Gait timing analysis-based post-spinal surgery efficacy evaluation method and device, and medium

Abstract

This invention discloses a method, device, and medium for evaluating the efficacy of spinal surgery based on gait temporal analysis. The method includes: acquiring the coordinates of the toe, heel, and hip at each time frame of a walking video of a patient after spinal surgery; obtaining the hip anchor point coordinates based on the hip coordinate distribution, and dividing the gait into multiple gait cycles based on the heel and toe coordinate distribution; calculating propulsion symmetry and walking coordination based on the heel, toe coordinates, and hip anchor point positions within each gait cycle; further obtaining gait stability and coordination stability based on the propulsion symmetry and walking coordination distribution across multiple cycles, and determining the degree of recovery; and finally evaluating the postoperative efficacy based on the degree of recovery. Through gait temporal analysis, the method can keenly capture bilateral limb movement differences and subtle fluctuations during cycles, objectively quantifying the recovery status. This effectively solves the problems of strong subjectivity and difficulty in distinguishing similar gait abnormalities in traditional assessment methods, improving the accuracy and reliability of spinal surgery efficacy evaluation.

Description

Technical Field

[0001] This invention relates to the field of medical image processing technology, and in particular to a method, device and medium for evaluating the efficacy of spinal surgery based on gait time sequence analysis. Background Technology

[0002] Postoperative rehabilitation assessment after spinal surgery is a crucial step in guiding patient rehabilitation training and assessing surgical outcomes. Studies have shown that postoperative spinal surgery patients often exhibit asymmetrical gait patterns, such as pelvic rotation stiffness and loss of spinal-pelvic linkage, due to surgical wound pain, incomplete muscle function recovery, or abnormal neurological control. These gait abnormalities not only affect patients' daily living abilities but may also indicate incomplete neurological function recovery or the presence of compensatory movement patterns.

[0003] Currently, clinical assessment of gait function after spinal surgery largely relies on the physician's subjective observation or simple spatiotemporal parameter measurements, such as gait speed, cadence, and stride length. While these methods reflect a patient's motor ability to some extent, they struggle to capture subtle temporal changes in gait and differences in bilateral limb coordination. In particular, the two pathological mechanisms of pelvic rotational stiffness and loss of spinal-pelvic linkage are extremely similar in gait presentation, making it difficult for conventional indicators to effectively distinguish between them, resulting in assessment results that lack specificity and objectivity.

[0004] In recent years, with the development of computer vision and pose estimation technologies, video-based human keypoint detection has provided a new technical means for gait analysis. However, existing methods mostly focus on gait recognition or action classification, lacking dedicated models for assessing postoperative rehabilitation status after spinal surgery. Furthermore, since abnormal neural control often manifests as subtle fluctuations between gait cycles, traditional methods struggle to extract stable features from time-series data, failing to effectively identify gait variations caused by neural control instability. This results in less objective postoperative rehabilitation assessments, limiting their application in postoperative efficacy evaluation. Summary of the Invention

[0005] This application provides a method, device, system, and medium for evaluating the efficacy of spinal surgery based on gait time sequence analysis. This solves the technical problem in the prior art that it cannot capture the subtle fluctuations caused by the instability of nerve control during long-term walking, resulting in inaccurate postoperative rehabilitation evaluation results.

[0006] To address the aforementioned technical problems, in a first aspect, embodiments of this application provide a method for evaluating the efficacy of spinal surgery based on gait temporal analysis, the method comprising: Obtain the coordinates of the toes, heels, and hips on both sides of each frame in a walking video of a patient after spinal surgery. Based on the hip coordinate distribution of different time frames, the hip anchor point coordinates of each time frame are obtained; based on the distribution of heel coordinates and toe coordinates of different time frames, multiple gait cycles composed of time frames are obtained. Based on the distribution of heel and toe coordinates at different time frames within each gait cycle, and the positional characteristics of the hip anchor point coordinates, the propulsion symmetry and walking coordination of each gait cycle are obtained; based on the distribution of propulsion symmetry and walking coordination in different gait cycles, gait stability and coordination stability are obtained, and the degree of rehabilitation is obtained. The postoperative efficacy was assessed based on the degree of recovery described.

[0007] Furthermore, obtaining the hip anchor point coordinates for each time frame includes: Obtain the center point of the hip: Obtain the average coordinates of the left and right hips on each time frame, and use them as the coordinates of the center point of the hips on each time frame. A temporal sequence is constructed based on the coordinates of the center point of the hip at different time frames in chronological order; The time series is subjected to a moving average filter to obtain the smoothed hip center point coordinates of each time frame, which are used as the hip anchor point coordinates.

[0008] Furthermore, obtaining multiple gait cycles composed of time frames includes: For each time frame, calculate the difference between the longitudinal coordinates of the left toe and the right heel, and record it as the longitudinal difference between the two feet. According to the time sequence, the longitudinal differences of the two feet in all time frames are used to form a difference sequence, and the minimum point in the sequence is obtained. The detected minimum point is used as the dividing time frame. All time frames are divided according to the dividing time frame, and the range formed by the consecutive time frames is taken as a gait cycle.

[0009] Furthermore, obtaining the propulsion symmetry for each gait cycle includes: For each time frame within any gait cycle, the first difference between the ordinates of the toe coordinates and the heel coordinates on the left and right sides is obtained respectively. The minimum value of the first difference among all time frames is selected, and the corresponding time frame is taken as the toe departure point. The maximum value of the first difference among all time frames is selected, and the corresponding time frame is taken as the heel contact point. The difference in the order of the corresponding time frames between the heel contact point and the toe departure point on each side is obtained as the propulsion duration; The difference in propulsion duration between the right and left sides is obtained, and the difference in propulsion duration is negatively correlated to obtain the propulsion symmetry.

[0010] Furthermore, obtaining the walking coordination for each gait cycle includes: For each time frame within any step cycle, calculate the absolute difference between the longitudinal coordinates of the heels on the left and right sides and the longitudinal coordinates of the hip anchor point, and use this as the relative distance on that side. Construct the relative distance sequence of different time frames within any step state period; select the minimum value in the relative distance sequence as the degree of leg extension; By performing a negative correlation mapping on the absolute differences, the walking coordination index of the gait cycle is obtained.

[0011] Furthermore, obtaining gait stability and coordination stability includes: Calculate the squared difference in propulsion symmetry between adjacent gait cycles, and obtain gait stability based on the mean of the squared differences across all adjacent cycles; Calculate the squared difference in walking coordination between adjacent gait cycles, and obtain coordination stability based on the mean of the squared differences across all adjacent cycles.

[0012] Furthermore, the degree of recovery achieved includes: The overall stability is obtained by weighting the gait stability and the coordination stability. The mean of propulsion symmetry within all gait cycles is calculated as the average propulsion symmetry, and the mean of walking coordination within all gait cycles is calculated as the average walking coordination. The first rehabilitation coefficient is obtained based on the degree to which the mean of the average propulsion symmetry and the average walking coordination is close to 1; The product of the overall stability and the first recovery coefficient is taken as the degree of recovery.

