Stress damage detection method and system based on artificial intelligence and medium

Abstract

The invention relates to a pressure damage detection method and system based on artificial intelligence, and a medium. The method comprises the following steps: respectively obtaining an infrared source image and a microwave source image corresponding to a target position; performing feature extraction on the infrared source image and the microwave source image to obtain corresponding infrared feature information and microwave feature information; based on an adversarial network, fusing the infrared feature information and the microwave feature information to obtain fused feature information; and based on the fused feature information, extracting a corresponding pressure damage feature value to obtain detection information of the pressure damage of the target position, and performing pressure damage grading on the target position based on the detection information. According to the method, the infrared image is combined with the microwave image, the images are subjected to feature extraction and feature fusion, the regular change of skin temperature distribution before and after the injury is detected by using an infrared technology, the hidden deep tissue injury is predicted, and grading is performed, so that the evaluation error of medical personnel is reduced, and the problem of non-uniform evaluation standards is solved.

Description

Technical Field

[0001] The present invention relates to the field of computer information, and in particular to a pressure injury auxiliary detection method, system and computer storage medium. Background Art

[0002] In clinical medicine, pressure injuries, also known as pressure sores, are caused by long-term pressure on local tissues, which leads to persistent ischemia, hypoxia, and malnutrition, resulting in tissue ulceration and necrosis. Pressure injuries have a long treatment cycle and seriously affect the quality of life of patients, which is a huge challenge facing medical care. How to detect and intervene in deep tissue injuries at an early stage is a necessary step to prevent skin pressure injuries. At present, the Braden pressure injury risk assessment form and visual and tactile inspection methods are commonly used in clinical assessment. Occasionally, ultrasound monitoring, infrared thermal imaging, or subcutaneous moisture measurement scanners are used for assessment.

[0003] The Braden pressure injury risk assessment form and visual tactile examination method currently commonly used in clinical practice are essentially based on tissue defects visible to the naked eye or directly palpated, and cannot obtain comprehensive information about the injury. This method has certain difficulties in accurately identifying pressure injuries.

[0004] The basic principle of ultrasound detection is that when local tissues are damaged, such as inflammation and edema, ultrasound detection will reveal some abnormal signs such as weakened local tissue reflectivity and low echo areas in the skin and subcutaneous tissues. However, ultrasound detection requires a bedside ultrasound machine, which is relatively expensive. Secondly, the pressure of the ultrasound probe will cause obvious differences in the hierarchical structure of the skin and subcutaneous tissue in the image, which places high demands on the operator's skills and is not widely used.

[0005] The basic principle of infrared detection is that due to ischemia or insufficient blood flow on the surface of the human skin, thermal imaging will show low temperatures, and if the deep tissue is damaged or inflamed, it will show high temperatures. However, infrared detection is limited to measuring the temperature close to the skin surface, and it is unlikely to detect damage to deep subcutaneous tissue through local temperature changes. The applicable scenarios are limited.

[0006] Subcutaneous moisture measurement scanners are based on the fact that early deep tissue damage is often accompanied by abnormal subcutaneous moisture values, and the water content of subcutaneous tissue changes dynamically with skin pathological damage. However, there are many factors affecting the detection of subcutaneous moisture scanners. The subcutaneous moisture value is affected by pressure, and there are confounding variables that affect the measurement results due to different patient positions, the duration of body pressure, and the pressure applied to the probe during measurement. In addition, there will be certain deviations in the judgment standards of different populations, which will lead to difficulties in promotion. Summary of the invention

[0007] Based on this, it is necessary to provide an artificial intelligence-based pressure injury detection method, system and medium to address the problems of large errors and inconsistent judgment standards caused by the above-mentioned manual detection of pressure injuries through instruments.

[0008] A pressure injury detection method based on artificial intelligence, comprising:

[0009] Respectively obtain infrared source images and microwave source images corresponding to the target position;

[0010] Extracting features from the infrared source image and the microwave source image respectively to obtain corresponding infrared feature information and microwave feature information respectively;

[0011] Based on the adversarial network, the infrared feature information and the microwave feature information are fused to obtain fused feature information;

[0012] Based on the fused feature information, the corresponding pressure injury feature value is extracted to obtain detection information of the pressure injury at the target position, and the pressure injury at the target position is graded based on the detection information.

