A step-by-step behavior detection device and damage classification method based on a ladder experiment

Abstract

The present application relates to a kind of based on lateral ladder experiment step-by-step behavior detection device and damage classification method, belong to biomedical engineering technical field, solve the behavior detection device in prior art not easy to install, poor stability, difficult to control, application range is narrow, heterogeneous damage classification is not accurate, and the problem of cumbersome classification process.The step-by-step behavior detection device of the present application includes support and main track;Main track includes test runway;Test runway is set up multiple groups;Multiple groups of test runway are sequentially arranged from the entrance of main track along the runway direction of main track until the exit of main track;Multiple pass-through pieces are arranged on each test runway;When using, according to the test runway of selecting corresponding experimental function and / or the setting pass-through piece of corresponding experimental function on each test runway is set according to experimental requirements.

Description

Technical Field

[0001] The present invention belongs to the field of biomedical engineering technology, and specifically relates to an animal experimental device, and more particularly to a step-by-step behavior detection device and an injury classification method based on a ladder experiment. Background Art

[0002] In behavioral experiments, the ladder test is a test that assesses the motor ability of animals by measuring the placement of the feet of animals (such as rats) on the steps as they walk along the ladder. However, existing behavioral detection devices are difficult to install, have poor stability, are difficult to control, and have a narrow range of applications. In addition, for spinal cord injury research, the current ladder device is fixed by inserting metal rods into cardboard, which cannot support some heavier experimental animals to complete forward tasks. And because the ladder requires experimental animals to complete crossing and grasping movements, it has the characteristics of high difficulty. In the case of more severe injuries, experimental animals cannot pass the steps and data cannot be collected. In summary, the current ladder experimental device is mainly used in the study of small animal models with mild injuries and has a narrow scope of application.

[0003] In addition, over the past few decades, researchers have proposed numerous treatments for spinal cord injury, which have entered the clinical stage. However, a complete cure or even effective treatment for spinal cord injury has yet to be found. This may be because previous studies have overlooked the clinical heterogeneity of spinal cord injuries, particularly the differences in the mechanisms of spinal cord injury between individual patients. Spinal cord injury is often the result of spinal column injury. Trauma can lead to spinal fracture-dislocation and / or burst fractures. Flexion-distraction and hyperextension are also common forms of injury. Different forms of spinal injury can cause different spinal cord injuries: burst fractures can cause vertebral fragments to enter the spinal canal, leading to spinal cord contusion; fracture-dislocations cause spinal fractures and ligamentous tears and displacement, clamping the spinal cord at the site of injury and causing spinal cord shearing; flexion-distraction and hyperextension can cause spinal cord ligament tears, joint separation, or fracture separation, stretching and lengthening the spinal cord, leading to spinal cord traction. In addition, the same type of spinal cord injury also includes a variety of different severities. Due to the limitations of experimental equipment and other conditions, current spinal cord injury research mostly focuses on comparative studies between spinal cord injury models and healthy models, while ignoring the heterogeneity between the three different spinal cord injury mechanisms and the heterogeneity between injuries of different severities. At the same time, there is currently no method that can accurately and easily classify heterogeneous injuries. Summary of the Invention

[0004] In view of the above problems, the present invention provides a step-by-step behavior detection device and injury classification method based on the ladder experiment, which solves the problems in the prior art that the behavior detection device is difficult to install, has poor stability, is difficult to control, has a narrow application range, inaccurate classification of heterogeneous injuries, and a cumbersome classification process.

[0005] The present invention provides a step-by-step behavior detection device based on a ladder experiment, which is used for behavioral detection of spinal cord injury heterogeneity, and includes a bracket and a main runway;

[0006] The main runway includes a test runway;

[0007] Multiple groups of test tracks are set up; the multiple groups of test tracks are arranged in sequence along the runway direction of the main track from the entrance to the main track to the exit of the main track; multiple pass pieces are set up on each group of test tracks; when in use, according to experimental requirements, the test track corresponding to the experimental function is selected and / or the pass piece corresponding to the experimental function is set on each test track;

[0008] Each test track includes a mounting plate, an upper mounting piece, and multiple mounting slots; the multiple mounting slots are linearly arranged on the mounting plate; the distance between two adjacent mounting slots is in, Represents the average height of all animals measured.

[0009] Optionally, the main runway further includes a baffle, and a through slot is provided on the baffle for inserting the test runway.