[0013] Secondly, this application provides a device for evaluating the efficacy of spinal surgery based on gait temporal analysis, comprising: The data acquisition module is used to acquire the coordinates of the toes, heels, and hips on the left and right sides of each frame in the walking video of a patient after spinal surgery. The cycle division module is used to obtain the hip anchor point coordinates of each time frame based on the hip coordinate distribution of different time frames, and to obtain multiple gait cycles composed of time frames based on the distribution of heel coordinates and toe coordinates of different time frames. The stability analysis module is used to obtain the propulsion symmetry and walking coordination of each gait cycle based on the distribution of the heel coordinates and toe coordinates of different time frames within each gait cycle, as well as the positional characteristics of the hip anchor point coordinates; and is used to obtain gait stability and coordination stability based on the propulsion symmetry and walking coordination distribution of different gait cycles. The rehabilitation assessment module is used to obtain the degree of rehabilitation based on the gait stability and coordination stability, and to evaluate the postoperative efficacy based on the degree of rehabilitation.

[0014] Thirdly, this application provides a spinal surgery efficacy evaluation device based on gait timing analysis, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method as described in any of the first aspects.

[0015] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the method as described in any of the first aspects. One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages: First, this invention achieves an objective and quantitative assessment of the rehabilitation status of patients after spinal surgery through gait temporal analysis. Addressing the shortcomings of traditional assessment methods that rely on subjective observation or simple spatiotemporal parameters and struggle to capture subtle gait changes, this invention extracts the temporal coordinates of key points on the toes, heels, and hips from walking videos. It obtains stable hip anchor point coordinates through moving average filtering and accurately divides the gait cycle based on the minimum point of the longitudinal difference between the two feet, providing a reliable temporal basis for subsequent analysis.

[0016] Secondly, this invention, by constructing propulsion symmetry and walking coordination indices, can keenly identify subtle differences in the duration of force exertion and the height of leg lift between the two limbs. Propulsion symmetry reflects the degree of symmetry in the force exertion phase of both legs by quantifying the difference in the duration from heel touchdown to toe lift. Walking coordination quantifies the consistency of the height lift during the leg swing by analyzing the vertical distance between the heel and the hip. These two indices characterize the manifestations of gait abnormalities from different dimensions, effectively distinguishing between force asymmetry caused by surgical trauma and leg lift incoordination caused by impaired balance function, thus compensating for the inability of conventional indices to distinguish similar gait abnormalities.

[0017] Third, by analyzing the fluctuations in propulsion symmetry and walking coordination across multiple gait cycles, this invention obtains gait stability and coordination stability, enabling the capture of subtle gait variations caused by neural control instability or pain compensation. Traditional methods struggle to extract stable features from time-series data, while this invention quantifies the periodic stability of gait patterns by calculating the squared differences between adjacent cycle indices and performing negative correlation mapping, providing a sensitive indicator for identifying the state of neural function recovery.

[0018] Fourth, this invention constructs a quantitative model of rehabilitation level by comprehensively considering gait stability, coordination stability, average propulsion symmetry, and average walking coordination. This model considers both the average symmetry of both limbs in terms of force exertion duration and elevation height, and the stability of gait patterns between cycles. The resulting rehabilitation level index has a value range of (0,1], which can intuitively reflect the patient's gait function recovery status. The graded assessment based on this index (good, moderate, poor) provides an objective basis for clinicians to develop personalized rehabilitation training plans.

[0019] Fifth, the method of this invention can be implemented using ordinary video acquisition equipment, without the need for expensive large-scale instruments, and has good clinical applicability. It obtains key point coordinates through a general posture estimation model, and combines this with a well-defined mathematical calculation model for analysis. The operation is simple, the results are objective, and it can be widely applied in clinical scenarios such as post-spinal surgery rehabilitation assessment and rehabilitation training effect monitoring.

[0020] In summary, this invention fully utilizes gait temporal information and achieves accurate quantitative assessment of the rehabilitation status of patients after spinal surgery through multi-dimensional feature extraction and periodic stability analysis. It effectively solves the technical problems of existing technologies, such as insufficient objectivity in assessment results and difficulty in capturing subtle gait fluctuations, and has significant clinical practical value. Attached Figure Description

[0021] Figure 1 This is a flowchart illustrating a method for evaluating the efficacy of spinal surgery based on gait temporal analysis, as described in an embodiment of the present invention. Figure 2 This is a schematic diagram of a spinal surgery efficacy assessment device based on gait timing analysis in an embodiment of the present invention; Figure 3 This is a schematic diagram of the structure of another spinal surgery efficacy evaluation device based on gait time sequence analysis in an embodiment of the present invention. Detailed Implementation

[0022] This application provides a method, device, system, and medium for evaluating the efficacy of spinal surgery based on gait time sequence analysis. This solves the technical problem in the prior art that it cannot capture the subtle fluctuations caused by the instability of nerve control during long-term walking, resulting in inaccurate postoperative rehabilitation evaluation results.

[0023] To better understand the above technical solutions, the following will provide a detailed explanation of the technical solutions in conjunction with the accompanying drawings and specific implementation methods.

[0024] Example 1

[0025] Figure 1This is a flowchart illustrating a method for evaluating the efficacy of spinal surgery based on gait temporal analysis, as described in an embodiment of the present invention. The method includes: Step S100: Obtain the coordinates of the toes, heels, and hips on the left and right sides of each frame in the walking video of the spinal surgery patient.

[0026] Specifically, in the embodiments of the present invention, considering that post-spinal surgery patients often experience gait abnormalities due to pelvic rotation stiffness and loss of spinal-pelvic linkage, and that conventional methods are insufficient to distinguish the specific cause, it is necessary to perform temporal analysis of the gait to accurately assess the rehabilitation status. To this end, the coordinates of key points of the feet and hips at each time frame must first be extracted from the walking video.

[0027] (a) Video capture The patient walks on a treadmill with a natural gait in a rehabilitation laboratory or clinical setting. Image acquisition equipment (such as a high-definition camera) is positioned 5-7 meters behind the patient to ensure complete capture of the patient's lower body movements. The frame rate is no less than 30 frames per second to ensure sufficient temporal resolution to capture gait details and ensure smooth and clear video footage. Video acquisition continuously records multiple gait cycles (typically no less than 10 cycles). To reduce data redundancy and balance computational efficiency, one frame is extracted from every two frames in the acquired video as the analysis frame, i.e., a two-frame interval between adjacent time frames (i.e., one frame is extracted every three frames). It should be noted that in other embodiments of the present invention, the frame rate (no less than 30 frames per second) and the acquisition interval (the original number of frames between two adjacent analysis frames) can be specifically set according to the specific circumstances, and are not limited or elaborated here.

[0028] (II) Image Preprocessing To facilitate subsequent image processing and ensure consistent analysis scale, the following preprocessing was performed on each frame image acquired at each time step: Distortion correction: Lens distortion correction is performed on the original image (i.e., each original frame extracted from the walking video that has not undergone any preprocessing, such as distortion correction, scale normalization, etc.) to eliminate geometric distortion caused by lens optical characteristics and ensure the accuracy of coordinate position. Specifically, to ensure the accuracy of foot and hip coordinates in subsequent gait analysis, the impact of lens distortion on image geometric accuracy needs to be eliminated. This embodiment employs a commonly used camera calibration method for distortion correction: using the same camera equipment as that used to acquire the patient's walking video, multiple checkerboard calibration images with different postures are pre-captured; corner coordinates within the checkerboard are extracted using a corner detection algorithm, and the camera's internal parameter matrix and distortion coefficients are calculated; after formally acquiring the patient's walking video, inverse distortion transformation is applied to each frame to obtain geometrically distortion-free images. It should be noted that distortion correction is a conventional preprocessing method in the field for ensuring image measurement accuracy; its specific implementation can be found in standard tools of computer vision libraries such as OpenCV, and will not be elaborated upon here. The core innovation of this invention lies in the subsequent gait temporal analysis method based on the corrected images.

