Automated evaluation method for postoperative results of plastic surgery

Abstract

This invention discloses an automated evaluation method for postoperative effects of plastic surgery, involving the interdisciplinary fields of medical aesthetics and computer vision. The method includes: defining several facial dynamic units and extracting motion vector sequences for each unit through motion analysis; calculating inter-unit synergy indicators to quantify overall facial coordination; further performing micro-expression recognition on facial videos to obtain emotion categories and intensities; when a preset emotion category is detected, tracing back its motion vector sequence within a time window and calculating the degree of motion abnormality deviation; combining emotion intensity, maximum deviation value, and temporal correlation with micro-expression events for weighted calculation to obtain a correlation score; and jointly judging the correlation score and synergy indicators to output a comprehensive evaluation report. This invention overcomes the limitations of traditional evaluation methods that cannot reflect dynamic functional recovery and psychological state, providing a multi-dimensional and comprehensive objective evaluation basis for the effects of complex plastic surgery.

Description

Technical Field

[0001] This invention relates to the field of medical aesthetics and computer vision interdisciplinary technology, specifically an automated evaluation method for postoperative effects of plastic surgery. Background Technology

[0002] With the rapid development of computer vision and artificial intelligence technologies, digital and intelligent assessment methods have gradually penetrated into the field of medical aesthetics. In recent years, automated assessment methods based on image processing and analysis have emerged. These methods provide objective data support for postoperative results by extracting key facial points, measuring specific angles and distances, and calculating symmetry indicators.

[0003] However, when faced with the increasing prevalence of complex plastic surgeries, such as performing rhinoplasty and blepharoplasty simultaneously, existing technologies have revealed significant limitations. These methods typically follow the logic of isolated analysis, with the core being to measure whether the shape of a single part meets the standard by collecting static images of the patient after surgery and using predefined, universal aesthetic standards.

[0004] This assessment method has three key flaws: First, it fails to capture the coordinated movement characteristics of facial muscle groups in dynamic expressions, resulting in a lack of assessment of postoperative neuromuscular function recovery; second, current technology lacks the ability to correlate macroscopic coordinated movement with instantaneous micro-expressions, making it difficult to identify subtle expression control disorders caused by surgery; and third, static analysis methods cannot reveal patients' potentially hidden subjective dissatisfaction stemming from overall disharmony, which often manifests through specific micro-expressions within specific time windows. More seriously, when patients undergo plastic surgery on multiple areas simultaneously, current methods cannot distinguish the differentiated impact of different surgical sites on overall facial dynamic coordination, nor can they quantify the spatiotemporal correlation between emotional triggering events and local movement abnormalities. This assessment blind spot may cause doctors to miss crucial postoperative recovery intervention opportunities or misjudge the actual aesthetic effect of the surgery.

[0005] To address the aforementioned issues, existing technologies urgently need improvement. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides an automated method for evaluating the postoperative effects of plastic surgery.

[0007] To achieve the above objectives, the technical solution of the present invention is as follows:

[0008] In a first aspect, the present invention discloses an automated evaluation method for postoperative effects of plastic surgery, comprising the following steps:

[0009] Acquire postoperative facial video data and 3D facial scan data of the patient. The facial video data includes dynamic images of the patient performing a standard facial expression sequence.

[0010] Several facial dynamic units are defined based on 3D facial scan data;

[0011] Based on facial video data, motion vector sequences of facial dynamic units are obtained through motion analysis;

[0012] Based on the motion vector sequence, the synergy index between the motion vector sequences of different facial dynamic units is calculated. The synergy index is used to characterize facial coordination.

[0013] Micro-expression analysis is performed on facial video data to identify micro-expression events and map them to corresponding emotion categories and intensity.

[0014] In response to the identification of micro-expression events of a preset emotion category, the motion vector sequence within a preset time window is traced back based on the occurrence time of the micro-expression event, and the abnormal deviation of motion of each facial dynamic unit within the time window is calculated.

[0015] A correlation score is obtained by weighting the intensity of emotion, the maximum value of the abnormal deviation of movement, and the temporal correlation between the abnormal deviation of movement and the micro-expression event.

[0016] A joint assessment report is generated based on correlation scores and synergy indicators.

[0017] Secondly, this invention discloses an automated evaluation system for postoperative effects in plastic surgery, comprising:

[0018] The data acquisition module is used to acquire postoperative facial video data and three-dimensional facial scan data of the patient. The video data includes dynamic images of the patient performing a standard facial expression sequence.

[0019] The unit definition module is used to define several facial dynamic units based on 3D facial scan data;

[0020] The motion analysis module is used to obtain the motion vector sequence of facial dynamic units based on facial video data through motion analysis;

[0021] The synergy analysis module is used to calculate the synergy index between motion vector sequences of different facial dynamic units based on the motion vector sequence. The synergy index is used to characterize facial coordination.

[0022] The micro-expression analysis module is used to perform micro-expression analysis on facial video data, identify micro-expression events and map them to corresponding emotion categories and intensity.

[0023] The micro-expression time backtracking module is used to respond to micro-expression events of preset emotion categories, backtrack the motion vector sequence within a preset time window based on the occurrence time of the micro-expression event, and calculate the abnormal deviation of motion of each facial dynamic unit within the time window;

[0024] The correlation score calculation module is used to calculate the correlation score by weighting the maximum value of the emotion intensity, the deviation of the movement abnormality, and the temporal correlation between the deviation of the movement abnormality and the micro-expression event.

[0025] The comprehensive evaluation module is used to make joint judgments based on correlation scores and synergy indicators, and output a comprehensive evaluation report.

[0026] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0027] 1. By combining the motion coordination analysis of the facial dynamics unit with micro-expression recognition, the system can simultaneously capture the coordinated movement ability of the facial muscle groups after surgery and the abnormal expression of emotions in an instant. This overcomes the limitations of traditional static assessment, which cannot reflect the dynamic functional recovery and psychological state, and provides a multi-dimensional and comprehensive objective assessment basis for the effect of composite plastic surgery.

[0028] 2. By retrospectively analyzing the time window triggered by micro-expression events and calculating the deviation of movement abnormalities, the system can establish the temporal and spatial correlation between emotional events and local muscle movement abnormalities. This mechanism can identify subtle neuromuscular control disorders caused by surgery and detect patients' potential hidden dissatisfaction in the early stages, thus providing doctors with a reference for intervention and repair.