[0013] In one preferred embodiment, the step of respectively acquiring the infrared source image and the microwave source image corresponding to the target position includes:

[0014] Acquire the infrared source image by using an infrared thermal imager;

[0015] The microwave source image is acquired by a microwave camera with a millimeter wave frequency of 50-70 Hz.

[0016] In one preferred embodiment, the extracting features of the infrared source image and the microwave source image respectively to obtain corresponding infrared feature information and microwave feature information respectively includes:

[0017] Feature extraction is performed on the infrared source image and the microwave source image based on a neural network.

[0018] In one preferred embodiment, the feature extraction of the infrared source image and the microwave source image is performed to obtain corresponding infrared feature information and microwave feature information, respectively, and further includes:

[0019] The infrared source image and the microwave source image are preprocessed.

[0020] In one preferred embodiment, the preprocessing of the infrared source image and the microwave source image includes:

[0021] The infrared source image and the microwave source image are subjected to image color level equalization processing, image noise reduction and image registration.

[0022] In a preferred embodiment, the image color scale equalization processing includes processing the grayscale values ​​of the infrared source image and the microwave source image by using a cumulative distribution function;

[0023] The image denoising comprises performing denoising processing on the image by a threshold filtering method;

[0024] The image registration includes searching for the same key point in multiple images to perform corresponding matching and superposition of the infrared source image and the microwave source image.

[0025] In one preferred embodiment, the infrared feature information and the microwave feature information are fused based on the adversarial network to obtain fused feature information, including:

[0026] The adversarial network has a generator and a discriminator. The generator is used to adopt local detail features and global semantic features of infrared source images and microwave source images to extract semantic information of infrared source images and microwave source images. Perceptual loss is added to the discriminator to make the data distribution of the fused image consistent with the source image.

[0027] In one preferred embodiment, extracting corresponding pressure injury characteristic values ​​based on the fused characteristic information to obtain detection information of pressure injury at the target location, and grading the pressure injury at the target location based on the detection information includes:

[0028] Importing the fused feature information into a convolutional network training model to mark the wound range at the target location;

[0029] Based on the marked wound range, the corresponding pressure injury characteristic value is extracted;

[0030] Based on the pressure injury characteristic value, the pressure injury classification is performed on the target position.

[0031] The method disclosed in the above-mentioned embodiment of the present invention utilizes infrared images in combination with microwave images, extracts features from the images respectively, and performs feature fusion. It fully utilizes infrared technology to detect the regular changes in skin temperature distribution before and after the injury occurs, and uses relative temperature difference to predict skin surface compression injuries. It also utilizes microwave technology to detect dermis thickness, density of tissue structure, etc., predicts hidden deep tissue injuries, and performs grading, which greatly reduces the evaluation errors of medical staff, solves the problem of inconsistent evaluation standards, and improves evaluation efficiency.

[0032] An artificial intelligence-based pressure injury detection system, comprising:

[0033] An image acquisition module, used to respectively acquire an infrared source image and a microwave source image corresponding to a target position;

[0034] A feature extraction module, used to extract features from the infrared source image and the microwave source image respectively, so as to obtain corresponding infrared feature information and microwave feature information respectively;

[0035] A feature fusion module, used to fuse the infrared feature information and the microwave feature information based on an adversarial network to obtain fused feature information;

[0036] The grading evaluation module is used to extract the corresponding pressure injury characteristic value based on the fused characteristic information to obtain the detection information of the pressure injury at the target position, and to grade the pressure injury at the target position based on the detection information.

[0037] The system disclosed in the above-mentioned embodiment of the present invention utilizes infrared images in combination with microwave images to extract features from the images respectively and perform feature fusion. It fully utilizes infrared technology to detect the regular changes in skin temperature distribution before and after the injury occurs, and uses relative temperature difference to predict skin surface compression injuries. It also utilizes microwave technology to detect dermis thickness, density of tissue structure, etc., predict hidden deep tissue injuries, and perform classification, which greatly reduces the evaluation errors of medical staff, solves the problem of inconsistent evaluation standards, and improves evaluation efficiency.