[0029] Scale normalization: This ensures all frames have a consistent spatial scale. Specifically, the distance between the patient and the camera may change during walking, leading to inconsistencies in the human body scale within the images. To eliminate this effect, the Euclidean distance from the shoulder joint to the hip joint in the first frame is calculated as a reference distance. For each subsequent frame, the shoulder-hip distance is adjusted to match the reference distance through overall scaling, thus ensuring a consistent spatial scale across all frames and eliminating scale differences caused by variations in the patient's distance from the camera. This step ensures the comparability of subsequent coordinate analyses.

[0030] Pose estimation and detection: Use a pose estimation model (such as MediaPipe) to detect human key points, including at least the shoulder joints, hip joints and foot joints (toes and heels) on both sides. The detection results are output in the form of pixel coordinates. After the preprocessing steps described above, a commonly used pose estimation model in this field (such as Google's open-source MediaPipePose model) performs joint detection on the preprocessed image. This model can output the image pixel coordinates of key points such as the hips, heels, and toes on both sides. It should be noted that this pose estimation model belongs to existing technology and is not an innovation of this invention. Therefore, through the above steps, the coordinates of the toes, heels, and hips on both sides of the image at each time frame are finally obtained, providing basic data for subsequent gait analysis.

[0031] Step S200: Based on the hip coordinate distribution of different time frames, obtain the hip anchor point coordinates of each time frame; based on the distribution of heel coordinates and toe coordinates of different time frames, obtain multiple gait cycles composed of time frames.

[0032] Specifically, step S200 includes: (a) Obtaining the coordinates of the crotch anchor point

[0033] During walking, the left and right hips swing alternately with the gait, but their center position remains relatively stable and can serve as a reference benchmark for gait analysis. Therefore, this step first calculates the mean of the left and right hip coordinates for each frame to obtain the hip center point for that frame; then, a moving average filter is applied to the temporal sequence of the hip center point to eliminate random jitter caused by keypoint detection and obtain stable hip anchor point coordinates. The specific implementation is as follows: Step S210a, obtain the center point of the hip: obtain the average value of the hip coordinates on the left and right sides of each time frame, and use it as the coordinates of the center point of the hip in each time frame. Step S220a, construct the time sequence: Let the ordinate of the hip center point in frame t be ht (this embodiment mainly focuses on the ordinate, because the vertical direction change is the most distinguishable in gait analysis), and construct the time sequence H = {ℎ1,ℎ2,…, ht,…} in chronological order based on the ordinate of the hip center point in frames at different times, where ht is the ordinate of the hip center point in frame t; Step S230a, moving average filtering to determine the crotch anchor point coordinates: Perform moving average filtering on the time series to obtain the smoothed crotch center point coordinates of each time frame, which are used as the crotch anchor point coordinates.

[0034] It should be noted that, considering that the hip lifting action during walking can cause the center point of the hip to shake, in order to obtain a smooth and stable reference trajectory, it is necessary to perform moving average filtering on the time sequence of the center point of the hip to eliminate the detected shaking and determine a stable reference coordinate, so as to calculate the relative motion of the foot more accurately.

[0035] Specifically, step S230a includes: using a moving average with a sliding window size of 5 and a step size of 1 to pair the sequence. H Smoothing is performed. For the first 4 frames (when the sliding window is not full), the smoothing value is the average of the current frame and all previous frames. Starting from the 5th frame, the smoothing value is the average of the current frame and the previous 4 frames (a total of 5 frames), that is, the average of all elements in each sliding window is used as the smoothing value of the frame at the end of the window.

[0036] In a specific application example, the ordinates (in cm) of the patient's hip center point in 30 frames are: 85.078, 85.257, 85.408, 85.298, 84.723, 84.629, 84.560, 85.146, 85.463, 85.399, 85.392, 84.925, 84.572, 84.610, 84.662, 84.965, 85.209, 85.440, 85.364, 85.048, 84.765, 84.577, 84.623, 84.798, 85.196, 85.468, 85.568, 85.277, 85.033. After moving average, the smoothed values ​​for the first four frames are: 85.078, 85.168, 85.248, and 85.260; the smoothed values ​​for frames 5 to 30 are: 85.228, 85.157, 85.031, 84.862, 84.779, 84.788, 84.936, 85.090, 85.256, 85.265, 85.150, 84.980, 84.832, 84.747, 84.804, 84.977, 85.128, 85.205, 85.165, 85.039, 84.875, 84.762, 84.792, 84.932, 85.131, 85.261, and 85.308. These smoothed coordinates are the coordinates of the crotch anchor point for the corresponding time frame.

[0037] (ii) Obtaining gait cycles

[0038] During walking, the feet move forward alternately. A complete gait cycle is formed when one heel leaves the ground and the other heel leaves the ground again. The temporal changes in heel and toe coordinates can reflect this periodic movement. Therefore, this step divides the time frame into periods based on the distribution of heel and toe coordinates at different time frames to obtain multiple gait cycles.

[0039] It should be noted that in the coordinate system of this application embodiment, a larger ordinate value indicates a higher position. During a normal gait cycle, when the right heel touches the ground, its ordinate reaches a local minimum. At this time, the left toe is in the swing phase, and its ordinate is relatively large. Therefore, the difference between the ordinate of the left toe and the right heel (defined as the longitudinal difference between the two feet) is positive. As the left leg swings forward to its highest point, the ordinate of the left toe reaches a local maximum, and the difference reaches its positive maximum. Subsequently, the right heel lifts up, and its ordinate gradually increases. As the left toe falls to the ground, its ordinate decreases, and the difference decreases accordingly. When the right leg swings to its highest point, the ordinate of the right heel reaches a local maximum, and the ordinate of the left toe touches the ground is at a local minimum, and the difference reaches its negative minimum. It can be seen that the minimum value (negative maximum) of the longitudinal difference between the two feet corresponds to the moment when the right heel leaves the ground (or the moment when the left toe touches the ground), and the time interval between two adjacent minimum values ​​constitutes a complete gait cycle.

[0040] It should be noted that the detection of minimum points can be achieved using various methods commonly used in the art. In the embodiments of the present invention, if the longitudinal difference between the two feet in a certain time frame is simultaneously less than the difference between its two adjacent time frames, then that time frame is determined to be a minimum point. In addition, algorithms such as peak detection, first-order difference sign change, or sliding window comparison can also be used to obtain minimum points. The specific implementation methods are well known to those skilled in the art and will not be described in detail here.

[0041] Based on the above principle, this embodiment divides gait cycles by detecting the minimum point of the longitudinal difference sequence of both feet. The specific steps are as follows: Preferably, in one embodiment of the present invention, the method for obtaining the gait period includes: Step S210b, calculate the longitudinal difference between the two feet: For each time frame, calculate the difference between the longitudinal coordinate of the left toe and the longitudinal coordinate of the right heel, and record it as the longitudinal difference between the two feet. ; in: L toe ( t ) indicates the first t The vertical coordinate of the left toe joint on the frame image. R heel ( t ) indicates the first t The vertical coordinate of the right heel joint on the frame image; t The time frame number is the time frame number, and its value range is... t =1,2,…, T , T This represents the total number of time frames.

[0042] Step S220b, construct the bipedal longitudinal difference sequence and detect the minimum point: construct the bipedal longitudinal differences of all time frames according to the time sequence to form a difference sequence {D (1), D (2),…, D ( t ),…},in, D ( t Let be the longitudinal difference between the two feet corresponding to the t-th frame image; use peak detection or first-order difference sign change method to obtain the minimum point in this sequence. If at a certain time frame... D ( t At the same time, it is less than the difference between its preceding and following adjacent frames, that is and D ( t )< D ( t +1), then mark that time frame as a minimum point; Step S230b, dividing the gait cycle: using the detected minimum point as the dividing time frame, dividing all time frames according to the dividing time frame, and taking the range formed by consecutive time frames as a gait cycle.

