Pain dynamic assessment and treatment linkage system based on three-dimensional somatosensory interaction

Abstract

The invention relates to the technical field of medical instruments, and particularly discloses a pain dynamic assessment and treatment linkage system based on three-dimensional somatosensory interaction, and the system comprises a human body modeling module which constructs a three-dimensional human body anatomical model; the somatosensory interaction module receives a pain area marking instruction and a pain intensity score; the infrared sensing module monitors human body infrared thermal imaging data of a patient in real time; the physiological monitoring module monitors a heart rate variability index and a skin electric response index of the patient based on the wearable device; the region marking module obtains a calibrated pain region of the patient based on the pain region marking instruction and an abnormal temperature region in the human body infrared thermal imaging data; the pain prediction module calibrates a pain area based on a pain intensity score, a heart rate variability index and a skin electric response index of the patient, and predicts a pain intensity prediction curve in a future period; the treatment linkage module generates a treatment parameter control instruction based on the pain trend prediction curve and the multi-dimensional pain feature vector; and accurate and dynamic treatment intervention is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medical devices, in particular to a pain dynamic assessment and treatment linkage system based on three-dimensional somatosensory interaction. BACKGROUND

[0002] In the medical field, the effective management of chronic pain has always been a key research direction. Pain not only brings physical suffering to patients, but also has a negative impact on their psychology and quality of life. For various chronic pains such as post-herpetic neuralgia, cancer pain, etc., accurate pain assessment and timely and effective treatment are crucial. With the development of technology, traditional pain assessment and treatment methods gradually fail to meet the growing demand for precision medicine. The pain dynamic assessment and treatment linkage system based on three-dimensional somatosensory interaction has emerged as the times require, and has important practical significance. In the future, as people's demand for the quality of medical services continues to improve, and technology continues to advance, this system has broad application prospects and can play an important role in more chronic pain treatment scenarios, promoting the development of intelligent and precise pain treatment.

[0003] However, the existing pain assessment and treatment technology simply uses a VAS score scale, which causes communication barriers and understanding difficulties for special groups such as the elderly, resulting in a longer time-consuming and less accurate pain assessment. Moreover, the traditional method cannot intuitively mark the specific pain location for patients, cannot fully obtain the spatial information of pain, cannot accurately predict the pain trend, and cannot adjust the treatment parameters in real time according to the real-time pain characteristics, leading to treatment delay and affecting the treatment effect. In addition, the existing assessment and treatment processes are relatively independent, and lack an integrated system to realize the efficient linkage of pain dynamic assessment and treatment, which cannot meet the comprehensive needs of precision medicine for pain management.

[0004] Therefore, the present application proposes a pain dynamic assessment and treatment linkage system based on three-dimensional somatosensory interaction. SUMMARY

[0005] The present application provides a pain dynamic assessment and treatment linkage system based on three-dimensional somatosensory interaction, which can provide a more intuitive and convenient pain assessment method for patients with the help of advanced three-dimensional somatosensory interaction technology, changing the relatively single and subjective assessment mode in the past. At the same time, through linkage with the treatment equipment, the organic combination of pain assessment and treatment is realized, which provides strong support for doctors to develop personalized treatment plans, and is expected to improve the effect and efficiency of chronic pain treatment.

[0006] The present application provides a pain dynamic assessment and treatment linkage system based on three-dimensional somatosensory interaction, comprising: a human body modeling module configured to construct a three-dimensional human body model including a layered structure of bones, nerves and internal organs of a patient and a pain sensitive area label associated with each anatomical part; a somatosensory interaction module configured to receive a pain area label instruction and a pain intensity score input by a user; an infrared sensing module configured to monitor infrared thermal imaging data of the patient in real time; a physiological monitoring module configured to monitor a heart rate variability indicator and a galvanic skin response indicator of the patient based on a wearable device; a region labeling module configured to obtain a calibrated pain area of the patient based on the pain area label instruction and an abnormal temperature area in the infrared thermal imaging data; a pain prediction module configured to predict a pain intensity prediction curve in a future period based on the pain intensity score, the heart rate variability indicator, the galvanic skin response indicator and the calibrated pain area of the patient; a treatment linkage module configured to generate a treatment parameter control instruction based on the pain trend prediction curve and a multi-dimensional pain feature vector.

[0007] Optionally, the human body modeling module comprises: a CT image data acquisition sub-module configured to acquire standard human body CT image data of the patient, and generate a basic model including bones, main nerve trunks and internal organs of the patient through a three-dimensional reconstruction algorithm; a pain sensitive area labeling sub-module configured to label the pain sensitive area of the basic model to obtain a three-dimensional human body model of the patient.

[0008] Optionally, the somatosensory interaction module comprises: a touch display screen configured to display the three-dimensional human body model and receive a first pain area label instruction input by the patient; a depth camera configured to capture a gesture action of the patient and translate the gesture action of the patient into a second pain area label instruction; an instruction fusion sub-module configured to determine the pain area label instruction input by the user based on the first pain area label instruction and / or the second pain area label instruction; a score receiving sub-module configured to receive the pain intensity score input by the user through an integrated VAS / NRS score slider.

[0009] Optionally, the region labeling module comprises: a region labeling sub-module configured to determine the pain area labeled by the patient based on the pain area label instruction; a fusion calibration sub-module configured to fuse and calibrate the pain area labeled by the patient and the abnormal temperature area in the infrared thermal imaging data to obtain the calibrated pain area of the patient.

[0010] Optionally, the fusion calibration sub-module comprises: a coordinate matrix conversion unit, configured to convert a pain area contour marked by the patient into a three-dimensional coordinate matrix to obtain a manual annotation three-dimensional coordinate matrix, and convert an abnormal temperature zone contour in the human infrared thermal imaging data into a three-dimensional coordinate matrix to obtain a thermal imaging three-dimensional coordinate matrix; an intersection matrix determination unit, configured to calculate a spatial intersection matrix of the manual annotation three-dimensional coordinate matrix and the thermal imaging three-dimensional coordinate matrix; a core area determination unit, configured to extract a first singular vector of the spatial intersection matrix through singular value decomposition, and determine a core pain area based on the first singular vector; a calibration-required area demarcation unit, configured to perform a first-order difference operation on the manual annotation three-dimensional coordinate matrix to obtain a gradient change matrix of the pain area contour marked by the patient, filter all boundary points with a gradient change rate exceeding a preset gradient change rate threshold as calibration-required edge area contour points based on the gradient change matrix, and determine a calibration-required edge area based on all the calibration-required edge area contour points; a temperature inflection point identification unit, configured to map the calibration-required edge area to a temperature gradient field in the human infrared thermal imaging data, fit a temperature distribution curve through a two-dimensional Gaussian function, and calculate a second derivative of the temperature distribution curve to determine all temperature change inflection points; a calibration-required area calibration unit, configured to correct the calibration-required edge area based on all the temperature change inflection points to obtain a calibrated edge area; a two-dimensional area fusion unit, configured to fuse the core pain area and the calibrated edge area to obtain a three-dimensional grid model of a calibrated pain area as the calibrated pain area of the patient.