[0038] A storage medium containing computer executable instructions, which, when executed by a computer processor, are used to perform the above-mentioned artificial intelligence-based pressure injury detection method.

[0039] The storage medium disclosed in the above-mentioned embodiment of the present invention executes the above-mentioned method, utilizes infrared images in combination with microwave images, extracts features from the images respectively, and performs feature fusion. It fully utilizes infrared technology to detect the regular changes in skin temperature distribution before and after the injury occurs, and uses relative temperature difference to predict skin surface compression injuries. It also utilizes microwave technology to detect dermis thickness, density of tissue structure, etc., predicts hidden deep tissue injuries, and performs grading, which greatly reduces the evaluation errors of medical staff, solves the problem of inconsistent evaluation standards, and improves evaluation efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 A schematic flow chart of a pressure injury detection method based on artificial intelligence disclosed in a first preferred embodiment of the present invention;

[0041] Figure 2 This is a module diagram of a pressure injury detection method based on artificial intelligence disclosed in the second preferred embodiment of the present invention. DETAILED DESCRIPTION

[0042] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0043] It should be noted that when an element is referred to as being "disposed on" another element, it may be directly on the other element or there may be a central element. When an element is considered to be "connected to" another element, it may be directly connected to the other element or there may be a central element at the same time. The terms "vertical", "horizontal", "left", "right" and similar expressions used herein are for illustrative purposes only and do not represent the only implementation method.

[0044] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used herein in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more related listed items.

[0045] like Figure 1 As shown, the first preferred embodiment of the present invention discloses a pressure injury detection method based on artificial intelligence, the method comprising:

[0046] S10: respectively obtaining an infrared source image and a microwave source image corresponding to the target position;

[0047] In this embodiment, the above step S10 acquires the infrared source image by an infrared thermal imager; and acquires the microwave source image by a microwave camera with a millimeter wave frequency of 50-70 Hz.

[0048] Medical infrared thermal imaging is an imaging device that combines medical technology and infrared camera technology. It assists diagnosis and treatment by recording images of the human body's thermal field. The human body is a natural biological heating body. Due to differences in anatomical structure, tissue metabolism, blood circulation and nerve state, the temperature of different parts of the body is different, forming different thermal fields. The infrared thermal imager uses an optical electronic system to filter, gather, modulate and photoelectrically convert the far-infrared light waves radiated by the human body into electrical signals, which are then converted into digital quantities through A / D. Then, through multimedia image processing technology, the temperature field of the human body is displayed in the form of a pseudo-color thermal map. A normal body state has a normal thermal map, and an abnormal body state has an abnormal thermal map. By comparing the similarities and differences between the two and combining them with clinical practice, a diagnosis can be made and the nature and extent of the disease can be inferred.

[0049] Medical microwave imaging is a technology that uses microwaves as information carriers to detect electromagnetic information of the human body. This technology first needs to measure the scattered field data after the microwaves pass through the object, and then use algorithms to reconstruct the data to obtain the structural information of the lesion. In this embodiment,

[0050] S20: extracting features from the infrared source image and the microwave source image respectively to obtain corresponding infrared feature information and microwave feature information respectively;

[0051] In this implementation, feature extraction is performed on the infrared source image and the microwave source image based on a neural network.

[0052] Specifically, in this embodiment, the method of combining multi-feature fusion and extreme learning machine is used to extract features of the above infrared source image. The extraction process of this method is: PCA, LBP and SIFT are used to extract features of the original infrared image respectively to obtain the corresponding feature vector. The three have good complementarity in describing the target characteristics. In order to further extract the discriminative power of the three and build feature redundancy, multiple set canonical correlation analysis (MCCA) is used to fuse and analyze the three types of features to obtain a single feature vector. The above is the feature extraction of the infrared source image.