[0043] Specifically, all time frames between two adjacent minimum points (including the start point, excluding the end point, or adjusted according to the definition) constitute a complete gait cycle.

[0044] It should be noted that during one gait cycle, the longitudinal difference between the two feet reaches its minimum when the right heel leaves the ground, then reaches its maximum when the left toe swings to its highest point, and then reaches its minimum again when the right heel leaves the ground. Therefore, the minimum point accurately corresponds to the start and end times of the gait cycle. This embodiment uses the minimum point as the basis for cycle division, ensuring that each gait cycle begins with the heel of the same side leaving the ground and ends with the next heel lift off the ground on the same side.

[0045] In a specific application example, the patient's 30 frames of data show the following ordinates (in cm) for the left toe: 0.045, 0.084, 0.034, 0.078, 0.101, 0.230, 0.607, 1.819, 3.866, 6.899, 10.977, 14.105, 15.386, 14.176, 10.864, 7.052, 3.896, -0.044, -0.027, 0.012, -0.027, 0.135, 0.285, 0.658, 1.771, 3.911, 7.258, 11.038, 14. 338, 15.642; the longitudinal coordinates of the right heel are: 10.707, 13.873, 15.163, 13.967, 10.754, 6.878, 3.816, 0.004, 0.004, -0.024, -0.010, 0.042, 0.228, 0.642, 1.691, 3.811, 7.034, 10.843, 14.089, 15.336, 13.997, 10.793, 7.087, 3.922, -0.050, 0.047, 0.010, -0.034, 0.043, 0.240. Calculate the longitudinal difference D(t) between the two feet, which can be obtained in the 3rd frame ( ), Frame 20 Minimum values ​​appear at points such as 3 and 19. Therefore, taking the 3rd frame as the starting point of the first cycle and the 20th frame as the starting point of the next cycle, the 3rd to 19th frames constitute the first gait cycle, the 20th frame to the next minimum point (if it exists) constitute the second cycle, and so on.

[0046] Step S300: Based on the distribution of heel coordinates and toe coordinates at different time frames within each gait cycle, and the positional characteristics of the hip anchor point coordinates, obtain the propulsion symmetry and walking coordination of each gait cycle; based on the distribution of propulsion symmetry and walking coordination of different gait cycles, obtain gait stability and coordination stability, and obtain the degree of rehabilitation.

[0047] Post-spinal surgery patients often experience differences in the duration of the push-off phase between their legs due to varying degrees of traction and stimulation of the soft tissues on both sides. Simultaneously, impaired balance may cause patients to subconsciously shift their weight towards the healthy side, resulting in reduced foot lift and limited knee raise on the affected side, while relatively increasing the degree of extension. To quantify these abnormalities, this step extracts two key indicators—propulsion symmetry and walking coordination—based on the vertical position changes of the heel and toes and the coordinates of the hip anchor point within each gait cycle.

[0048] (a) Advancing the acquisition of symmetry Advancing symmetry is used to quantify the consistency of duration of the push-off phase in both legs. Within one gait cycle, the vertical positional relationship between the toes and heels is analyzed for each side.

[0049] Preferably, in one embodiment of the present invention, the method for obtaining advancement symmetry includes: Step S310a, identify key event frames: For each time frame within any step cycle, obtain the first difference between the ordinates of the toe coordinates and the heel coordinates on the left and right sides respectively, select the minimum value of the first difference among all time frames, and take the corresponding time frame as the toe departure point; select the maximum value of the first difference among all time frames, and take the corresponding time frame as the heel contact point. Step S320a, calculate the propulsion duration: obtain the order difference of the corresponding time frames between the heel contact point and the toe departure point on each side, and use it as the propulsion duration on that side; Step S330a, determine propulsion symmetry: obtain the difference in propulsion duration between the right and left sides, and perform a negative correlation mapping on the difference in propulsion duration to obtain a propulsion symmetry index to reflect propulsion symmetry.

[0050] Calculate the duration of right-side advance TR Duration of advancement on the left TL absolute difference By performing a negative correlation mapping on this difference, the propulsion symmetry index of this gait cycle is obtained. Bla Negative correlation mapping can be expressed in reciprocal form. (in (For extremely small positive numbers, avoid division by zero), or in exponential function form. The mapped index range is (0,1]. A larger value indicates a more consistent duration of propulsion on both sides and better symmetry. To facilitate horizontal comparisons across multiple cycles, the propulsion symmetry of all cycles can be linearly normalized to its maximum and minimum values, mapping it to the [0,1] interval. The specific formula is as follows:

[0051] in, Blai For the first i The original symmetry index of each cycle, Blai ′ is the normalized propagation symmetry, with a range of [0,1], used for direct comparison between different periods.

[0052] Specifically, it should be noted that the difference refers to the absolute value of the difference. For example, the difference in the order of time frames refers to the absolute value of the difference between the order of time frames; the difference in the propagation duration between the right and left sides refers to the absolute value of the difference in the propagation duration between the right and left sides.

[0053] When the heel touches the ground, the larger the toe's ordinate and the smaller the heel's ordinate, the larger the positive value of the first difference. When the toe leaves the ground, the smaller the toe's ordinate and the larger the heel's ordinate, the smaller the negative value of the first difference. Therefore, the minimum value of the first difference among all time frames is selected as the toe-off point, and the maximum value of the first difference among all time frames is selected as the heel-touch point. The more consistent the propulsion duration, the more similar the force exertion time of both legs, the smaller the duration difference, and the greater the symmetry. The more inconsistent the propulsion duration, the greater the difference, and the smaller the propulsion symmetry. When the right side is longer than the left side, the right side exerts force for a longer time, and the postoperative wound has a greater impact on the left leg. When the right side is shorter than the left side, the left side exerts force for a longer time, and the postoperative wound has a greater impact on the right leg.

[0054] Between multiple gait cycles, propulsive symmetry reflects the degree of force exerted on both sides. The closer the propulsive symmetry, the more consistent the force exertion, and the greater the gait stability. Walking coordination reflects the regularity of walking posture between both sides; the more consistent the elevation trend, the greater the coordination stability. If postoperative wound pain causes patients to consciously control their walking posture, it can lead to irregular gait changes between adjacent gait cycles, resulting in significant differences in propulsive symmetry and walking coordination between adjacent gait cycles. Combining propulsive symmetry and walking coordination reflects gait stability across multiple gait cycles, which helps in analyzing rehabilitation. By analyzing the distribution of propulsive symmetry and walking coordination across different gait cycles, gait stability and coordination stability can be obtained, along with the degree of rehabilitation.

[0055] In some embodiments of the present invention, this can be achieved by taking the reciprocal or an exponential function with the natural constant as its base. Negative correlation mapping is performed. When calculating the reciprocal, to avoid the difference being zero and the formula being meaningless, a very small positive number with consistent dimensions needs to be added to the denominator. The value of this number is specifically set according to the range of values ​​of the denominator, which will not be limited or elaborated here. Normalization is performed by linear normalization or normalization function, such as maximum and minimum value normalization. Maximum and minimum value normalization is performed for the analysis of multiple gait cycles. The specific methods are well known to those skilled in the art and will not be elaborated here.