[0011] Optionally, the intersection matrix determination unit comprises: a distance threshold determination sub-unit, configured to set a maximum distance threshold and a minimum distance threshold based on a precision requirement corresponding to the pain intensity score input by the user; an Euclidean distance determination sub-unit, configured to calculate a three-dimensional Euclidean distance between each boundary point in the manual annotation three-dimensional coordinate matrix and each boundary point in the thermal imaging three-dimensional coordinate matrix; a dense point set determination sub-unit, configured to collect all boundary point pairs corresponding to a three-dimensional Euclidean distance not exceeding the maximum distance threshold to obtain an initial intersection point set, and determine whether there is a boundary point pair corresponding to a three-dimensional Euclidean distance not exceeding the minimum distance threshold in the initial intersection point set, and if so, collect the boundary point pair corresponding to the three-dimensional Euclidean distance not exceeding the minimum distance threshold in the initial intersection point set to obtain a dense point set; The intersection point set determination subunit is configured to: regarding all the dense point sets whose corresponding spatial densities exceed the corresponding density threshold value as true dense point sets, determining a true dense region based on each true dense point set, and deleting all the boundary points in all the true dense point sets except the boundary point closest to the geometric center of the corresponding true dense region from the initial intersection point set to obtain a final intersection point set; The intersection matrix determination subunit is configured to generate a spatial intersection matrix of the manually labeled three-dimensional coordinate matrix and the thermographic three-dimensional coordinate matrix based on the three-dimensional coordinates of each boundary point in the final intersection point set.

[0012] Optionally, the core region determination unit comprises: The distance-to-others determination subunit is configured to calculate the three-dimensional Euclidean distances between each boundary point in the spatial intersection matrix and all the boundary points other than the current boundary point in the spatial intersection matrix as the distance-to-others of the current boundary point. The inner point set determination subunit is configured to: from all the boundary points in the spatial intersection matrix, screen all the boundary points whose distance-to-others are greater than the distance-to-others threshold value and collect them to obtain an inner point set. The main direction vector determination subunit is configured to extract the first singular vector of the spatial intersection matrix by singular value decomposition as the main direction vector of the core pain region. The main direction interval determination subunit is configured to: take the three-dimensional average coordinates of the inner point set as a starting point, extend along the direction and the corresponding opposite direction of the main direction vector of the core pain region to obtain an extending straight line, determine all the intersection points of the extending straight line and the inner point set and collect them as a main direction intersection set, and determine the main direction interval based on the main direction intersection set. The core pain region determination subunit is configured to determine a pain direction standard body of the current limb region, take the main direction interval as an axis, and combine the distribution range of the inner point set in the plane perpendicular to the main direction vector of the core pain region to construct a core pain region in the shape of the pain direction standard body.

[0013] Optionally, the calibration-required region labeling unit comprises: The first-order difference operation subunit is configured to perform first-order difference operation on the manually labeled three-dimensional coordinate matrix to obtain a gradient change matrix of the pain region profile labeled by the patient. The gradient change rate calculation subunit is configured to: take the ratio of the length of each row gradient vector in the gradient change matrix to the temperature difference between the corresponding two boundary points as the gradient change rate. The calibration-required region labeling subunit is configured to: from the pain region profile labeled by the patient, screen all the boundary points whose gradient change rates exceed the preset gradient change rate threshold value as calibration-required edge region contour points, and determine a calibration-required edge region based on all the calibration-required edge region contour points.

[0014] Optionally, the pain prediction module comprises: a feature construction submodule, configured to align time series of the patient's pain intensity score, heart rate variability index and skin electric response index, and combine the patient's calibrated pain area to construct a multi-dimensional pain feature vector; a pain prediction submodule, configured to process the multi-dimensional pain feature vector by using the trained LSTM neural network model to obtain a pain intensity prediction curve in a future period.

[0015] Optionally, the treatment linkage module comprises: an instruction determination submodule, configured to generate a treatment parameter control instruction based on the pain trend prediction curve and the multi-dimensional pain feature vector; a linkage treatment submodule, configured to control the spinal cord electrical stimulator and the intrathecal morphine pump based on the Bluetooth wireless connection established with the spinal cord electrical stimulator and the intrathecal morphine pump and the treatment parameter control instruction; wherein the treatment parameters of the spinal cord electrical stimulator include stimulation frequency, pulse width and stimulation intensity; and the treatment parameter of the intrathecal morphine pump is drug infusion rate.

[0016] The present application has the following beneficial effects over the prior art: a three-dimensional human anatomy model is constructed, which contains hierarchical structures of bones, nerves and organs and is marked with pain sensitive areas, providing an intuitive and accurate human structure reference for subsequent pain assessment. The pain area marking instruction and intensity score input by the user enable the patient to directly express their own pain perception. The human infrared thermal imaging data are monitored in real time, which can assist in determining the pain area from the perspective of temperature change. The heart rate variability and skin electric response indicators are obtained through the wearable device, which reflect the patient's pain state from the physiological aspect. The pain area is calibrated by combining the pain area marking instruction with the abnormal temperature area of the infrared thermal imaging, improving the accuracy of pain area determination. The pain intensity curve in the future period is predicted based on multiple data, realizing the prospective grasp of the pain development trend. The treatment parameter control instruction is generated according to the pain trend prediction curve and the multi-dimensional pain feature vector, closely linking pain assessment and treatment, realizing precise and dynamic treatment intervention, and comprehensively improving the scientificity and effectiveness of pain assessment and treatment.

[0017] Other features and advantages of the present application will be set forth in the following description, and in part will become apparent to those skilled in the art from the description, or can be learned by practice of the application. The objects and other advantages of the present application can be realized and attained by particularly pointed out in the application.

[0018] The technical solutions of the present application will be further described in detail below with the help of the accompanying drawings and examples. BRIEF DESCRIPTION OF DRAWINGS

[0019] The accompanying drawings are included to provide a further understanding of the application, and are incorporated in and constitute a part of this specification, illustrate embodiments of the application, and together with the description serve to explain the principles of the application. In the drawings: Figure 1 A schematic diagram of a pain dynamic assessment and treatment linkage system based on three-dimensional somatosensory interaction in an embodiment of the present application; Figure 2 A sub-module flowchart of a human modeling module in an embodiment of the present application; Figure 3 A sub-module flowchart of a somatosensory interaction module in an embodiment of the present application. DETAILED DESCRIPTION

[0020] The preferred embodiments of the present application will be described herein below with reference to the accompanying drawings, in which it should be understood that the preferred embodiments described herein are intended for the purpose of illustration and explanation only. It is not intended to limit the present application in any manner.

[0021] As shown in Figure 1 , the present application provides an embodiment of a pain dynamic assessment and treatment linkage system based on three-dimensional somatosensory interaction, comprising: a human modeling module for constructing a three-dimensional human anatomy model comprising a layered structure of a patient's skeleton, nerves and internal organs, and a pain-sensitive area marker associated with each anatomical site; a somatosensory interaction module for receiving a pain area marking instruction and a pain intensity score input by a user; an infrared sensing module for real-time monitoring of human infrared thermal imaging data of the patient; a physiological monitoring module for monitoring heart rate variability and skin electrical response indicators of the patient based on wearable devices; a region marking module for obtaining a calibrated pain area of the patient based on the pain area marking instruction and an abnormal temperature area in the human infrared thermal imaging data; a pain prediction module for predicting a pain intensity prediction curve in a future period based on the pain intensity score, the heart rate variability indicator, the skin electrical response indicator, and the calibrated pain area of the patient; a treatment linkage module for generating a treatment parameter control instruction based on the pain trend prediction curve and a multi-dimensional pain feature vector.

