A method and system for implementing trauma healing monitoring

Abstract

The application discloses a method and system for realizing trauma healing monitoring. The method comprises the following steps: scanning a trauma site, and obtaining a temperature data set, a point cloud data set, a pulse wave signal and a serial number i of the trauma area after processing; reproducing a three-dimensional model through the point cloud data set to obtain wound three-dimensional data, dividing three areas to obtain a wound type; grouping the temperature data set and the point cloud data set according to the three areas and generating a function AC, and generating a function BC from the pulse wave signal; comparing the AC and the BC in each partition two by two to obtain AC difference values and BC difference values, storing the AC difference values and the BC difference values and the wound three-dimensional data in O[i], and calculating a difference rate Omega of adjacent serial number data of the data set O; inputting the BC difference values and the AC difference values into a neural adversarial network to generate new CC difference values, querying a medical database corresponding to the wound type to obtain an evaluation parameter G, establishing a recovery function, generating a predicted evaluation parameter G2, and comparing the G and the G2; and visualizing and outputting. The application can more comprehensively monitor a wound condition.

Description

Technical Field

[0001] This invention relates to the fields of image processing, computer vision, and health monitoring, and specifically to a method and system for monitoring trauma healing. Background Technology

[0002] With technological advancements and increased emphasis on health, the demand for routine health monitoring has grown unprecedentedly. However, conventional clinical medical examination equipment is expensive, highly specialized, and its data is overly technical, leading users to rely heavily on medical professionals for interpretation and resulting in lengthy waiting times for results, hindering its widespread adoption in daily life. Existing home-based self-monitoring health data devices primarily focus on contact-based internal medicine data monitoring, while research on trauma monitoring equipment is relatively lacking. In scenarios such as military exercises or remote areas, situations arise where injuries require high-precision surgical diagnosis, but specialized clinical medical equipment and personnel cannot arrive on-site promptly. In these situations, convenient and accurate surgical data monitoring devices can provide timely and effective diagnostic assistance, helping arriving rescue personnel quickly assess the injured person's condition and provide appropriate emergency treatment.

[0003] One current technology is a skin blood perfusion imaging method based on imaging photoplethysmography (IPPG), proposed in the paper "Non-contact Skin Blood Perfusion Imaging Based on IPPG". This method utilizes the LK optical flow method to dynamically track ROI feature points, reducing motion artifact noise in ROI videos and significantly improving IPPG signal quality. It employs Spearman correlation coefficients to perform correlation topographic imaging of the ROI, obtaining blood perfusion distribution images. A P-value is proposed as a quantification index for blood perfusion images, with the AC value as a reference. The proposed method achieves an accuracy of 81.82%, demonstrating high overall imaging accuracy. However, this method has limitations: its overall accuracy can be improved, it cannot directly diagnose trauma based solely on blood perfusion imaging data, and the data is susceptible to external lighting conditions.

[0004] The second existing technology is a 3D scanning method based on dual optical combs, proposed in the paper "Simulation of Measurement Accuracy of 3D Scanning System Based on Dual Optical Combs". This method uses a pair of phase-locked optical combs with a small repetition frequency difference as the light source for ranging. In the dual optical comb ranging method, the two optical combs serve as the signal light and the local oscillator light, respectively, maintaining good coherence and exhibiting a small repetition frequency difference. The signal light is divided into a reference light and a measurement light, emitted to the reflector and the target object, respectively. The two reflected beams are combined at the beam splitter and interfere with the local oscillator light after passing through another beam splitter. By employing asynchronous optical sampling, the interference of the signal light and the local oscillator light generates a downsampled signal. Through continuous scanning by the dual-axis galvanometer, all point cloud information within the measured space can be obtained. The disadvantages of this method are the high technical difficulty of 3D modeling, the high cost of hardware infrastructure, and the fact that high-precision imaging of the 3D point cloud will simultaneously image the uneven areas of the external injury area, making it difficult to analyze the external injury situation in detail from the point cloud data.

[0005] The third existing technology is a method proposed in the paper "A Preliminary Study on the Infrared Thermographic Characteristics of Primary Immune Thrombocytopenic Pulmonary Disease in Children," which uses an infrared detection device to detect the infrared thermal radiation of the human body. This method processes the infrared thermal radiation data of the human body using a computer and displays it in different colors to form a temperature distribution map that can be intuitively perceived and analyzed, thereby indirectly reflecting changes in the body's metabolism and blood circulation. It has the characteristics of being non-invasive, radiation-free, comprehensive, sensitive, and inexpensive. The disadvantages of this method are that it can only measure the relative temperature differences between different temperature zones, which is too limiting for research on specific diseases in patients and non-patients. It cannot accurately and intuitively reflect the temperature parameters in the study of trauma data, and the temperature parameters are sensitive to environmental factors such as temperature and humidity. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of existing methods and propose a method and system for monitoring wound healing. The main problem addressed by this invention is how to achieve comprehensive technological utilization, simultaneously introducing multiple parameters, and simultaneously using temperature, IPPG blood perfusion imaging, and dual-comb three-dimensional scanning modeling data for mutual reference and correction. Simultaneously, it utilizes data from the wound of the patient and surrounding intact areas as reference comparison data, introducing two reference coordinate axes—time and space—to more comprehensively monitor the wound condition.

[0007] To address the above problems, this invention proposes a method for monitoring traumatic injury healing, the method comprising:

[0008] The dataset of the injured body parts was obtained by scanning. After detection by an infrared thermometer, 3D modeling by a 3D laser scanner, and IPPG signal processing of the video signal in the dataset of the injured body parts, the temperature dataset Tnew, point cloud dataset Snew, original pulse wave signal Mnew and data measurement sequence number i of the injured area were obtained.