[0010] Optionally, the baffles are symmetrically arranged on both sides of the test track; the front baffle is made of transparent material, and the rear baffle is made of opaque material.

[0011] Optionally, a reflector is also included.

[0012] Optionally, the upper mounting member cover is arranged on the mounting plate; and the mounting groove is arranged on the upper surface of the mounting plate.

[0013] Optionally, a fixing member is further included for detachably mounting the test track and / or the passage member in the through slot.

[0014] Optionally, a lifting frame is further included to control the linear lifting of the runway plane.

[0015] Optionally, an animal return bin is further included, which is arranged at the exit end of the main runway.

[0016] In another aspect, the present invention provides a method for classifying spinal cord injury heterogeneity, comprising the following steps:

[0017] Step S1, obtaining a plurality of healthy test animals and test animals with spinal cord injuries;

[0018] Step S2: setting the interval step difficulty parameters of adjacent pass pieces on the main runway; placing the pass pieces on the test runway according to the interval step difficulty parameters; and obtaining the pattern coefficient of the tested animal's movement on the main runway;

[0019] Step S3, placing a plurality of healthy test animals and test animals with spinal cord injuries on a horizontal main runway in sequence for horizontal passage; recording horizontal behavior images of all test animals on the main runway during horizontal passage;

[0020] Step S4: raising the main runway to a preset lifting angle; placing all healthy animals on the main runway in sequence for the nth group to pass; placing all animals with spinal cord injuries on the main runway in sequence for the kth group to pass;

[0021] Step S5, based on the passage experiment in step S4, obtaining the maximum ascending angle that the healthy test animal can climb and the maximum ascending injured ladder plane angle that the test animal with spinal cord injury can climb;

[0022] Step S6, obtaining an ascending injury angle difficulty coefficient based on the maximum ascending incline angle that the healthy animal can climb and the maximum ascending injury plane angle that the animal with spinal cord injury can climb;

[0023] Step S7, obtaining the descending action images and descending injury angle difficulty coefficients of the spinal cord injured animals that cannot complete the k-th group of passage in S6;

[0024] Step S8, randomly extracting multiple frames of images from the action image data of steps S3, S5 and S7; and creating a database using the multiple frames of images;

[0025] Step S9, randomly selecting multiple frames of images from the database obtained in step S8 to form a labeled data set; selecting multiple frames of images in the labeled data set, labeling the body parts of the animal to be tracked, and obtaining a guidance labeled data set;

[0026] Step S10, checking the accuracy of the labels obtained in S9; if the labels are correct, a verification label data set is obtained, and the process proceeds to step S11;

[0027] Step S11, classifying the spinal cord injury heterogeneity categories of the animals tested with spinal cord injury in the verification label data set to obtain an injury category data set;

[0028] Step S12, obtaining an image analysis network model based on the damage category dataset in S11;

[0029] Step S13, acquiring multiple frames of images in steps S3, S5 and S7 that are not extracted by step S8;

[0030] Step S14, using the image analysis network model S11 to analyze the multiple frames of image data extracted in step S13, using the distance between two adjacent steps in each frame of the image data as a unit; marking the body part of the test animal to be tracked; and obtaining a statistical image of the movement trajectory of the body part of the test animal to be tracked based on all the marks;

[0031] Step S15, based on the pattern coefficient obtained in step S2, the ascending injury angle difficulty coefficient obtained in step S6, the descending injury angle difficulty coefficient obtained in step S7, and the motion trajectory statistical image of the body part to be tracked by the test animal obtained in step S14, obtaining an error image and a correct stepping image of the body part to be tracked by the test animal;

[0032] Step S16, interpolating the error image and the correct stepping image of the body part to be tracked by the test animal to obtain an interpolated image;

[0033] Step S17, obtaining a motion feature image of the body part to be tracked of the test animal based on the interpolated image;

[0034] Step S18: performing cluster analysis on the motion feature images of the body part to be tracked of the test animal based on the motion feature images to obtain an injury classification result.

[0035] Compared with the prior art, the present invention has at least the following beneficial effects:

[0036] (1) The ladder device of the present invention is easy to install and stable, and provides a setting method for the ladder and the test runway for experimental animal subjects of different sizes and different experiments, which can adapt to related research on different types of experimental animal subjects and different experiments.

[0037] (2) The present invention can adjust the plane inclination angle of the main runway, so that the main runway can be freely changed and fixed within a larger angle range, which is used to study the stepping conditions of experimental animal subjects under different slopes and expand the scope of application.