[0022] In this embodiment, each anatomical site is associated with a corresponding pain-sensitive area marker: when constructing a three-dimensional human anatomy model, different anatomical sites in the model, such as bones, nerves, organs, etc., are marked with areas that are prone to pain. For example, the nerve pain-sensitive area is marked with a red semi-transparent layer, the skeletal muscle pain-sensitive area is marked with a blue semi-transparent layer, and the internal organ pain-sensitive area is marked with a yellow semi-transparent layer. In this way, when the patient marks the pain area on the model, the pain source can be more accurately reflected by combining these pre-set sensitive area markers. For example, if the patient marks the pain area on the back, and the area is a skeletal muscle pain-sensitive area, the doctor can preliminarily judge that the pain may be related to the skeletal muscle, combined with the blue marker.

[0023] In this embodiment, pain intensity scoring: the patient inputs a value of 0-10 through the VAS / NRS scoring slider integrated in the somatosensory interaction module to represent the degree of pain he / she feels. A score of 0 represents no pain, and a score of 10 represents extreme pain. For example, if the patient feels that the pain is 5 points, he / she will slide the slider to the corresponding position, and the scoring data will be used to assess the patient's current pain intensity.

[0024] In this embodiment, real-time monitoring of patient's human infrared thermal imaging data: using an infrared sensing module, the patient's body is continuously monitored to obtain thermal imaging data of the temperature distribution on the human body surface. Because the temperature of some parts of the body may change abnormally when they are in pain, real-time monitoring can capture this temperature change information to assist in pain assessment. For example, if a patient has inflammation-induced pain in a certain part of the body, that part may show an increase in temperature on the infrared thermal imaging map.

[0025] In this embodiment, heart rate variability and galvanic skin response indicators: these two physiological indicators are monitored by wearable devices. The heart rate variability (HRV) indicator reflects the changes in the interval between heartbeats, for example, by calculating the difference in time interval between adjacent R waves, and by analyzing the variation pattern of the heart rate interval over a period of time, the dynamic changes of the autonomic nervous system in regulating the heart can be assessed, which indirectly reflects the body's stress response to pain. The galvanic skin response (GSR) indicator reflects the activity of human sweat glands by measuring the skin conductance, for example, when a patient experiences tension due to pain, the sweat gland activity increases, the skin conductance rises, and the GSR indicator value becomes larger, thus reflecting the patient's pain state from a physiological perspective.

[0026] In this embodiment, abnormal temperature zones in human infrared thermal imaging data are defined as areas where the temperature deviates from the normal range within a certain degree. For example, if the normal body temperature range is set to 36.0-37.2℃, and the thermal imaging data shows that the temperature of a certain area is higher than 37.7℃ or lower than 35.5℃ (deviating from the normal range by more than ±0.5℃; the specific threshold can be adjusted according to the actual situation), then that area is marked as an abnormal temperature zone. These areas may be related to the patient's pain area and are used to assist in calibrating the patient's marked pain area.

[0027] In this embodiment, the future period refers to the future time period in which the pain prediction module predicts the intensity of the patient's pain, which is usually the next 24 hours in this embodiment.

[0028] In this embodiment, the pain intensity prediction curve is obtained by the pain prediction module using a trained LSTM neural network model to process the constructed multidimensional pain feature vector. This curve shows the predicted pain intensity change trend for each hour in the next 24 hours.

[0029] In this embodiment, the treatment parameter control instructions are generated by the treatment linkage module based on the pain trend prediction curve and multidimensional pain feature vector, and are used to control the treatment parameters of the spinal cord stimulator and the intrathecal morphine pump. For example, when the current VAS / NRS score is >7 and is predicted to continue to rise within the next hour, an emergency adjustment mode is triggered, increasing the stimulation frequency of the spinal cord stimulator by 20% and the stimulation intensity by 15% (not exceeding the safety threshold); and increasing the infusion rate of the intrathecal morphine pump by 10% (each adjustment not exceeding 0.1 ml / h). When the VAS / NRS score is <3 and persists for more than 2 hours, a maintenance adjustment mode is triggered, reducing the stimulation frequency of the spinal cord stimulator by 10% and the infusion rate of the intrathecal morphine pump by 5%. After all parameter adjustment instructions are generated, they must be confirmed by medical staff via a mobile terminal before execution. In emergency situations, a 30-second automatic execution timeout can be set.

[0030] like Figure 2 As shown, a human body modeling module is proposed to construct an accurate and realistic three-dimensional human anatomical model of the patient, including: The CT image data acquisition submodule is used to acquire standard human CT image data of patients and generate a basic model containing the patient's bones, major nerve trunks and internal organs through a three-dimensional reconstruction algorithm. The pain-sensitive area annotation submodule is used to annotate the pain-sensitive areas of the base model to obtain a three-dimensional human anatomical model of the patient.

[0031] In this embodiment, standard human CT image data of the patient is collected: a professional CT scanning device is used to scan the patient's whole body or specific parts. During the scanning process, X-rays penetrate the patient's body from multiple angles, and the detector receives the attenuation information of the rays after passing through the body. These attenuation information varies according to the difference in the degree of X-ray absorption by different tissues, such as bone, which absorbs more X-rays and shows weaker signals on the detector, while soft tissue absorbs less and shows relatively stronger signals. The device converts these received signals into digital data, which constitutes the patient's standard human CT image data.

[0032] In this embodiment, a basic model containing the patient's skeleton, main nerve trunks and internal organs is generated by a three-dimensional reconstruction algorithm: based on the collected standard human CT image data of the patient, first, the data is preprocessed to remove noise interference and enhance the clarity and contrast of the image. Then, different tissues in the CT image data, such as bone, nerve trunk, internal organ, etc., are separated and identified using segmentation algorithms. For example, based on the difference in gray value of different tissues in the CT image, a suitable gray threshold is set to separate the bone tissue from other tissues. For nerve trunks and internal organs, more complex machine learning or deep learning algorithms may be required for identification and segmentation. After segmentation, according to the spatial relationship of each tissue, the segmented tissues are reconstructed in three-dimensional space by three-dimensional reconstruction methods such as surface reconstruction or volume rendering, to generate a basic model containing the patient's skeleton, main nerve trunks and internal organs. For example, the segmented bone tissue is reconstructed into a three-dimensional bone model according to its actual spatial position in the body, and combined with the nerve trunk and internal organ models reconstructed in the same way to form a complete basic model.

[0033] In this embodiment, the basic model is marked with pain sensitive areas to obtain a three-dimensional human anatomy model of the patient: according to medical knowledge and clinical experience, the generated basic model is marked with pain sensitive areas. For nerve pain sensitive areas, a red translucent layer is overlaid on the corresponding nerve distribution area in the basic model; for skeletal muscle pain sensitive areas, a blue translucent layer is marked on the relevant parts of the skeleton and muscles; for internal organ pain sensitive areas, a yellow translucent layer is marked on the corresponding position of the internal organs. For example, in the basic model, the sciatic nerve distribution area is covered with a red translucent layer, indicating that this area is a nerve pain sensitive area; the lumbar spine and surrounding muscle parts are marked with a blue translucent layer, representing the skeletal muscle pain sensitive area; the stomach area is marked with a yellow translucent layer, indicating that it is an internal organ pain sensitive area. Through such marking, a three-dimensional human anatomy model of the patient with pain sensitive area marking is finally obtained.

