysis of elbow joint (a);

[0083] Figure 13-1 It is a shoulder range of motion diagram;

[0084] Figure 13-2 It is the elbow range of motion diagram;

[0085] Figure 13-3 It is a map of the range of motion of the palm;

[0086] Fig.14 It is the distance change diagram between the tip of thumb and the tip of index finger;

[0087] Fig.15 It is a schematic diagram of the structure of the use state of the shelf of the MAP motion evaluation panel of the present invention;

[0088] Fig.16 It is a schematic structural diagram of a first embodiment of a feeding device of the present invention;

[0089] Fig.17 It is a schematic structural diagram of a second embodiment of the feeding device of the present invention. DETAILED DESCRIPTION

[0090] The above content is a detailed description of this patent in combination with specific implementation methods, and it cannot be determined that the specific implementation methods of this patent are limited to the above description. For ordinary technicians in the technical field to which this patent belongs, without departing from the concept of this patent, several substitutions or modifications made to the above-described implementation methods should be deemed to belong to the protection scope of this patent. In the description of this specification, in this specification, the specific features, structures, materials or characteristics described can be combined in a suitable manner in any one or more embodiments or examples. In the absence of contradictions, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples. Although the embodiments of the present invention and their advantages have been described in detail, it should be understood that various changes, substitutions and modifications can be made herein without departing from the scope of protection of the patent application.

[0091] A comprehensive evaluation of upper limb motor ability of non-human primate experimental animals, including:

[0092] (I) Obtaining multi-angle video collection scenes of non-human primate experimental animals eating (e.g. Figure 1), in which A represents the existing MAP motion evaluation panel (rectangle), B represents the existing MAP motion evaluation panel (circle), and C represents the food placed on the feeding test device. In actual use, A, B, and C can be replaced as needed to perform relevant tests; in this embodiment, taking the MAP motion evaluation panel as an example, it mainly includes:

[0093] (1) Fig.15 As shown, the non-human primate experimental animals were transferred to a behavioral test cage 1 which was a 6-sided cube with a size of 50cm*50cm*50cm;

[0094] (2) A slot 3 is provided on the front side of the behavior test cage 1 for fixing a transparent baffle 2, and the transparent baffle 2 is designed with different feeding ports 4 (i.e., the feeding ports 4 are opened at different positions), and the animal can be controlled to use the left or right hand to raise or lower the food by replacing different fixed transparent baffles 2, and three digital cameras are installed next to the behavior test cage, which can simultaneously record the entire process of the animal's feeding behavior from the front, side and top view, so as to obtain a multi-angle feeding test video;

[0095] (3) The feeding test device includes: a feeding device (such as a Brinkman feeding device or a feeding device as in the present invention); Figures 16-17 The feeding device shown in the figure) and the MAP movement assessment panel are fixed in a self-designed system behavior test cage to conduct feeding tests to obtain comprehensive movement behavior data of the upper limbs of non-human primate experimental animals;

[0096] The MAP motion assessment panel is a commonly used test device. When used, it is used in conjunction with the present invention. Fig.15 The shelf 8 shown in the figure has a plurality of clamping holes 7 on the top plate of the shelf 8 for clamping the MAP motion assessment panel. The position of the MAP motion assessment panel can be adjusted and fixed according to the actual situation to achieve tests of different distances and difficulties. The usage scenarios are as follows: Fig.11 shown.

[0097] (4) After the test, the non-human primate experimental animals were returned to the breeding room, and the test environment and equipment were disinfected and sterilized.

[0098] (ii) Use the Upper Limb Comprehensive Motor Assessment Scale to conduct quantifiable evaluation of upper limb gross motor skills and finger fine motor skills;

[0099] Based on the multi-angle and multi-scene feeding test videos obtained, the upper limb comprehensive movement scoring scale was used to conduct a systematic upper limb comprehensive movement scoring of non-human primate experimental animals.

