Acupuncture point recognition system based on artificial intelligence

Abstract

The invention discloses an acupuncture point recognition system based on artificial intelligence, which belongs to the technical field of intelligent medical treatment and comprises a data construction module, an intelligent acupoint matching recommendation module, a three-dimensional acupoint positioning module and a recognition report generation module. An intelligent acupoint matching recommendation method combining meridian acupoint association reasoning and multi-factor comprehensive evaluation is adopted, systematic acupoint matching reasoning is carried out according to a meridian conduction relation on the basis of integrating patient chief complaint symptoms, accompanying symptoms and individual signs, and a target acupoint combination conforming to individual differences is generated; a three-dimensional acupoint positioning method combining acupoint matching prior constraint and thermal anomaly analysis is adopted, on the premise that a target acupoint is determined, effective convergence is conducted on the individual body surface range, fine correction is conducted on the acupoint position in combination with the body surface temperature change characteristics, and therefore a three-dimensional acupoint positioning result better conforming to the individual characteristics and the meridian trend of a patient is obtained.

Description

Technical Field

[0001] This invention belongs to the field of smart medical technology, specifically referring to an acupuncture point recognition system based on artificial intelligence. Background Technology

[0002] The AI-based acupuncture point recognition system employs intelligent analysis technology that combines traditional Chinese medicine meridian and acupoint theory with multi-source sensory information from the human body surface. It extracts features, models spatially, and comprehensively analyzes patient symptom information and multimodal data from the body surface to generate target acupoints and their spatial recognition results on the individual's body surface. This enables assisted identification and three-dimensional positioning of acupuncture points, aiming to reduce reliance on personal experience in acupuncture point selection and improve accuracy and consistency. Consequently, it provides reliable technical support for clinical acupuncture applications, teaching and training, and intelligent assisted diagnosis and treatment.

[0003] However, existing acupuncture point identification methods have several technical problems. Firstly, relying solely on a single symptom or fixed point templates for point selection makes it difficult to comprehensively reflect the coexistence of multiple symptoms in patients. Secondly, they fail to adequately consider the overall connection of meridians and individual differences in the body surface, resulting in insufficient targeting and poor stability of recommended points. Thirdly, point location relies excessively on standard proportions or anatomical landmarks, making it difficult to adapt to differences in patient body type. Furthermore, the lack of comprehensive consideration of the current point selection scheme and the functional response of the body surface during actual point selection can easily lead to excessively large location ranges, scattered location results, or even deviations from the meridian pathways. Summary of the Invention

[0004] To address the above issues and overcome the shortcomings of existing technologies, this invention provides an artificial intelligence-based acupuncture point recognition system. It creatively employs an intelligent acupuncture point recommendation method that combines meridian-acupuncture point correlation reasoning with multi-factor comprehensive evaluation. Based on a comprehensive analysis of the patient's chief complaint, accompanying symptoms, and individual physical signs, it systematically reasons about acupuncture points according to meridian conduction relationships, generating target acupuncture point combinations that conform to individual differences. Furthermore, it creatively employs a three-dimensional acupuncture point localization method that combines prior constraints on acupuncture point selection with thermal anomaly analysis. Under the premise of clearly identifying the target acupuncture points, it achieves effective convergence to the individual's body surface area and performs fine-tuning of acupuncture point locations based on body surface temperature change characteristics, thereby obtaining three-dimensional acupuncture point localization results that better match the patient's individual characteristics and meridian pathways.

[0005] The technical solution adopted by the present invention is as follows: The acupuncture point recognition system based on artificial intelligence provided by the present invention includes a data construction module, an intelligent acupuncture point recommendation module, a three-dimensional acupuncture point positioning module, and a recognition report generation module;

[0006] The data construction module is used to construct data, obtain a multi-source basic dataset, and send the multi-source basic dataset to the intelligent acupoint recommendation module and the three-dimensional acupoint positioning module.

[0007] The intelligent acupoint recommendation module is used for intelligent acupoint recommendation. Based on a multi-source basic dataset, it adopts an intelligent acupoint recommendation method that combines meridian acupoint association reasoning and multi-factor comprehensive evaluation to obtain an acupoint recommendation scheme, and sends the acupoint recommendation scheme to the three-dimensional acupoint positioning module and the recognition report generation module.