[0034] As shown in Figure 3 , a multi-mode and convenient somatosensory interaction is realized, and a somatosensory interaction module is proposed, which includes: a touch display screen for displaying a three-dimensional human anatomy model and receiving a first pain area marking instruction input by a patient; a depth camera for capturing a gesture action of the patient and translating the gesture action of the patient into a second pain area marking instruction; an instruction fusion sub-module for determining a pain area marking instruction input by the user based on the first pain area marking instruction and / or the second pain area marking instruction; a score receiving sub-module for receiving a pain intensity score input by the user through an integrated VAS / NRS score slider.

[0035] In this embodiment, the touch display screen presents a constructed 1:1 scale three-dimensional human anatomy model, which includes the layered structure of bones, nerves and organs, in the somatosensory interaction module. The patient can directly click or circle the part of the body where they feel pain on the touch display screen with their fingers, and these operations form the first pain area marking instruction. For example, if the patient clicks the chest position on the touch display screen, this click operation conveys the information that the chest may have a pain area to the system.

[0036] In this embodiment, the depth camera tracks the gesture action of the patient in front of the three-dimensional human anatomy model in real time. For example, the patient may use their fingers to gesture a circle in the air to indicate the pain area. After the depth camera captures this gesture action, it converts the gesture action into a second pain area marking instruction that the system can understand through specific image recognition and conversion algorithms (similar to matching the coordinates, shape and other features of the gesture action with pre-defined action templates), so that the system can clearly indicate the pain area indicated by the patient.

[0037] In this embodiment, the instruction fusion sub-module analyzes and processes the first pain area marking instruction (from the touch operation) and the second pain area marking instruction (from the gesture translation). If the patient performs both touch operation and gesture action, and the areas indicated by the two are similar, the system may combine the information of the two and take the overlapping or similar area as the finally determined pain area marking instruction; if only one of the two exists, the system directly takes the instruction as the pain area marking instruction input by the user. For example, if the patient clicks the leg on the touch screen and simultaneously gestures a circle in the air to enclose the same position on the leg, the system will determine the leg as the pain area marked by the patient.

[0038] In this embodiment, the pain intensity score input by the user is received through the integrated VAS / NRS score slider: on the touch display screen of the somatosensory interaction module, the VAS / NRS score slider is integrated. The patient selects the corresponding score by sliding the slider according to the degree of pain he feels within the range of 0-10 points. For example, if the patient feels the pain degree is 6 points, he slides the slider to the position corresponding to 6 points, and the system receives this score as the pain intensity score input by the patient, thereby quantifying the current pain degree of the patient.

[0039] To accurately obtain the calibrated pain area of the patient, a region marking module is proposed, comprising: The region marking sub-module is used to determine the pain area marked by the patient based on the pain area marking instruction. The fusion calibration sub-module is used to fuse and calibrate the pain area marked by the patient and the abnormal temperature area in the human infrared thermal imaging data to obtain the calibrated pain area of the patient.

[0040] In this embodiment, the pain area marked by the patient is determined based on the pain area marking instruction: after the patient inputs the pain area marking instruction through the touch display screen or the depth camera in the somatosensory interaction module, the system first acquires the coordinate information in the instruction. If it is the first pain area marking instruction generated by the touch operation, the system directly reads the coordinates on the three-dimensional human anatomy model corresponding to the finger touch screen position; if it is the second pain area marking instruction translated by the gesture action, the coordinate information on the corresponding model is also included in the instruction captured and translated by the depth camera.

[0041] Based on these coordinate information, the system accurately maps the two-dimensional screen coordinates or the coordinates translated by the gesture action into the spatial coordinate system of the three-dimensional human anatomy model through a spatial mapping algorithm. For example, the system pre-sets the conversion relationship between the screen coordinates and the three-dimensional model spatial coordinates, and according to this relationship, the point of touching the screen or the virtual point pointed by the gesture is corresponded to the actual position in the three-dimensional model.

[0042] Then, the system determines the specific area on the three-dimensional human anatomy model according to the acquired coordinates. If it is a click operation, the system determines the pain area according to the pre-set range (such as a spherical area with a radius of 5mm) centered on the click coordinates; if it is a circle selection or smearing operation, the system determines the pain area according to the range enclosed by the operation trajectory. In this way, the pain area marked by the patient is determined based on the pain area marking instruction, providing a basis for subsequent pain assessment and calibration.

[0043] To realize high-precision fusion calibration of the pain area, a fusion calibration sub-module is proposed, comprising: a coordinate matrix conversion unit configured to convert a pain region contour marked by a patient into a three-dimensional coordinate matrix, to obtain a manually-annotated three-dimensional coordinate matrix, and to convert an abnormal temperature zone contour in human infrared thermal imaging data into a three-dimensional coordinate matrix, to obtain a thermal imaging three-dimensional coordinate matrix; an intersection matrix determination unit configured to calculate a spatial intersection matrix of the manually-annotated three-dimensional coordinate matrix and the thermal imaging three-dimensional coordinate matrix; a core region determination unit configured to extract a first singular vector of the spatial intersection matrix through singular value decomposition, and to determine a core pain region based on the first singular vector; a calibration-required region demarcation unit configured to perform first-order difference operation on the manually-annotated three-dimensional coordinate matrix, to obtain a gradient change matrix of the pain region contour marked by the patient, to screen all boundary points with a gradient change rate exceeding a preset gradient change rate threshold as calibration-required edge region contour points based on the gradient change matrix, and to determine a calibration-required edge region based on all the calibration-required edge region contour points; a temperature inflection point identification unit configured to map the calibration-required edge region to a temperature gradient field in the human infrared thermal imaging data, to fit a temperature distribution curve through a two-dimensional Gaussian function, and to calculate a second derivative of the temperature distribution curve to determine all temperature change inflection points; a calibration-required region calibration unit configured to correct the calibration-required edge region based on all the temperature change inflection points, to obtain a calibrated edge region; a two-dimensional region fusion unit configured to fuse the core pain region and the calibrated edge region, to obtain a three-dimensional mesh model of a calibrated pain region as the calibrated pain region of the patient.

[0044] In this embodiment, the pain region contour marked by the patient is converted into a three-dimensional coordinate matrix to obtain a manually-annotated three-dimensional coordinate matrix: the system acquires the pain region contour marked by the patient on the three-dimensional human anatomy model through touch or gesture. For this contour, the coordinate values of each point on the contour in the three-dimensional human anatomy model coordinate system are extracted in a certain order, and the coordinates of each point contain three dimensions (x, y, z). These coordinate values are arranged in order to form a matrix, with the number of contour points as the rows and 3 as the columns, to obtain the manually-annotated three-dimensional coordinate matrix.

[0045] In this embodiment, the abnormal temperature zone contour in the human infrared thermal imaging data is converted into a three-dimensional coordinate matrix to obtain a thermal imaging three-dimensional coordinate matrix: the contour of the abnormal temperature zone is identified from the human infrared thermal imaging data. Similarly, for each point on the contour, its coordinates (x, y, z) in the same spatial coordinate system as the three-dimensional human anatomy model are determined. These coordinate values are arranged in order to form a matrix, with the number of contour points as the rows and 3 as the columns, to obtain the thermal imaging three-dimensional coordinate matrix.