[0009] When the data measurement sequence number i is not zero, the following four steps are performed:

[0010] The first step is to input the point cloud dataset Snew, reproduce the three-dimensional model of the trauma area, extract the wound image using the video signal, and calculate the maximum width w, maximum length l, maximum depth d, average depth and depth variance of the wound. Then, divide the trauma area h1, the boundary area h2 and the normal skin area h3, and obtain the wound type.

[0011] The second step involves grouping the data in the temperature dataset Tnew and the point cloud dataset Snew according to the trauma area h1, the boundary area h2, and the normal skin area h3, and generating corresponding infrared thermography thermal imaging images and functions AC. Correlation imaging is performed on the wound images to obtain functions BC, and optical pulse signal extraction results and imaging corresponding to the trauma area h1, the boundary area h2, and the normal skin area h3 are generated.

[0012] The third step is to perform pairwise comparison analysis on the functions AC and BC to obtain three AC difference images and three BC difference images. The AC difference images and BC difference images in the same comparison area are merged and recorded as Δ1, Δ2, and Δ3. These images, along with the maximum width w, maximum length l, and maximum depth d of the wound, are stored in the dataset O[i]. If the data measurement sequence number i>1, the difference rate Ω between the two datasets with adjacent sequence numbers in dataset O[1] to O[i] is calculated.

[0013] The fourth step involves performing coordinate alignment and adversarial analysis on the AC difference image and BC difference image. The BC difference image is used as the original data and the AC difference image is used as the corrected data. The data is then input into the neural adversarial network for analysis and comparison to generate the CC difference image. Based on the CC difference image, the maximum width w, the maximum length l, and the maximum depth d of the wound, the medical database corresponding to the wound type is queried to obtain the wound severity assessment parameter G. When the data measurement sequence number i > 1, the difference rate Ω and the data measurement sequence number i are used to perform regression analysis to establish a recovery function and generate the predicted wound severity assessment parameter G2. G[i] and G2[i-1] are compared. If the difference between the two is outside the normal range, the alarm signal is recorded as 1; otherwise, it is recorded as 0.

[0014] When the data measurement sequence number i is not zero, the maximum width w, maximum length l, maximum depth d of the wound, the CC difference image, the wound severity assessment parameter G, and the alarm signal are visualized and output.

[0015] Preferably, the dataset of the injured body parts obtained by scanning is then processed by an infrared thermometer, a 3D laser scanner for 3D modeling, and IPPG signal processing of the video signals in the dataset to obtain the temperature dataset Tnew, point cloud dataset Snew, original pulse wave signal Mnew, and data measurement sequence number i of the injured area. Specifically:

[0016] The temperature dataset Tnew consists of the number of high-temperature patches, the number of low-temperature patches, and area parameters.

[0017] The point cloud dataset Snew is a point cloud dataset obtained using dual optical comb 3D modeling;

[0018] The raw pulse wave signal Mnew is a preliminary pulse wave signal obtained by analyzing the video signal using the LK optical flow method;

[0019] If the body part dataset does not contain any external injuries, the measurement number i is recorded as 0. If there are external injuries, the existing database is searched. If there are no previous wound dataset records for the body part, the measurement number i is recorded as 1. If there are previous wound dataset records, the measurement number i is recorded as the existing maximum value + 1.

[0020] Preferably, the point cloud dataset Snew is input to reconstruct the three-dimensional model of the trauma area. The wound image is extracted using the video signal, and the maximum width w, maximum length l, maximum depth d, average depth, and depth variance of the wound are calculated. The trauma area h1, boundary area h2, and normal skin area h3 are then divided, and the wound type is determined. Specifically:

[0021] A three-dimensional coordinate system was established for the point cloud dataset Snew, and the coordinate system was corrected by measuring the point cloud dataset with index i=0, i.e., when there is no external injury to the skin.

[0022] The point cloud dataset Snew is divided into blocks, and the depth coordinates, variance, and maximum value of each block are calculated. The maximum value is recorded as the maximum depth d. Statistical analysis is performed on continuous blocks with similar depths to obtain the width and length of the three-dimensional wound.

[0023] A two-dimensional plane coordinate system is established by extracting wound images from the video signal. The width and length of the planar wound are determined based on the color distribution of pixels in the wound image. The width and length of the planar wound are corrected using the width and length of the three-dimensional wound to obtain the maximum width w and the maximum length l of the wound.

[0024] Based on the depth variance of each block, the block can be classified into trauma area h1, boundary area h2, and normal skin area h3;

[0025] The wound type is determined based on the maximum width w, maximum length l, maximum depth d, average depth, and depth variance of the wound.

[0026] Preferably, the data in the temperature dataset Tnew and the point cloud dataset Snew are grouped according to the trauma region h1, boundary region h2, and normal skin region h3, and corresponding infrared thermography thermal imaging images and functions AC are generated. Correlation imaging is performed on the wound images to obtain function BC, and optical pulse signal extraction results and imaging corresponding to the trauma region h1, boundary region h2, and normal skin region h3 are generated. Specifically:

[0027] Infrared thermography thermal imaging images are generated based on the temperature dataset. Points in the image represent the temperature at the corresponding location, and the function AC is the average temperature within the corresponding temperature patch.