[0038] (3) The present invention introduces neural network learning and k-means cluster analysis, and provides a behavioral step image heterogeneity analysis method based on artificial intelligence neural network learning. The classification results are accurate and the classification process is easy to operate.

[0039] (4) The present invention pioneered the use of a horizontal ladder device as an effective tool for studying the heterogeneity of different types of damage, which can easily and accurately classify different types of heterogeneous damage. The classification results are accurate and the classification process is easy to operate. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] The drawings are only for purposes of illustrating particular embodiments and are not to be considered limiting of the invention.

[0041] Figure 1 It is a front view of the detection device of the present invention;

[0042] Figure 2 is a top view of the detection device of the present invention;

[0043] Figure 3 is a side view of the detection device of the present application;

[0044] Figure 4 is a top view of the detection device of the present application in a lifted state;

[0045] Figure 5 is a partial schematic view of the test runway of the present application.

[0046] Reference signs:

[0047] 1. support, 2. test runway, 3. mirror, 4. animal return bin. DETAILED DESCRIPTION

[0048] In order to enable a clearer understanding of the above-mentioned objects, features and advantages of the present application, the present application will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict. In addition, the present application can also be implemented in other ways different from those described herein, and therefore, the protection scope of the present application is not limited by the specific embodiments disclosed below.

[0049] One specific embodiment of the present application, as Figure 1-5 disclosed a step-by-step behavior detection device based on a ladder experiment, for behavior detection of spinal cord injury heterogeneity, comprising a support 1, a main runway 2 and a mirror 3;

[0050] The main runway 2 comprises a baffle and a test runway; the baffles are symmetrically arranged on both sides of the test runway to prevent the measured animals from falling when moving on the test runway;

[0051] The baffle is processed with a through slot by a laser engraving machine, for inserting the test runway;

[0052] The test runway is provided with multiple groups; the multiple groups of test runways are arranged in sequence from the entrance of the main runway 2 along the runway direction of the main runway until the exit of the main runway 2; multiple passing members are arranged on each group of test runways; multiple through slots are provided; different functional test runways can be inserted in different through slots and / or different functional passing members can be arranged on each test runway according to different experimental requirements.

[0053] Optionally, the front baffle is made of transparent material, further, high-transparency acrylic plate, glass or carbonate, etc., and the rear baffle is made of opaque dark acrylic, which facilitates observation of the action posture of the measured animals while improving the wear resistance and scratch resistance of the baffle.

[0054] Optionally, obtain multiple animals to be tested, obtain the height a (unit: mm), width b (unit: mm) and weight m (unit: g) of multiple animals to be tested, and obtain the average height of the multiple animals to be tested The average width b and the average weight m. Preferably, the test animal is a rat.

[0055] Optionally, the baffle length is To ensure a reasonable and sufficient runway length; the baffle height is set to 5*b_ to ensure that the experimental subject cannot step out of the runway range; at the same time, the baffle thickness is set to its length That is a / 20, which controls the weight while preventing it from shaking.

[0056] Preferably, the baffle has a length of 1600 mm, a height of 200 mm, and a thickness of 8 mm.

[0057] Optionally, the test track includes two sets of mounting plates, upper mounting members, and mounting grooves; the two sets of mounting plates and the upper mounting members are symmetrically arranged in grooves of symmetrically arranged baffles; a plurality of mounting grooves are provided on the upper surface of the mounting plates; the plurality of mounting grooves are linearly arranged on the mounting plates; and the upper mounting member cover is provided on the mounting plates;

[0058] Furthermore, it also includes a fixing piece for detachably installing the test track and / or the pass piece in the through groove; the fixing piece increases the flexibility of the detection device, so that the test track and / or the pass piece can be easily disassembled and assembled, thereby adapting to various experimental needs.

[0059] Preferably, the fixing member includes two groups of high-strength nylon-boron magnets with opposite polarities; the fixing member is arranged in the through slot, the high-strength nylon-boron magnet of one polarity is arranged below the mounting plate, and the high-strength nylon-boron magnet of the other polarity is arranged between the high-strength magnet of one polarity and the lower plane of the baffle through slot; the mounting slot is a semicircular groove, and the two ends of the through member are respectively mounted on the symmetrically arranged mounting slots.