[0008] The three-dimensional acupoint positioning module is used for three-dimensional acupoint positioning. Based on a multi-source basic dataset and acupoint recommendation scheme, it adopts a three-dimensional acupoint positioning method that combines acupoint prior constraints and thermal anomaly analysis to obtain acupoint positioning information, and sends the acupoint positioning information to the recognition report generation module.

[0009] The identification report generation module is used to generate an acupuncture point identification report.

[0010] Furthermore, the data construction is used to provide a data foundation for intelligent acupoint recommendation and three-dimensional acupoint positioning. Specifically, it involves collecting and organizing patient symptom data, body surface perception data, and traditional Chinese medicine knowledge data to construct a multi-source basic dataset.

[0011] The patient symptom data includes the patient's chief complaint symptoms, accompanying symptoms, and physical signs and parameters.

[0012] The body surface sensing data includes the patient's body surface infrared thermal imaging data and three-dimensional body sensing data.

[0013] Furthermore, the intelligent acupoint recommendation includes the following steps: symptom and sign structured modeling, acupoint association reasoning, acupoint optimization, and acupoint scheme generation;

[0014] The structured modeling of symptoms and signs specifically involves semantic parsing of the patient's chief complaint symptoms and accompanying symptoms, mapping them to a preset TCM symptom semantic space, constructing a symptom feature vector, and constructing an individual sign vector by normalizing the patient's sign parameters. Then, the symptom feature vector and the individual sign vector are concatenated to obtain a patient state representation vector.

[0015] The acupoint association reasoning specifically involves constructing a knowledge graph of meridians and acupoints, activating corresponding symptom nodes based on the patient state representation vector, performing hierarchical propagation reasoning with the symptom meridian acupoint path as the main reasoning path, calculating the symptom-related activation intensity of each acupoint node, and then selecting acupoint nodes with symptom-related activation intensity greater than the activation intensity threshold to generate a candidate acupoint set.

[0016] The construction of the meridian and acupoint knowledge graph is specifically based on traditional Chinese medicine theory data. It involves establishing a knowledge graph structure that includes symptom nodes, meridian nodes, and acupoint nodes, and establishing symptom-meridian association edges, meridian-acupoint membership edges, and symptom-acupoint association edges between nodes to form a meridian and acupoint knowledge graph.

[0017] The activation of corresponding symptom nodes based on the patient state representation vector specifically involves using each symptom semantic dimension component in the patient state representation vector as the initial activation intensity of the corresponding symptom node.

[0018] The acupoint optimization is used to determine the comprehensive priority of each acupoint in the candidate acupoint set. Specifically, it involves combining symptom-related activation intensity and introducing clinically common use evaluation factors and acupoint synergy evaluation factors to conduct a multi-dimensional comprehensive evaluation of the candidate acupoints and calculate the comprehensive priority score.

[0019] The acupoint matching scheme is generated by sorting candidate acupoints from high to low based on their comprehensive priority scores, and then generating a recommended acupoint matching scheme.

[0020] Furthermore, the three-dimensional acupoint localization includes the following steps: three-dimensional modeling of the body surface, extraction of candidate regions for thermal anomalies, individual proportion localization modeling, refined three-dimensional localization, and output of localization results;

[0021] The aforementioned three-dimensional body surface modeling is used to construct an individualized three-dimensional body surface model. Specifically, it involves synchronizing and spatially registering the patient's infrared thermal imaging data and three-dimensional somatosensory data to establish a unified three-dimensional coordinate system for the body surface and generating an individualized three-dimensional body surface model.

[0022] The extraction of thermal anomaly candidate regions specifically involves taking each acupoint in the recommended acupoint combination scheme as the target acupoint, extracting the body surface region related to the corresponding meridian direction from the patient's infrared thermal imaging data based on the meridian affiliation of the target acupoint, performing body surface temperature field analysis, calculating the body surface temperature gradient and local temperature difference offset, constructing a thermal anomaly index, and then extracting thermal anomaly regions according to preset thermal anomaly judgment conditions. By mapping the thermal anomaly regions onto an individualized three-dimensional body surface model, a three-dimensional candidate region set of the target acupoint is formed.