[0056] Taking the aforementioned gait cycle (frames 3 to 20) as an example, the vertical coordinates of the right toe within this cycle are 10.732, 13.823, 15.218, 13.856, 10.717, 6.915, 3.809, -0.11, 0.045, -0.005, -0.027, 0.078, 0.207, 0.705, 1.814, 3.855, 7.028, and 10.815 (in cm); the vertical coordinates of the left heel are 0.008, -0.038, 0.242, 0.666, 1.801, 3.86, 7.209, 10.891, 14.202, 15.541, 14.111, 10.909, and 7. 081, 3.847, 0.03, -0.026, -0.036, -0.103 (unit: cm); Analyzing each side (left and right), the maximum first difference on the right side is 6.978 in frame 6 and the minimum is -7.061 in frame 19, with a propulsion time of 13; the maximum first difference on the left side is 3.866 in frame 17 and the minimum is 3.992 in frame 10, with a propulsion time of 7; the difference in propulsion time between the right and left sides is 6 frames. After reciprocal mapping (adding a very small positive number), the propulsion symmetry is approximately 0.1667 (Bla=1 / 6≈0.1667). This value is small, indicating a significant difference in the force application time between the two sides, and the propulsion time on the right side is longer than that on the left side, suggesting that the left side may be more affected by surgery.

[0057] (ii) Acquisition of walking coordination Walking coordination is used to quantify the consistency of leg lift height during the swinging motion. Within the gait cycle, the change in the vertical position of the heel relative to the hip reflects the degree of leg extension; the closer the bilateral extension, the more coordinated the walking posture. In one embodiment of the invention, the method for obtaining walking coordination includes: Step S310b, calculate the relative distance: For each limb, obtain the relative distance between the heel coordinate and the hip anchor point coordinate at each time frame.

[0058] In this embodiment, the relative distance is defined as the absolute value of the difference between the longitudinal coordinate of the heel and the longitudinal coordinate of the hip anchor point, i.e. ( G L ( t The number () indicates the relative distance between the left heel and the hip anchor point. L heel ( t ) is the first t The vertical coordinate of the left heel joint point on the frame image. At For the first t (vertical coordinates of the crotch anchor point in the frame image) and ( G R ( tThe number () indicates the relative distance between the right heel and the hip anchor point. Rheel ( t ) is the first t The vertical coordinate of the right heel joint point on the frame image. At For the first t (The vertical coordinate of the crotch anchor point in the frame image). It should be noted that the relative distance can also be calculated using other distance metrics such as Euclidean distance or Manhattan distance. The specific implementation is well known to those skilled in the art and will not be elaborated here.

[0059] Step S320b, determine the degree of leg extension: construct the relative distance sequence of frames at different times within any step state period; select the minimum value in the relative distance sequence as the degree of leg extension; Within a gait cycle, the minimum value in the relative distance sequence corresponds to the moment when the heel is raised to its highest point (closest to the hip), and this minimum value reflects the maximum extension of the leg within that cycle. Therefore, the minimum value of the left relative distance sequence is denoted as the left leg extension. E L =min t G L ( t (mint: calculates the minimum value for all time frames t within the period), similarly, the degree of extension of the right leg can be obtained. E R =min t G R ( t ).

[0060] Step S330b, determine walking coordination: obtain the absolute difference in the degree of leg extension between the right and left sides. Regarding the absolute difference Δ E By performing negative correlation mapping, the walking coordination index of the gait cycle is obtained. Cor .

[0061] Negative correlation mapping can be expressed in reciprocal form. (in (For extremely small positive numbers, avoid division by zero), or in exponential function form. Therefore, the mapped index range is (0,1], where a larger value indicates a more consistent degree of bilateral extension and better walking coordination. To facilitate lateral comparisons across multiple cycles, the walking coordination of all cycles can be linearly normalized to the maximum and minimum values, mapping it to the [0,1] interval.

[0062] It should be noted that the smaller the relative distance between the heel coordinate and the hip anchor point coordinate, the closer the heel is to the hip, and the less the leg extension (i.e., the greater the contraction and the higher the leg lift). If there is a significant difference in the extension of the two legs, it indicates that one leg is lifted significantly higher than the other, resulting in poor coordination of the walking posture. This asymmetry may stem from compensatory gait adopted by postoperative patients due to pain or impaired balance.

[0063] Between multiple gait cycles, walking coordination and propulsion symmetry together reflect the stability of gait patterns. Propulsion symmetry reflects the similarity in the duration of force exertion on both sides, while walking coordination reflects the similarity in the height of elevation on both sides. Significant fluctuations in propulsion symmetry or walking coordination between adjacent cycles may indicate that the patient is deliberately adjusting their gait due to postoperative wound pain, leading to irregular gait changes. Therefore, combining the distribution of propulsion symmetry and walking coordination across cycles can quantify gait stability and provide a basis for assessing rehabilitation progress.

[0064] (III) Acquisition of gait stability and coordination stability Preferably, in one embodiment of the present invention, the method for obtaining gait stability and coordination stability includes: For any data in propulsion symmetry and walking coordination, data stability is obtained based on the difference between different gait cycles and the subsequent gait cycle. Data stability is the gait stability corresponding to propulsion symmetry or the coordination stability corresponding to walking coordination. The difference in data is negatively correlated with data stability.

[0065] It should be noted that the difference in data represents the absolute value of the difference, reflecting the degree of deviation between adjacent gait cycles. The larger the absolute value of the difference, the more inconsistent the data between adjacent gait cycles, the more inconsistent the changes in walking, the more likely there is the influence of pain, the less consistent the walking coordination, and the lower the data stability; therefore, the difference in data is negatively correlated with data stability.

[0066] In one embodiment of the present invention, the squared difference between different gait cycles and the next gait cycle is obtained, the mean of the squared difference between all gait cycles and the next gait cycle is calculated, and a negative correlation mapping is performed as data stability; if the data corresponds to propulsion symmetry, gait stability is obtained, and if the data corresponds to walking coordination, coordination stability is obtained.

[0067] In this embodiment, the calculation steps for data stability are as follows: First, calculate the squared difference of this index between all adjacent gait cycles, that is, for the ... i The first cycle and the first i +1 cycle, calculate ,in a iIndicates the first i The index values ​​for each period (which could be propulsion symmetry or walking coordination) are then calculated; next, the mean of the squared differences between all adjacent periods is calculated; finally, a negative correlation mapping is applied to this mean to obtain data stability. The negative correlation mapping uses an exponential function with the natural constant as the base, i.e., data stability. ,in, N This represents the total number of gait cycles. After mapping... Q The value range is (0,1], and the larger the value, the more stable the indicator is during the week.

[0068] Specifically, when ai To advance symmetry Bla i When the data stability is obtained, the gait stability is the same as the stability of the gait. Sw ;when ai For walking coordination Cor i When the stability of the obtained data is known as coherent stability, then the stability of the data is called coherent stability. Pw .Right now:

[0069] in, Bla i+1 For the first i Propulsion symmetry with +1 gait cycle, Cor i+1 For the first i Walking coordination over +1 gait cycle, subscript i This is the gait cycle index; the range of values ​​for this index is... i =1,2,…, (Requires pairwise comparisons, there are) (for adjacent periods), where, N This represents the total number of gait cycles.

[0070] It should be noted that in other embodiments of the present invention, the negative correlation mapping can also be implemented in the form of a reciprocal. However, it should be noted that when the mean is zero, a very small positive number needs to be added to the denominator to avoid division by zero. The specific implementation is well known to those skilled in the art and will not be described in detail here.

[0071] The gait stability and coordination stability obtained through the above methods can be further combined for analysis to assess the patient's gait health status.