[0060] Optionally, a lower ladder frame is further included, which is arranged below the high-strength neodymium boron magnet and is used to support the bracket 1 and the main runway 2, and the reflector 3 is arranged inside the lower ladder frame.

[0061] Optionally, the pass piece is a rod made of metal, and a step is formed between two adjacent pass pieces; the pass piece is inserted into the installation slot, and a magnetic fixing piece is used to limit the degree of rotation of the pass piece in the installation slot.

[0062] Preferably, the diameters of both ends of the pass member are smaller than the diameter of the middle section, so that both ends can be firmly embedded in the mounting groove and limit left and right displacement to avoid displacement or falling off during experiments.

[0063] Furthermore, the distance between two adjacent mounting slots is In use, the interval ladder difficulty parameters are set in a cycle of 1, 4 and 9 levels, and 1, 4 and 9 levels respectively represent the intervals between adjacent passages of 1, 2 and 3 times the unit interval of the installation groove (ie 1 times, 2 times, 3 times the unit interval of the installation groove)

[0064] Further, the passage is a cylindrical rod; the cross-sectional area of the rod is linearly related to the body weight of the animal being tested, thereby ensuring the strength of the passage.

[0065] Optionally, a mirror is obliquely arranged below the test runway for reflecting the behavior of the animal being tested below the test runway when the animal being tested moves on the test runway to the observation position.

[0066] Optionally, it further comprises an animal return bin 4 arranged at the exit end of the main runway 2; an auxiliary runway is arranged between the animal return bin 4 and the main runway 2 for guiding the animal being tested into the animal return bin 4, and further, the auxiliary runway can be inserted into other accessories for behavioral experiments to meet the needs of laboratory animal models for various behavioral tests.

[0067] Optionally, referring to the attached Figure 4-5 It further comprises a lifting frame which is detachably arranged at the entrance end or exit end of the test runway for controlling the linear lifting of the runway plane.

[0068] Optionally, the lifting frame is a detachable electric cylinder; the use of the lifting frame expands the application range of the experimental type and introduces new experimental variables. Further, the electric cylinder system can be controlled by remote control, avoiding the dangers, instability and errors of manual lifting.

[0069] Optionally, the lifting frame is a scissor lifting frame.

[0070] Optionally, referring to the attached Figure 4 The lifting frame is connected to the lower ladder frame through a connecting shaft to ensure that the upper end of the lifting frame closely fits the lower surface of the lower ladder frame, and is connected by a t-shaped nut and a bolt.

[0071] The step-by-step behavior detection device based on the horizontal ladder experiment provided by the present application is a behavioral detection device which is convenient to install and flexible to replace, has the functions of adjusting the rotation degree of the horizontal ladder steps around the shaft and the horizontal plane angle of the horizontal ladder. In addition, the modular design and plug-in and plug-out function of the device enable it to adapt to different experimental requirements and meet the needs of laboratories for various behavioral tests of animal models. At the same time, the introduction of a remotely controllable electric cylinder system further improves the stability and operability of the device. The embodiments of the present application can be widely applied in the field of scientific research, especially in the field of behavioral research, and have important application value.

[0072] Further, the detection device is used to study the heterogeneity of spinal cord injury.

[0073] There may be some alternatives and improvements in the proposed technical solution. The following are some possible alternatives and improvements: (1) The baffle can also use other transparent materials, such as glass or polycarbonate (PC), to improve wear resistance or scratch resistance. (2) The test runway fixing method can also use bolts, snaps or magnetic locks to enhance stability and safety. (3) The horizontal ladder group design, the current horizontal ladder group design includes semicircular grooves and metal steps, and other shapes and structures can be considered, such as square grooves or stainless steel steps, to meet different experimental requirements or improve durability. (4) The lifting frame control method can also use hydraulic or pneumatic systems to provide greater force and more precise control; sensors and feedback mechanisms can also be added to achieve more precise control and automated operation.

[0074] The merits of these alternatives and improvements depend on the specific application requirements and practical circumstances. The optimal solution should achieve the device's functionality while providing improved performance and efficiency. A suboptimal solution may have limitations or compromises in certain areas but still achieve improvements. An average solution may be feasible but may not offer the same significant performance or efficiency improvements as other solutions.