[0053] Multi-feature fusion refers to combining multiple different types of features to obtain a more comprehensive and representative feature representation. In infrared image target classification, commonly used features include principal component analysis (PCA), local binary pattern (LBP) and scale-invariant feature transform (SIFT). These features describe the pixel distribution, local texture and feature point information of the target in the infrared image, respectively, and reflect the characteristics of the target from different aspects, so they have complementary advantages. The basic idea of ​​this method is to improve the accuracy of target classification by fusing multiple features, and at the same time use the fast learning and good generalization ability of extreme learning machine for classification.

[0054] In this embodiment, a microwave imaging recognition method based on on-site programmable deep learning is used to extract features from the microwave source image. This method combines programmable metasurface materials with deep learning networks to achieve real-time data acquisition and data processing, thereby achieving real-time microwave imaging and target recognition.

[0055] In more detail, the above recognition method uses FPGA to switch the encoding of programmable Metasurface n times to obtain n groups of different data. Then, the received data is processed by Tikhonov regularization algorithm to reconstruct the image. The reconstructed image is enhanced by DeepNIS network. The CNN network in the deep learning network is trained. The CNN network is used for target recognition and classification. In this process, FPGA is used to control the switching of different Metasurface encodings to obtain data, so that the acquisition efficiency is high. At the same time, a single receiver at a fixed position is used to receive data, and the system structure is simple and the cost is low. The deep learning network used (including DeepNIS network and CNN network) is based on sample drive, which can well capture image features, has good imaging quality, high target recognition and classification accuracy, and strong anti-noise ability.

[0056] In this embodiment, before extracting features from the infrared source image and the microwave source image, the step S20 further includes preprocessing the infrared source image and the microwave source image, specifically including:

[0057] The infrared source image and the microwave source image are subjected to image color level equalization processing, image noise reduction and image registration.

[0058] Specifically, the image color level equalization processing includes processing the grayscale values ​​of the infrared source image and the microwave source image through a cumulative distribution function; the image denoising includes denoising the image through a threshold filtering method; the image registration includes matching and superimposing the infrared source image and the microwave source image by searching for the same key point in the image through multiple images.

[0059] The above-mentioned image color level equalization processing aims to improve the contrast and brightness distribution of the image to make it more consistent with the visual perception of the human eye. The cumulative distribution function (CDF) plays a key role in the image color level equalization processing. It is a cumulative version of the histogram, which represents the total frequency of each gray level and all gray levels below it. The calculation of CDF is based on the original histogram and is obtained by accumulating each value of the histogram. The first value of the CDF is the first value of the histogram, the second value of the CDF is the sum of the first two values ​​of the histogram, and so on, until all gray levels are processed. The quality of the image can be significantly improved by color level equalization processing. In grayscale images, equalization processing makes the details of dark and bright areas more clearly visible. In color images, equalization processing makes the color distribution more uniform, improving the overall visual effect of the image.

[0060] S30: Based on the adversarial network, the infrared feature information and the microwave feature information are fused to obtain fused feature information;

[0061] The adversarial network has a generator and a discriminator. The generator is used to adopt local detail features and global semantic features of infrared source images and microwave source images to extract semantic information of infrared source images and microwave source images. Perceptual loss is added to the discriminator to make the data distribution of the fused image consistent with the source image.

[0062] S40: Based on the fused feature information, extract the corresponding pressure injury feature value to obtain detection information of the pressure injury at the target position, and perform pressure injury classification on the target position based on the detection information.

[0063] Specifically, the above step S40 imports the fused feature information into the convolutional network training model to mark the wound range of the target position; based on the marked wound range, extracts the corresponding pressure injury characteristic value; based on the pressure injury characteristic value, performs pressure injury classification on the above target position.

[0064] The method disclosed in the above-mentioned embodiment of the present invention utilizes infrared images in combination with microwave images, extracts features from the images respectively, and performs feature fusion. It fully utilizes infrared technology to detect the regular changes in skin temperature distribution before and after the injury occurs, and uses relative temperature difference to predict skin surface compression injuries. It also utilizes microwave technology to detect dermis thickness, density of tissue structure, etc., predicts hidden deep tissue injuries, and performs grading, which greatly reduces the evaluation errors of medical staff, solves the problem of inconsistent evaluation standards, and improves evaluation efficiency.