[0029] 3. By generating comprehensive evaluation reports, it is possible not only to distinguish postoperative effect conclusions of different levels, but also to locate potential problem areas in a fine-grained manner, providing doctors with dual reference for postoperative functional recovery and aesthetic effects, thereby improving the timeliness of intervention and supporting the design of personalized rehabilitation and secondary repair plans. Attached Figure Description

[0030] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0031] Figure 1 This is an overall block diagram of the method in Embodiment 1 of the present invention;

[0032] Figure 2 This is a flowchart of the method according to Embodiment 1 of the present invention;

[0033] Figure 3 This is an overall block diagram of the system in Embodiment 2 of the present invention. Detailed Implementation

[0034] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0035] In traditional postoperative outcome assessment systems, isolated analytical logic leads to a lack of correlation between facial dynamic coordination and instantaneous emotional responses. These systems rely solely on static images to extract geometric features, failing to capture the motor coordination of facial muscle groups within standard expression sequences, and neglecting the temporal coupling between micro-expression events and motor abnormalities. When dealing with complex plastic surgery cases, this deficiency causes a significant discrepancy between assessment results and the patient's subjective experience. Specifically, the morphological parameters recorded by the system may meet preset standards, but the patient may still report a subjective feeling of facial stiffness or unnaturalness.

[0036] For example, in the postoperative evaluation scenario of combined rhinoplasty and ophthalmological plastic surgery, the system acquires nasal bridge height and eyelid symmetry data through 3D scanning and calculates facial golden ratio parameters based on static images. When the patient performs a standard smiling expression, the system fails to detect the phase delay of the motion vectors of the zygomaticus major and orbicularis oris muscles, and also misses the negative emotional micro-expressions corresponding to slight tremors at the corners of the mouth. The resulting evaluation report shows that the facial symmetry index meets the standard, but in actual clinical follow-up, it was found that the patient's expression was stiff due to abnormal muscle coordination, and there was anxiety caused by the postoperative results not meeting expectations.

[0037] If the above issues are not addressed, the assessment system will be unable to identify functional defects caused by neuromuscular control dysregulation, resulting in a lack of crucial biomechanical basis for postoperative recovery planning. The absence of dynamic coordination indicators can mask abnormal compensatory phenomena at key nodes in the facial motor chain, delaying the optimal intervention time for secondary repair surgery. Decoupling analysis of micro-expressions and motor abnormalities will reduce the corrective value of emotional state for postoperative outcome assessment, causing systematic errors between objective data and patient subjective satisfaction.

[0038] To address the aforementioned issues, this application first recognizes that existing evaluation methods rely solely on static image analysis, failing to capture the coordinated movement characteristics of facial muscle groups in dynamic expressions, and neglecting the temporal correlation between micro-expression events and motion anomalies. To resolve this, this application attempts to combine 3D facial scanning with dynamic video analysis. It establishes a motion analysis framework by defining facial dynamic units and introduces a time window backtracking mechanism to couple micro-expression recognition with motion vector anomaly detection. Furthermore, by constructing a coordination index and a correlation scoring model, it achieves joint analysis of dynamic coordination and emotional response, thereby overcoming the limitations of isolated evaluation.

[0039] Example 1:

[0040] like Figures 1-2 As shown, the automated evaluation method for postoperative results of plastic surgery includes the following steps:

[0041] Postoperative facial video data and 3D facial scan data of the patient were acquired. The facial video data included dynamic images of the patient performing a standard facial expression sequence. Specifically, the facial video data refers to the dynamic image data of the patient performing a standard facial expression sequence after surgery, recorded by a camera device. This can be achieved using a high frame rate camera combined with facial marker tracking technology to capture the continuous movement characteristics of facial muscle groups during facial expression changes. The 3D facial scan data refers to the 3D geometric model data of the patient's facial structure obtained by a 3D scanner. This can be achieved using structured light scanning or laser scanning technology to accurately delineate the anatomical boundaries of facial dynamic units.

[0042] Several facial dynamic units are defined based on three-dimensional facial scan data. Among them, a facial dynamic unit refers to a local region with independent motion function that is divided based on the three-dimensional facial structure. Specifically, it can be defined by combining the distribution of muscle attachment points with motion correlation analysis, which is used to establish a refined spatial framework for facial motion analysis.

[0043] Based on facial video data, motion vector sequences of facial dynamic units are obtained through motion analysis. The motion vector sequence refers to the displacement change data of facial dynamic units in the time dimension extracted by computer vision algorithms. Specifically, it can be implemented using optical flow or feature point tracking algorithms to quantify the spatiotemporal characteristics of facial muscle movement.

[0044] Based on the motion vector sequence, the synergy index between the motion vector sequences of different facial dynamic units is calculated. The synergy index is used to characterize facial coordination. The synergy index refers to the correlation measure between motion sequences of different facial regions calculated by statistical methods. Specifically, it can be implemented using Pearson correlation coefficient or mutual information entropy algorithm, and is used to evaluate the degree of coordination of facial muscle groups in expression.

[0045] Micro-expression analysis is performed on facial video data to identify micro-expression events and map them to corresponding emotion categories and intensity. Micro-expression events refer to the instantaneous activation state of facial action units that lasts for a shorter period than regular expressions. Specifically, a deep learning-based micro-expression recognition model can be used to capture unconscious emotional signals revealed by patients.

[0046] In response to the identification of micro-expression events of a preset emotion category, the motion vector sequence within a preset time window is traced back based on the occurrence time of the micro-expression event, and the motion abnormality deviation of each facial dynamic unit within the time window is calculated. The motion abnormality deviation refers to the standardized difference value calculated by comparing real-time motion data with global benchmark data. Specifically, it can be implemented by using the Z-score standardization method combined with sliding window technology to detect the deviation between local muscle movement and overall expression pattern.

[0047] A correlation score is calculated by weighting the maximum value of the emotional intensity, the deviation of the movement abnormality, and the temporal correlation between the deviation of the movement abnormality and the micro-expression event. Temporal correlation refers to the sequential relationship between the micro-expression event and the movement abnormality established through time series analysis, which can be implemented using cross-correlation analysis or event synchronicity detection algorithms to verify the causal relationship between emotional triggering and movement abnormality. The correlation score is a quantitative evaluation value generated by fusing multi-dimensional features, which can be implemented using linear weighting or machine learning regression models to comprehensively reflect the degree of correlation between emotional intensity and movement abnormality.

[0048] The comprehensive assessment report is generated by combining correlation scores and synergy indicators. The comprehensive assessment report refers to the classification assessment conclusion generated by the decision rule engine. Specifically, it can be implemented by using threshold judgment combined with logical decision tree to output multi-dimensional postoperative assessment results that include facial coordination and emotional correlation.

