A system and method for identifying low back pain based on electromyography and muscle oxygenation combined array information

Abstract

This invention discloses a system and method for identifying low back pain based on a combined electromyography (EMG) and muscle oxygenation (MOO) array. The system includes: a data generation module for generating EMG and MOO topographic maps based on EMG and MOO information from the lower back; a feature processing module for extracting features from the EMG and MOO topographic maps to obtain EMG and MOO topographic map feature data, wherein the EMG and MOO topographic map feature data are morphological features of areas of high activity; and an analysis and diagnosis module for inputting the EMG and MOO topographic map feature data into a trained neural network model to obtain a low back pain category identification result. This invention improves the objectivity and accuracy of low back pain detection and classification, thereby improving the treatment and rehabilitation effects of low back pain.

Description

Technical Field

[0001] This invention relates to the field of biomedical engineering technology, and more specifically, to a system and method for recognizing low back pain based on electromyography and muscle oxygenation combined array information. Background Technology

[0002] Low back pain (LBP) refers to pain in the lower back, lumbosacral region, and sacroiliac region, sometimes accompanied by referred pain or radiating pain in the lower limbs. It usually manifests in the lower lumbar spine and lumbosacral and sacroiliac regions, hence it is also called "lower back pain" or "back pain". Because chronic low back pain is persistent and recurrent, it requires long-term rehabilitation treatment and seriously affects the patient's normal life and work.

[0003] In the prior art, patent application CN201480057113.3 discloses a system and method for restoring lumbar spine muscle function. This system includes electrodes coupled to an implantable pulse generator (IPG), a handheld activator transmitting stimulation commands to the IPG, and an external programmer transmitting programming data to the IPG, wherein a programmable controller generates programming data to stimulate tissue based on the stimulation commands. The system may include a software-based programming system running on a computer, allowing the therapist to program and adjust stimulation parameters. However, this system only utilizes electrical stimulation to restore lumbar spine muscle function to alleviate back pain and does not address back pain classification and diagnosis.

[0004] Patent application CN201710001886.5 discloses an array-type electromyography (EMG) signal acquisition device and a lower back pain auxiliary diagnostic system. The array-type EMG signal acquisition device includes a waist belt and a wireless transmission module. Multiple electrodes arranged in an array are embedded on the inner side of the waist belt. These electrodes are connected to the wireless transmission module via embedded wires. The electrodes are used to acquire surface EMG signals from the lumbar region and transmit them to the wireless transmission module, which then emits the surface EMG signals radioly. However, this system only acquires back EMG information based on an array-type electrode distribution and analyzes the EMG information to determine if lower back pain is present. The accuracy of the acquired EMG signals is limited, and the system only extracts relevant characteristic parameters from the back EMG signals to determine back pain symptoms, offering limited reference value for clinical diagnosis.

[0005] Patent application CN202210081133.0 discloses an automatic assessment method for limb dysfunction in patients with low back pain based on multi-source features. First, a U-Net neural network is used to automatically segment the multifidus and erector spinae muscles in lumbar MRI images, and the Otsu's method is used to extract imaging features such as muscle cross-sectional area and fat infiltration. Second, commonly used clinical assessment scales for low back pain are selected to extract scale features. Then, a Support Vector Machine Recursive Feature Elimination (SVM-RFE) algorithm is used to filter all features, reducing feature redundancy and determining the optimal feature combination. Finally, the Support Vector Machine (SVM) method in machine learning is used to construct an assessment model for limb dysfunction in patients with low back pain, enabling automatic assessment and discrimination of limb dysfunction. However, this approach only extracts relevant features from MRI images of the multifidus and erector spinae muscles of the lumbar region and combines them with assessment scales for low back pain to assess the degree of limb dysfunction in patients with low back pain; it does not involve the collection of physiological information such as electromyography or the diagnostic analysis of low back pain.

[0006] Analysis reveals that current diagnostic methods for low back pain largely rely on patient complaints and physician experience, lacking objective testing and classification standards, which easily leads to misdiagnosis and affects treatment outcomes. Diagnosing low back pain using physiological electrical signals or medical imaging often involves assessing muscle state through electromyography (EMG) signals in the lower back or through medical imaging of the lower back muscles. However, this reliance on single-mode physiological information testing or subjective assessment scales by low back pain patients cannot accurately and effectively diagnose low back pain. Research indicates that the occurrence of low back pain is highly correlated with neuromuscular bioelectrical activity and blood flow. Current diagnostic methods for low back pain have the following shortcomings: firstly, the relationship between lower back muscle contraction patterns and neuromuscular electrophysiological-blood oxygenation information characteristics is unclear; secondly, the accurate mapping relationship between neuromuscular electrophysiological-blood oxygenation information characteristics and low back pain categories is unclear. Summary of the Invention

[0007] The purpose of this invention is to overcome the shortcomings of the prior art and provide a system and method for recognizing low back pain based on electromyography and muscle oxygenation combined array information.

[0008] According to a first aspect of the present invention, a low back pain recognition system based on electromyography and muscle oxygenation combined array information is provided. The system includes:

[0009] Data generation module: used to generate electromyography and muscle oxygenation topography maps based on electromyography and muscle oxygenation information of the lower back;

[0010] Feature processing module: used to extract features from the electromyography topography map and the oxygenation topography map to obtain electromyography topography map feature data and oxygenation topography map feature data, wherein the electromyography topography map feature data and the oxygenation topography map feature data are morphological features of strong activity areas;

[0011] Analysis and Diagnosis Module: This module is used to input the electromyography topographic feature data and the muscle oxygenation topographic feature data into a trained neural network model to obtain the low back pain category recognition result.