[0065] like Figure 2 As shown, the second preferred embodiment of the present invention discloses an artificial intelligence-based pressure injury detection system 100, and the system 100 includes an image acquisition module 110, a feature acquisition module 120, a feature fusion module 130, and a grading evaluation module 140.

[0066] The image acquisition module 110 is used to respectively acquire an infrared source image and a microwave source image corresponding to a target position;

[0067] In this embodiment, the image acquisition module 110 acquires the infrared source image through an infrared thermal imager and acquires the microwave source image through a microwave camera with a millimeter wave frequency of 50-70 Hz.

[0068] Medical infrared thermal imaging is an imaging device that combines medical technology and infrared camera technology. It assists diagnosis and treatment by recording images of the human body's thermal field. The human body is a natural biological heating body. Due to differences in anatomical structure, tissue metabolism, blood circulation and nerve state, the temperature of different parts of the body is different, forming different thermal fields. The infrared thermal imager uses an optical electronic system to filter, gather, modulate and photoelectrically convert the far-infrared light waves radiated by the human body into electrical signals, which are then converted into digital quantities through A / D. Then, through multimedia image processing technology, the temperature field of the human body is displayed in the form of a pseudo-color thermal map. A normal body state has a normal thermal map, and an abnormal body state has an abnormal thermal map. By comparing the similarities and differences between the two and combining them with clinical practice, a diagnosis can be made and the nature and extent of the disease can be inferred.

[0069] Medical microwave imaging is a technology that uses microwaves as information carriers to detect electromagnetic information of the human body. This technology first needs to measure the scattered field data after the microwaves pass through the object, and then use algorithms to reconstruct the data to obtain the structural information of the lesion. In this embodiment,

[0070] The feature extraction module 120 is used to extract features from the infrared source image and the microwave source image respectively, so as to obtain corresponding infrared feature information and microwave feature information respectively.

[0071] In this implementation, the feature extraction module 120 extracts features from the infrared source image and the microwave source image based on a neural network.

[0072] Specifically, in this embodiment, the method of combining multi-feature fusion and extreme learning machine is used to extract features of the above infrared source image. The extraction process of this method is: PCA, LBP and SIFT are used to extract features of the original infrared image respectively to obtain the corresponding feature vector. The three have good complementarity in describing the target characteristics. In order to further extract the discriminative power of the three and build feature redundancy, multiple set canonical correlation analysis (MCCA) is used to fuse and analyze the three types of features to obtain a single feature vector. The above is the feature extraction of the infrared source image.

[0073] Multi-feature fusion refers to combining multiple different types of features to obtain a more comprehensive and representative feature representation. In infrared image target classification, commonly used features include principal component analysis (PCA), local binary pattern (LBP) and scale-invariant feature transform (SIFT). These features describe the pixel distribution, local texture and feature point information of the target in the infrared image, respectively, and reflect the characteristics of the target from different aspects, so they have complementary advantages. The basic idea of ​​this method is to improve the accuracy of target classification by fusing multiple features, and at the same time use the fast learning and good generalization ability of extreme learning machine for classification.

[0074] In this embodiment, a microwave imaging recognition method based on on-site programmable deep learning is used to extract features from the microwave source image. This method combines programmable metasurface materials with deep learning networks to achieve real-time data acquisition and data processing, thereby achieving real-time microwave imaging and target recognition.

[0075] In more detail, the above recognition method uses FPGA to switch the encoding of programmable Metasurface n times to obtain n groups of different data. Then, the received data is processed by Tikhonov regularization algorithm to reconstruct the image. The reconstructed image is enhanced by DeepNIS network. The CNN network in the deep learning network is trained. The CNN network is used for target recognition and classification. In this process, FPGA is used to control the switching of different Metasurface encodings to obtain data, so that the acquisition efficiency is high. At the same time, a single receiver at a fixed position is used to receive data, and the system structure is simple and the cost is low. The deep learning network used (including DeepNIS network and CNN network) is based on sample drive, which can well capture image features, has good imaging quality, high target recognition and classification accuracy, and strong anti-noise ability.