[0023] The individual proportion positioning modeling specifically involves identifying a set of key bony landmarks in the human body from three-dimensional somatosensory data, constructing an individualized proportion scale model based on the spatial distance relationship between the key bony landmarks, and then mapping the relative proportion position parameters of the target acupoint in the standard human body model to the individualized proportion scale model according to the standard proportion inch definition, thereby obtaining the individualized proportion prediction position of the target acupoint.

[0024] The refined 3D localization specifically involves calculating thermal feature consistency scores, proportional consistency scores, and meridian direction consistency scores for candidate points within a set of 3D candidate regions, and then performing weighted fusion to obtain a multimodal consistency score. Finally, the candidate point with the largest multimodal consistency score is selected as the final 3D localization result for the corresponding target acupoint.

[0025] The proportional consistency score uses the individualized proportional prediction location of the target acupoint as a spatial prior constraint.

[0026] The location result output specifically involves performing the three-dimensional modeling of the body surface, the extraction of thermal anomaly candidate regions, the individual proportion location modeling, and the refined three-dimensional location to obtain the final three-dimensional location result of the target acupoint on the individual's body surface. Based on the multimodal consistency score corresponding to the final three-dimensional location result and the spatial distribution characteristics of the three-dimensional candidate regions, a reliability assessment and standardization processing are performed to obtain acupoint location information.

[0027] Furthermore, the identification report generation specifically involves summarizing and integrating the acupoint recommendation scheme and acupoint location information to generate an acupoint identification report, which is used to provide a reference for acupoint confirmation, location verification, and visual auxiliary operation during the acupuncture diagnosis and treatment process.

[0028] The beneficial effects achieved by the present invention using the above solution are as follows:

[0029] (1) In response to the technical problems in the existing acupuncture point identification process, which is that the selection of acupoints is based on a single symptom or a fixed acupoint template, which is difficult to fully reflect the coexistence of multiple symptoms in patients and fails to fully consider the overall connection of meridians and individual differences in the body surface, resulting in insufficient targeting and poor stability of recommended acupoints, this solution creatively adopts an intelligent acupoint recommendation method that combines meridian acupoint association reasoning and multi-factor comprehensive evaluation. Based on the comprehensive patient's chief complaint symptoms, accompanying symptoms and individual signs, it performs systematic acupoint reasoning based on the meridian conduction relationship and generates a target acupoint combination that conforms to individual differences.

[0030] (2) In response to the technical problems in the existing acupuncture point identification process, such as the excessive reliance on standard proportions or anatomical landmarks for point location, the difficulty in adapting to the differences in body shape of different patients, and the lack of comprehensive consideration of the current point selection scheme and the body surface functional response in the actual point selection process, which easily leads to an excessively large location range, scattered location results, or even deviation from the meridian circulation law, this scheme creatively adopts a three-dimensional point location method that combines a priori constraints of point selection with thermal anomaly analysis. Under the premise of clearly identifying the target point, it effectively converges the individual body surface range and combines the body surface temperature change characteristics to finely correct the point location, thereby obtaining a three-dimensional point location result that is more in line with the individual characteristics of the patient and the meridian direction. Attached Figure Description

[0031] Figure 1 A schematic diagram of the modules of the artificial intelligence-based acupuncture point recognition system provided by the present invention;

[0032] Figure 2 A flowchart illustrating the intelligent acupoint recommendation module;

[0033] Figure 3 This is a flowchart illustrating the three-dimensional acupoint positioning module.

[0034] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof. Detailed Implementation

[0035] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0036] In the description of this invention, it should be understood that the terms "upper", "lower", "front", "rear", "left", "right", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0037] Example 1, see Figure 1 The acupuncture point recognition system based on artificial intelligence provided by the present invention includes a data construction module, an intelligent acupuncture point recommendation module, a three-dimensional acupuncture point positioning module, and a recognition report generation module.

[0038] The data construction module is used to construct data, obtain a multi-source basic dataset, and send the multi-source basic dataset to the intelligent acupoint recommendation module and the three-dimensional acupoint positioning module.