[0046] In this embodiment, the first singular vector of the spatial intersection matrix is extracted by singular value decomposition: singular value decomposition is performed on the spatial intersection matrix. Singular value decomposition will decompose the matrix into the product of three matrices, i.e. A = UΣV T where Σ is a diagonal matrix, and the values on the diagonal are singular values arranged in descending order. The first singular vector is the first column vector of the U matrix, which reflects the main directional characteristics of the area represented by the spatial intersection matrix.

[0047] In this embodiment, the gradient change matrix of the pain area contour marked by the patient is obtained by performing first-order difference operation on the manually marked three-dimensional coordinate matrix: each row of the manually marked three-dimensional coordinate matrix represents the three-dimensional coordinates of a point on the pain area contour. First-order difference operation is performed on the matrix, i.e. the difference between the coordinates of adjacent rows is calculated, and the resulting difference values form a new matrix, which is the gradient change matrix of the pain area contour marked by the patient. It reflects the change of coordinates between points on the contour, and embodies the steepness and direction change of the contour.

[0048] In this embodiment, a preset gradient change rate threshold is set: according to the actual pain assessment requirements and experience, a numerical value is set as the threshold of the gradient change rate. The gradient change rate refers to the ratio of the length of the gradient vector in each row of the gradient change matrix to the temperature difference between the corresponding two boundary points. For example, the threshold is set to 0.5.

[0049] In this embodiment, the edge region to be calibrated is determined based on all the edge region contour points to be calibrated: the edge region contour points to be calibrated are connected, and the edge region to be calibrated is determined according to their distribution range and shape.

[0050] In this embodiment, the edge region to be calibrated is mapped into the temperature gradient field in the human infrared thermal imaging data, and the second derivative of the temperature distribution curve is calculated to determine all the temperature change inflection points by fitting the temperature distribution curve with a two-dimensional Gaussian function: the coordinates of the edge region to be calibrated are mapped into the corresponding temperature gradient field in the human infrared thermal imaging data to obtain the temperature distribution data of the region in the temperature gradient field. Then, the two-dimensional Gaussian function is used to fit these temperature distribution data to obtain a smooth temperature distribution curve. The second derivative of the curve is calculated, and when the second derivative is 0 and the first derivative has an extreme value at this point, the corresponding point is the temperature change inflection point. All temperature change inflection points are determined in this way, and these inflection points represent the positions where the degree of temperature change changes.

[0051] In this embodiment, the calibration edge region is obtained by correcting the edge region to be calibrated based on all temperature change inflection points: the contour of the edge region to be calibrated is adjusted according to the coordinates of the determined temperature change inflection points. For example, if a certain temperature change inflection point is located near a certain segment of the contour of the edge region to be calibrated, it indicates that this segment of the contour may need to be corrected according to the temperature change. The possible correction method is to move this segment of the contour a certain distance in the direction of the temperature change inflection point, or to change the shape of this segment of the contour so that it is more consistent with the actual situation reflected by the temperature change, thereby obtaining the calibration edge region.

[0052] In this embodiment, the three-dimensional mesh model of the calibration pain region is obtained by merging the core pain region and the calibration edge region: the calibration edge region and the core pain region are spliced and merged according to their positional relationship in the three-dimensional space. For example, the calibration edge region surrounds the periphery of the core pain region, and after combining the two, a three-dimensional mesh model is generated on this merged region using a three-dimensional mesh generation algorithm, and the mesh cell length is set to be not greater than 2 mm. This three-dimensional mesh model is the calibration pain region, which more accurately reflects the actual pain region of the patient, providing a more accurate basis for subsequent pain assessment and treatment.

[0053] To improve the accuracy of pain region calibration, an intersection matrix determination unit is proposed, which includes: A distance threshold determination subunit is configured to set a maximum distance threshold and a minimum distance threshold based on the precision requirement corresponding to the pain intensity score input by the user; A Euclidean distance determination subunit is configured to calculate the three-dimensional Euclidean distance between each boundary point in the manually labeled three-dimensional coordinate matrix and each boundary point in the thermographic three-dimensional coordinate matrix; A dense point set determination subunit is configured to collect all boundary point pairs corresponding to a three-dimensional Euclidean distance not exceeding the maximum distance threshold to obtain an initial intersection point set, and determine whether there are boundary point pairs corresponding to a three-dimensional Euclidean distance not exceeding the minimum distance threshold in the initial intersection point set. If so, the initial intersection point set is collected to obtain a dense point set; An intersection point set determination subunit is configured to regard the dense point set corresponding to a spatial density exceeding the corresponding density threshold in all dense point sets as a true dense point set, determine a true dense region based on each true dense point set, and delete all boundary points in the initial intersection point set except the boundary point closest to the geometric center of the corresponding true dense region to obtain a final intersection point set; An intersection matrix determination subunit is configured to generate a spatial intersection matrix of the manually labeled three-dimensional coordinate matrix and the thermographic three-dimensional coordinate matrix based on the three-dimensional coordinates of each boundary point in the final intersection point set.

[0054] In this embodiment, the maximum distance threshold and the minimum distance threshold are set based on the precision requirement corresponding to the pain intensity score input by the user: different pain intensity scores input by the user correspond to different precision requirements for positioning the pain area. Generally speaking, the higher the pain intensity score, the more intense the patient's pain perception, and at this time, the pain area positioning is more accurate. For example, when the pain intensity score is 8-10, the maximum distance threshold can be set to 2 mm and the minimum distance threshold to 0.5 mm; if the score is 3-7, the maximum distance threshold can be set to 3 mm and the minimum distance threshold to 1 mm. In this way, the threshold is adjusted according to the pain intensity score, which can reasonably screen the intersection points under different pain levels and improve the accuracy of pain area calibration.

[0055] In this embodiment, the three-dimensional Euclidean distance is the square root of the sum of the coordinate differences of the same dimensions of two points in space. In this embodiment, the distance between each boundary point in the manually labeled three-dimensional coordinate matrix and each boundary point in the thermal imaging three-dimensional coordinate matrix is calculated to determine the proximity of the two points in three-dimensional space and determine whether they can belong to the intersection points.

[0056] In this embodiment, the spatial density refers to the number of points in a certain spatial volume. In this embodiment, for a group of points (such as some points in the initial intersection point set), a spatial volume containing these points is first determined , for example, a spherical volume with a certain point as the center and a radius of . Then the number of points in the volume is counted and the spatial density .

[0057] In this embodiment, the density threshold is a spatial density numerical standard set according to the actual pain area calibration requirements and experience. For example, the density threshold is set to 5 points / mm 3 .

[0058] In this embodiment, the true dense region is determined based on each true dense point set: for each true dense point set that satisfies the spatial density greater than the density threshold, the true dense region is determined based on the distribution of these points in three-dimensional space. For example, by calculating the convex hull of the point set, a convex polygon (convex polyhedron in three-dimensional space) containing all the points is obtained, and the region surrounded by the convex polyhedron is the true dense region. Or take the two points farthest apart in the point set as the axis, and combine the distance ranges of other points to the axis to determine a region similar to a cylindrical or elliptical cylindrical region as the true dense region, which can better cover the actual region represented by the true dense point set.