[0028] The wound image is traversed using a sliding window to obtain the Spearman correlation coefficients of the signals S and R of each window and the global spatial pixel average signal, respectively, for correlation topographic imaging. Points in this image represent pulse waves at corresponding locations. The function BC corresponds to the Spearman correlation coefficient, and its expression is as follows:

[0029]

[0030] Where N is the number of samples of signal S, and the signals S are sorted in ascending order, S' is the sorted set of S, i.e., S' n For S n The rank in S, Let S′ be the mean of S′. Similarly, R′ is the sorted set of R. n For R n In R, the level Let R' be the mean.

[0031] Preferably, the functions AC and BC are subjected to pairwise comparison analysis in partitions to obtain three AC difference images and three BC difference images, specifically:

[0032] The function AC is partitioned and compared pairwise. First, the average temperature of the normal skin region h3 is calculated, and then the temperature difference rate between the trauma region h1 and the boundary region h2 is calculated to obtain three AC difference images.

[0033] The function BC is divided into two partitions for pairwise comparison. First, the average BC value of the normal skin region h3 is calculated, and then the difference rate of BC values ​​between the trauma region h1 and the boundary region h2 is calculated to obtain three BC difference images.

[0034] Preferably, the maximum width w, maximum length l, maximum depth d of the wound, the CC difference image, the wound severity assessment parameter G, and the alarm signal are visualized and output, specifically as follows:

[0035] The wound image is divided into injury severity zones based on the wound severity assessment parameter G, and wound images with zone color markings and annotations are generated by combining the wound images.

[0036] Output the image, the maximum width w, the maximum length l, and the maximum depth d of the wound. If the alarm signal is 1, output "healing is normal"; if the alarm signal is 0, output "healing is abnormal".

[0037] Accordingly, the present invention also provides a system for monitoring wound healing, comprising:

[0038] The data input unit is used to scan and obtain the dataset of the injured body parts. After being detected by an infrared thermometer, 3D modeled by a 3D laser scanner, and processed by IPPG signal processing of the video signal in the dataset of the injured body parts, the temperature dataset Tnew, point cloud dataset Snew, original pulse wave signal Mnew and data measurement sequence number i of the injured area are obtained.

[0039] The wound data calculation and judgment unit performs the following four steps when the data measurement sequence number i is not zero: Step 1: Input the point cloud dataset Snew, reproduce the three-dimensional model of the trauma area, extract the wound image using the video signal, and calculate the maximum width w, maximum length l, maximum depth d, average depth, and depth variance of the wound. Divide the wound area into trauma area h1, boundary area h2, and normal skin area h3, and obtain the wound type. Step 2: Group the data in the temperature dataset Tnew and the point cloud dataset Snew according to the trauma area h1, boundary area h2, and normal skin area h3, and generate corresponding infrared thermography thermal imaging images and function AC. Perform correlation imaging on the wound image and obtain function BC. Generate the optical pulse signal extraction results and imaging corresponding to the trauma area h1, boundary area h2, and normal skin area h3. Step 3: Perform pairwise comparison analysis on the function AC and the function BC, obtaining three AC difference images and three BC difference images, and group the same comparison areas... The AC difference image and BC difference image are merged and recorded as Δ1, Δ2, and Δ3. They are stored in dataset O[i] along with the maximum width w, maximum length l, and maximum depth d of the wound. If the data measurement sequence number i>1, the difference rate Ω between the two datasets with adjacent sequence numbers in dataset O[1] to O[i] is calculated. In the fourth step, the AC difference image and BC difference image are subjected to coordinate alignment and data adversarial analysis. The BC difference image is used as the original data and the AC difference image is used as the corrected data. The data is input into the neural adversarial network for analysis and comparison to generate the CC difference image. Based on the CC difference image, the maximum width w, maximum length l, and maximum depth d of the wound, the medical database corresponding to the wound type is queried to obtain the wound severity assessment parameter G. When the data measurement sequence number i>1, the difference rate Ω and the data measurement sequence number i are subjected to regression analysis to establish a recovery function and generate the predicted wound severity assessment parameter G2. G[i] and G2[i-1] are compared. If the difference between the two is outside the normal range, the alarm signal is recorded as 1; otherwise, it is recorded as 0.

[0040] The visualization output unit is used to visualize the maximum width w, maximum length l, maximum depth d, CC difference image, wound severity assessment parameter G, and alarm signal when the data measurement sequence number i is not zero.

[0041] Implementing this invention has the following beneficial effects:

[0042] This invention proposes a method and system for monitoring wound healing. The invention comprehensively utilizes multiple technologies, simultaneously introducing multiple parameters. It uses temperature, IPPG blood perfusion imaging, and dual-comb 3D scanning modeling data for mutual reference and correction. Simultaneously, it uses data from the patient's own wound and surrounding intact areas as reference comparison data, introducing two reference coordinate axes—time and space—to more comprehensively monitor wound conditions. Attached Figure Description

[0043] Figure 1 This is a flowchart of a method for monitoring trauma healing according to an embodiment of the present invention;

[0044] Figure 2 This is a structural diagram of a system for monitoring trauma healing according to an embodiment of the present invention. Detailed Implementation

[0045] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0046] Figure 1 This is a flowchart of a method for monitoring wound healing according to an embodiment of the present invention, such as... Figure 1 As shown, the method includes:

[0047] S1, scan to obtain the dataset of the injured body parts, after detection by infrared thermometer, three-dimensional modeling by three-dimensional laser scanner, and IPPG signal processing of video signals in the dataset of the injured body parts, the temperature dataset Tnew, point cloud dataset Snew, original pulse wave signal Mnew and data measurement sequence number i of the injured area are obtained;

[0048] S2, when the data measurement sequence number i is not zero, the following four steps are performed:

[0049] S2-1, Input the point cloud dataset Snew, reproduce the three-dimensional model of the trauma area, extract the wound image using the video signal, and calculate the maximum width w, maximum length l, maximum depth d, average depth and depth variance of the wound, and divide the trauma area h1, boundary area h2 and normal skin area h3, and obtain the wound type.