[0075] Another embodiment of the present invention discloses a method for classifying spinal cord injury heterogeneity, using the aforementioned step-by-step behavioral detection device, and specifically comprising the following steps:

[0076] Step S1, obtaining a plurality of healthy test animals and test animals of the same sex, similar weight and similar body shape;

[0077] Step S2: setting the interval step difficulty parameter between adjacent pass elements on the main runway; obtaining the pattern difficulty parameter γ1 based on the interval step difficulty parameter; placing the pass elements on the mounting slots based on the interval step difficulty parameter; and obtaining the pattern coefficient α of the animal under test moving on the main runway;

[0078] Optionally, the interval step difficulty parameter is set as 1, 4 and 9 as a set of cycles, and 1, 4 and 9 respectively represent the interval between adjacent pass pieces. (i.e. 1 times, 2 times, 3 times the unit interval of the installation slot); the difficulty parameters of all interval steps on the main runway are accumulated to obtain the pattern difficulty parameter γ1.

[0079] Preferably, the pattern difficulty parameter γ1∈[100,297].

[0080] Furthermore, when placing the pass pieces, pass pieces are set at both ends of each test track, thereby ensuring that the total difficulty parameter remains unchanged when switching in units of test tracks.

[0081] The pattern coefficient of the animal under test moving on the main runway is α=γ1-λ, where λ is the movement coefficient. Preferably, λ=99.

[0082] Step S3, placing a plurality of healthy test animals and test animals with spinal cord injuries on a horizontal main runway in sequence for horizontal passage; recording the behavioral images of all test animals on the main runway during horizontal passage;

[0083] Optionally, the tested animal enters the animal return chamber 4 after reaching the end of the main runway from the starting point.

[0084] Optionally, if a tested animal cannot complete the horizontal passage, the tested animal is removed and replaced with a new healthy tested animal of the same sex, similar weight and similar size.

[0085] Optionally, the behavior image is a video.

[0086] Optionally, the behavior image includes the behavior image observed by the symmetrically arranged side baffles corresponding to the step-type behavior detection device, and the behavior image observed by the reflector obliquely arranged below the test track.

[0087] Optionally, a high-resolution, high-frame-rate camera can be mounted far enough away from the device to capture the entire device and the required images within its field of view. It is important to ensure that the camera lens is at the same height as the steps and captures the footage at a zero-degree angle to ensure the accuracy of the collected data.

[0088] Step S4: raising the main runway by a preset raising angle θ; placing all healthy animals on the main runway in sequence for the nth group to pass; placing all animals with spinal cord injuries on the main runway in sequence for the kth group to pass;

[0089] Step S5, based on the passage experiment in step S4, obtaining the maximum ascending angle that the healthy test animal can climb and the maximum ascending injured ladder plane angle that the test animal with spinal cord injury can climb;

[0090] For healthy animals, if all healthy animals can reach the end of the main runway from the starting point, the nth group of passage is completed, and n=n+1 is set, and the process returns to step S4; if one of the healthy animals cannot complete the nth group of passage, the behavioral test of the healthy animals is terminated, and the ascending behavior image of the healthy animals and the corresponding maximum ascending angle that the healthy animals can climb are obtained.

[0091] For the animals with spinal cord injury, if all the animals with known spinal cord injury can reach the end of the main runway from the starting point, the kth group of passage is completed, and k=k+1 is set, and the process returns to step S4; if one of the animals with known spinal cord injury cannot complete the kth group of passage, the behavioral test of the animal with spinal cord injury is terminated, and the maximum upward incline angle that the animal with spinal cord injury can climb is recorded. and rising injury behavior images;

[0092] Step S6, obtaining an ascending injury angle difficulty coefficient based on the maximum ascending incline angle that the healthy animal can climb and the maximum ascending injury plane angle that the animal with spinal cord injury can climb;

[0093] The expression of the difficulty coefficient of the ascending injury angle is:

[0094]

[0095] Wherein, β represents the difficulty coefficient of the ascending injury angle of the spinal cord injury animal.

[0096] Step S7, obtaining the descending action images and descending injury angle difficulty coefficients of the spinal cord injured animals that cannot complete the k-th group of passage in S6;

[0097] The horizontal main runway was tilted downward at a preset downward angle θ'. The spinal cord injured animals that could not complete the kth group of passage in S6 were placed on the downward inclined main runway to conduct the downward inclined passage experiment and record the average time taken. After the test is completed, the main runway will continue to descend at a preset downward angle θ' to conduct a downward passage test and record the average time taken Repeat the above operation until the spinal cord injured animal shows unstable grasping of the steps due to the large downward angle and a series of involuntary movement behaviors under the influence of gravity; compare the average time taken for the downward slope passage test of each group Get selected time Selected time The corresponding angle is the descending passage angle θ M , descending passage angle θ M It is considered to be the most suitable descending angle for passage; based on the preset descending angle θ' and the descending angle θ M Get the difficulty coefficient of the descent damage angle; and get the descent passage angle θ M The image of the descending movement of the animal with spinal cord injury.