[0059] In this embodiment, the geometric center of the true dense area: for a determined true dense area, calculate its geometric center. If the true dense area is a convex polyhedron composed of a set of points, first calculate the sum of the coordinates of all points, and then divide by the number of points n to obtain the geometric center coordinates.

[0060] In this embodiment, based on the three-dimensional coordinates of each boundary point in the final intersection point set, a spatial intersection matrix of the manually labeled three-dimensional coordinate matrix and the thermal imaging three-dimensional coordinate matrix is generated: the final intersection point set contains the possible intersection points of the manually labeled pain area and the thermal imaging abnormal temperature area after filtering out redundancies. The three-dimensional coordinates of these points are arranged in order to form a matrix, with the number of intersection points as the rows and 3 as the columns. This matrix is the spatial intersection matrix of the manually labeled three-dimensional coordinate matrix and the thermal imaging three-dimensional coordinate matrix.

[0061] To accurately determine the core pain area, a core area determination unit is proposed, including: A distance from it determination sub-unit is used to calculate the three-dimensional Euclidean distance between each boundary point in the spatial intersection matrix and all the remaining boundary points in the spatial intersection matrix except the current boundary point, as the distance from it of the current boundary point. An inner point set determination sub-unit is used to filter out all boundary points whose distance from it is greater than the distance from it threshold value from all boundary points in the spatial intersection matrix and aggregate them to obtain an inner point set. A main direction vector determination sub-unit is used to extract the first singular vector of the spatial intersection matrix through singular value decomposition as the main direction vector of the core pain area. A main direction interval determination sub-unit is used to extend along the direction and corresponding opposite direction of the main direction vector of the core pain area from the three-dimensional average coordinates of the inner point set as the starting point to obtain an extended straight line, determine all intersection points of the extended straight line and the inner point set and aggregate them as a main direction intersection point set, and determine the main direction interval based on the main direction intersection point set. A core pain area determination sub-unit is used to determine the pain direction standard body of the current limb area, take the main direction interval as the axis, and combine the distribution range of the inner point set in the plane perpendicular to the main direction vector of the core pain area to construct a core pain area in the shape of a pain direction standard body.

[0062] In this embodiment, the distance from it threshold value: this is a pre-set distance value used to filter the inner point set from all boundary points in the spatial intersection matrix. For example, it is set to 5mm.

[0063] In this embodiment, the three-dimensional average coordinates of the inner point set are taken as the starting point, and the extending straight lines are obtained by extending along the direction of the main direction vector of the core pain area and the corresponding opposite direction. All intersection points of the extending straight lines and the inner point set are determined and collected as the main direction intersection set. The main direction interval is determined based on the main direction intersection set: first, the average value of the three-dimensional coordinates of all points of the inner point set is calculated to obtain a three-dimensional average coordinate point. Take this point as the starting point, and extend straight lines along the direction of the main direction vector v and its opposite direction -v obtained by singular value decomposition. Then, find all intersection points of the two extending straight lines and the inner point set, and collect these intersection points to form the main direction intersection set. In the main direction intersection set, determine the two intersection points farthest away from each other. The interval formed by the line segment between the two intersection points is the main direction interval. For example, the three-dimensional average coordinates of the inner point set are (10, 10, 10), and the main direction vector is (1, 0, 0). Then, the intersection points of the extending straight lines along the direction and the opposite direction with the inner point set may be (5, 10, 10), (15, 10, 10), etc., and the main direction interval may be the interval determined by the line segment from (5, 10, 10) to (15, 10, 10), which reflects the range of the core pain area in the main direction.

[0064] In this embodiment, the pain direction standard body of the current limb region is determined: according to medical knowledge and clinical experience, the pain of different limb regions often has specific direction and morphological characteristics, which are summarized as pain direction standard body. For example, the neuralgia of the arm usually has a long strip shape along the nerve distribution, and the corresponding pain direction standard body may be a long strip model; while the internal pain of the abdomen may be more inclined to be approximately spherical distribution with a certain internal organ as the center, and the corresponding pain direction standard body is a spherical shape. By judging the limb part where the current pain area is located, the pain direction standard body model corresponding thereto is selected to provide a rough shape reference framework for constructing the core pain area.

[0065] In this embodiment, the main direction interval is taken as the axis, and the distribution range of the inner point set on the plane perpendicular to the main direction vector of the core pain area is combined to construct the core pain area of the standard body shape of pain direction: taking the previously determined main direction interval as the central axis of the core pain area. Then, considering the distribution of the inner point set on the plane perpendicular to the main direction vector, for example, calculating the projection of the inner point set on these planes, determining the distribution range of the projection points, such as finding the point in the projection points farthest from the central axis, to determine the radius or size in the vertical direction. Taking the main direction interval as the axis and the size in the vertical direction as the basis, combined with the shape of the previously determined standard body of pain direction (such as long strip, spherical shape, etc.), the model of the core pain area is constructed. For example, if the main direction interval is a line segment along the x-axis direction, and the projection of the inner point set on the plane perpendicular to the x-axis shows that the distribution range is a circle with a radius of 3 mm, and the standard body of pain direction of the current limb area is a long strip, then a long strip core pain area with a radius of 3 mm is constructed, which is taken as the axis, so that it is more in line with the characteristics of the actual pain area.

[0066] To accurately calibrate the calibration region, a calibration region calibration unit is proposed, which includes: A first-order difference operator unit is configured to perform first-order difference operation on the manually annotated three-dimensional coordinate matrix to obtain a gradient change matrix of the pain region profile annotated by the patient; A gradient change rate calculation subunit is configured to take the ratio of the length of each row gradient vector in the gradient change matrix to the temperature difference between the corresponding two boundary points as the gradient change rate; A calibration region calibration subunit is configured to filter out all boundary points with a gradient change rate exceeding a preset gradient change rate threshold from the pain region profile annotated by the patient as calibration edge region profile points, and determine a calibration edge region based on all calibration edge region profile points.

[0067] In this embodiment, the ratio of the length of each row gradient vector in the gradient change matrix to the temperature difference between the corresponding two boundary points is taken as the gradient change rate: in the gradient change matrix G, each row (Δx i ,Δy i ,Δz i ) constitutes a gradient vector. The length of the gradient vector represents the actual distance change of adjacent boundary points in three-dimensional space. At the same time, the temperature values of the two adjacent boundary points in the human infrared thermal imaging data are obtained, and the temperature difference is calculated. The gradient change rate is equal to the ratio of the length to the temperature difference, which reflects the degree of change of the pain region profile in space corresponding to the unit temperature change, and is used to measure the relationship between the profile change and the temperature change.

[0068] To effectively predict the development trend of pain, a pain prediction module is proposed, which includes: The feature construction submodule is configured to align time series of the pain intensity score, the heart rate variability index, and the galvanic skin response index of the patient, and construct a multi-dimensional pain feature vector in combination with the calibrated pain region of the patient. The pain prediction submodule is configured to process the multi-dimensional pain feature vector by using the trained LSTM neural network model to obtain a pain intensity prediction curve in a future period.