[0050] S2-2, group the data in the temperature dataset Tnew and the point cloud dataset Snew according to the trauma area h1, the boundary area h2, and the normal skin area h3, and generate the corresponding infrared thermography thermal imaging image and function AC. Perform correlation imaging on the wound image and obtain function BC. Generate the optical pulse signal extraction results and imaging corresponding to the trauma area h1, the boundary area h2, and the normal skin area h3.

[0051] S2-3, perform pairwise comparison analysis on the functions AC and BC to obtain three AC difference images and three BC difference images. Merge the AC difference images and BC difference images in the same comparison area and record them as Δ1, Δ2, and Δ3. Store them in the dataset O[i] along with the maximum width w, maximum length l, and maximum depth d of the wound. If the data measurement sequence number i>1, calculate the difference rate Ω between the two datasets with adjacent sequence numbers in dataset O[1] to O[i].

[0052] S2-4, perform coordinate alignment and data adversarial analysis on the AC difference image and BC difference image. Input the BC difference image as the original data and the AC difference image as the corrected data into the neural adversarial network for analysis and comparison to generate the CC difference image. Based on the CC difference image, the maximum width w, the maximum length l, and the maximum depth d of the wound, query the medical database corresponding to the wound type to obtain the wound severity assessment parameter G. When the data measurement sequence number i>1, perform regression analysis on the difference rate Ω and the data measurement sequence number i to establish a recovery function and generate the predicted wound severity level assessment parameter G2. Compare G[i] and G2[i-1]. If the difference between the two is outside the normal range, the alarm signal is recorded as 1; otherwise, it is recorded as 0.

[0053] S3, when the data measurement sequence number i is not zero, the maximum width w, maximum length l, maximum depth d of the wound, the CC difference image, the wound severity assessment parameter G, and the alarm signal are visualized and output.

[0054] Step S1 is as follows:

[0055] S1-1, the temperature dataset Tnew, which includes the number of high-temperature patches, the number of low-temperature patches, and area parameters;

[0056] S1-2, the point cloud dataset Snew is a point cloud dataset obtained by using dual optical comb 3D modeling;

[0057] S1-3, the original pulse wave signal Mnew is a preliminary pulse wave signal obtained by analyzing the video signal using the LK optical flow method;

[0058] S1-4, if the body part dataset does not contain any external injuries, the measurement sequence number i is recorded as 0. If there are external injuries, the existing database is searched. If there are no previous wound dataset records for the body part, the measurement sequence number i is recorded as 1. If there are previous wound dataset records, the measurement sequence number i is recorded as the existing maximum value + 1.

[0059] Step S2-1 is as follows:

[0060] S2-1-1: After extracting the point cloud dataset Snew, a three-dimensional coordinate system is established with a distance of 10mm from the wound as the reference. The coordinates of the point set within the measurement range are established with an accuracy of 10 micrometers. In order to eliminate the influence of uneven skin surface on the data, the coordinates of the point set stored in the database under normal and non-traumatic conditions are compared and overlapped. The overlapping uneven skin data are unified into flat coordinates. Note that the measurement angle and the establishment position must be the same during processing. The coordinate system is corrected by measuring the point cloud dataset when the sequence number i=0, that is, when there is no external injury to the skin. This step is mainly to consider that the original uneven area of ​​the skin surface may interfere with the judgment of the skin surface wound data under high precision. In this step, the coordinate point cloud data recorded by Snew is used for matching, and the overlapping part of the coordinates in i=0 is called to eliminate the depressions at the same coordinate position.

[0061] S2-1-2, the point cloud dataset Snew is divided into blocks, and the depth coordinates, variance and maximum value of each block are calculated. The maximum value is recorded as the maximum depth d. Statistical analysis is performed on continuous blocks with similar depths to obtain the width and length of the three-dimensional wound.

[0062] S2-1-3, using the video signal to extract the wound image and establish a two-dimensional plane coordinate system, determining the width and length of the plane wound based on the color distribution of the pixels in the wound image, and correcting the width and length of the plane wound with the width and length of the three-dimensional wound to obtain the maximum width w and the maximum length l of the wound;

[0063] S2-1-4, based on the depth variance of each block, the block can be classified into trauma area h1, boundary area h2, and normal skin area h3;

[0064] S2-1-5, the wound type is obtained based on the maximum width w, maximum length l, maximum depth d, average depth, and depth variance of the wound.

[0065] Step S2-2 is as follows:

[0066] S2-2-1, Generate an infrared thermographic image based on the temperature dataset. The points in the image represent the temperature at the corresponding location, and the function AC is the average temperature within the corresponding temperature patch.

[0067] S2-2-2, the wound image is traversed using a sliding window to obtain the Spearman correlation coefficients of each window signal S and the global spatial pixel average signal R for correlation topographic imaging. Points in this image represent pulse waves at corresponding locations. The function BC corresponds to the Spearman correlation coefficient, and its expression is:

[0068]

[0069] Where N is the number of samples of signal S, and the signals S are sorted in ascending order, S' is the sorted set of S, i.e., S' n For S n The rank in S, Let S′ be the mean of S′. Similarly, R′ is the sorted set of R. n For R n In R, the level For R ′ The mean.