[0098] The expression of the difficulty coefficient of the descent angle is:

[0099]

[0100] Preferably, the preset downward tilt angle is 5°.

[0101] Step S8: randomly extracting multiple frames of images from the image data obtained in steps S3, S5, and S7; and creating a database using the multiple frames of images.

[0102] Optionally, an image k-means clustering method is used to extract multiple frames of images from the image data.

[0103] Step S9: randomly select multiple frames of images from the database obtained in step 8 to form a labeled data set; select multiple frames of images in the labeled data set, label the body parts of the animal to be tracked, and obtain a guided labeled data set.

[0104] Preferably, when labeling, the main joints (such as hip joints, elbow joints, ankle joints) and main positioning parts (such as the tip of the nose, shoulders, and tail) are marked; further, the detailed joints are also marked (such as: the forefoot and the hindfoot; further, the carpal bones, metatarsal bones, first phalanges and fingertips of the longest toe (middle finger) of the forefoot and the hindfoot).

[0105] Step S10, check the accuracy of the marks obtained in S9, and check whether the marks of the body parts of the tested animal to be tracked are correct; if the marks are correct, obtain the verification mark data set and enter step S11; if the marks are incorrect, correct the wrong marks until they are correct.

[0106] Step S11 , classifying the spinal cord injury heterogeneity categories of the animals under test with spinal cord injury in the verification label data set to obtain an injury category data set.

[0107] The spinal cord injury heterogeneity categories of the tested animals in the validation labeled dataset were classified by multiple spinal cord injury heterogeneity experts.

[0108] Furthermore, the classification of spinal cord injury heterogeneity includes the type of injury and the severity of the injury.

[0109] Step S12, obtaining an image analysis network model based on the damage category dataset in S11;

[0110] The damage category dataset of S11 is divided into a training dataset and a test dataset; the network model is trained using the training dataset; and the performance of the trained network model is evaluated using the test dataset until the error reaches a stable level to obtain the image analysis network model.

[0111] The specific steps for using the test dataset to evaluate the performance of the network model are as follows:

[0112] manual labeling and prediction labeling using the image analysis network on the test data set;

[0113] Calculate the average Euclidean error (MAE) between the manual labeling and the prediction labeling of the test database of the image analysis network; use the average Euclidean error (MAE) to measure the performance of the image analysis network;

[0114] If the image analysis network cannot well generalize to unseen data in the evaluation and analysis step, extract the images with poor prediction labels, and manually move the prediction labels to their ideal positions. Further, when manually moving the prediction labels to their ideal positions, a set of additional annotated images are created at the same time, and the annotated images are combined with the training data to obtain an updated training data set; use the updated training data set to evaluate the performance of the network model, and repeat the operation until an image analysis network model that meets the generalization requirement is obtained.

[0115] Step S13, obtain the multiple frames of images that are not extracted by step S8 in steps S3, S5 and S7;

[0116] Step S14, analyze the multiple frames of image data extracted in step S13 using the image analysis network model of S11, in units of the distance between two adjacent steps in the image data; label the body parts to be tracked of the test animal; and obtain a motion trajectory statistical image of the body parts to be tracked of the test animal based on all the labels.

[0117] If the main joint and main positioning part labeling method is used in this process, the derived statistical image is an x-y coordinate-time image corresponding to the joint and part; if detailed joint labeling is also introduced, in addition to the x-y coordinate-time image, the statistical image will also derive the measurement values of the distance between each joint of the foot in the detailed joint and / or the motion variance of each joint and the frame number used for gripping the pass member.

[0118] Step S15, based on the motion trajectory statistical image of the body parts to be tracked of the test animal obtained in S14, obtain the error image and correct stepping image of the body parts to be tracked of the test animal;

[0119] Multiply the output value of the motion trajectory statistical image of the main body runway placed horizontally and multiple times with a preset lifting angle θ by a coefficient αβ to obtain a coefficient motion trajectory statistical image, and multiply the output value of the motion trajectory statistical image of the main body runway placed multiple times with a preset descending angle θ' by a descending angle difficulty coefficient γ to obtain a coefficient motion trajectory statistical image.