[0100] Scoring is done based on upper limb movements. Different scores can determine upper limb muscle strength and finger fine motor ability. These behaviors will be divided into 7 levels (such as Figure 2 ), these 7 levels include:

[0101] Level 1: When the feeding motivation exists, but the affected upper limb does not have any activity, the upper limb score is 0, and the corresponding upper limb muscle strength is also 0. Since the fingers do not have any activity, there is no fine motor ability of the fingers, and the fingers have no score;

[0102] Level 2: When the feeding motivation exists, but only the upper limb muscle tremors or contractions are observed, the upper limb score is 1, and the corresponding upper limb muscle strength is 1. Since only the upper limb muscles are trembling or contracting, there is no fine motor ability of the fingers, and the fingers have no score;

[0103] Level 3: When the affected upper limb shows feeding intention movements (shoulder, upper arm, forearm, etc. lifting or flexion and extension movements), but the upper limb fails to extend out of the test cage, the upper limb score is 2, and the corresponding upper limb muscle strength is 2. Since only the upper limb shows feeding intention movements, but the fingers do not move, there is no fine motor ability of the fingers, and the fingers have no score;

[0104] Level 4: When the upper limb on the affected side shows a feeding action and stretches out of the cage (the upper limb can resist gravity), and the finger does not touch the food (the finger does not move), the upper limb score is 3, and the corresponding upper limb muscle strength is 3. Since the finger does not touch the food, the finger score is 0;

[0105] Level 5: When the affected upper limb shows feeding action and stretches out of the cage, the finger touches the food but fails to complete the feeding action, the upper limb score is 4, and the corresponding upper limb muscle strength is 3+. Since the finger touches the food but fails to complete the feeding action, the finger score is 1;

[0106] Level 6: When the affected upper limb shows feeding action and stretches out of the cage, the finger touches the food and successfully retrieves the food, but it is not as good as normal, the upper limb score is 5, and the corresponding upper limb muscle strength is 4 or 4+. Since the finger touches the food and successfully retrieves the food, the finger score is 2;

[0107] Level 7: When it can take food quickly and continuously for many times as in the normal state, its upper limb score is 6, and its corresponding upper limb muscle strength is 5. Since its fingers can take food quickly and continuously for many times, its finger score is 3.

[0108] like Figure 3The figure shows the evaluation of the upper limbs of crab-eating macaques before and two weeks after stroke based on the upper limb comprehensive motor assessment scale. The evaluation results of the crab-eating macaques performing 20 feeding actions were as follows: before the stroke, the upper limb score was 6, and the corresponding upper limb muscle strength was 5. Since the fingers could quickly and continuously take food for many times, the finger score was 3; two weeks after the stroke, the highest upper limb score was 5, and the corresponding upper limb muscle strength was 4 or 4+. Since the fingers touched the food and successfully retrieved the food, the finger score was 3.

[0109] This scoring system is specially designed to comprehensively evaluate the various motor behaviors of the upper limbs of animals during the feeding process and movement operations, covering the movements of the arms, wrists, shoulders and fingers, including reaching, grasping, raising, and flexing and extending fingers to feed. This upper limb comprehensive movement evaluation scale summarizes the muscle joint movement ability and separate movement evaluation in the first three levels of scores by applying the upper limb raising, lowering, and stretching movements of model animals in the test scene (such as non-human primate experimental animals after stroke disease modeling). The last four levels evaluate wrist stability and finger flexion and extension by applying the model animals in the test scene to grasp food and other movements. The evaluation is scientific, reasonable and objective.

[0110] In addition, the scoring system also includes the assessment of upper limb muscle strength and finger fine motor skills to more comprehensively measure the overall motor ability of animals. Through the magnitude scoring, researchers can objectively evaluate the strength and stability of animals when performing upper limb movements. This comprehensive evaluation method aims to provide a comprehensive and accurate quantitative standard for the upper limb movement characteristics of experimental animals, thereby providing a solid data foundation for further research.