[0070] Steps S2-3 are as follows:

[0071] S2-3-1, the function AC is partitioned and compared pairwise, that is, first the average temperature of the normal skin area h3 is calculated, and then the temperature difference rate of the trauma area h1 and the boundary area h2 is calculated to obtain three AC difference images.

[0072] S2-3-2, the function BC is partitioned and compared pairwise. That is, the average BC value of the normal skin region h3 is first calculated, and then the difference rate of BC values ​​between the trauma region h1 and the boundary region h2 is calculated to obtain three BC difference images.

[0073] Step S3 is as follows:

[0074] S3-1, Divide the wound image into injury severity partitions according to the wound severity assessment parameter G, and generate wound images with partition color markings and annotations based on the wound images;

[0075] S3-2, output the image, the maximum width w, the maximum length l, and the maximum depth d of the wound. If the alarm signal is 1, output normal healing; if the alarm signal is 0, output abnormal healing.

[0076] Accordingly, the present invention also provides a method and system for monitoring wound healing, such as... Figure 2 As shown, it includes:

[0077] Data input unit 1 is used to scan and obtain the dataset of the injured body parts. After being detected by an infrared thermometer, 3D modeled by a 3D laser scanner, and processed by IPPG signal processing of the video signal in the dataset of the injured body parts, the temperature dataset Tnew, point cloud dataset Snew, original pulse wave signal Mnew and data measurement sequence number i of the injured area are obtained.

[0078] Specifically, the temperature dataset Tnew consists of the number of high-temperature patches, the number of low-temperature patches, and area parameters;

[0079] The point cloud dataset Snew is a point cloud dataset obtained using dual optical comb 3D modeling;

[0080] The raw pulse wave signal Mnew is a preliminary pulse wave signal obtained by analyzing the video signal using the LK optical flow method;

[0081] If the body part dataset does not contain any external injuries, the measurement number i is recorded as 0. If there are external injuries, the existing database is searched. If there are no previous wound dataset records for the body part, the measurement number i is recorded as 1. If there are previous wound dataset records, the measurement number i is recorded as the existing maximum value + 1.

[0082] The wound data calculation and judgment unit 2 is used to perform the following four steps when the data measurement sequence number i is not zero: First, input the point cloud dataset Snew, reproduce the three-dimensional model of the trauma area, extract the wound image using the video signal, and calculate the maximum width w, maximum length l, maximum depth d, average depth, and depth variance of the wound. Then, divide the trauma area h1, boundary area h2, and normal skin area h3, and obtain the wound type. Second, group the data in the temperature dataset Tnew and the point cloud dataset Snew according to the trauma area h1, boundary area h2, and normal skin area h3, and generate corresponding infrared thermography thermal imaging images and functions AC. Perform correlation imaging on the wound image and obtain function BC. Generate the optical pulse signal extraction results and imaging corresponding to the trauma area h1, boundary area h2, and normal skin area h3. Third, perform pairwise comparison analysis on the functions AC and BC, obtaining three AC difference images and three BC difference images, and grouping the same comparison areas... The AC difference image and BC difference image of the domain are merged and recorded as Δ1, Δ2, Δ3. They are stored in dataset O[i] along with the maximum width w, maximum length l, and maximum depth d of the wound. If the data measurement sequence number i>1, the difference rate Ω between the two datasets with adjacent sequence numbers in dataset O[1] to O[i] is calculated. In the fourth step, the AC difference image and BC difference image are subjected to coordinate alignment and data adversarial analysis. The BC difference image is used as the original data and the AC difference image is used as the corrected data. The data is input into the neural adversarial network for analysis and comparison to generate the CC difference image. Based on the CC difference image, the maximum width w, maximum length l, and maximum depth d of the wound, the medical database corresponding to the wound type is queried to obtain the wound severity assessment parameter G. When the data measurement sequence number i>1, the difference rate Ω and the data measurement sequence number i are subjected to regression analysis to establish a recovery function and generate the predicted wound severity assessment parameter G2. G[i] and G2[i-1] are compared. If the difference between the two is outside the normal range, the alarm signal is recorded as 1; otherwise, it is recorded as 0.

[0083] Specifically, a three-dimensional coordinate system is established for the point cloud dataset Snew, and the coordinate system is corrected by measuring the point cloud dataset with index i=0, i.e., when there is no external injury to the skin.

[0084] The point cloud dataset Snew is divided into blocks, and the depth coordinates, variance, and maximum value of each block are calculated. The maximum value is recorded as the maximum depth d. Statistical analysis is performed on continuous blocks with similar depths to obtain the width and length of the three-dimensional wound.

[0085] A two-dimensional plane coordinate system is established by extracting wound images from the video signal. The width and length of the planar wound are determined based on the color distribution of pixels in the wound image. The width and length of the planar wound are corrected using the width and length of the three-dimensional wound to obtain the maximum width w and the maximum length l of the wound.

[0086] Based on the depth variance of each block, the block can be classified into trauma area h1, boundary area h2, and normal skin area h3;

[0087] The wound type is obtained based on the maximum width w, maximum length l, maximum depth d, average depth, and depth variance of the wound.

[0088] Infrared thermography thermal imaging images are generated based on the temperature dataset. Points in the image represent the temperature at the corresponding location, and the function AC is the average temperature within the corresponding temperature patch.