[0039] The intelligent acupoint recommendation module is used for intelligent acupoint recommendation. Based on a multi-source basic dataset, it adopts an intelligent acupoint recommendation method that combines meridian acupoint association reasoning and multi-factor comprehensive evaluation to obtain an acupoint recommendation scheme, and sends the acupoint recommendation scheme to the three-dimensional acupoint positioning module and the recognition report generation module.

[0040] The three-dimensional acupoint positioning module is used for three-dimensional acupoint positioning. Based on a multi-source basic dataset and acupoint recommendation scheme, it adopts a three-dimensional acupoint positioning method that combines acupoint prior constraints and thermal anomaly analysis to obtain acupoint positioning information, and sends the acupoint positioning information to the recognition report generation module.

[0041] The identification report generation module is used to generate an acupuncture point identification report.

[0042] Example 2, see Figure 1 This embodiment is based on the above embodiment. The data construction is used to provide a data foundation for intelligent acupoint recommendation and three-dimensional acupoint positioning. Specifically, it involves collecting and organizing patient symptom data, body surface perception data, and traditional Chinese medicine knowledge data to construct a multi-source basic dataset.

[0043] The patient symptom data includes patient complaints, accompanying symptoms, and vital signs collected through manual entry, consultation systems, or smart terminals; after collection, the patient symptom data is standardized, encoded, and stored in a formatted manner.

[0044] The body surface sensing data includes infrared thermal imaging data and three-dimensional somatosensory data of the patient's body surface acquired through infrared thermal imaging equipment and three-dimensional somatosensory acquisition equipment; after acquisition, the body surface sensing data is subjected to noise filtering and coordinate calibration processing.

[0045] The TCM knowledge data includes data on the association between symptoms, meridians, and acupoints; data on the attribution of acupoints to meridians; data on the functions and indications of acupoints; and data on the definition of standard inch proportions.

[0046] Example 3, see Figure 1 and Figure 2 This embodiment is based on the above embodiment. The intelligent acupoint recommendation is used to generate target acupoints under the individualized symptom conditions of patients, including the following steps: symptom and sign structured modeling, meridian and acupoint association reasoning, acupoint optimization, and acupoint scheme generation.

[0047] The structured modeling of symptoms and signs is used to transform the patient's chief complaint symptoms, accompanying symptoms, and sign parameters into a unified patient state representation. Specifically, it involves semantic parsing of the patient's chief complaint symptoms and accompanying symptoms and mapping them to a preset TCM symptom semantic space to construct a symptom feature vector. Then, it involves normalizing the patient's sign parameters to construct an individual sign vector. Finally, it involves concatenating the symptom feature vector and the individual sign vector to obtain the patient state representation vector.

[0048] The preset TCM symptom semantic space can be specifically set based on the standardized classification and hierarchical relationship of symptoms in the TCM theoretical system, and is used to map each symptom to a corresponding semantic dimension.

[0049] The acupoint association reasoning is used to perform symptom-related reasoning on acupoints based on patient state representation. Specifically, it involves constructing a knowledge graph of meridians and acupoints, activating corresponding symptom nodes according to the patient state representation vector, using the symptom meridian acupoint path as the main reasoning path for hierarchical propagation reasoning, calculating the symptom-related activation intensity of each acupoint node, and then selecting acupoint nodes with symptom-related activation intensity greater than the activation intensity threshold to generate a candidate acupoint set.

[0050] The construction of the meridian and acupoint knowledge graph is specifically based on traditional Chinese medicine theory data. It involves establishing a knowledge graph structure that includes symptom nodes, meridian nodes, and acupoint nodes, and establishing symptom-meridian association edges, meridian-acupoint membership edges, and symptom-acupoint association edges between nodes to form a meridian and acupoint knowledge graph.

[0051] The activation of corresponding symptom nodes based on the patient state representation vector specifically involves using each symptom semantic dimension component in the patient state representation vector as the initial activation intensity of the corresponding symptom node. This activates symptom nodes in the meridian acupoint knowledge graph that are related to the patient's chief complaint and accompanying symptoms, providing initial activation conditions for subsequent propagation reasoning.