[0111] The method for collecting and analyzing upper limb motor ability data of non-human primate experimental animals of the present invention specifically includes the following steps:

[0112] (I) Obtaining the comprehensive exercise time of the upper limbs

[0113] like Fig.16 As shown, a first embodiment of a feeding test device is provided, wherein the feeding test device comprises a transparent front plate 9, a left side plate 15 and a right side plate; when in use, the transparent front plate 9 is inserted into the Fig.15 The second slot 5 of the behavior test cage 1 shown (when in use, you can put Fig.15The transparent baffle 2 in the transparent front panel is removed), in some embodiments, a bottom support 6 is provided at the bottom of the second slot 5; a first feeding port 11 is provided on the left and right sides of the transparent front panel 9 (in some embodiments, a drop-down door can also be provided at the first feeding port 11 to guide the non-human primate experimental animal to take food with the left hand or the right hand); a left side panel 15 is provided on the left side of the transparent front panel 9, and a right side panel is provided on the right side thereof; the left side panel 15 and the right side panel are provided between the two first feeding ports 11; the second feeding port 12 is provided on the left side panel 15 and the right side panel respectively; A food shelf 13 is arranged between the two first feeding ports 11; a group of first timing sensors 10 are respectively arranged at the two first feeding ports 11, and the signal input module of the timing module can be an infrared sensing module or a light sensing module. When the light sensing module is used, a grating is used for reflection, and the light source is correspondingly arranged on one side of the feeding port. When feeding, the upper limbs block the light source, causing the light intensity of the photosensitive semiconductor to change and obtain a signal; a group of second timing sensors 14 are respectively arranged at the two second feeding ports 12; the first timing sensors and the second timing sensors are connected to the circuit board, and the circuit board transmits the data to the data acquisition panel to realize automatic data acquisition;

[0114] in:

[0115] (1) The timing starts when the non-human primate experimental animal's hand extends from the first feeding port, and the time when the hand reaches the second feeding port is the non-human primate experimental animal's upper limb gross motor extension time;

[0116] (2) When the hand of the non-human primate experimental animal obtains food and extends from the second feeding port, the time starts, and when the hand returns to the first feeding port, the time is the return time of the gross motor movement of the upper limb of the non-human primate experimental animal;

[0117] (3) The time it takes for the non-human primate experimental animal to extend its upper limbs and the time it takes for the non-human primate experimental animal to return to its upper limbs are summed up as the time it takes for the non-human primate experimental animal to perform gross motor movements of the upper limbs; (4) The timing starts when the hand of the non-human primate experimental animal passes through the second feeding port, and the time it takes when the hand leaves the second feeding port after taking the food is recorded as the time it takes for fine motor movements of the fingers.

[0118] Recording these times can be used to analyze the time taken by animals for gross upper limb movements and fine finger movements when feeding. This precise recording method is not only convenient for monitoring the upper limb reaction ability of non-human primate experimental animals, but also provides more detailed data support for analyzing the hand dexterity and movement strategies of animals, helping us to more deeply understand the behavioral patterns of animals in the feeding process.

[0119] In order to control non-human primate experimental animals to use upper limbs to raise or lower to eat, the present invention also provides a second embodiment of the feeding test device, the structure of which is as follows: Fig.17 As shown, the difference between this structure and the first embodiment is that: the height of the feeding opening is adjustable; long openings 22 are respectively provided along the upper and lower positions on the left and right sides of the transparent front plate 9, and then a slot 16 is provided on the inner side of the long opening 22, and a plug board 19 is movably provided in the slot 16, and a feeding opening 20 is provided on the plug board 19. At this time, the timing sensor 10-1 corresponding to the first timing sensor 10 in the first embodiment can be integrated on the plug board 19, and the power line / signal line is set longer and connected to the circuit board; the height of the first feeding opening is changed by adjusting the upper and lower positions of the plug board 19, so that the non-human primate experimental animal can be controlled to raise or lower its upper limbs for feeding; in order to facilitate fixation, in this embodiment, a plurality of screw holes 18 are provided on the slot 16, and then screws 17 are provided, and the plug board 19 is pressed by tightening the screws 17, so as to fix the position of the plug board 19. By adopting this implementation, it is possible to achieve induced control of non-human primate experimental animals using upper limb raising or lowering to eat, thereby ensuring the smooth implementation of relevant tests.