[0089] The wound image is traversed using a sliding window to obtain the Spearman correlation coefficients of the signals S and R of each window and the global spatial pixel average signal, respectively, for correlation topographic imaging. Points in this image represent pulse waves at corresponding locations. The function BC corresponds to the Spearman correlation coefficient, and its expression is as follows:

[0090]

[0091] Where N is the number of samples of signal S, and the signals S are sorted in ascending order, S' is the sorted set of S, i.e., S' n For S n The rank in S, Let S′ be the mean of S′. Similarly, R′ is the sorted set of R. n For R n In R, the level Let R' be the mean;

[0092] The function AC is partitioned and compared pairwise. First, the average temperature of the normal skin region h3 is calculated, and then the temperature difference rate between the trauma region h1 and the boundary region h2 is calculated to obtain three AC difference images.

[0093] The function BC is divided into two partitions for pairwise comparison. First, the average BC value of the normal skin region h3 is calculated, and then the difference rate of BC values ​​between the trauma region h1 and the boundary region h2 is calculated to obtain three BC difference images.

[0094] The visualization output unit 3 is used to visualize the maximum width w, maximum length l, maximum depth d, CC difference image, wound severity assessment parameter G and alarm signal when the data measurement sequence number i is not zero.

[0095] Specifically, the wound image is divided into injury severity zones according to the wound severity assessment parameter G, and wound images with zone color markings and annotations are generated in combination with the wound images;

[0096] Output the image, the maximum width w, the maximum length l, and the maximum depth d of the wound. If the alarm signal is 1, output "healing is normal"; if the alarm signal is 0, output "healing is abnormal".

[0097] Therefore, this invention achieves comprehensive utilization of technology, simultaneously introducing multiple parameters, and simultaneously using temperature, IPPG blood perfusion imaging, and dual-comb three-dimensional scanning modeling data for mutual reference and correction. At the same time, it uses data from the wound of the trauma data source and the surrounding intact area as reference comparison data, introducing two reference coordinate axes of time and space to more comprehensively monitor the wound condition.

[0098] The method and system for monitoring trauma healing provided by the embodiments of the present invention have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for monitoring wound healing, characterized in that, The method includes: The dataset of the injured body parts was obtained by scanning. After detection by an infrared thermometer, 3D modeling by a 3D laser scanner, and IPPG signal processing of the video signal in the dataset of the injured body parts, the temperature dataset Tnew, point cloud dataset Snew, original pulse wave signal Mnew and data measurement sequence number i of the injured area were obtained. When the data measurement sequence number i is not zero, the following four steps are performed: The first step is to input the point cloud dataset Snew, reproduce the three-dimensional model of the trauma area, extract the wound image using the video signal, and calculate the maximum width w, maximum length l, maximum depth d, average depth and depth variance of the wound. Then, divide the trauma area h1, the boundary area h2 and the normal skin area h3, and obtain the wound type. The second step involves grouping the data in the temperature dataset Tnew and the point cloud dataset Snew according to the trauma area h1, the boundary area h2, and the normal skin area h3, and generating corresponding infrared thermography thermal imaging images and functions AC. The points in the thermal imaging images represent the temperature at the corresponding locations, and the function AC corresponds to the average temperature within the temperature patch. Correlation imaging is performed on the wound images to obtain the function BC, which corresponds to the Spearman correlation coefficient. The optical pulse signal extraction results and imaging corresponding to the trauma area h1, the boundary area h2, and the normal skin area h3 are then generated. The third step is to perform pairwise comparison analysis on the functions AC and BC to obtain three AC difference images and three BC difference images. The AC difference images and BC difference images in the same comparison area are merged and recorded as Δ1, Δ2, and Δ3. These images, along with the maximum width w, maximum length l, and maximum depth d of the wound, are stored in the dataset O[i]. If the data measurement sequence number i>1, the difference rate Ω between the two datasets with adjacent sequence numbers in dataset O[1] to O[i] is calculated. The fourth step involves performing coordinate alignment and adversarial analysis on the AC difference image and BC difference image. The BC difference image is used as the original data and the AC difference image is used as the corrected data. The data is then input into the neural adversarial network for analysis and comparison to generate the CC difference image. Based on the CC difference image, the maximum width w, the maximum length l, and the maximum depth d of the wound, the medical database corresponding to the wound type is queried to obtain the wound severity assessment parameter G. When the data measurement sequence number i > 1, the difference rate Ω and the data measurement sequence number i are used to perform regression analysis to establish a recovery function and generate the predicted wound severity assessment parameter G2. G[i] and G2[i-1] are compared. If the difference between the two is outside the normal range, the alarm signal is recorded as 1; otherwise, it is recorded as 0. When the data measurement sequence number i is not zero, the maximum width w, maximum length l, maximum depth d of the wound, the CC difference image, the wound severity assessment parameter G, and the alarm signal are visualized and output.

2. The method for monitoring wound healing as described in claim 1, characterized in that, The scan yields a dataset of the injured body parts. After detection by an infrared thermometer, 3D modeling by a 3D laser scanner, and IPPG signal processing of the video signals from the dataset, the following data are obtained: temperature dataset Tnew, point cloud dataset Snew, original pulse wave signal Mnew, and data measurement sequence number i of the injured area. The temperature dataset Tnew consists of the number of high-temperature patches, the number of low-temperature patches, and area parameters. The point cloud dataset Snew is a point cloud dataset obtained using dual optical comb 3D modeling; The raw pulse wave signal Mnew is a preliminary pulse wave signal obtained by analyzing the video signal using the LK optical flow method; If the body part dataset does not contain any external injuries, the measurement number i is recorded as 0. If there are external injuries, the existing database is searched. If there are no previous wound dataset records for the body part, the measurement number i is recorded as 1. If there are previous wound dataset records, the measurement number i is recorded as the existing maximum value + 1.