[0049] As a preferred embodiment, the solution of this application is specifically implemented as follows:

[0050] First, a high-speed camera is used to acquire facial video data of the patient performing a standard expression sequence, and a 3D structured light scanner is used to acquire 3D point cloud data of the patient's face; the standard expression sequence includes basic expressions such as natural state, smiling, frowning, and surprise.

[0051] Based on 3D point cloud data, facial dynamic units are defined in key facial areas such as the forehead, between the eyebrows, around the eyes, the sides of the nose, and the corners of the mouth. Each unit consists of several 3D coordinate points used to track the movement of the local area.

[0052] Image processing and feature point tracking are performed on facial video data to extract the motion trajectory of each facial dynamic unit in the entire expression sequence, resulting in a three-dimensional motion vector sequence with time as the independent variable.

[0053] The correlation coefficients between motion vector sequences of different facial dynamic units are calculated to construct a synergy index matrix; each element in the matrix represents the degree of motion synergy between two units.

[0054] By using deep learning models to perform micro-expression analysis on facial video data, the occurrence time, duration, and facial regions involved in micro-expression events are identified, and these are mapped to corresponding emotion categories (such as joy, disgust, surprise, etc.) and emotion intensity values.

[0055] When a preset emotion category, such as a micro-expression event of negative emotion, is detected, the motion vector sequence within a preset time window (e.g., 500 milliseconds) is traced back based on the time of its occurrence. For each facial dynamics unit, the degree of deviation between the motion vector amplitude within that time window and the average amplitude of the entire expression sequence is calculated to obtain the motion abnormality deviation.

[0056] The correlation score is calculated by weighting the emotional intensity of the micro-expression event, the maximum value of the deviation of the motion abnormality in each unit, and the temporal correlation between the motion abnormality and the micro-expression event. The temporal correlation is quantified by calculating the time difference between the peak of the motion abnormality and the occurrence time of the micro-expression event.

[0057] Finally, a comprehensive judgment is made based on the correlation score and the synergy index matrix. For example, when the correlation score exceeds a threshold and there are unit pairs with low synergy, it may indicate that there is incoordination in certain facial areas. An evaluation report is generated based on the judgment results, including facial coordination analysis and identification of potential problem areas.

[0058] Through the above-described approach, this application achieves an objective assessment of facial dynamic coordination after plastic surgery. By combining facial motion analysis and micro-expression recognition, dynamic abnormalities that are difficult to detect through static image analysis can be captured. The correlation scoring model establishes a link between instantaneous emotional reactions and facial motion abnormalities, helping to identify potential problems that may cause patient discomfort. The synergy index assesses the motion coordination of different facial regions from a holistic perspective. This dynamic, multi-dimensional assessment method improves the accuracy of evaluating the effects of complex plastic surgeries, providing a more comprehensive basis for developing personalized postoperative recovery plans. Simultaneously, this method also provides objective indicators for the timely detection and intervention of potential postoperative complications, which is beneficial for improving long-term patient satisfaction.

[0059] In some of the solutions mentioned above in this application, facial coordination is evaluated through coordination indices. However, if only a single coordination index is used in the calculation process, the local incoordination problem may be masked by the overall index due to the difference in coordination between different facial regions, and the coordination anomaly between specific facial dynamic units cannot be accurately identified.

[0060] This application further proposes a calculation process for the synergy index, which includes calculating the correlation coefficient between two motion vector sequences representing different facial dynamic units to obtain a set of synergy sub-indices. The synergy index is a set of synergy indices composed of all synergy sub-indices.

[0061] The correlation coefficient is calculated based on the motion vector sequences of two pairs of corresponding facial dynamic units, and the linear correlation between the sequences is quantified by statistical methods. The number of synergy sub-indicators depends on the total number of facial dynamic units, and each sub-indicator corresponds to the synergy between a pair of units. The synergy index set is constructed by storing all sub-indicators according to a preset data structure to form a multi-dimensional synergy feature matrix.

[0062] Specifically, after extracting the motion vector sequences of each dynamic unit from the facial video data, the Pearson correlation coefficient is calculated for each pair of units' motion vector sequences. For example, when the total number of facial dynamic units is N, C(N,2) synergy sub-indices can be generated, with each sub-indicator ranging from -1 to 1. The closer the value is to 1, the higher the synchronicity of the two units' motion. By integrating the sub-indices into a set, the independent synergy information between different unit pairs can be preserved. In the subsequent joint judgment process, each sub-indicator in the synergy index set can independently participate in the logical judgment, avoiding the over-smoothing effect of a single index on the overall synergy assessment. Thus, when the synergy sub-indicator value of a certain pair of units is significantly lower than other sub-indicators, the specific facial region with abnormal synergy can be quickly located, providing fine-grained data support for comprehensive assessment.

[0063] As a preferred embodiment, the solution of this application is specifically implemented as follows:

[0064] The calculation process of the synergy index involves calculating the correlation coefficient between two motion vector sequences representing different facial dynamic units, resulting in a set of synergy sub-indices. The synergy index is a set of synergy indices composed of all synergy sub-indices.

[0065] Specifically, the process begins by acquiring facial video data of the patient performing a standard facial expression sequence. Motion analysis is then used to obtain the motion vector sequences of each facial dynamics unit. Next, two different facial dynamics units are selected, and their respective motion vector sequences are extracted. The Pearson correlation coefficient is calculated between these two sequences to obtain a synergy sub-indicator. This process is repeated, calculating correlation coefficients for all possible pairs of facial dynamics units to obtain a set of synergy sub-indicators. Finally, all synergy sub-indicators are combined into a synergy index set, which serves as the overall synergy index.

[0066] For example, suppose we define five facial dynamic units: forehead, eyebrows, eyes, nose, and mouth. Pairing these five units together yields ten synergy sub-indices. The calculated synergy sub-indice values ​​range from -1 to 1, where 1 represents perfect positive correlation, -1 represents perfect negative correlation, and 0 represents no correlation. The set of these ten synergy sub-indices constitutes the final synergy index.

[0067] Through the above technical solution, this application can quantitatively assess the coordination of movements in different facial regions. This objectively reflects the overall degree of facial coordination, providing a reliable basis for subsequent comprehensive evaluation. Furthermore, by analyzing the numerical distribution of each sub-indicator in the coordination index set, it is possible to identify which facial regions have poor coordination, providing guidance for doctors to make targeted adjustments.