[0120] (ii) Use DeepLabCut to mark key points and obtain the coordinate value table csv file; DeepLabCut is a posture estimation tool based on transfer learning, which is characterized by the ability to accurately detect key points with relatively little training data.

[0121] The DeepLabCut program is used to obtain the coordinate numerical table csv file of each key point of the upper limb of non-human primates in the feeding process on the two-dimensional plane, which mainly includes the following steps:

[0122] (1) Extract frames from the video and annotate them (e.g. Figure 6 ): First, extract representative frames from the video data. These frames should cover the greatest diversity, including different upper limb feeding movements, different individuals, and various lighting conditions. Then, these frames are consistently annotated to ensure accuracy and consistency of the annotations.

[0123] (2) Generate training data set: Based on the annotated images, generate an annotated data set for training. This step includes integrating the annotation information with the image data to construct the input set of the deep learning model.

[0124] (3) Training a deep neural network: Use the generated training dataset to train a deep neural network model. The goal of this model is to predict the key point positions of the upper limb joints based on the input image. The training process includes multiple iterations to optimize the model parameters to minimize the prediction error.

[0125] (4) Automatic extraction: After training, transfer learning is used to use the model to predict new unlabeled video data and extract the locations of key points of the upper limbs. This includes the 2D marked points, and a csv file is generated to record the coordinate values ​​of each marked point on the two-dimensional plane (such as Figure 7 ) and the coordinate diagram of each marking point (such as Figure 8 ).

[0126] (III) Use the pose3d package in Matlab to perform 3D construction, mainly by obtaining multiple 2D data from different camera angles (such as front, side, and top view) to construct a 3D data (such as Fig. 9 ), that is, install camera 1 on the front of the behavior test cage to collect the y-axis and z-axis motion trajectories of the animal's feeding process from the front perspective (that is, take the animal's body as a reference object to obtain the motion trajectory of the upper limbs in the left and right and up and down height directions); install camera 2 on the side of the behavior test cage to collect the x-axis and z-axis motion trajectories of the animal's feeding process from the side perspective (that is, take the animal's body as a reference object to obtain the motion trajectory of the upper limbs in the front and back and up and down height directions); install camera 3 above the behavior test cage to collect the x-axis and y-axis motion trajectories of the animal's feeding process from a top-down perspective (that is, take the animal's body as a reference object to obtain the motion trajectory of the upper limbs in the front and back and left and right directions), and then perform 3D construction, which can more intuitively observe the motion space trajectory of non-human primates. The 3D motion space trajectory construction process mainly includes the following operations:

[0127] (1) Data import: Import the 2D key point coordinates extracted by DeepLabCut into Matlab;

[0128] (2) Camera calibration: In order to perform 3D reconstruction, the camera needs to be calibrated to obtain the camera's intrinsic and extrinsic parameters. Open 'checkerboardPattern.pdf' in Matlab and print the checkerboard, and fix the checkerboard on a stable surface for easy calibration;

[0129] (3) 3D pose estimation: Use the functions in the pose3d package to combine the camera calibration results and the 2D key point coordinates to perform 3D pose estimation. The pose3d package provides a variety of methods to achieve this, such as direct linear transformation (DLT), perspective n-point problem (PnP), etc.

[0130] (4) 3D trajectory generation: Connect the 3D postures in consecutive frames to form the 3D spatial motion trajectory of the upper limb key points.