[0072] (iv) Obtaining the degree of rehabilitation Gait stability and coordination stability reflect the fluctuations of a patient's gait across multiple cycles from different dimensions. Combining the two allows for a comprehensive assessment of the overall stability of the gait. Based on this, the patient's recovery level can be quantified by combining the average propulsion symmetry and average walking coordination within each cycle. The specific methods for obtaining these parameters are as follows: Preferably, in one embodiment of the present invention, the method for obtaining the degree of recovery includes: Step S310c, calculate the overall stability: Gait stability Sw This reflects the consistency of the alternating force exertion rhythm of both legs between adjacent cycles, and the coordination and stability. Pw This reflects the consistency of leg lift height between adjacent cycles during each step. Higher values ​​for both indicate a more stable gait pattern and less impact from muscle protective spasms or neural control instability on the patient.

[0073] This embodiment uses a weighted summation method to obtain the overall stability. Sta :

[0074] in, α To preset the first weight, β To preset the second weight, satisfying α + β =1. Considering the postoperative recovery period after spinal surgery, changes in bilateral symmetry are often the most direct and earliest compensatory response, while coordination reflects more the joint linkage within one limb. Therefore, this embodiment sets... α =0.6, β =0.4, to more sensitively capture gait fluctuations caused by pain. It should be noted that in other embodiments, the preset weights can also be obtained by training a machine learning model; the specific method is not limited here.

[0075] Step S320c, obtain average propulsion symmetry and average walking coordination: propulsion symmetry over all gait cycles Bla i and walking coordination Cor i Calculate the mean values ​​separately to obtain the average propulsion symmetry. and average walking coordination :

[0076] in: Blai Indicates the first i The propulsion symmetry index value of the gait cycle is used to quantify the _th gait cycle. iThe consistency of the duration of the push-off phase of both legs within each cycle, with a value range of (0,1]. A larger value indicates more symmetrical force exertion on both sides within that cycle; Cori represents the walking coordination index value of the i-th gait cycle, used to quantify the consistency of leg lift height during the swing process within the i-th cycle, with a value range of (0,1]. A larger value indicates more coordinated leg lift on both sides within that cycle; subscript i This is the index of the gait cycle, with a value range of... i =1,2,…, N , N The total number of gait cycles. Average propulsion symmetry. and average walking coordination It reflects the patient's overall level of symmetry and coordination throughout the walking process.

[0077] Step S330c, calculate the first recovery coefficient: First, calculate the arithmetic mean of the two means above, and then find the absolute difference between it and 1:

[0078] The smaller this difference, the closer the patient's average gait is to a normal state. A negative correlation mapping is applied to Δ to obtain the first rehabilitation coefficient. R 1:

[0079] in, This is an exponential function with the natural constant as its base. Through this mapping, R The range of 1 is (0,1], and the smaller Δ is (i.e., the closer to normal), the better. R The closer 1 is to 1. It should be noted that in other embodiments, the reciprocal form can also be used for negative correlation mapping, but it should be noted that when Δ=0, a very small positive number needs to be added to the denominator. The specific implementation is well known to those skilled in the art and will not be described in detail here.

[0080] Step S340c, calculate the degree of recovery: Multiplying the overall stability by the first recovery coefficient yields the final recovery level, Res.

[0081] degree of recovery Res The value range is (0,1], and a larger value indicates better recovery of the patient's gait function. Specifically, the more consistent the force exertion and leg lift height on both sides, the more symmetrical the average propulsion. and average walking coordination The closer to 1, the greater the first rehabilitation coefficient; the more stable the gait pattern between adjacent cycles, the greater the overall stability and the greater the degree of rehabilitation.

[0082] Through the above steps, the symmetry, coordination and stability of gait are integrated, and a quantitative assessment of the rehabilitation degree of patients after spinal surgery is achieved.

[0083] Step S400: Assess the postoperative efficacy based on the degree of recovery.

[0084] degree of recovery Res This study quantifies the recovery status of lower limb motor function in patients after spinal surgery, objectively reflecting their current gait function. Based on the numerical values ​​of the recovery level, postoperative efficacy can be graded and assessed, thus providing a basis for clinical rehabilitation decisions.

[0085] In one embodiment of the present invention, two preset thresholds are set according to the degree of recovery to divide the postoperative recovery status into three levels: like Res ≥ Th 1. Determined to be at the first level of rehabilitation, indicating good recovery; like Th 2≤ Res < Th 1. Determined as Level 2 rehabilitation, indicating moderate rehabilitation; like Res < Th 2. The assessment is at the third level of rehabilitation, indicating poor rehabilitation.

[0086] Among them, the preset first threshold Th 1 is greater than the preset second threshold Th 2. The rehabilitation effect of the first rehabilitation level is better than that of the second rehabilitation level, and the second rehabilitation level is better than that of the third rehabilitation level. Based on clinical experience, this embodiment is set as follows: Th 1 = 0.8 ,Th 2 = 0.6. It should be noted that in other embodiments, these two thresholds may be adjusted according to the specific application scenario or patient population characteristics, and are not limited here.

[0087] Through the aforementioned tiered assessment, doctors can gain a clear understanding of a patient's recovery status and develop or dynamically adjust targeted rehabilitation training plans accordingly. For example, the intensity of intervention can be appropriately reduced for patients at the first rehabilitation level, while patients at the third rehabilitation level require enhanced rehabilitation treatment and close monitoring. This helps improve rehabilitation efficiency and achieve personalized rehabilitation management.

[0088] The following example, taken three months after posterior lumbar decompression, fusion and internal fixation surgery, illustrates the specific implementation process of the method of the present invention and verifies its effectiveness.

[0089] Step S100: Obtain walking video and key point coordinates The patient walked on a treadmill at a comfortable speed selected by the patient within the rehabilitation laboratory. A high-definition camera was positioned approximately 6 meters behind the patient, capturing video of the walking process at a frame rate of 60 frames per second. To balance computational efficiency, every other frame was extracted from the original video for analysis, resulting in an analysis frame rate of 30 frames per second, with an interval of 0.033 seconds between adjacent analysis frames. A total of 300 frames were extracted and analyzed, covering approximately 10 seconds of the patient's walking process.

[0090] After distortion correction and scale normalization preprocessing for each extracted image frame, the MediaPipe pose estimation model was used to detect keypoints, obtaining the coordinates of the toes, heels, and hips on the left and right sides of each image. The original coordinate data (unit: pixels, after scale normalization) of some keyframes are shown in the table below:

[0091] Step S200: Obtain the coordinates of the hip anchor point and divide the gait cycle. 1. Obtain the coordinates of the crotch anchor point: First, calculate the ordinate of the center point of the hip at each time frame, which is the average of the ordinates of the left and right hips. For example, the ordinate of the center point of the hip in frame 1 is (85.10 + 85.05) / 2 = 85.075.

[0092] Then, a moving average filter with a window size of 5 was applied to the time sequence of the ordinates of the hip center point in all 300 frames to obtain smoothed hip anchor point ordinates. The smoothing results for the first 5 frames are as follows: Smoothing value for frame 1: 85.075 Smoothing value for frame 2: (85.075 + 85.225) / 2 = 85.150 Smoothing value for frame 3: (85.075 + 85.225 + 85.425) / 3 = 85.242 Smoothing value for frame 4: (85.075 + 85.225 + 85.425 + 85.275) / 4 = 85.250 Smoothing value for frame 5: (85.075 + 85.225 + 85.425 + 85.275 + 84.725) / 5 = 85.145 Subsequent frames are processed in the same way to obtain the coordinates of the crotch anchor point in each frame. These coordinates serve as a stable reference benchmark for subsequent analysis.