[0068] This application further proposes a process for calculating the motion anomaly deviation of each facial dynamics unit within a time window, including: for each facial dynamics unit, extracting its motion vector amplitude sequence within a preset time window, where the motion vector amplitude is the magnitude of the motion vector; calculating the degree of deviation between the motion vector amplitude sequence and a reference amplitude sequence, using this deviation as the motion anomaly deviation of the facial dynamics unit within the time window, where the reference amplitude sequence is the average sequence of motion vector amplitudes of the facial dynamics unit over the entire standard expression sequence time period, and the calculation formula is:

[0069]

[0070] in, For abnormal deviation of motion, The mean of the motion vector amplitude sequence within a preset time window. This is the global mean of the motion vector amplitude of this unit across the entire standard facial expression sequence. This represents the global standard deviation of the motion vector amplitude of this unit across the entire standard facial expression sequence.

[0071] In extracting the motion vector amplitude sequence within a preset time window, the start point and duration of the time window are triggered by the occurrence time of the micro-expression event, ensuring that the analysis scope covers the range of motion anomalies that may be affected by emotions. The reference amplitude sequence is constructed based on global statistics within a complete standard expression cycle. By calculating the global mean and standard deviation, an individualized dynamic benchmark is established. The deviation calculation adopts standardization processing, comparing the mean motion amplitude within the time window with the global benchmark to eliminate the interference of individual motion amplitude differences on anomaly judgment.

[0072] Specifically, upon detecting a micro-expression event, the system automatically extracts motion vector data within a preset time window prior to the event, extracts the amplitude sequence of each dynamic unit during that time period, and calculates its mean. By With global statistics and Substituting into the formula, we obtain the standardized deviation D. This value reflects the degree of deviation of the movement amplitude within the current time window from the normal facial expression cycle; a larger value indicates a higher probability of abnormality. This quantitative method can effectively distinguish between persistent movement abnormalities caused by poor postoperative recovery and normal facial expression fluctuations, improving the objectivity and accuracy of the assessment results.

[0073] Through the aforementioned technical solution, this application can accurately quantify the degree of motion abnormality of facial dynamic units within a specific time window. This allows for the capture of subtle changes in facial muscle groups before and after micro-expression events, providing objective evidence for subsequent analysis of facial coordination and emotional expression. This statistically based anomaly detection method avoids errors caused by subjective judgment, improving the accuracy and reliability of the assessment. Furthermore, by comparing motion characteristics within a local time window with global motion characteristics, instantaneous abnormal motion patterns can be effectively identified, providing important clues for revealing potential discomfort or dissatisfaction in patients.

[0074] This application further proposes that in the calculation process of the correlation score, the temporal correlation between the motion abnormality deviation degree and the micro-expression event specifically includes: in response to the identification of multiple micro-expression events within a preset time window, the absolute difference between the occurrence time of each micro-expression event and the peak time of the motion abnormality deviation degree is weighted and averaged using its emotional intensity as the weight to obtain the weighted average time difference.

[0075] The preset time window duration is set to a range of 300 to 800 milliseconds based on the periodic characteristics of the standard facial expression sequence; emotional intensity is quantified using a normalized value between 0 and 1; peak time is detected using a sliding window method with a window length of 50 milliseconds and a step size of 10 milliseconds; the weighted average is calculated using the following formula:

[0076]

[0077] in Let be the emotional intensity value of the i-th micro-expression event. Let be the occurrence time of the i-th event. This represents the peak time of abnormal motion deviation.

[0078] Specifically, when multiple micro-expression events exist in facial video data, the peak time point of motion anomaly deviation is first detected using the sliding window method. For each detected micro-expression event, the absolute difference between its occurrence time and peak time is calculated, and this difference is multiplied by the normalized emotional intensity value of the corresponding event as a weight. The sum of the weighted differences of all events divided by the total weight yields the weighted average time difference, reflecting the overall temporal correlation. For example, when two micro-expression events are detected with emotional intensities of 0.8 and 0.5, and time differences of 20 milliseconds and 50 milliseconds, respectively, the weighted average time difference is (0.8×20+0.5×50) / (0.8+0.5)=34.6 milliseconds. This indicator, by introducing a weighting factor of emotional intensity, makes micro-expression events with high emotional intensity contribute more to temporal correlation, thus more accurately characterizing the temporal matching degree between motion anomalies and emotional expression.

[0079] Through the above technical solution, this application can more accurately assess the temporal correlation between micro-expression events and facial movement abnormalities. By considering the occurrence time and emotional intensity of multiple micro-expression events and comparing them with the peak time of deviation from the movement abnormality, a comprehensive, weighted temporal correlation index can be obtained. This method not only considers the temporal proximity but also incorporates emotional intensity as a weight in the calculation, thus more comprehensively reflecting the relationship between micro-expressions and facial movement abnormalities. This helps improve the accuracy and reliability of postoperative evaluation of plastic surgery results, providing more valuable feedback information for doctors and patients.

[0080] This application further proposes a process for making a joint judgment based on the correlation score and the aforementioned synergy indicators and outputting a comprehensive evaluation report, including:

[0081] If the correlation score is greater than the first score threshold, and there is at least one synergy sub-indicator in the synergy index set whose value is less than the preset synergy threshold, then the first type of evaluation report will be generated and output.

[0082] If the correlation score is less than the second scoring threshold and the values ​​of all synergy sub-indicators in the synergy index set are greater than the synergy threshold, then a second type of evaluation report will be generated and output.

[0083] If none of the above conditions are met, a third type of evaluation report will be generated and output.

[0084] The first, second, and third categories of assessment reports are used to indicate different postoperative outcome assessment conclusions.

[0085] Specifically, the first scoring threshold can be set to 0.8, the second scoring threshold can be set to 0.2, and the synergy threshold can be set to 0.6. The system first determines whether the correlation score is greater than 0.8. If so, it further checks whether there are sub-indicators with values ​​less than 0.6 in the synergy indicator set. If so, a first-type assessment report is output, indicating that an objective synergy deficiency has been found and is associated with subjective negative emotions. If the correlation score is less than 0.2 and all synergy sub-indicators are greater than 0.6, a second-type assessment report is output, indicating excellent overall synergy and high patient psychological satisfaction. If neither of the above two conditions is met, a third-type assessment report is output, indicating that the overall facial synergy is average and there is potential subjective dissatisfaction that is separate from the objective indicators, requiring further observation and follow-up.

[0086] Through the above technical solution, this application achieves a comprehensive evaluation based on multi-dimensional indicators, combining the correlation score obtained from micro-expression analysis with the synergy index of facial dynamics units to comprehensively evaluate the effect of plastic surgery. This avoids the one-sided judgment that may result from a single indicator, improving the accuracy and reliability of the evaluation results. Furthermore, by setting different thresholds and evaluation categories, the evaluation results are made more detailed and specific, providing strong support for doctors to formulate subsequent treatment plans.