[0069] In this embodiment, the pain intensity score, the heart rate variability index, and the galvanic skin response index of the patient are aligned in time series, and a multi-dimensional pain feature vector is constructed in combination with the calibrated pain region of the patient: the pain intensity score, the heart rate variability (HRV) index, and the galvanic skin response (GSR) index data of the patient at different time points are collected. Since these data may differ in collection time, time series alignment is performed to make each time point correspond to the data of the three indexes at the same time. For example, it is assumed that data is collected once an hour, and the data of different indexes at each whole point is arranged to correspond. For the calibrated pain region, features such as the pain region area, the overlap degree of the pain region and the nerve-intensive area, and the three-dimensional center coordinates of the pain region are extracted. Then, the pain intensity score, the HRV index, and the GSR index data after time series alignment and the features of the calibrated pain region are integrated together to form a multi-dimensional pain feature vector. For example, the constructed vector may be in the form of [pain intensity score, HRV time domain index, GSR peak value, pain region area, overlap degree of pain region and nerve-intensive area, x value of three-dimensional center coordinates of pain region, y value of three-dimensional center coordinates of pain region, and z value of three-dimensional center coordinates of pain region].

[0070] In this embodiment, the trained LSTM neural network model is used to process the multi-dimensional pain feature vector to obtain a pain intensity prediction curve in a future period: the constructed multi-dimensional pain feature vector is sequentially input into the trained LSTM neural network model in time order. The structure inside the model will process these input data layer by layer, learn the time series features and patterns in the data through the memory unit and the gating mechanism, and capture the relationship between the pain intensity and other related indexes changing over time. For example, the model may learn the potential relationship between the change of the heart rate variability index and the change of the future pain intensity. After the model processing, the pain intensity prediction value corresponding to each time point in the future period (such as the next 24 hours) is output. Finally, these prediction values are generated into a continuous pain intensity prediction curve through a specific curve fitting method (such as cubic spline curve fitting), which intuitively shows the trend of the future pain intensity changing over time.

[0071] In this embodiment, the trained LSTM neural network model: first, at least 500 cases of chronic pain patients' historical data are collected, including hourly recorded pain area, VAS / NRS score, HRV, GSR and corresponding treatment parameters. These data are divided into training set and validation set according to the proportion of 7:3. The input layer dimension of the model is set to 4 (corresponding to the pain intensity score, HRV index, GSR index and calibrated pain area in the multi-dimensional feature vector), the hidden layer contains 32 neurons, which are connected by complex weights to perform nonlinear transformation on the input data, and mine the potential features and relationships in the data. The output layer is the pain intensity prediction value for the next 24 hours. The Adam optimizer is used to adjust the weight parameters of the model to make the prediction results of the model as close to the true value as possible. The training period is set to 100 rounds, and in the training process, the model continuously adjusts the weights according to the training set data. After each round of training, the performance of the model is evaluated using the validation set data, and the mean absolute error of the validation set is calculated. When the mean absolute error of the validation set is less than 1.0, it is considered that the model training has achieved good results, and the training is stopped, and the final trained LSTM neural network model is obtained. The model can effectively process the input multi-dimensional pain feature vector and predict the future pain intensity.

[0072] To realize the precise linkage of pain assessment and treatment, a treatment linkage module is proposed, which includes: The instruction determination sub-module is used to generate treatment parameter control instructions based on the pain trend prediction curve and the multi-dimensional pain feature vector. The linkage treatment sub-module is used to control the spinal cord electrical stimulator and the intrathecal morphine pump based on the Bluetooth wireless connection established with the spinal cord electrical stimulator and the intrathecal morphine pump and the treatment parameter control instructions. The treatment parameters of the spinal cord electrical stimulator include stimulation frequency, pulse width and stimulation intensity. The treatment parameter of the intrathecal morphine pump is the drug infusion rate.

[0073] In this embodiment, the treatment parameter control instructions are generated based on the pain trend prediction curve and the multi-dimensional pain feature vector: the system generates treatment parameter control instructions according to the established rules based on these data.

[0074] For example, when the pain trend prediction curve shows that the pain intensity continues to rise in the next 1 hour, and the current VAS / NRS score is greater than 7, while the multi-dimensional pain feature vector indicates that the patient's physiological indicators (such as heart rate variability, skin electrical response) also show corresponding abnormal changes, the emergency adjustment mode is triggered. For the spinal cord electrical stimulator, the stimulation frequency is increased by 20% according to the rule, that is, if the current stimulation frequency is f, the adjusted frequency is f x (1 + 20%); the stimulation intensity is increased by 15% (but not more than the safety threshold, assuming the safety threshold is Imax, if the current stimulation intensity is I, the adjusted intensity is I x (1 + 15%), and I x (1 + 15%) ≤ Imax must be met). For the intrathecal morphine pump, the drug infusion rate is increased by 10% (single adjustment does not exceed 0.1 ml / h, assuming the current infusion rate is v, the adjusted rate is v x (1 + 10%), if v x (1 + 10%) - v > 0.1 ml / h, the adjusted rate is v + 0.1 ml / h).

[0075] For example, when the VAS / NRS score is less than 3 and lasts for more than 2 hours, while the multi-dimensional pain feature vector reflects that the patient's physiological state is relatively stable, the maintenance adjustment mode is triggered. For the spinal cord electrical stimulator, the stimulation frequency is reduced by 10%, that is, the adjusted frequency is the current frequency f multiplied by (1 - 10%); for the intrathecal morphine pump, the drug infusion rate is reduced by 5%, that is, the adjusted rate is the current rate v multiplied by (1 - 5%).

[0076] By comprehensively analyzing the pain trend prediction curve and the multi-dimensional pain feature vector, and according to these rules, the system generates treatment parameter control instructions for the spinal cord electrical stimulator and the intrathecal morphine pump to achieve precise and dynamic treatment intervention.

[0077] Obviously, those skilled in the art can make various modifications and variations to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application also intends to include these modifications and variations.

Claims

1. A pain dynamic assessment and treatment linkage system based on three-dimensional somatosensory interaction, characterized in that, include: The human modeling module is used to construct a three-dimensional human anatomical model that includes the patient's bone, nerves and organs in a layered structure and with pain-sensitive area markers associated with each anatomical part. The somatosensory interaction module is used to receive user input commands for marking pain areas and pain intensity scores; Infrared sensing module, used to monitor the patient's human infrared thermal imaging data in real time; A physiological monitoring module is used to monitor patients' heart rate variability and skin conductance parameters based on wearable devices; The region marking module is used to obtain the patient's calibrated pain region based on the pain region marking instructions and abnormal temperature areas in human infrared thermal imaging data; The pain prediction module is used to predict the pain intensity prediction curve for future cycles based on the patient's pain intensity score, heart rate variability index, skin conductance index, and calibrated pain area. The treatment linkage module is used to generate treatment parameter control instructions based on pain trend prediction curves and multidimensional pain feature vectors.

2. The pain dynamic assessment and treatment linkage system based on three-dimensional somatosensory interaction according to claim 1, characterized in that, The human body modeling module includes: The CT image data acquisition submodule is used to acquire standard human CT image data of patients and generate a basic model containing the patient's bones, major nerve trunks and internal organs through a three-dimensional reconstruction algorithm. The pain-sensitive area annotation submodule is used to annotate the pain-sensitive areas of the base model to obtain a three-dimensional human anatomical model of the patient.

3. The pain dynamic assessment and treatment linkage system based on three-dimensional somatosensory interaction according to claim 1, characterized in that, The motion-sensing interaction module includes: A touchscreen display is used to show a three-dimensional human anatomical model and receive the patient's input of a first pain area marking instruction; A depth camera is used to capture the patient's hand gestures and translate them into instructions for marking the second pain area; The instruction fusion submodule is used to determine the pain area marking instruction input by the user based on the first pain area marking instruction and / or the second pain area marking instruction; The rating receiving submodule is used to receive the pain intensity rating input by the user via an integrated VAS / NRS rating slider.