3. The method for monitoring wound healing as described in claim 1, characterized in that, The input point cloud dataset Snew is used to reconstruct the 3D model of the trauma area. Wound images are extracted using the video signal, and the maximum width w, maximum length l, maximum depth d, average depth, and depth variance of the wound are calculated. The trauma area h1, boundary area h2, and normal skin area h3 are then divided, and the wound type is determined. Specifically: A three-dimensional coordinate system was established for the point cloud dataset Snew, and the coordinate system was corrected by measuring the point cloud dataset with index i=0, i.e., when there was no external injury to the skin. The point cloud dataset Snew is divided into blocks, and the depth coordinates, variance, and maximum value of each block are calculated. The maximum value is recorded as the maximum depth d. Statistical analysis is performed on continuous blocks with similar depths to obtain the width and length of the three-dimensional wound. A two-dimensional plane coordinate system is established by extracting wound images from the video signal. The width and length of the planar wound are determined based on the color distribution of pixels in the wound image. The width and length of the planar wound are corrected using the width and length of the three-dimensional wound to obtain the maximum width w and the maximum length l of the wound. Based on the depth variance of each block, the block can be classified into trauma area h1, boundary area h2, and normal skin area h3; The wound type is determined based on the maximum width w, maximum length l, maximum depth d, average depth, and depth variance of the wound.

4. The method for monitoring wound healing as described in claim 1, characterized in that, The data in the temperature dataset Tnew and the point cloud dataset Snew are grouped according to the trauma region h1, boundary region h2, and normal skin region h3, and corresponding infrared thermographic images and functions AC are generated. Points in the thermal imaging images represent the temperature at the corresponding location, and function AC corresponds to the average temperature within the temperature patch. Correlation imaging is performed on the wound image to obtain function BC, which corresponds to the Spearman correlation coefficient. The optical pulse signal extraction results and images corresponding to the trauma region h1, boundary region h2, and normal skin region h3 are generated. Specifically: Generate an infrared thermographic image based on the temperature dataset; The wound image is traversed using a sliding window to obtain the Spearman correlation coefficients of the signals S and R of each window and the global spatial pixel average signal, respectively, for correlation topographic imaging. Points in this image represent pulse waves at corresponding locations. The function BC is expressed as: ， Where N is the number of samples of signal S, and the signals S are sorted in ascending order, with S' serving as the sorted set of S, i.e. for The rank in S, for The mean, similarly, As a sorted set of R, for In R, the level for The mean.

5. The method for monitoring wound healing as described in claim 1, characterized in that, The step involves performing pairwise comparison analysis on the functions AC and BC, resulting in three AC difference images and three BC difference images. Specifically: The function AC is partitioned and compared pairwise. First, the average temperature of the normal skin region h3, the trauma region h1, and the boundary region h2 is calculated. Then, the difference rate of the average temperature between h3 and h1, h3 and h2, and h1 and h2 is calculated to obtain three AC difference images. The function BC is divided into two partitions for pairwise comparison. First, the average BC value of the normal skin region h3, the trauma region h1, and the boundary region h2 is calculated. Then, the difference rate of the average BC values ​​between h3 and h1, h3 and h2, and h1 and h2 is calculated to obtain three BC difference images.

6. A system for monitoring wound healing, characterized in that, The system includes: The data input unit is used to scan and obtain the dataset of the injured body parts. After being detected by an infrared thermometer, 3D modeled by a 3D laser scanner, and processed by IPPG signal processing of the video signal in the dataset of the injured body parts, the temperature dataset Tnew, point cloud dataset Snew, original pulse wave signal Mnew and data measurement sequence number i of the injured area are obtained. The wound data calculation and judgment unit performs the following four steps when the data measurement sequence number i is not zero: Step 1: Input the point cloud dataset Snew, reproduce the three-dimensional model of the trauma area, extract the wound image using the video signal, and calculate the maximum width w, maximum length l, maximum depth d, average depth, and depth variance of the wound. Divide the wound area into trauma area h1, boundary area h2, and normal skin area h3, and obtain the wound type. Step 2: Group the data in the temperature dataset Tnew and the point cloud dataset Snew according to the trauma area h1, boundary area h2, and normal skin area h3, and generate corresponding infrared thermography thermal imaging images and functions AC. Points in the thermal imaging image represent the temperature at the corresponding location, and function AC corresponds to the average temperature within the temperature patch. Perform correlation imaging on the wound image and obtain function BC, which corresponds to the Spearman correlation coefficient. Generate the optical pulse signal extraction results and imaging corresponding to the trauma area h1, boundary area h2, and normal skin area h3. Step 3: Perform pairwise comparison analysis on functions AC and BC to obtain... Three AC difference images and three BC difference images are obtained. The AC difference images and BC difference images with the same comparison area are merged and recorded as Δ1, Δ2, and Δ3. These are stored in dataset O[i] along with the maximum width w, maximum length l, and maximum depth d of the wound. If the data measurement sequence number i>1, the difference rate Ω between two datasets with adjacent sequence numbers in dataset O[1] to O[i] is calculated. In the fourth step, the AC difference images and BC difference images are subjected to coordinate alignment and data adversarial analysis. The BC difference image is used as the original data and the AC difference image is used as the correction. Data is input into a neural adversarial network for analysis and comparison to generate a CC difference image. Based on the CC difference image, the maximum width w, the maximum length l, and the maximum depth d of the wound, the medical database corresponding to the wound type is queried to obtain the wound severity assessment parameter G. When the data measurement number i>1, the difference rate Ω and the data measurement number i are used to perform regression analysis to establish a recovery function and generate the predicted wound severity assessment parameter G2. G[i] and G2[i-1] are compared. If the difference between the two is outside the normal range, the alarm signal is recorded as 1; otherwise, it is recorded as 0. The visualization output unit is used to visualize the maximum width w, maximum length l, maximum depth d, CC difference image, wound severity assessment parameter G, and alarm signal when the data measurement sequence number i is not zero.