[0052] The formula for calculating the symptom-related activation intensity of each acupoint node is as follows:

[0053] ;

[0054] ;

[0055] In the formula, Am j Let be the symptom-related activation intensity of the j-th meridian node, where j is the meridian node index, i is the symptom node index, and n(j) is the set of neighboring nodes of the j-th meridian node. i w is the initial activation intensity of the i-th symptom node. i,j Aa is the association weight between the i-th symptom node and the j-th meridian node. k is the symptom-related activation intensity of the k-th acupoint node, where k is the acupoint node index, n(k) is the set of neighbor nodes of the k-th acupoint node, and w j,k It is the membership weight of the j-th meridian node and the k-th acupoint node;

[0056] The activation intensity threshold can be specifically set by analyzing the distribution of acupoint activation intensity based on historical clinical acupoint combination data, and selecting a fixed threshold, percentile threshold, or adaptive threshold as the activation intensity threshold.

[0057] The acupoint optimization is used to determine the comprehensive priority of each acupoint in the candidate acupoint set. Specifically, it involves combining symptom-related activation intensity and introducing clinically common use evaluation factors and acupoint synergy evaluation factors to conduct a multi-dimensional comprehensive evaluation of the candidate acupoints and calculate the comprehensive priority score.

[0058] The calculation formula for the clinically commonly used evaluation factor is as follows:

[0059] ;

[0060] In the formula, UF cN is the clinical frequency evaluation factor for the c-th candidate acupoint, where c is the first index of the candidate acupoint, and N is the frequency evaluation factor for the c-th candidate acupoint. c is the number of times the c-th candidate acupoint is used in the clinical experience database, d is the second index of the candidate acupoint, Acp is the set of candidate acupoints, and N is the number of times the c-th candidate acupoint is used. d This represents the number of times the d-th candidate acupoint is used in the clinical experience database.

[0061] The calculation formula for the acupoint synergy evaluation factor is as follows:

[0062] ;

[0063] In the formula, CF c is the evaluation factor for acupoint synergy, |Acp| is the number of candidate acupoints, sim(·) is the similarity measurement function, and the similarity measurement function can be calculated using methods including but not limited to cosine similarity or Euclidean distance similarity, fea c It is the feature vector of the c-th candidate acupoint, fea d It is the feature vector of the d-th candidate acupoint. The feature vector of the candidate acupoint can be obtained by feature encoding the information of the meridian to which the acupoint belongs, its functional and indication attributes, and its structural relationship in the meridian and acupoint knowledge graph.

[0064] The formula for calculating the comprehensive priority score is as follows:

[0065] ;

[0066] In the formula, Score c It is the comprehensive priority score of the c-th candidate acupoint. It is the first weight in the comprehensive evaluation, Aa c It represents the symptom-related activation intensity of the c-th candidate acupoint. It is the second weight in the comprehensive evaluation. It is the third weight in the comprehensive evaluation;

[0067] The acupoint combination scheme is generated by sorting candidate acupoints from high to low according to their comprehensive priority scores, and generating a recommended acupoint combination scheme, including main acupoints, auxiliary acupoints, alternative acupoints, and information on the meridian affiliation, functions and indications, and synergistic acupoint combinations for each acupoint.

[0068] By performing the above operations, this solution addresses the technical problems in existing acupuncture point identification processes, where point selection is based solely on a single symptom or a fixed point template, making it difficult to comprehensively reflect the coexistence of multiple symptoms in patients and failing to fully consider the overall connection of meridians and individual differences in the body surface, resulting in insufficient targeting and poor stability of recommended point points. This solution creatively adopts an intelligent point selection recommendation method that combines meridian and point correlation reasoning with multi-factor comprehensive evaluation. Based on a comprehensive analysis of the patient's chief complaint symptoms, accompanying symptoms, and individual signs, it performs systematic point selection reasoning based on meridian conduction relationships and generates target point combinations that conform to individual differences.

[0069] Example 4, see Figure 1 and Figure 3 This embodiment is based on the above embodiment. The three-dimensional acupoint positioning is used to achieve high-precision three-dimensional spatial positioning of acupoints on an individual's body surface under the premise that the target acupoints are known. It includes the following steps: three-dimensional modeling of the body surface, extraction of thermal anomaly candidate regions, individual proportional positioning modeling, refined three-dimensional positioning, and output of positioning results.