[0131] (IV) The JupyterNotebook (Anaconda3) program was used to analyze the movement speed of the upper limb marked key points in the behavior cage. The movement depth of the key points was obtained mainly through the following operations:

[0132] (1) Read the csv file and skip the first three rows: Call the read_csv() function to read the csv file in the specified path, skiprows=3. The first three rows in the file do not have valid data information, so they are skipped and the data is stored in the DataFrame variable df.

[0133] (2) Extracting time series and coordinate columns: The time series and x and y coordinate columns of two points are extracted from the DataFrame for subsequent analysis.

[0134] (3) Calculate the position change and velocity: Call the function of the NumPy library to calculate the velocity between two consecutive points;

[0135] (4) Display the speed graph: Call the plot function to display the drawn speed graph (such as Fig.10 ).

[0136] (V) Use Jupyter Notebook (Anaconda3) to obtain the range of motion of the palm, elbow, and shoulder, as well as the elbow joint angle formed by the upper arm, forearm, and elbow caused by upper limb activity (e.g. Fig.11 ) to determine the gross motor ability of non-human primates' upper limbs. Jupyter obtains the angle and range of motion of the joint points of each marker through the following operations:

[0137] (1) Calculate joint angles (JointAngles):

[0138] Use the dlc2kinematics.compute_joint_angles function to calculate the joint angles based on the DataFrame (df) and joint dictionary (joints_dict) output from DeepLabCut;

[0139] (2) Plot joint angles: Use the dlc2kinematics.plot_joint_angles function to plot joint angles and show how the elbow angle changes over time (e.g. Fig.12 );

[0140] (3) Calculate the activity range: Use the dlc2kinematics.compute_correlation function to calculate the point distance matrix and obtain the correlation distance of each point to other points, and then draw the correlation matrix;

[0141] (4) Draw the activity optical flow map and output the activity range map: Draw the calculated correlation matrix and obtain the activity direction and activity range of the palm, elbow, and shoulder (as shown in Figure 13).

[0142] (VI) using a Matlab program to obtain the continuous change in the distance between the fingertips of the non-human primates during the feeding process to determine the fine motor ability of the fingers of the non-human primates; the Matlab program performs the following operations on the coordinate data of the thumb tip and the index finger tip:

[0143] (1) Data import: read data from the csv file in the specified path through the csvread function;

[0144] (2) Data extraction: The X-coordinate, Y-coordinate and corresponding likelihood information of the thumb and index finger tips were extracted;

[0145] (3) Data screening: By screening the confidence level, only the data points with the higher confidence level of both being greater than or equal to 0.9 are retained;

[0146] (4) Distance calculation: Calculate the Euclidean distance between the tip of the thumb and the tip of the index finger when the confidence level is greater than or equal to 0.9;

[0147] Euclidean distance: Where d is the Euclidean distance, n is the dimension of the space, and x and y are the horizontal and vertical coordinate values ​​of two points in the vector space.

[0148] (5) Display results: Output the calculated distance and draw two scatter plots, showing the coordinates of the thumb tip and index finger tip, and a line graph showing the change of the distance between the thumb tip and index finger tip over time (e.g. Fig.14 ), that is, the change in data points.

[0149] Through DeepLabCut, JupyterNotebook (Anaconda3), Matlab and other programs and comprehensive algorithms, the spatial trajectory, movement time, movement speed, movement complexity, etc. of the upper limbs of non-human primates during feeding are automatically analyzed, thereby realizing automated analysis of the comprehensive movement of the upper limbs, providing a more accurate evaluation method for related disease research, and being able to quantify the comprehensive movement of the upper limbs of non-human primates. The intermediate results obtained by the present invention are of great significance for evaluating experimental studies such as stroke, especially providing an objective and effective evaluation basis for related experimental studies such as disease models and their efficacy evaluation.