[0087] In some of the schemes described above in this application, the synergy index is calculated based on the correlation coefficient between motion vector sequences of different facial dynamic units. However, during the dynamic changes of facial expressions, the synergy index may fluctuate due to the instability of muscle movement in local time periods, which leads to a decrease in the reliability of the evaluation results.

[0088] This application further proposes a temporal stability verification mechanism after calculating the synergy index, including: for each synergy sub-index, extracting its motion vector sequence within the entire cycle of the standard facial expression sequence; dividing the sequence into multiple equal-length sub-segments, calculating the local synergy value of the synergy sub-index within each sub-segment; calculating the standard deviation of the local synergy values ​​of all sub-segments as the temporal stability score of the synergy sub-index; if the temporal stability score of a certain synergy sub-index is higher than a preset stability threshold, then marking the synergy sub-index as an unstable unit from the synergy index set; when making joint judgments based on the synergy index set, assigning a weight coefficient to each synergy sub-index in the synergy index set, and using the weighted synergy index set for joint judgments; wherein, the weight coefficient of the synergy sub-index marked as an unstable unit is lower than that of the unmarked synergy sub-index.

[0089] The temporal stability verification mechanism calculates local synergy values ​​by dividing the sequence into equal-length sub-segments. The number of sub-segments is determined based on the total duration of the standard facial expression sequence; for example, if the total duration is in the seconds range, it is divided into 1 to 2 sub-segments. The local synergy value is calculated using the same correlation coefficient algorithm as the global synergy sub-indicator, but the data range is limited to the sub-segments. The standard deviation is calculated using an unbiased estimation formula, and the preset stability threshold is set based on the statistical distribution of historical data, for example, taking a normal distribution range of 1 / 2 the standard deviation. The weighting coefficients employ a linear decreasing strategy.

[0090] Specifically, during the execution of a standard facial expression sequence, facial muscle movements may exhibit periodic instability due to postoperative recovery. By dividing the full-cycle motion vector sequence into equal-length sub-segments (e.g., dividing a 1-second sequence into 2-second sub-segments), the synergy sub-indicator is recalculated within each sub-segment. If the fluctuation of a certain synergy sub-indicator exceeds a preset threshold across different sub-segments, it indicates that the indicator is significantly affected by local muscle movement abnormalities. By calculating the standard deviation of the synergy values ​​for each sub-segment, for example, if the standard deviation is higher than a certain value, it is identified as an unstable unit, and its weight coefficient is reduced in subsequent joint judgments, effectively suppressing evaluation bias caused by abnormal data in local segments. This dynamic weight adjustment mechanism allows the comprehensive evaluation report to focus more on temporally stable synergy indicators, improving the robustness of the evaluation results.

[0091] As a preferred embodiment, the solution of this application is specifically implemented as follows:

[0092] After calculating the synergy index, a temporal stability verification mechanism is introduced. For each synergy sub-index, its motion vector sequence over the entire period of the standard facial expression sequence is extracted. This sequence is divided into multiple equal-length sub-segments, for example, the entire sequence is evenly divided into 10 sub-segments. The local synergy value of the synergy sub-index within each sub-segment is calculated. The standard deviation of the local synergy values ​​across all sub-segments is calculated as the temporal stability score of the synergy sub-index.

[0093] Furthermore, a preset stability threshold of 0.2 is set. If the time-series stability score of a certain synergy sub-indicator is higher than 0.2, then that synergy sub-indicator is marked as an unstable unit from the synergy indicator set. When making a joint judgment based on the synergy indicator set, each synergy sub-indicator in the synergy indicator set is assigned a weight coefficient. Specifically, unmarked synergy sub-indicators are assigned a weight coefficient of 1.0, and synergy sub-indicators marked as unstable units are assigned a weight coefficient of 0.5. The weighted synergy indicator set is then used for joint judgment.

[0094] Through the above technical solution, this application introduces a temporal stability verification mechanism, improving the reliability of the synergy index. By analyzing the changes of the synergy sub-indicators over different time periods, unstable facial dynamic units are identified, and their weights are reduced in subsequent judgments. This avoids misjudgments caused by unstable movements in certain facial areas, improving the accuracy and robustness of the overall assessment results. Simultaneously, this mechanism provides doctors with more detailed facial dynamic analysis information, helping to develop more precise postoperative rehabilitation plans.

[0095] This application further proposes to strengthen the process by establishing a spatial coupling coefficient matrix after calculating the motion anomaly deviation.

[0096] The predefined spatial coupling coefficient matrix describes the physiological correlation strength between each facial dynamics unit and each micro-expression action unit. Corresponding correlation strength row vectors are extracted based on the action units activated by the current micro-expression event. The spatial similarity between the vector formed by the motion abnormality deviation of all facial dynamics units and the correlation strength row vector is calculated. This spatial similarity is then introduced as a new weight into the correlation score calculation formula to obtain a strengthened correlation score. For example, the spatial coupling coefficient matrix predefines the physiological correlation strength between units using anatomical data; the correlation strength row vector reflects the coupling relationship between units related to the current micro-expression event; and the spatial similarity calculation uses a vector cosine similarity algorithm.

[0097] Specifically, after a micro-expression event is identified, the system extracts the corresponding row vector from a predefined matrix based on the event type. This row vector contains the association strength values ​​between each dynamic unit and the current micro-expression action unit. Subsequently, the motion abnormality deviation of each unit is compiled into a vector, and its similarity is calculated with the row vector. This similarity value is used to adjust the association score, so that the abnormality deviation of units with a stronger physiological association with the current micro-expression has a greater impact on the score. For example, when a micro-expression event involves an eye action unit, the deviation of the physiologically associated forehead dynamic unit receives a higher weight through similarity weighting, thus more accurately reflecting the motion abnormality caused by the event. Therefore, the enhanced association score can more accurately characterize the association strength between micro-expression events and motion abnormalities, improving the reliability of the comprehensive assessment.

[0098] As a preferred embodiment, the solution of this application is specifically implemented as follows:

[0099] After calculating the deviation of abnormal movements, a spatial coupling coefficient matrix is ​​established for enhanced processing. First, a spatial coupling coefficient matrix is ​​predefined to describe the physiological correlation strength between each facial dynamic unit and each micro-expression movement unit. This matrix can be obtained through statistical analysis of a large amount of clinical data, reflecting the anatomical connections between muscle groups in different facial regions.

[0100] Specifically, the rows of the matrix represent facial dynamics units, and the columns represent micro-expression units. Each element in the matrix has a value between 0 and 1, with larger values ​​indicating a stronger physiological correlation between the two. For example, the coefficient between dynamics units in the eyebrow area and units expressing surprise may be higher, while the coefficient between dynamics units in the corner of the mouth area and units expressing blinking may be lower.