[0120] According to the peak value of the body parts to be tracked of the test animal in the coefficient motion trajectory statistical image, obtain the peak height threshold and the peak-to-peak distance; and determine the longitudinal coordinate y of the correct stepping step of the test animal in combination with the actual height of the main body runway correct ;

[0121] According to the coefficient motion trajectory statistics image, the valley value of the body part to be tracked by the test animal is obtained. The valley value height threshold and the distance between valleys are obtained. The vertical coordinate of all valleys is less than the vertical coordinate y of the correct step. correct Error image; get the valley value ordinate greater than or equal to the correct step ordinate y correct The correct stepping image.

[0122] Furthermore, when detailed joint markers are introduced, the lateral and vertical distances between each adjacent foot detailed joint marker are obtained: d x =|x label1 -x label2 |,d y =|y label1 -y label2 |;

[0123] Among them, d x Indicates the lateral distance between adjacent foot detail joint markers; d y Indicates the longitudinal distance between adjacent foot detail joint markers; x label1 represents the horizontal coordinate of the first foot; x label2 represents the horizontal coordinate of the second foot; y label1 represents the vertical coordinate of the first foot; y label2 Indicates the ordinate of the second foot.

[0124] First, we screened the detailed images of the grasped pass for investigation by ensuring that the distance from the fingertip to the wrist bone was less than the palm length and that the number of vertical coordinates of the foot landmarks located at the vertical coordinate of the pass was ≥ 1. For these detailed images, we derived the mean and variance of the horizontal and vertical distances of all adjacent foot landmarks as subsequent consideration criteria.

[0125] Step S16, interpolating the error image and the correct stepping image of the body part to be tracked by the detected animal to obtain an interpolated image;

[0126] The error images and correct stepping images of S15 are counted, and one-dimensional interpolation is performed to obtain an interpolated image.

[0127] Step S17, obtaining a motion feature image of the body part to be tracked of the test animal based on the interpolated image;

[0128] Align the interpolated images obtained in S16 according to the peak values ​​to obtain a waveform line graph of the body part to be tracked of the test animal;

[0129] Fitting multiple waveform line graphs to obtain a waveform average line graph of the body part to be tracked;

[0130] The waveform average line graph is used as the motion characteristic image of the body part to be tracked of the test animal.

[0131] Furthermore, when introducing detailed foot joint markers, it is also necessary to obtain feature grasping patterns. The specific steps are as follows:

[0132] The mean and variance of the horizontal and vertical distances of adjacent foot detail joint markers, the moving mean and variance of the horizontal and vertical coordinates of all foot detail joint markers, and the number of frames required to complete grasping (i.e., the sum of all consecutive frames satisfying the requirement that the distance from the fingertip to the wrist bone is less than the palm length and the number of vertical coordinates of the foot marker located at the pass element is ≥1) were obtained. Paired T-test was used for cluster analysis to fit the characteristic grasping patterns of heterogeneous injury types.

[0133] Step S18 , performing cluster analysis on the motion feature images of the body part to be tracked of the test animal based on the motion feature images to obtain an injury classification result.

[0134] Furthermore, for the case of introducing detailed foot joint markers, cluster analysis is performed on the motion feature images and characteristic grasping patterns of the body parts to be tracked by the test animals based on the motion feature images to obtain injury classification results.

[0135] To achieve uniform scaling, the vertical units are multiplied by a specific scale, usually chosen to be relevant to the subject's kinematics, such as the reciprocal of the subject's shoulder height. This will produce a profile of the kinematic characteristics of the subject with a specific injury (or healthy condition) for subsequent study.

[0136] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by any technician familiar with this technical field within the technical scope disclosed by the present invention should be covered by the scope of protection of the present invention.