[0093] 2. Divide gait cycles: Calculate the longitudinal difference between the two feet in each frame, D(t) = left toe ordinate - right heel ordinate. For example, in frame 1, D(1) = 0.05 - 10.70 = -10.65; in frame 3, D(3) = 0.03 - 15.16 = -15.13.

[0094] Construct the difference sequence over the entire time series and detect all local minima. According to the detection, the local minima appear in frames 3, 20, 37, ..., 293.

[0095] Using these minimum points as periodic division points, consecutive time frames are divided into multiple gait cycles. For example, frames 3 to 19 (with frame 20 as the starting point of the next cycle) constitute the first gait cycle (cycle 1), frames 20 to 36 constitute the second gait cycle (cycle 2), and so on. A total of 10 complete gait cycles are divided.

[0096] Step S300: Calculate the indicators and periodic stability within the period. 1. Calculate propulsion symmetry (taking period 1 as an example): Period 1: Contains frames 3 to 19.

[0097] Identify key events: For each frame within the cycle, calculate the toe-heel ordinates for the right and left sides.

[0098] Right side: The maximum difference is in frame 7 (approximately 6.98), corresponding to the right heel contact point; the minimum difference is in frame 19 (approximately -7.06), corresponding to the right toe departure point. Right side propulsion duration TR = frame 19 - frame 7 = 12 (frames).

[0099] Left side: The maximum difference is in frame 17 (approximately 3.87), corresponding to the left heel contact point; the minimum difference is in frame 10 (approximately -3.99), corresponding to the left toe departure point. Left side propulsion duration TL = frame 17 - frame 10 = 7 (frames).

[0100] Symmetry calculation: The absolute difference in duration ΔT = |12 - 7| = 5 frames. The symmetry index Bla1 is set to 0.2. After normalization to the maximum and minimum values ​​(over all 10 cycles), the final normalized Bla1' = 0.21 is obtained.

[0101] Performing the same calculation on all 10 cycles yields the normalized propulsion symmetry sequence Bla' = [0.21, 0.55, 0.68, 0.72, 0.70, 0.65, 0.59, 0.62, 0.75, 0.78].

[0102] 2. Walking coordination (taking cycle 1 as an example): Period 1: Contains frames 3 to 19.

[0103] Calculate the relative distance: For each frame, calculate |heel coordinate - hip anchor coordinate|. The hip anchor coordinate is obtained from step S200.

[0104] The minimum value of the left relative distance sequence appears in frame 14 (approximately 0.02), which corresponds to the left leg extension degree E. L = 0.02.

[0105] The minimum value of the right-side relative distance sequence appears in frame 7 (approximately 0.01), which corresponds to the right leg extension degree E. R = 0.01.

[0106] Coordination calculation: The absolute difference in the extension of the right and left legs, ΔE = |0.02 - 0.01| = 0.01. Walking coordination Cor1 = 0.99. After normalization to the maximum and minimum values ​​(all 10 cycles), the final normalized Cor1' = 0.98 is obtained.

[0107] The same calculation was performed on all 10 cycles to obtain the normalized walking coordination sequence Cor' = [0.98,0.92, 0.88, 0.85, 0.86, 0.89, 0.91, 0.93, 0.96, 0.97].

[0108] 3. Calculate gait stability and coordination stability: Gait stability Sw: Calculate the squared difference in propulsion symmetry between adjacent cycles and take the mean.

[0109] Differential squared sequence: (0.21-0.55) 2 =0.1156, (0.55-0.68) 2 =0.0169, ..., (0.75-0.78) 2 =0.0009. There are 9 values ​​in total.

[0110] Gait stability Sw = 0.966.

[0111] Coordination stability Pw: Calculate the squared difference in coordination between adjacent cycles and take the mean.

[0112] Differential squared sequence: (0.98-0.92) 2 =0.0036, (0.92-0.88) 2 =0.0016, ..., (0.96-0.97) 2 =0.0001.

[0113] Coordination stability Pw = 0.999.

[0114] 4. Calculate the degree of recovery (Res): Overall stability Sta: Sta = 0.6×Sw + 0.4×Pw = 0.6×0.966 + 0.4×0.999 =0.979.

[0115] Average propulsion symmetry: = (0.21+0.55+...+0.78) / 10 = 0.625.

[0116] Average walking coordination: = (0.98+0.92+...+0.97) / 10 = 0.915.

[0117] The first recovery coefficient R1 is: Δ = |(0.625+0.915) / 2 - 1| = |0.77 - 1| = 0.23. R1 =exp(-0.23) ≈ 0.795.

[0118] Recovery progress: .

[0119] Step S400: Evaluate postoperative efficacy The patient's final calculated recovery level, Res, was 0.778. Based on the preset thresholds Th1=0.8 and Th2=0.6: Res = 0.778 < 0.8 and > 0.6.

[0120] Therefore, the patient's rehabilitation level was determined to be Level 2 (intermediate rehabilitation).

[0121] Therapeutic effect analysis and verification: The calculation results show that: The average indicators are acceptable but asymmetrical: the average walking coordination (0.915) is higher than the average propulsion symmetry (0.625), indicating that the height of the patient's legs lifted during gait is relatively consistent, but there is a significant asymmetry in the duration of the push-off force (especially in cycle 1, the right side has a significantly longer force exertion time than the left side, suggesting that the left side has limited force exertion after surgery).

[0122] Good periodic stability: Gait stability (0.966) and coordination stability (0.999) are both close to 1, indicating that although there is asymmetry in force exertion, the patient's gait pattern remains stable in continuous cycles without significant fluctuations, indicating that neural control is basically stable and there is no obvious compensatory gait adjustment.

[0123] Comprehensive assessment and clinical comparison: The overall recovery level was 0.778, which is at a moderate level. This is consistent with the subjective assessment results of the clinicians. The doctors believe that although the patient has regained the ability to walk independently 3 months after surgery, gait asymmetry still exists (shorter stance phase time on the affected side), and recommend continued targeted muscle strength training and gait symmetry training.

[0124] This application example fully demonstrates the entire process from video data acquisition, key point extraction, gait cycle segmentation, multi-dimensional index calculation to final efficacy evaluation. Through data-driven and quantitative analysis, this invention successfully captured "force asymmetry" and "leg lift incoordination," which are difficult to distinguish with conventional indicators, and provided objective and accurate rehabilitation assessment results for clinical practice, verifying the effectiveness and clinical value of the method of this invention.

[0125] In summary, this invention analyzes the temporal changes of key points in the feet and hips during walking videos of patients after spinal surgery. First, it obtains stable hip anchor point coordinates and divides gait cycles. Then, it extracts propulsion symmetry and walking coordination indicators within each cycle, quantifying the consistency of force exertion duration and leg lift height between the two legs. Based on this, it further analyzes the fluctuations of these indicators between cycles to obtain gait stability and coordination stability, and comprehensively derives the degree of recovery. Finally, it objectively assesses the postoperative efficacy based on the degree of recovery. This invention fully utilizes gait temporal information, enabling it to keenly capture bilateral limb movement differences and subtle fluctuations within cycles. It effectively solves the problems of strong subjectivity and difficulty in distinguishing similar gait abnormalities in traditional assessment methods, thus improving the accuracy and reliability of postoperative efficacy assessment for spinal surgery.