Claims

1. A method for comprehensive evaluation of upper limb motor ability of non-human primate experimental animals, characterized in that: include: Evaluation of upper limb comprehensive motor ability The upper limb comprehensive movement evaluation scale is used to evaluate the feeding movements of non-human primates; quantifiable evaluation Based on the feeding movements of the fingers, wrists, arms and shoulders during the feeding process, a quantifiable evaluation of the gross motor skills of the upper limbs and the fine motor skills of the fingers was conducted.

2. The method for comprehensive evaluation of upper limb motor ability of non-human primate experimental animals according to claim 1, characterized in that: The upper limb comprehensive motor ability evaluation includes: (1) Setting up the behavioral test cage The behavior test cage is provided with a plurality of transparent baffles, and the feeding ports of the plurality of transparent baffles are provided at different positions; by replacing different transparent baffles, the non-human primate experimental animal is controlled to use the left hand or the right hand and to raise or lower the upper limbs to take food; a plurality of cameras are installed in front of the behavior test cage, and the whole process of the feeding behavior of the non-human primate experimental animal is recorded from the front, side and top view at the same time, so as to obtain a multi-angle feeding test video; (2) Combined feeding test device Assembling a feeding test device onto the behavior test cage to conduct a feeding test, so as to obtain comprehensive upper limb motor behavior data of non-human primate experimental animals and multi-angle and multi-scene feeding test videos; The quantifiable evaluation includes: Based on the multi-angle and multi-scene feeding test videos obtained, the upper limb comprehensive movement scoring scale was used to conduct a systematic upper limb comprehensive movement scoring of non-human primate experimental animals.

3. The method for comprehensive evaluation of upper limb motor ability of non-human primate experimental animals according to claim 2, characterized in that: The upper limb comprehensive motor score includes: The upper limb movements of non-human primate experimental animals were divided into 7 levels to determine their upper limb muscle strength and finger fine motor ability; The 7 levels include: Level 1: When the feeding motivation exists, but the affected upper limb does not have any activity, the upper limb score is 0, and the corresponding upper limb muscle strength is also 0. Since the fingers do not have any activity, there is no fine motor ability of the fingers, and the fingers have no score; Level 2: When the feeding motivation exists, but only the upper limb muscle tremors or contractions are observed, the upper limb score is 1, and the corresponding upper limb muscle strength is 1. Since only the upper limb muscles are trembling or contracting, there is no fine motor ability of the fingers, and the fingers have no score; Level 3: When the affected upper limb shows feeding intention movements, such as raising the shoulder, upper arm, or forearm, or flexing and extending the limb, but the upper limb fails to extend out of the test cage, the upper limb score is 2, and the corresponding upper limb muscle strength is 2. Since only the upper limb shows feeding intention movements, but the fingers do not move, there is no fine motor ability of the fingers, and the fingers have no score; Level 4: When the affected upper limb shows a feeding action and stretches out of the cage, but the finger does not touch the food, the upper limb score is 3, and the corresponding upper limb muscle strength is 3. Since the finger does not touch the food, the finger score is 0; Level 5: When the affected upper limb shows feeding action and stretches out of the cage, the finger touches the food but fails to complete the feeding action, the upper limb score is 4, and the corresponding upper limb muscle strength is 3+. Since the finger touches the food but fails to complete the feeding action, the finger score is 1; Level 6: When the affected upper limb shows feeding action and stretches out of the cage, the finger touches the food and successfully retrieves the food, but it is not as good as normal, the upper limb score is 5, and the corresponding upper limb muscle strength is 4 or 4+. Since the finger touches the food and successfully retrieves the food, the finger score is 2; Level 7: When it can take food quickly and continuously for many times as in the normal state, its upper limb score is 6, and its corresponding upper limb muscle strength is 5. Since its fingers can take food quickly and continuously for many times, its finger score is 3.