7. The system for monitoring external wound healing as described in claim 6, characterized in that, The data input unit is used to scan and obtain a dataset of the injured body parts. After detection by an infrared thermometer, 3D modeling by a 3D laser scanner, and IPPG signal processing of the video signal in the dataset of the injured body parts, the following data are obtained: temperature dataset Tnew, point cloud dataset Snew, original pulse wave signal Mnew, and data measurement sequence number i of the injured area. Specifically: The temperature dataset Tnew consists of the number of high-temperature patches, the number of low-temperature patches, and area parameters. The point cloud dataset Snew is a point cloud dataset obtained using dual optical comb 3D modeling; The raw pulse wave signal Mnew is a preliminary pulse wave signal obtained by analyzing the video signal using the LK optical flow method; If the body part dataset does not contain any external injuries, the measurement number i is recorded as 0. If there are external injuries, the existing database is searched. If there are no previous wound dataset records for the body part, the measurement number i is recorded as 1. If there are previous wound dataset records, the measurement number i is recorded as the existing maximum value + 1.

8. The system for monitoring wound healing as described in claim 6, characterized in that, The wound data calculation and judgment unit is used to perform the following four steps when the data determination sequence number i is not zero: First, input the point cloud dataset Snew, reproduce the three-dimensional model of the trauma area, extract the wound image using the video signal, and calculate the maximum width w, maximum length l, maximum depth d, average depth, and depth variance of the wound. Then, divide the trauma area h1, boundary area h2, and normal skin area h3, and obtain the wound type. Second, group the data in the temperature dataset Tnew and the point cloud dataset Snew according to the trauma area h1, boundary area h2, and normal skin area h3, and generate corresponding infrared thermography thermal imaging images and functions AC. Points in the thermal imaging image represent the temperature at the corresponding location, and function AC corresponds to the average temperature within the temperature patch. Perform correlation imaging on the wound image and obtain function BC, which corresponds to the Spearman correlation coefficient. Generate the optical pulse signal extraction results and imaging corresponding to the trauma area h1, boundary area h2, and normal skin area h3. Third, perform pairwise comparison analysis on the functions AC and BC to obtain... Three AC difference images and three BC difference images are used. The AC difference images and BC difference images with the same comparison area are merged and recorded as Δ1, Δ2, and Δ3. These are stored in dataset O[i] along with the maximum width w, maximum length l, and maximum depth d of the wound. If the data measurement index i>1, the difference rate Ω between two datasets with adjacent indices in dataset O[1] and O[i] is calculated. In the fourth step, the AC difference images and BC difference images are subjected to coordinate alignment and data adversarial analysis. The BC difference image is used as the original data and the AC difference image is used as the correction data. The neural adversarial network is used for analysis and comparison to generate a CC difference image. Based on the CC difference image, the maximum width w, maximum length l, and maximum depth d of the wound, the medical database corresponding to the wound type is queried to obtain the wound severity assessment parameter G. When the data measurement number i > 1, the difference rate Ω and the data measurement number i are used to perform regression analysis to establish a recovery function and generate the predicted wound severity assessment parameter G2. G[i] and G2[i-1] are compared. If the difference between the two is outside the normal range, the alarm signal is recorded as 1; otherwise, it is recorded as 0. Specifically: A three-dimensional coordinate system was established for the point cloud dataset Snew, and the coordinate system was corrected by measuring the point cloud dataset with index i=0, i.e., when there was no external injury to the skin. The point cloud dataset Snew is divided into blocks, and the depth coordinates, variance, and maximum value of each block are calculated. The maximum value is recorded as the maximum depth d. Statistical analysis is performed on continuous blocks with similar depths to obtain the width and length of the three-dimensional wound. A two-dimensional plane coordinate system is established by extracting wound images from the video signal. The width and length of the planar wound are determined based on the color distribution of pixels in the wound image. The width and length of the planar wound are corrected using the width and length of the three-dimensional wound to obtain the maximum width w and the maximum length l of the wound. Based on the depth variance of each block, the block can be classified into trauma area h1, boundary area h2, and normal skin area h3; The wound type is obtained based on the maximum width w, maximum length l, maximum depth d, average depth, and depth variance of the wound. Generate an infrared thermographic image based on the temperature dataset; The wound image is traversed using a sliding window to obtain the Spearman correlation coefficients of the signals S and R of each window and the global spatial pixel average signal, respectively, for correlation topographic imaging. Points in this image represent pulse waves at corresponding locations. The function BC is expressed as: ， Where N is the number of samples of signal S, and the signals S are sorted in ascending order, with S' serving as the sorted set of S, i.e. for The rank in S, for The mean, similarly, As a sorted set of R, for In R, the level for The mean; The function AC is partitioned and compared pairwise. First, the average temperature of the normal skin region h3, the trauma region h1, and the boundary region h2 is calculated. Then, the difference rate of the average temperature between h3 and h1, h3 and h2, and h1 and h2 is calculated to obtain three AC difference images. The function BC is divided into two partitions for pairwise comparison. First, the average BC value of the normal skin region h3, the trauma region h1, and the boundary region h2 is calculated. Then, the difference rate of the average BC values ​​between h3 and h1, h3 and h2, and h1 and h2 is calculated to obtain three BC difference images.

Images