4. The pain dynamic assessment and treatment linkage system based on three-dimensional somatosensory interaction according to claim 1, characterized in that, The region marking module includes: The region marking submodule is used to determine the pain region marked by the patient based on the pain region marking instructions; The fusion calibration submodule is used to fuse and calibrate the patient's marked pain area and the abnormal temperature area in the human infrared thermal imaging data to obtain the patient's calibrated pain area.

5. The pain dynamic assessment and treatment linkage system based on three-dimensional somatosensory interaction according to claim 4, characterized in that, The integrated calibration submodule includes: The coordinate matrix transformation unit is used to convert the outline of the pain area marked by the patient into a three-dimensional coordinate matrix to obtain a manually annotated three-dimensional coordinate matrix, and at the same time, convert the outline of the abnormal temperature area in the human infrared thermal imaging data into a three-dimensional coordinate matrix to obtain a thermal imaging three-dimensional coordinate matrix. The intersection matrix determination unit is used to calculate the spatial intersection matrix between the manually labeled 3D coordinate matrix and the thermal imaging 3D coordinate matrix. The core region determination unit is used to extract the first singular vector of the spatial intersection matrix through singular value decomposition, and to determine the core pain region based on the first singular vector. The calibration unit for the region to be calibrated is used to perform first-order difference operations on the manually annotated three-dimensional coordinate matrix to obtain the gradient change matrix of the contour of the pain region annotated by the patient, and to select all boundary points whose gradient change rate exceeds the preset gradient change rate threshold as contour points of the edge region to be calibrated based on the gradient change matrix, and to determine the edge region to be calibrated based on all the contour points of the edge region to be calibrated. The temperature inflection point identification unit is used to map the edge region to be calibrated onto the temperature gradient field in the human infrared thermal imaging data, and to determine all temperature change inflection points by fitting the temperature distribution curve with a two-dimensional Gaussian function and calculating the second derivative of the temperature distribution curve. The calibration unit for the area to be calibrated is used to correct the edge area to be calibrated based on all temperature change inflection points to obtain the calibration edge area. Two-dimensional region fusion unit is used to fuse the core pain region with the calibration edge region to obtain a three-dimensional mesh model of the calibration pain region as the patient's calibration pain region.

6. The pain dynamic assessment and treatment linkage system based on three-dimensional somatosensory interaction according to claim 5, characterized in that, The intersection matrix determines the unit, including: The distance threshold determination subunit is used to set the maximum and minimum distance thresholds based on the accuracy requirements corresponding to the pain intensity score input by the user. The Euclidean distance determination sub-unit is used to calculate the three-dimensional Euclidean distance between each boundary point in the manually labeled three-dimensional coordinate matrix and each boundary point in the thermal imaging three-dimensional coordinate matrix. The dense point set determination sub-unit is used to summarize all boundary point pairs whose corresponding three-dimensional Euclidean distance does not exceed the maximum distance threshold to obtain the initial intersection point set, and to determine whether there are boundary point pairs in the initial intersection point set whose corresponding three-dimensional Euclidean distance does not exceed the minimum distance threshold. If so, the boundary point pairs in the initial intersection point set whose corresponding three-dimensional Euclidean distance does not exceed the minimum distance threshold are summarized to obtain the dense point set. The intersection point set determines the sub-unit, which is used to gather all dense points. Dense point sets whose spatial density exceeds the corresponding density threshold are regarded as true dense point sets. True dense regions are determined based on each true dense point set. All boundary points in all true dense point sets except the boundary point closest to the geometric center of the corresponding true dense region are removed from the initial intersection point set to obtain the final intersection point set. The intersection matrix determines the sub-units, which are used to generate the spatial intersection matrix of manually labeled 3D coordinate matrices and thermal imaging 3D coordinate matrices based on the 3D coordinates of each boundary point in the final intersection point set.

7. The pain dynamic assessment and treatment linkage system based on three-dimensional somatosensory interaction according to claim 6, characterized in that, The core area is defined by the following units: The distance-to-it sub-unit is used to calculate the three-dimensional Euclidean distance between each boundary point in the spatial intersection matrix and all remaining boundary points in the spatial intersection matrix except for the current boundary point, based on the spatial intersection matrix, and to serve as all distances to it for the current boundary point; The interior point set determines the sub-unit, which is used to select all boundary points in the spatial intersection matrix that are at a distance greater than a distance threshold from the boundary points and summarize them to obtain the interior point set; The principal direction vector determines the sub-unit, which is used to extract the first singular vector of the spatial intersection matrix through singular value decomposition, and serves as the principal direction vector of the core pain region. The main direction interval is determined by the sub-unit, which is used to extend along the direction of the main direction vector of the core pain area and the corresponding opposite direction, starting from the three-dimensional average coordinates of the inner point set, to obtain the extension line, determine all the intersection points of the extension line with the inner point set and summarize them as the main direction intersection point set, and determine the main direction interval based on the main direction intersection point set. The core pain region determination sub-unit is used to determine the standard body of pain direction in the current limb region. Using the main direction interval as the axis, and combining the distribution range of the inner point set on the plane perpendicular to the main direction vector of the core pain region, the core pain region in the shape of the standard body of pain direction is constructed.

8. The pain dynamic assessment and treatment linkage system based on three-dimensional somatosensory interaction according to claim 6, characterized in that, The calibration unit for the area to be calibrated includes: The first-order difference operation subunit is used to perform first-order difference operations on the manually annotated 3D coordinate matrix to obtain the gradient change matrix of the contour of the pain area annotated by the patient. The gradient change rate calculation sub-unit is used to take the ratio of the magnitude of the gradient vector in each row of the gradient change matrix to the temperature difference between the corresponding two boundary points as the gradient change rate. The calibration region subunit is used to select all boundary points in the pain region contour marked by the patient as the edge region contour points that have a gradient change rate exceeding a preset gradient change rate threshold, and to determine the edge region to be calibrated based on all the edge region contour points to be calibrated.

9. The pain dynamic assessment and treatment linkage system based on three-dimensional somatosensory interaction according to claim 1, characterized in that, The pain prediction module includes: The feature construction submodule is used to perform time-series alignment of the patient's pain intensity score, heart rate variability index, and skin conductance index, and to construct a multidimensional pain feature vector by combining the patient's calibrated pain area. The pain prediction submodule is used to process multidimensional pain feature vectors using a trained LSTM neural network model to obtain a pain intensity prediction curve for future periods.

10. The pain dynamic assessment and treatment linkage system based on three-dimensional somatosensory interaction according to claim 1, characterized in that, The treatment linkage module includes: The instruction determination submodule is used to generate treatment parameter control instructions based on the pain trend prediction curve and multidimensional pain feature vector. The linkage therapy submodule is used to control the spinal cord stimulator and the intrathecal morphine pump based on the Bluetooth wireless connection established with the spinal cord stimulator and the intrathecal morphine pump and the control commands for treatment parameters. The treatment parameters of the spinal cord stimulator include stimulation frequency, pulse width, and stimulation intensity. The therapeutic parameters of an intrathecal morphine pump are the drug infusion rate.

Images