[0101] When a micro-expression event is detected, the corresponding correlation strength row vector is extracted from the spatial coupling coefficient matrix based on the action unit activated by the event. This vector reflects the theoretical correlation between the current micro-expression and various facial regions.

[0102] Furthermore, the spatial similarity between the vector formed by the motion anomaly deviations of all facial dynamic units and the aforementioned correlation strength row vector is calculated. Cosine similarity can be used here. A high spatial similarity between two vectors indicates a strong match between the observed motion anomalies and the theoretically predicted micro-expression correlation regions, increasing the credibility of a causal relationship between micro-expressions and facial motion anomalies.

[0103] Finally, the calculated spatial similarity is introduced as a new weight into the association score calculation formula. For example, the original association score can be multiplied by (1 + α * spatial similarity), where α is an adjustable weighting coefficient. This results in an enhanced association score that more comprehensively reflects the degree of association between micro-expressions and facial movement abnormalities.

[0104] Through the aforementioned technical solution, this application fully utilizes knowledge of facial anatomy, incorporating the spatial distribution characteristics of micro-expressions and facial movements into its consideration. This method not only improves the accuracy and reliability of the correlation score but also effectively reduces interference caused by random movements in other facial areas. Furthermore, this approach provides richer and more precise input information for subsequent comprehensive evaluation, contributing to the scientific rigor and persuasiveness of the overall evaluation results.

[0105] This application further proposes a time consistency verification mechanism after obtaining the correlation score: defining an extended time window, the duration of which is longer than the preset time window; calculating the time-shifted cross-correlation function between the deviation of motion abnormality and the intensity of emotion within the extended time window; determining whether the peak value of the time-shifted cross-correlation function at zero time shift exceeds the preset correlation threshold; if so, confirming that the correlation between the micro-expression event and motion abnormality on which the current calculation is based is reliable, and outputting the calculated correlation score; otherwise, determining that the correlation between the micro-expression event and motion abnormality on which the current calculation is based is unreliable, and prompting to obtain the patient's next facial video data.

[0106] The extended time window is set to be more than twice the length of the preset time window to cover dynamic changes over a longer period of time. The time-shift cross-correlation function calculates the correlation coefficient at different time offsets using a sliding window algorithm, and the peak value at zero time shift reflects the synchronicity of the two in time. The correlation threshold is obtained through historical data statistics, for example, set to 0.7. When the correlation is unreliable, the system automatically triggers a data resampling command to avoid erroneous evaluations based on occasional correlations.

[0107] Specifically, the extended time window allows for the capture of abnormal motion patterns over a broader timeframe before and after micro-expression events. By calculating the time-shifted cross-correlation function, the correlation between motion abnormality deviation and emotional intensity at different time shifts can be quantified. If the peak value at zero time shift is significantly higher than the threshold, it indicates stable temporal synchronicity between the two, thus confirming a reliable association. For example, when the peak emotional intensity and peak motion abnormality deviation corresponding to a micro-expression event are highly synchronized within the extended time window, the cross-correlation function produces a peak value of 0.85 at zero time shift, exceeding the threshold of 0.7, and the system determines the association to be reliable. If the peak value of the cross-correlation function is lower than the threshold, it indicates that the current data may be affected by noise or by chance, requiring data re-collection to eliminate the influence of chance. Thus, the time consistency verification mechanism effectively improves the credibility of the association score and ensures the accuracy of the evaluation conclusion.

[0108] As a preferred embodiment, the specific implementation of this application is as follows: After calculating the correlation score, an extended time window is defined, the duration of which is set to twice the preset time window. Based on the starting time point of this extended time window, the motion abnormality deviation sequence and the emotion intensity sequence are simultaneously extracted. After normalizing the two sequences, the time-shift cross-correlation function is calculated using the sliding window method, where the sliding step size is set to the single-frame time interval of the video data, and the maximum time shift range is set to one-third of the total duration of the extended time window. By traversing all possible time shifts, the set of cross-correlation values ​​of the two sequences is calculated. The cross-correlation value at zero time shift is obtained by calculating the Pearson correlation coefficient of the two sequences in a time-aligned state. When the cross-correlation value at zero time shift exceeds a preset threshold, a verification pass signal is triggered, allowing the current correlation score to enter the subsequent comprehensive evaluation process; if it does not exceed the threshold, a data re-acquisition instruction is generated, which triggers the system to re-acquire video data when the patient performs the next round of standard expression sequences.

[0109] Through the above technical solution, this application effectively solves the problem of misjudgment of the correlation between micro-expression events and motion abnormalities in the prior art. By introducing time-shifted cross-correlation analysis within an extended time window, the coupling stability between emotion-triggered events and motion abnormalities in the time dimension can be verified, avoiding false associations caused by single, accidental events. This mechanism retains the true correlation signal while filtering out noise interference due to time dimension mismatch, thereby improving the credibility and clinical guidance value of postoperative effect evaluation conclusions.

[0110] This application further proposes extracting bilaterally symmetrical facial dynamic unit pairs, defining left and right symmetrical unit groups based on three-dimensional facial scan data; calculating the amplitude difference ratio and temporal phase difference of each symmetrical unit pair in a standard expression sequence according to the motion vector sequence; counting the number of unit pairs whose amplitude difference ratio exceeds a preset asymmetry threshold or whose temporal phase difference exceeds a preset phase tolerance, and calculating their proportion of the total symmetrical unit pairs, denoted as the asymmetric unit ratio; summing the amplitude difference ratio and temporal phase difference of all exceeding unit pairs, normalizing them respectively, and then summing them to obtain the asymmetry intensity index; using the asymmetric unit ratio and asymmetry intensity index as symmetry quantification parameters, and jointly judging them with the synergy index and correlation score to generate a comprehensive evaluation report.

[0111] The definition of symmetrical unit groups is achieved through the geometric symmetry axis in 3D facial scan data, ensuring the symmetry of unit pairs in anatomical structure. The motion amplitude difference ratio is calculated using a relative difference measure to avoid the influence of absolute differences on the evaluation results. The temporal phase difference is calculated using the peak time point difference to capture the temporal synchronicity of bilateral motion. The combined use of asymmetric unit proportion and intensity indicators can simultaneously reflect the breadth and depth of asymmetry.