[0076] In this embodiment, before extracting features from the infrared source image and the microwave source image, the feature extraction module 120 further includes preprocessing the infrared source image and the microwave source image, specifically including:

[0077] The feature extraction module 120 performs image color level equalization, image noise reduction and image registration on the infrared source image and the microwave source image.

[0078] Specifically, the image color level equalization processing includes processing the grayscale values ​​of the infrared source image and the microwave source image through a cumulative distribution function; the image denoising includes denoising the image through a threshold filtering method; the image registration includes matching and superimposing the infrared source image and the microwave source image by searching for the same key point in the image through multiple images.

[0079] The above-mentioned image color level equalization processing aims to improve the contrast and brightness distribution of the image to make it more consistent with the visual perception of the human eye. The cumulative distribution function (CDF) plays a key role in the image color level equalization processing. It is a cumulative version of the histogram, which represents the total frequency of each gray level and all gray levels below it. The calculation of CDF is based on the original histogram and is obtained by accumulating each value of the histogram. The first value of the CDF is the first value of the histogram, the second value of the CDF is the sum of the first two values ​​of the histogram, and so on, until all gray levels are processed. The quality of the image can be significantly improved by color level equalization processing. In grayscale images, equalization processing makes the details of dark and bright areas more clearly visible. In color images, equalization processing makes the color distribution more uniform, improving the overall visual effect of the image.

[0080] The feature fusion module 130 is used to fuse the infrared feature information and the microwave feature information based on an adversarial network to obtain fused feature information;

[0081] The adversarial network has a generator and a discriminator. The generator is used to adopt local detail features and global semantic features of infrared source images and microwave source images to extract semantic information of infrared source images and microwave source images. Perceptual loss is added to the discriminator to make the data distribution of the fused image consistent with the source image.

[0082] The grading evaluation module 140 is used to extract the corresponding pressure injury feature value based on the fused feature information to obtain the detection information of the pressure injury at the target position, and to grade the pressure injury at the target position based on the detection information.

[0083] Specifically, the grading evaluation module 140 imports the fused feature information into the convolutional network training model to mark the wound range of the target position; based on the marked wound range, extracts the corresponding pressure injury characteristic value; based on the pressure injury characteristic value, performs pressure injury grading on the target position.

[0084] The system disclosed in the above-mentioned embodiment of the present invention utilizes infrared images in combination with microwave images to extract features from the images respectively and perform feature fusion. It fully utilizes infrared technology to detect the regular changes in skin temperature distribution before and after the injury occurs, and uses relative temperature difference to predict skin surface compression injuries. It also utilizes microwave technology to detect dermis thickness, density of tissue structure, etc., predict hidden deep tissue injuries, and perform classification, which greatly reduces the evaluation errors of medical staff, solves the problem of inconsistent evaluation standards, and improves evaluation efficiency.

[0085] A storage medium containing computer executable instructions, which, when executed by a computer processor, are used to perform the above-mentioned artificial intelligence-based pressure injury detection method.

[0086] The storage medium disclosed in the above-mentioned embodiment of the present invention executes the above-mentioned method, utilizes infrared images in combination with microwave images, extracts features from the images respectively, and performs feature fusion. It fully utilizes infrared technology to detect the regular changes in skin temperature distribution before and after the injury occurs, and uses relative temperature difference to predict skin surface compression injuries. It also utilizes microwave technology to detect dermis thickness, density of tissue structure, etc., predicts hidden deep tissue injuries, and performs grading, which greatly reduces the evaluation errors of medical staff, solves the problem of inconsistent evaluation standards, and improves evaluation efficiency.

[0087] It should be noted that the storage medium disclosed above may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, device or device. In the present disclosure, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, in which a computer-readable program code is carried. This propagated data signal may take a variety of forms, including but not limited to an electromagnetic signal, an optical signal, or any suitable combination of the above. The computer readable signal medium may also be any computer storage medium other than a computer readable storage medium, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0088] In some embodiments, the client and the server may communicate using any currently known or future developed network protocol such as HTTP (Hyper Text Transfer Protocol), and may be interconnected with any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network ("LAN"), a wide area network ("WAN"), an internet (e.g., the Internet), and a peer-to-peer network (e.g., an ad hoc peer-to-peer network), as well as any currently known or future developed network.