[0070] The aforementioned three-dimensional body surface modeling is used to construct an individualized three-dimensional body surface model. Specifically, it involves synchronizing and spatially registering the patient's infrared thermal imaging data and three-dimensional somatosensory data to establish a unified three-dimensional coordinate system for the body surface and generating an individualized three-dimensional body surface model.

[0071] The extraction of thermal anomaly candidate regions is used to screen three-dimensional candidate regions related to target acupoints within the body surface area. Specifically, each acupoint in the recommended acupoint combination scheme is taken as the target acupoint. Based on the meridian affiliation of the target acupoint, the body surface region related to the corresponding meridian direction is extracted from the patient's infrared thermal imaging data. The body surface temperature field is analyzed, the body surface temperature gradient and local temperature difference offset are calculated, and a thermal anomaly index is constructed. Then, thermal anomaly regions are extracted according to preset thermal anomaly judgment conditions. By mapping the thermal anomaly regions to an individualized three-dimensional body surface model, a set of three-dimensional candidate regions for the target acupoint is formed.

[0072] The formula for calculating the thermal anomaly index is as follows:

[0073] ;

[0074] In the formula, H(x) is the thermal anomaly index at candidate point x, and x is the position of the first candidate point. It represents the temperature gradient of the body surface temperature field at candidate point x, and ||·|| is the L2 norm operator. It is the local temperature difference offset of candidate point x;

[0075] The calculation formula for the set of three-dimensional candidate regions forming the target acupoint is as follows:

[0076] ;

[0077] In the formula, R is the set of three-dimensional candidate regions for the target acupoint. It is the thermal anomaly threshold;

[0078] The individual proportion positioning modeling is used to obtain the proportional prediction position of the target acupoint on the individual's body surface. Specifically, it identifies a set of key bony landmarks in the human body in three-dimensional somatosensory data, constructs an individualized proportion scale model based on the spatial distance relationship between the key bony landmarks, and then maps the relative proportion position parameters of the target acupoint in the standard human body model to the individualized proportion scale model according to the standard proportion inch definition to obtain the individualized proportional prediction position of the target acupoint.

[0079] The refined three-dimensional positioning is used to determine the final three-dimensional spatial location of the target acupoint within the candidate region. Specifically, within the set of three-dimensional candidate regions, thermal feature consistency score, proportional consistency score, and meridian direction consistency score are calculated for each candidate point, and weighted fusion is performed to obtain a multimodal consistency score. Then, the candidate point with the largest multimodal consistency score is selected as the final three-dimensional positioning result of the corresponding target acupoint.

[0080] The thermal feature consistency score is used to evaluate the stability and significance of candidate points in the thermal anomaly candidate region, and the calculation formula is as follows:

[0081] ;

[0082] In the formula, C1(x) is the thermal feature consistency score, max is the maximum value function, y is the position of the second candidate point, and H(y) is the thermal anomaly index at candidate point y.

[0083] The proportional consistency score uses the individualized proportional prediction location of the target acupoint as a spatial prior constraint, and the calculation formula is as follows:

[0084] ;

[0085] In the formula, C2(x) is the proportionality consistency score, exp(·) is the natural exponential function, and x (p) It is a personalized proportional prediction of the target acupoint location;

[0086] The formula for calculating the consistency score of the meridian pathways is as follows:

[0087] ;

[0088] In the formula, C3(x) is the consistency score of meridian pathway, and d M (x) is the minimum spatial distance from candidate point x to the center line of the meridian to which the target acupoint belongs;

[0089] The formula for calculating the multimodal consistency score is as follows:

[0090] ;

[0091] In the formula, C final (x) is the multimodal consistency score. It is the thermal feature consistency weight. It is a proportional consistency weight. It is the consistency weight of the meridian pathways;

[0092] The positioning result output specifically involves performing the three-dimensional modeling of the body surface, the extraction of thermal anomaly candidate regions, the individual proportion positioning modeling, and the refined three-dimensional positioning to obtain the final three-dimensional positioning result of the target acupoint on the individual's body surface. Based on the multimodal consistency score and spatial distribution characteristics of the three-dimensional candidate regions corresponding to the final three-dimensional positioning result, a credibility assessment and standardization processing are performed to obtain acupoint positioning information, including the three-dimensional spatial coordinates of the target acupoint and the positioning error estimation range.