[0112] Specifically, during the execution of the standard facial expression sequence, the motion vector amplitudes of bilateral symmetrical unit pairs are acquired in real time. At each time point, the motion vector amplitudes of the left and right units are input into the difference ratio calculation formula to generate a dynamic difference ratio sequence. After the peak time point of the motion vector of the same unit pair is identified, the time difference between the left and right peaks is calculated as the phase difference. By statistically analyzing the number of unit pairs exceeding the limit, the proportion of asymmetrical units is calculated, reflecting the overall range of symmetry anomalies. The sum of the normalized difference ratio and the phase difference forms an asymmetry intensity index, quantifying the severity of the asymmetry phenomenon. When generating a comprehensive evaluation report, the symmetry quantification parameter, along with the synergy index and correlation score, participates in the decision-making logic. For example, when the proportion of asymmetrical units exceeds a threshold and the synergy index is below the standard, a specific evaluation conclusion is triggered. By introducing the symmetry quantification parameter, the evaluation system can identify coordination defects caused by bilateral muscle movement asymmetry, compensating for the shortcomings of relying solely on the synergy index.

[0113] As a preferred embodiment, the solution of this application is implemented as follows: Three-dimensional facial scan data is imported into a data processing system, and the face is divided into left and right symmetrical regions using a preset symmetry plane. Around anatomical landmarks such as the nasal alar base, corner of the mouth, and outer canthus, bilaterally symmetrical facial dynamic unit pairs are defined based on muscle attachment points and motion trajectories. For example, the left and right levator alar muscles form a symmetrical unit pair. After preprocessing the motion vector sequence, a timestamp-aligned motion amplitude sequence is extracted for each symmetrical unit pair. Within a standard expression execution cycle, the amplitude difference ratio between the left and right units is calculated at 10-millisecond intervals. When the instantaneous amplitude of the left levator alar muscle is 3.2 mm while that of the right is 2.8 mm, the difference ratio is (3.2-2.8) / 3.2 = 0.125. The temporal phase difference is obtained by detecting the difference in the peak times of the motion vectors of the left and right units. If the left unit reaches its peak at 1.25 seconds and the right unit reaches its peak at 1.28 seconds, the phase difference is 30 milliseconds. The system iterates through all symmetric unit pairs, counts the number of anomalous unit pairs with a difference ratio exceeding 0.15 or a phase difference exceeding 50 milliseconds, and calculates their proportion among all symmetric unit pairs. The difference ratio and phase difference of the anomalous units are normalized and then weighted and summed to generate an asymmetry intensity index. These parameters are integrated into the comprehensive evaluation algorithm, participating in the decision-making logic along with the synergy index and correlation score.

[0114] Through the above technical solution, this application effectively solves the problem that existing technologies cannot detect latent coordination disorders caused by bilateral asymmetry of movement. By quantifying the differences in motion amplitude and temporal phase deviation of symmetrical unit pairs, subtle bilateral motor incoordination caused by surgical procedures or nerve damage can be identified. Combining multi-dimensional analysis of asymmetry proportion and intensity indicators significantly improves the accuracy of facial dynamic coordination assessment, expanding postoperative effect evaluation from a single static indicator to the dynamic symmetry dimension, thereby more accurately predicting the patient's potential subjective discomfort risk.

[0115] Example 2:

[0116] like Figure 3 As shown, the automated evaluation system for postoperative results of plastic surgery includes:

[0117] The data acquisition module is used to acquire postoperative facial video data and three-dimensional facial scan data of the patient. The video data includes dynamic images of the patient performing a standard facial expression sequence.

[0118] The unit definition module is used to define several facial dynamic units based on 3D facial scan data;

[0119] The motion analysis module is used to obtain the motion vector sequence of facial dynamic units based on facial video data through motion analysis;

[0120] The synergy analysis module is used to calculate the synergy index between motion vector sequences of different facial dynamic units based on the motion vector sequence. The synergy index is used to characterize facial coordination.

[0121] The micro-expression analysis module is used to perform micro-expression analysis on facial video data, identify micro-expression events and map them to corresponding emotion categories and intensity.

[0122] The micro-expression time backtracking module is used to respond to micro-expression events of preset emotion categories, backtrack the motion vector sequence within a preset time window based on the occurrence time of the micro-expression event, and calculate the abnormal deviation of motion of each facial dynamic unit within the time window;

[0123] The correlation score calculation module is used to calculate the correlation score by weighting the maximum value of the emotion intensity, the deviation of the movement abnormality, and the temporal correlation between the deviation of the movement abnormality and the micro-expression event.

[0124] The comprehensive evaluation module is used to make joint judgments based on correlation scores and synergy indicators, and output a comprehensive evaluation report.

[0125] The above description is merely an example and illustration of the structure of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the structure of the invention or exceed the scope defined in the claims, all of which should fall within the protection scope of the present invention.

[0126] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0127] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. Method for automated postoperative outcome assessment in orthopedic surgery, characterized in that, The method comprises the following steps: obtaining postoperative facial video data of a patient and three-dimensional facial scan data, the facial video data comprising dynamic images of the patient performing a standard facial expression sequence; defining a plurality of facial dynamics units according to the three-dimensional facial scan data; obtaining a sequence of motion vectors of the facial dynamics units through motion analysis according to the facial video data; calculating a coordination index between the sequences of motion vectors of different facial dynamics units, the coordination index being used to represent facial coordination; performing micro-expression analysis on the facial video data to identify micro-expression events and map them to corresponding emotion categories and emotion intensities; in response to identifying a micro-expression event of a preset emotion category, backtracking the sequence of motion vectors within a preset time window of the micro-expression event and calculating motion abnormal deviation degrees of the facial dynamics units within the time window; performing weighted calculation on the emotion intensity, the maximum value of the motion abnormal deviation degrees and the time correlation between the motion abnormal deviation degrees and the micro-expression event to obtain a correlation score; jointly judging the correlation score and the coordination index and outputting a comprehensive evaluation report; after calculating the motion abnormal deviation degrees, further comprising reinforcing processing by establishing a spatial coupling coefficient matrix: predefining a spatial coupling coefficient matrix, the spatial coupling coefficient matrix being used to describe the physiological correlation strength between the facial dynamics units and the micro-expression action units; extracting a corresponding correlation strength row vector from the spatial coupling coefficient matrix according to the action unit activated by the current micro-expression event; calculating the spatial similarity between the vector composed of the motion abnormal deviation degrees of all facial dynamics units and the correlation strength row vector; introducing the spatial similarity as a new weighted term into the calculation formula of the correlation score to obtain a reinforced correlation score.

2. The orthopaedic post-operative outcome automated assessment method of claim 1, wherein: The calculation process of the coordination index comprises: calculating the correlation coefficient between two sequences of motion vectors representing different facial dynamics units to obtain a set of coordination sub-indices; the coordination index is a coordination index set composed of all coordination sub-indices.