4. A method for collecting and analyzing upper limb motor ability data of non-human primate experimental animals, characterized in that: The method comprises collecting and analyzing the upper limb motor ability of the non-human primate experimental animal as claimed in claim 1, and the specific steps are as follows: S1: Using a feeding test device to obtain the time taken by the non-human primate's upper limbs and fine fingers to take food, the comprehensive movement time of the upper limbs is obtained; S2: The DeepLabCut program is used to obtain the coordinate numerical table csv file of each key point of the upper limb of non-human primates in the feeding process on the two-dimensional plane; S3: Then, the pose3d package in Matlab is used for 3D construction, and a three-dimensional motion situation is constructed through multiple 2D data to obtain a 3D motion space trajectory, so as to observe the motion space trajectory of non-human primates more intuitively; S4: Use JupyterNotebook program to analyze the movement speed of key points of upper limb markers such as shoulder, upper arm, forearm and hand, so as to realize automatic evaluation of comprehensive upper limb movement of non-human primates; S5: Use JupyterNotebook program to obtain elbow joint angles and palm, elbow, and shoulder motion ranges, thereby achieving automated evaluation of upper limb gross motor skills in nonhuman primates; S6: Use Matlab program to analyze finger movement data to obtain finger key point distance change data, thereby realizing automated evaluation of non-human primate finger fine movement.

5. The method for collecting and analyzing upper limb motor ability data of non-human primate experimental animals according to claim 4, characterized in that: Acquisition of the upper limb comprehensive movement time described in S1: A feeding test device is provided, and the feeding test device includes a transparent front plate, a left plate and a right plate; a first feeding port is provided on each of the left and right sides of the transparent front plate; a left plate is provided on the left side of the transparent front plate, and a right plate is provided on the right side thereof; the left plate and the right plate are provided between the two first feeding ports; a second feeding port is provided on the left plate and the right plate, respectively; a food shelf is provided between the left plate and the right plate; a first timing sensor is provided at the first feeding port; a second timing sensor is provided at the second feeding port; the first timing sensor and the second timing sensor are connected to a circuit board, and the circuit board transmits data to a data acquisition panel to realize automatic data acquisition; in: (1) The timing starts when the non-human primate experimental animal's hand extends from the first feeding port, and the time when the hand reaches the second feeding port is the non-human primate experimental animal's upper limb gross motor extension time; (2) When the hand of the non-human primate experimental animal obtains food and extends from the second feeding port, the time starts, and when the hand returns to the first feeding port, the time is the return time of the gross motor movement of the upper limb of the non-human primate experimental animal; (3) The time taken for the non-human primate experimental animal to perform gross motor movements of the upper limbs is summed up to the time taken for the non-human primate experimental animal to perform gross motor movements of the upper limbs; (4) The timing starts when the hand of the non-human primate experimental animal passes through the second feeding port and leaves the second feeding port after taking the food. The time taken for this process is recorded as the time taken for fine movement of the fingers.

6. The method for collecting and analyzing upper limb motor ability data of non-human primate experimental animals according to claim 4, characterized in that: The acquisition of the coordinate value table csv file in S2 mainly includes the following steps: (1) Extract and annotate frames from the video: First, extract representative frames from the video data; these frames should cover the greatest diversity, including different upper limb feeding movements, different individuals, and various brightness conditions; then, annotate these frames consistently to ensure the accuracy and consistency of the annotations; (2) Generate training data set: Based on the annotated images, generate an annotated data set for training; this step includes integrating the annotation information with the image data to construct the input set of the deep learning model; (3) Training a deep neural network: Using the generated training dataset, a deep neural network model is trained; the goal of the model is to predict the key point positions of the upper limb joints based on the input image; The training process involves multiple iterations of optimizing model parameters to minimize prediction error; (4) Automatic extraction: After training is completed, the model is used to predict the unlabeled new video data using transfer learning to extract the positions of the key points of the upper limbs; this includes the 2D marked points, generating a csv file that records the coordinate values ​​of each marked point on the two-dimensional plane and a coordinate diagram of each marked point.