[0093] By performing the above operations, this solution addresses the technical problems in existing acupuncture point identification processes, such as over-reliance on standard proportions or anatomical landmarks for point location, difficulty in adapting to differences in patient body shape, and lack of comprehensive consideration of the current point selection scheme and surface functional response during actual point selection, which can easily lead to excessively large location ranges, scattered location results, or even deviations from the meridian pathways. This solution creatively adopts a three-dimensional acupuncture point location method that combines prior constraints on point selection with thermal anomaly analysis. Under the premise of clearly identifying the target acupuncture point, it effectively converges the individual's surface range and performs fine correction of the acupuncture point location by combining the characteristics of surface temperature changes, thereby obtaining three-dimensional acupuncture point location results that are more consistent with the individual characteristics of the patient and the meridian pathways.

[0094] Example 5, see Figure 1 This embodiment is based on the above embodiment. The generation of the identification report specifically involves summarizing and integrating the acupoint recommendation scheme and acupoint location information to generate an acupoint identification report, which is used to provide a reference for acupoint confirmation, location verification and visual auxiliary operation in the acupuncture diagnosis and treatment process.

[0095] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0096] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention.

[0097] The present invention and its embodiments have been described above. This description is not restrictive, and the accompanying drawings are only one embodiment of the present invention; the actual structure is not limited thereto. In conclusion, if those skilled in the art are inspired by this description and design similar structures and embodiments without departing from the spirit of the invention, such designs should fall within the protection scope of the present invention.

Claims

1. An acupuncture point recognition system based on artificial intelligence, characterized in that: It includes a data construction module, an intelligent acupoint recommendation module, a three-dimensional acupoint positioning module, and a recognition report generation module; The data construction module is used to construct data, obtain a multi-source basic dataset, and send the multi-source basic dataset to the intelligent acupoint recommendation module and the three-dimensional acupoint positioning module. The intelligent acupoint recommendation module is used for intelligent acupoint recommendation. Based on a multi-source basic dataset, it adopts an intelligent acupoint recommendation method that combines meridian acupoint association reasoning and multi-factor comprehensive evaluation to obtain an acupoint recommendation scheme, and sends the acupoint recommendation scheme to the three-dimensional acupoint positioning module and the recognition report generation module. The intelligent acupoint recommendation includes the following steps: symptom and sign structured modeling, meridian and acupoint association reasoning, acupoint optimization, and acupoint scheme generation; The three-dimensional acupoint positioning module is used for three-dimensional acupoint positioning. Based on a multi-source basic dataset and acupoint recommendation scheme, it adopts a three-dimensional acupoint positioning method that combines acupoint prior constraints and thermal anomaly analysis to obtain acupoint positioning information, and sends the acupoint positioning information to the recognition report generation module. The three-dimensional acupoint localization includes the following steps: three-dimensional modeling of the body surface, extraction of candidate regions for thermal anomalies, individual proportion localization modeling, refined three-dimensional localization, and output of localization results; The identification report generation module is used to generate an acupuncture point identification report.

2. The acupuncture point recognition system based on artificial intelligence according to claim 1, characterized in that: The structured modeling of symptoms and signs specifically involves semantic parsing of the patient's chief complaint symptoms and accompanying symptoms, mapping them to a preset TCM symptom semantic space, constructing a symptom feature vector, and constructing an individual sign vector by normalizing the patient's sign parameters. Then, the symptom feature vector and the individual sign vector are concatenated to obtain a patient state representation vector. The acupoint association reasoning specifically involves constructing a knowledge graph of meridians and acupoints, activating corresponding symptom nodes based on the patient state representation vector, performing hierarchical propagation reasoning with the symptom meridian acupoint path as the main reasoning path, calculating the symptom-related activation intensity of each acupoint node, and then selecting acupoint nodes with symptom-related activation intensity greater than the activation intensity threshold to generate a candidate acupoint set. The construction of the meridian and acupoint knowledge graph is specifically based on traditional Chinese medicine theory data. It involves establishing a knowledge graph structure that includes symptom nodes, meridian nodes, and acupoint nodes, and establishing symptom-meridian association edges, meridian-acupoint membership edges, and symptom-acupoint association edges between nodes to form a meridian and acupoint knowledge graph. The activation of corresponding symptom nodes based on the patient state representation vector specifically involves using each symptom semantic dimension component in the patient state representation vector as the initial activation intensity of the corresponding symptom node.