[0089] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0090] The above-mentioned embodiments only express several implementation methods of the present invention, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A pressure injury detection method based on artificial intelligence, characterized in that: include: Respectively obtain infrared source images and microwave source images corresponding to the target position; Extracting features from the infrared source image and the microwave source image respectively to obtain corresponding infrared feature information and microwave feature information respectively; Based on the adversarial network, the infrared feature information and the microwave feature information are fused to obtain fused feature information; Based on the fused feature information, the corresponding pressure injury feature value is extracted to obtain detection information of the pressure injury at the target position, and the pressure injury at the target position is graded based on the detection information.

2. The artificial intelligence-based pressure injury detection method according to claim 1, characterized in that: The step of respectively acquiring the infrared source image and the microwave source image corresponding to the target position includes: Acquire the infrared source image by using an infrared thermal imager; The microwave source image is acquired by a microwave camera with a millimeter wave frequency of 50-70 Hz.

3. The artificial intelligence-based pressure injury detection method according to claim 1, characterized in that: The extracting features of the infrared source image and the microwave source image respectively to obtain corresponding infrared feature information and microwave feature information respectively includes: Feature extraction is performed on the infrared source image and the microwave source image based on a neural network.

4. The artificial intelligence-based pressure injury detection method according to claim 1, characterized in that: The extracting features of the infrared source image and the microwave source image to obtain corresponding infrared feature information and microwave feature information respectively also includes: The infrared source image and the microwave source image are preprocessed.

5. The artificial intelligence-based pressure injury detection method according to claim 1, characterized in that: The preprocessing of the infrared source image and the microwave source image comprises: The infrared source image and the microwave source image are subjected to image color level equalization processing, image noise reduction and image registration.

6. The artificial intelligence-based pressure injury detection method according to claim 1, characterized in that: The image color scale equalization processing includes processing the grayscale values ​​of the infrared source image and the microwave source image by using a cumulative distribution function; The image denoising comprises performing denoising processing on the image by a threshold filtering method; The image registration includes searching for the same key point in multiple images to perform corresponding matching and superposition of the infrared source image and the microwave source image.

7. The artificial intelligence-based pressure injury detection method according to claim 1, characterized in that: Based on the adversarial network, the infrared feature information and the microwave feature information are fused to obtain fused feature information, including: The adversarial network has a generator and a discriminator. The generator is used to adopt local detail features and global semantic features of infrared source images and microwave source images to extract semantic information of infrared source images and microwave source images. Perceptual loss is added to the discriminator to make the data distribution of the fused image consistent with the source image.

8. The artificial intelligence-based pressure injury detection method according to claim 1, characterized in that: The extracting corresponding pressure injury characteristic values ​​based on the fused characteristic information to obtain detection information of the pressure injury at the target position, and grading the pressure injury at the target position based on the detection information, includes: Importing the fused feature information into a convolutional network training model to mark the wound range at the target location; Based on the marked wound range, the corresponding pressure injury characteristic value is extracted; Based on the pressure injury characteristic value, the pressure injury classification is performed on the target position.

9. An artificial intelligence-based pressure injury detection system, characterized in that: include: An image acquisition module, used to respectively acquire an infrared source image and a microwave source image corresponding to a target position; A feature extraction module, used to extract features from the infrared source image and the microwave source image respectively, so as to obtain corresponding infrared feature information and microwave feature information respectively; A feature fusion module, used to fuse the infrared feature information and the microwave feature information based on an adversarial network to obtain fused feature information; The grading evaluation module is used to extract the corresponding pressure injury characteristic value based on the fused characteristic information to obtain the detection information of the pressure injury at the target position, and to grade the pressure injury at the target position based on the detection information.

10. A storage medium containing computer executable instructions, characterized in that: The computer executable instructions, when executed by a computer processor, are used to execute an artificial intelligence-based pressure injury detection method as described in any one of claims 1-8.

Images