7. The method for collecting and analyzing upper limb motor ability data of non-human primate experimental animals according to claim 4, characterized in that: In S3, a plurality of 2D data are obtained by recording the front, side and top views of the behavior test cage to construct a 3D data, that is, a first camera is installed on the front of the behavior test cage to collect and obtain the motion trajectory of the y-axis and z-axis during the animal's feeding process from the front perspective; a second camera is installed on the side of the behavior test cage to collect and obtain the motion trajectory of the x-axis and z-axis during the animal's feeding process from the side perspective; a third camera is installed above the behavior test cage to collect and obtain the motion trajectory of the x-axis and y-axis during the animal's feeding process from the top perspective, and then 3D construction is performed. The construction process of the 3D motion space trajectory mainly includes the following operations: (1) Data import: Import the 2D key point coordinates extracted by DeepLabCut into Matlab; (2) Camera calibration: In order to perform 3D reconstruction, the camera needs to be calibrated to obtain the camera's intrinsic and extrinsic parameters; open 'checkerboardPattern.pdf' in Matlab and print the checkerboard, and fix the checkerboard on a stable surface for easy calibration; (3) 3D pose estimation: Use the functions in the pose3d package to perform 3D pose estimation by combining the camera calibration results and 2D key point coordinates. (4) 3D trajectory generation: Connect the 3D postures in consecutive frames to form the 3D spatial motion trajectory of the upper limb key points.

8. The method for collecting and analyzing upper limb motor ability data of non-human primate experimental animals according to claim 4, characterized in that: The steps for analyzing the movement speed of the upper limb key points in S4 are as follows: (1) Read the csv file and skip the first three rows: Call the read_csv() function to read the csv file in the specified path, skiprows=3. The first three rows in the file do not have valid data information, so they are skipped and the data is stored in the DataFrame variable df. (2) Extracting time series and coordinate columns: The time series and x and y coordinate columns of two points are extracted from the DataFrame for subsequent analysis. (3) Calculate the position change and velocity: Call the function of the NumPy library to calculate the velocity between two consecutive points; (4) Display the speed graph: Call the plot function to display the drawn speed graph.

9. The method for collecting and analyzing upper limb motor ability data of non-human primate experimental animals according to claim 4, characterized in that: The S5 comprises the following steps: (1) Calculate joint angles JointAngles: Use the dlc2kinematics.compute_joint_angles function to calculate the joint angles based on the DataFrame output from DeepLabCut and the joint dictionary joints_dict; (2) Plot joint angles: Use the dlc2kinematics.plot_joint_angles function to plot joint angles and show how the elbow angle changes over time. (3) Calculate the activity range: Use the dlc2kinematics.compute_correlation function to calculate the point distance matrix and obtain the correlation distance of each point to other points, and then draw the correlation matrix; (4) Draw the activity optical flow map and output the activity range map: Draw the calculated correlation matrix and obtain the activity direction and range of the palm, elbow, and shoulder.

10. The method for collecting and analyzing upper limb motor ability data of non-human primate experimental animals according to claim 4, characterized in that: The S6 includes: (1) Data import: read data from the csv file in the specified path through the csvread function; (2) Data extraction: The X-coordinates, Y-coordinates and corresponding confidence information of the thumb and index finger tips were extracted; (3) Data screening: By screening the confidence level, only the data points with the higher confidence level of both being greater than or equal to 0.9 are retained; (4) Distance calculation: Calculate the Euclidean distance between the tip of the thumb and the tip of the index finger when the confidence level is greater than or equal to 0.9; Euclidean distance: Where d is the Euclidean distance, n is the dimension of the space, x and y are the horizontal and vertical coordinates of two points in the vector space; (5) Display results: Output the calculated distance and draw two scatter plots, which respectively show the coordinates of the thumb tip and the index finger tip, and a line graph, which shows the change of the distance between the thumb tip and the index finger tip over time, that is, the change of data points.

Images