3. The orthopaedic post-operative outcome automated assessment method of claim 1, wherein: The process of calculating the motion abnormal deviation degrees of the facial dynamics units within the calculation time window comprises: for each facial dynamics unit, extracting the sequence of motion vector magnitudes thereof within the preset time window; wherein the motion vector magnitude is the module length of the motion vector; calculating the deviation degree of the sequence of motion vector magnitudes from a reference magnitude sequence, taking the deviation degree as the motion abnormal deviation degree of the facial dynamics unit within the time window; the reference magnitude sequence is the mean sequence of the motion vector magnitudes of the facial dynamics unit within the entire standard facial expression sequence time period; the calculation formula is: ； wherein, is the motion abnormal deviation degree, is the mean of the motion vector magnitude sequence within the preset time window, is the global mean of the motion vector magnitude of the unit in the whole standard expression sequence, is the global standard deviation of the motion vector magnitude of the unit in the whole standard expression sequence.

4. The post-orthopaedic surgical outcome automated assessment method of claim 1, wherein: In the calculation process of the correlation score, the time correlation between the motion abnormal deviation degrees and the micro-expression event specifically comprises: in response to identifying a plurality of micro-expression events within the preset time window, calculating the absolute difference between the occurrence time of each micro-expression event and the peak time of the motion abnormal deviation degrees, and performing weighted average on the absolute difference with the emotion intensity as the weight to obtain a weighted average time difference.

5. The post-orthopaedic surgical outcome automated assessment method according to claim 2, wherein: The process of jointly judging according to the correlation score and the cooperativity indicators and outputting a comprehensive evaluation report comprises: if the correlation score is greater than a first score threshold and there is at least one cooperativity sub-indicator in the cooperativity indicator set whose value is less than a preset cooperativity threshold, a first type of evaluation report is generated and outputted; if the correlation score is less than a second score threshold and all cooperativity sub-indicators in the cooperativity indicator set have values greater than the cooperativity threshold, a second type of evaluation report is generated and outputted; if none of the above conditions is met, a third type of evaluation report is generated and outputted.

6. The post-orthopaedic surgical outcome automated assessment method according to claim 2, wherein: After calculating the cooperativity indicators, a timing stability checking mechanism is introduced: for each cooperativity sub-indicator, the motion vector sequence of the cooperativity sub-indicator in the whole cycle of the standard expression sequence is extracted; the sequence is divided into multiple equal-length sub-periods, and the local cooperativity value of the cooperativity sub-indicator in each sub-period is calculated; the standard deviation of the local cooperativity values of all sub-periods is calculated as the timing stability score of the cooperativity sub-indicator; if the timing stability score of a cooperativity sub-indicator is higher than a preset stability threshold, the cooperativity sub-indicator is marked as an unstable unit in the cooperativity indicator set; when jointly judging according to the cooperativity indicator set, each cooperativity sub-indicator in the cooperativity indicator set is given a weight coefficient, and the cooperativity indicator set after weighting is used for joint judgment; wherein the weight coefficient of the cooperativity sub-indicator marked as an unstable unit is lower than that of the unmarked cooperativity sub-indicator.

7. The post-orthopaedic surgical outcome automated assessment method of claim 1, wherein: After obtaining the correlation score, a time consistency checking mechanism is introduced: an extended time window is defined, and the length of the extended time window is greater than that of a preset time window; the time-shift cross-correlation function of the motion abnormality deviation and the emotional intensity in the extended time window is calculated; it is judged whether the peak value of the time-shift cross-correlation function at zero time shift exceeds a preset correlation threshold; if yes, it is confirmed that the correlation between the micro-expression event and the motion abnormality on which the calculation is based is reliable, and the correlation score obtained by calculation is outputted; otherwise, it is determined that the correlation between the micro-expression event and the motion abnormality on which the calculation is based is unreliable, and the patient is prompted to obtain the next facial video data.

8. The post-orthopaedic surgical outcome automated assessment method according to claim 3, wherein: Before outputting the comprehensive evaluation report, the following steps are further included: extracting bilateral symmetrical facial dynamics unit pairs, defining left-right symmetrical unit groups based on three-dimensional facial scan data; calculating the motion amplitude difference ratio and the timing phase difference of each symmetrical unit pair in the standard expression sequence according to the motion vector sequence; the calculation formula of the motion amplitude difference ratio is: ； wherein is a magnitude difference ratio, and are the motion vector magnitudes of the left and right units, respectively, at the current time point. the timing phase difference is obtained by calculating the absolute difference of the peak time points of the motion vectors of the left and right units in the same symmetrical unit pair; counting the number of unit pairs whose amplitude difference ratio exceeds a preset asymmetry threshold or whose timing phase difference exceeds a preset phase tolerance in all symmetrical unit pairs, calculating the proportion of the unit pairs in the total symmetrical unit pairs, and recording the proportion as the asymmetry unit proportion; accumulating the amplitude difference ratio and the timing phase difference of all out-of-limit unit pairs, normalizing and summing them to obtain the asymmetry intensity indicator; The asymmetry unit proportion and asymmetry intensity index are taken as symmetry quantization parameters, and are combined with the cooperativity index and the correlation score to generate a comprehensive evaluation report.

9. An automated post-orthopedic surgical outcome assessment system, characterized by: The use of the automatic postoperative effect evaluation method for orthopedic surgery as claimed in any one of claims 1-8, comprising: a data acquisition module for acquiring postoperative facial video data and three-dimensional facial scan data of a patient, the video data including dynamic images of the patient performing a standard expression sequence; a unit definition module for defining a plurality of facial dynamics units according to the three-dimensional facial scan data; a motion analysis module for obtaining a motion vector sequence of the facial dynamics units through motion analysis according to the facial video data; a cooperativity analysis module for calculating a cooperativity index between the motion vector sequences of different facial dynamics units according to the motion vector sequence, the cooperativity index being used to represent facial coordination; a micro-expression analysis module for performing micro-expression analysis on the facial video data, identifying micro-expression events, and mapping the micro-expression events to corresponding emotion categories and emotion intensities; a micro-expression time backtracking module for, in response to identifying a micro-expression event of a preset emotion category, backtracking the motion vector sequence within a preset time window of the micro-expression event according to the occurrence time of the micro-expression event, and calculating a motion abnormal deviation degree of each facial dynamics unit within the time window; a correlation score calculation module for calculating a correlation score by weighting the emotion intensity, the maximum of the motion abnormal deviation degree, and the time correlation between the motion abnormal deviation degree and the micro-expression event; a comprehensive evaluation module for combining the correlation score and the cooperativity index to output a comprehensive evaluation report.

Images