Claims

1. A method for classifying spinal cord injury heterogeneity using a step-by-step behavioral detection device based on a ladder test, characterized in that: The invention discloses a step-by-step behavioral detection device based on a horizontal ladder experiment, which is used for behavioral detection of spinal cord injury heterogeneity, and comprises a bracket and a main runway; the main runway comprises a test runway; a plurality of test runways are provided; the plurality of test runways are arranged in sequence along the runway direction of the main runway from the entrance of the main runway to the exit of the main runway; a plurality of pass pieces are provided on each group of test runways; when in use, a test runway corresponding to an experimental function is selected according to experimental requirements and / or a pass piece corresponding to an experimental function is provided on each test runway; each group of test runways comprises a mounting plate, an upper mounting piece and a plurality of mounting slots; the plurality of mounting slots are linearly arranged on the mounting plate; the distance between two adjacent mounting slots is in, represents the average height of all animals measured; The specific steps for injury classification include: Step S1, obtaining a plurality of healthy test animals and test animals with spinal cord injuries; Step S2: setting the interval step difficulty parameters of adjacent pass pieces on the main runway; placing the pass pieces on the test runway according to the interval step difficulty parameters; and obtaining the pattern coefficient of the tested animal's movement on the main runway; Step S3, placing a plurality of healthy test animals and test animals with spinal cord injuries on a horizontal main runway in sequence for horizontal passage; recording horizontal behavior images of all test animals on the main runway during horizontal passage; Step S4: raising the main runway to a preset lifting angle; placing all healthy animals on the main runway in sequence for the nth group to pass; placing all animals with spinal cord injuries on the main runway in sequence for the kth group to pass; Step S5, based on the passage experiment in step S4, obtaining the maximum ascending angle that the healthy test animal can climb and the maximum ascending injured ladder plane angle that the test animal with spinal cord injury can climb; Step S6, obtaining an ascending injury angle difficulty coefficient based on the maximum ascending incline angle that the healthy animal can climb and the maximum ascending injury plane angle that the animal with spinal cord injury can climb; Step S7, obtaining the descending action images and descending injury angle difficulty coefficients of the spinal cord injured animals that cannot complete the k-th group of passage in S6; Step S8, randomly extracting multiple frames of images from the action image data of steps S3, S5 and S7; and creating a database using the multiple frames of images; Step S9, randomly selecting multiple frames of images from the database obtained in step S8 to form a labeled data set; selecting multiple frames of images in the labeled data set, labeling the body parts of the animal to be tracked, and obtaining a guidance labeled data set; Step S10, checking the accuracy of the labels obtained in S9; if the labels are correct, a verification label data set is obtained, and the process proceeds to step S11; Step S11, classifying the spinal cord injury heterogeneity categories of the animals tested with spinal cord injury in the verification label data set to obtain an injury category data set; Step S12, obtaining an image analysis network model based on the damage category dataset in S11; Step S13, acquiring multiple frames of images in steps S3, S5 and S7 that are not extracted by step S8; Step S14, using the image analysis network model S11 to analyze the multiple frames of image data extracted in step S13, using the distance between two adjacent steps in each frame of the image data as a unit; marking the body part of the test animal to be tracked; and obtaining a statistical image of the movement trajectory of the body part of the test animal to be tracked based on all the marks; Step S15, based on the pattern coefficient obtained in step S2, the ascending injury angle difficulty coefficient obtained in step S6, the descending injury angle difficulty coefficient obtained in step S7, and the motion trajectory statistical image of the body part to be tracked by the test animal obtained in step S14, obtaining an error image and a correct stepping image of the body part to be tracked by the test animal; Step S16, interpolating the error image and the correct stepping image of the body part to be tracked by the test animal to obtain an interpolated image; Step S17, obtaining a motion feature image of the body part to be tracked of the test animal based on the interpolated image; Step S18: performing cluster analysis on the motion feature images of the body part to be tracked of the test animal based on the motion feature images to obtain an injury classification result.

2. According to the damage classification method according to claim 1, the main runway also includes a baffle, which is provided with a through slot for inserting the test runway.

3. The damage classification method according to claim 2, characterized in that: The baffles are symmetrically arranged on both sides of the test track; the front baffle is made of transparent material and the rear baffle is made of opaque material.

4. The damage classification method according to claim 1, characterized in that: Also includes reflector.

5. The damage classification method according to claim 1, characterized in that: The upper mounting piece cover is arranged on the mounting plate; the mounting groove is arranged on the upper surface of the mounting plate.

6. The damage classification method according to claim 2, characterized in that: It also includes a fixing piece for detachably mounting the test track and / or the passage piece in the through slot.

7. The damage classification method according to claim 2, characterized in that: Also included are lifting frames for controlling the linear lift of the runway surface.

8. The damage classification method according to claim 2, characterized in that: It also includes an animal return bin, which is arranged at the exit end of the main runway.

Images