[0126] Example 2

[0127] Based on the same inventive concept as the spinal surgery efficacy assessment method based on gait temporal analysis in the foregoing embodiments, the present invention also provides a spinal surgery efficacy assessment device based on gait temporal analysis, such as... Figure 2 As shown, the device includes: The data acquisition module 201 is used to acquire the coordinates of the toes, heels and hips on the left and right sides of each frame in the walking video of a patient after spinal surgery. The cycle division module 202 is used to obtain the hip anchor point coordinates of each time frame according to the hip coordinate distribution of different time frames, and to obtain multiple gait cycles composed of time frames according to the distribution of heel coordinates and toe coordinates of different time frames. The stability analysis module 203 is used to obtain the propulsion symmetry and walking coordination of each gait cycle based on the distribution of the heel coordinates and toe coordinates of different time frames within each gait cycle, as well as the positional characteristics of the hip anchor point coordinates; and to obtain gait stability and coordination stability based on the distribution of propulsion symmetry and walking coordination of different gait cycles. The rehabilitation assessment module 204 is used to obtain the degree of rehabilitation based on the gait stability and coordination stability, and to evaluate the postoperative efficacy based on the degree of rehabilitation.

[0128] Example 3

[0129] Based on the same inventive concept as the spinal surgery efficacy evaluation method based on gait temporal analysis in the foregoing embodiments, the present invention also provides a spinal surgery efficacy evaluation device based on gait temporal analysis, which stores a computer program that, when executed by a processor, implements any step of the spinal surgery efficacy evaluation method based on gait temporal analysis described above.

[0130] Among them, Figure 3 In this document, a bus architecture (represented by bus 300) is used. Bus 300 may include any number of interconnected buses and bridges, linking various circuits including one or more processors represented by processor 302 and memory represented by memory 304. Bus 300 may also link various other circuits such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and therefore will not be described further herein. Bus interface 306 provides an interface between bus 300 and receiver 301 and transmitter 303. Receiver 301 and transmitter 303 may be the same element, i.e., a transceiver, providing a unit for communicating with various other devices over a transmission medium.

[0131] The processor 302 is responsible for managing the bus 300 and general processing, while the memory 304 can be used to store the data used by the processor 302 when performing operations.

[0132] Example 4

[0133] Based on the same inventive concept as the spinal surgery efficacy evaluation method based on gait timing analysis in the foregoing embodiments, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in Embodiment 1.

Claims

1. A method for evaluating the efficacy of spinal surgery based on gait temporal analysis, characterized in that, The method includes: Obtain the coordinates of the toes, heels, and hips on both sides of each frame in a walking video of a patient after spinal surgery. Based on the hip coordinate distribution of different time frames, the hip anchor point coordinates of each time frame are obtained; based on the distribution of heel coordinates and toe coordinates of different time frames, multiple gait cycles composed of time frames are obtained. Based on the distribution of heel and toe coordinates at different time frames within each gait cycle, and the positional characteristics of the hip anchor point coordinates, the propulsion symmetry and walking coordination of each gait cycle are obtained; based on the distribution of propulsion symmetry and walking coordination in different gait cycles, gait stability and coordination stability are obtained, and the degree of rehabilitation is obtained. The postoperative efficacy was assessed based on the degree of recovery described.

2. The method for evaluating the efficacy of spinal surgery based on gait temporal analysis as described in claim 1, characterized in that, Obtaining the hip anchor point coordinates for each time frame includes: The average coordinates of the hips on the left and right sides of each frame are obtained and used as the coordinates of the center point of the hips in each frame. A temporal sequence is constructed based on the coordinates of the center point of the hip at different time frames in chronological order; The time series is subjected to a moving average filter to obtain the smoothed hip center point coordinates of each time frame, which are used as the hip anchor point coordinates.

3. The method for evaluating the efficacy of spinal surgery based on gait temporal analysis as described in claim 1, characterized in that, The process of obtaining multiple gait cycles composed of time frames includes: For each time frame, calculate the difference between the longitudinal coordinates of the left toe and the right heel, and record it as the longitudinal difference between the two feet. According to the time sequence, the longitudinal differences of the two feet in all time frames are used to form a difference sequence, and the minimum point in the sequence is obtained. The detected minimum point is used as the dividing time frame. All time frames are divided according to the dividing time frame, and the range formed by the consecutive time frames is taken as a gait cycle.

4. The method for evaluating the efficacy of spinal surgery based on gait temporal analysis as described in claim 1, characterized in that, Obtaining the propulsion symmetry for each gait cycle includes: For each time frame within any gait cycle, the first difference between the ordinates of the toe coordinates and the heel coordinates on the left and right sides is obtained respectively. The minimum value of the first difference among all time frames is selected, and the corresponding time frame is taken as the toe departure point. The maximum value of the first difference among all time frames is selected, and the corresponding time frame is taken as the heel contact point. The difference in the order of the corresponding time frames between the heel contact point and the toe departure point on each side is obtained as the propulsion duration; The difference in propulsion duration between the right and left sides is obtained, and the difference in propulsion duration is negatively correlated to obtain the propulsion symmetry.

5. The method for evaluating the efficacy of spinal surgery based on gait temporal analysis as described in claim 1, characterized in that, Obtaining the walking coordination for each gait cycle includes: For each time frame within any step cycle, calculate the absolute difference between the longitudinal coordinates of the heels on the left and right sides and the longitudinal coordinates of the hip anchor point, and use this as the relative distance on that side. Construct the relative distance sequence of different time frames within any step state period; select the minimum value in the relative distance sequence as the degree of leg extension; By performing a negative correlation mapping on the absolute differences, the walking coordination index of the gait cycle is obtained.

6. The method for evaluating the efficacy of spinal surgery based on gait temporal analysis as described in claim 1, characterized in that, The acquisition of gait stability and coordination stability includes: Calculate the squared difference in propulsion symmetry between adjacent gait cycles, and obtain gait stability based on the mean of the squared differences across all adjacent cycles; Calculate the squared difference in walking coordination between adjacent gait cycles, and obtain coordination stability based on the mean of the squared differences across all adjacent cycles.

7. The method for evaluating the efficacy of spinal surgery based on gait temporal analysis as described in claim 1, characterized in that, The degree of recovery achieved includes: The overall stability is obtained by weighting the gait stability and the coordination stability. The mean of propulsion symmetry within all gait cycles is calculated as the average propulsion symmetry, and the mean of walking coordination within all gait cycles is calculated as the average walking coordination. The first rehabilitation coefficient is obtained based on the degree to which the mean of the average propulsion symmetry and the average walking coordination is close to 1; The product of the overall stability and the first recovery coefficient is taken as the degree of recovery.

8. A device for evaluating the efficacy of spinal surgery based on gait temporal analysis, characterized in that, include: The data acquisition module is used to acquire the coordinates of the toes, heels, and hips on the left and right sides of each frame in the walking video of a patient after spinal surgery. The cycle division module is used to obtain the hip anchor point coordinates of each time frame based on the hip coordinate distribution of different time frames, and to obtain multiple gait cycles composed of time frames based on the distribution of heel coordinates and toe coordinates of different time frames. The stability analysis module is used to obtain the propulsion symmetry and walking coordination of each gait cycle based on the distribution of the heel coordinates and toe coordinates of different time frames within each gait cycle, as well as the positional characteristics of the hip anchor point coordinates; and is used to obtain gait stability and coordination stability based on the propulsion symmetry and walking coordination distribution of different gait cycles. The rehabilitation assessment module is used to obtain the degree of rehabilitation based on the gait stability and coordination stability, and to evaluate the postoperative efficacy based on the degree of rehabilitation.

9. A device for evaluating the efficacy of spinal surgery based on gait temporal analysis, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1 to 7.

Images