3. The acupuncture point recognition system based on artificial intelligence according to claim 2, characterized in that: The acupoint optimization is used to determine the comprehensive priority of each acupoint in the candidate acupoint set. Specifically, it involves combining symptom-related activation intensity and introducing clinically common use evaluation factors and acupoint synergy evaluation factors to conduct a multi-dimensional comprehensive evaluation of the candidate acupoints and calculate the comprehensive priority score. The acupoint matching scheme is generated by sorting candidate acupoints from high to low based on their comprehensive priority scores, and then generating a recommended acupoint matching scheme.

4. The acupuncture point recognition system based on artificial intelligence according to claim 3, characterized in that: The aforementioned three-dimensional body surface modeling is used to construct an individualized three-dimensional body surface model. Specifically, it involves synchronizing and spatially registering the patient's infrared thermal imaging data and three-dimensional somatosensory data to establish a unified three-dimensional coordinate system for the body surface and generating an individualized three-dimensional body surface model. The extraction of thermal anomaly candidate regions specifically involves taking each acupoint in the recommended acupoint combination scheme as the target acupoint, extracting the body surface region related to the corresponding meridian direction from the patient's infrared thermal imaging data based on the meridian affiliation of the target acupoint, performing body surface temperature field analysis, calculating the body surface temperature gradient and local temperature difference offset, constructing a thermal anomaly index, and then extracting thermal anomaly regions according to preset thermal anomaly judgment conditions. By mapping the thermal anomaly regions onto an individualized three-dimensional body surface model, a three-dimensional candidate region set for the target acupoint is formed.

5. The acupuncture point recognition system based on artificial intelligence according to claim 4, characterized in that: Individual proportion positioning modeling specifically involves identifying a set of key bony landmarks in the human body from three-dimensional somatosensory data, constructing an individualized proportion scale model based on the spatial distance relationship between these key bony landmarks, and then mapping the relative proportion position parameters of the target acupoint in the standard human body model to the individualized proportion scale model according to the standard proportion definition, thereby obtaining the individualized proportion prediction position of the target acupoint.

6. The acupuncture point recognition system based on artificial intelligence according to claim 5, characterized in that: The refined 3D localization specifically involves calculating thermal feature consistency scores, proportional consistency scores, and meridian direction consistency scores for candidate points within a 3D candidate region set, and then performing weighted fusion to obtain a multimodal consistency score. The candidate point with the largest multimodal consistency score is then selected as the final 3D localization result for the corresponding target acupoint. The proportional consistency score uses the individualized proportional prediction position of the target acupoint as a spatial prior constraint. The location result output specifically involves performing the three-dimensional modeling of the body surface, the extraction of thermal anomaly candidate regions, the individual proportion location modeling, and the refined three-dimensional location to obtain the final three-dimensional location result of the target acupoint on the individual's body surface. Based on the multimodal consistency score corresponding to the final three-dimensional location result and the spatial distribution characteristics of the three-dimensional candidate regions, a reliability assessment and standardization processing are performed to obtain acupoint location information.

7. The acupuncture point recognition system based on artificial intelligence according to claim 6, characterized in that: The data construction is used to provide a data foundation for intelligent acupoint recommendation and three-dimensional acupoint positioning. Specifically, it involves collecting and organizing patient symptom data, body surface perception data, and traditional Chinese medicine knowledge data to construct a multi-source basic dataset. The patient symptom data includes the patient's chief complaint symptoms, accompanying symptoms, and physical signs and parameters. The body surface sensing data includes the patient's body surface infrared thermal imaging data and three-dimensional body sensing data.

8. The acupuncture point recognition system based on artificial intelligence according to claim 7, characterized in that: The identification report is generated by summarizing and integrating the acupoint recommendation scheme and acupoint location information to generate an acupoint identification report, which is used to provide a reference for acupoint confirmation, location verification and visual auxiliary operation in the acupuncture diagnosis and treatment process.

Images