[0012] According to a second aspect of the present invention, a method for identifying low back pain based on electromyography and muscle oxygenation combined array information is provided. The method includes the following steps:

[0013] Electromyography (EMG) and oxygenation (OO) topography maps are generated based on EMG and OO data from the lower back.

[0014] Feature extraction is performed on the electromyography topography map and the oxygenation topography map to obtain electromyography topography map feature data and oxygenation topography map feature data, wherein the electromyography topography map feature data and the oxygenation topography map feature data are morphological features of strong activity areas;

[0015] The electromyography topography feature data and the muscle oxygenation topography feature data are input into a trained neural network model to obtain the low back pain category recognition result.

[0016] Compared with the prior art, the advantages of this invention are that it proposes a low back pain recognition system and method based on electromyography and muscle oxygenation composite array information. By determining the correspondence between the dual-mode physiological information characteristics of low back pain neuromuscular activity and pain, it improves the objectivity and accuracy of low back pain detection and classification, and improves the treatment and rehabilitation effects of low back pain, thus providing technical support for the accurate and objective diagnosis and treatment of low back pain.

[0017] Other features and advantages of the invention will become clear from the following detailed description of exemplary embodiments of the invention with reference to the accompanying drawings. Attached Figure Description

[0018] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments of the invention and, together with their description, serve to explain the principles of the invention.

[0019] Figure 1 This is a block diagram of a low back pain recognition system based on electromyography and muscle oxygenation combined array information according to an embodiment of the present invention;

[0020] Figure 2 This is a schematic diagram of electromyography of the lumbar and back muscles according to an embodiment of the present invention;

[0021] Figure 3This is a schematic diagram of the oxygenation topography of the lumbar and back muscles according to an embodiment of the present invention;

[0022] Figure 4 This is a flowchart of a threshold-based morphological feature extraction method according to an embodiment of the present invention;

[0023] Figure 5 This is a diagram of a multilayer feedforward neural network framework according to an embodiment of the present invention;

[0024] Figure 6 This is a diagram of a supervised deep convolutional neural network framework according to an embodiment of the present invention;

[0025] Figure 7 This is a flowchart of a low back pain recognition method based on electromyography and muscle oxygenation combined array information according to an embodiment of the present invention. Detailed Implementation

[0026] Various exemplary embodiments of the present invention will now be described in detail with reference to the accompanying drawings. It should be noted that, unless otherwise specifically stated, the relative arrangement, numerical expressions, and values ​​of the components and steps set forth in these embodiments do not limit the scope of the invention.

[0027] The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit the invention or its application or use.

[0028] Techniques, methods, and equipment known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques, methods, and equipment should be considered part of the specification.

[0029] In all the examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not as limitations. Therefore, other examples of exemplary embodiments may have different values.

[0030] It should be noted that similar labels and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be discussed further in subsequent figures.

[0031] See Figure 1As shown, the low back pain recognition system based on electromyography (EMG) and muscle oxygenation (MOO) composite array information proposed in this invention includes a data generation module 110, a feature processing module 120, and an analysis and diagnosis module 130. The data generation module 110 generates EMG and MOO topographic maps based on EMG and MOO information from the low back. The feature processing module 120 extracts optimal physiological information map feature parameters with strong pain category separability. The analysis and diagnosis module 130 uses a neural network self-learning model to generate a mapping relationship between the neuromuscular electrophysiological-blood oxygenation information features of the low back and the low back pain category, thereby establishing a pain classification model and achieving accurate and objective classification of low back pain.

[0032] In one embodiment, the data generation module 110 generates electromyographic topographic maps and muscle oxygenation topographic maps of the lower back based on electromyographic and muscle oxygenation information of the lower back, using methods such as cubic spline interpolation, to obtain high-resolution maps of electromyographic and muscle oxygenation information of the lower back.

[0033] Specifically, cubic spline interpolation is performed on the electromyographic (EMG) information of the lumbar and back regions, and this information is combined with EMG feature values ​​from multiple locations in the lumbar and back regions to form a matrix structure. The EMG feature values ​​include the root mean square (RMS) value, autoregressive model coefficients (AR), mean frequency (MNF) value, and median frequency (MDF) value. Each element in the matrix is ​​transformed to the range [0,1] using a linearization method, and then the new elements are mapped to a three-primary-color light mode (RGB). Finally, this can be displayed as a visualized topographic map, i.e., an EMG topographic map. Figure 2 As shown, electromyographic topography can include RMS topography, AR topography, MDF topography, etc.

[0034] Figure 3 This is a schematic diagram of muscle oxygenation topography in the lower back. Specifically, by performing three interpolation calculations on the muscle oxygenation information of the lower back and its deoxygenated hemoglobin (HB), oxyhemoglobin (HBO), and total hemoglobin (HBT), the distribution of muscle blood oxygen concentration in the entire lower back test area can be obtained. This enables near-infrared spectral imaging of blood oxygenation information to obtain muscle oxygenation topography, including HB topography, HBO topography, and HBT topography. Through interpolation calculations, computational efficiency can be improved based on the obtained electromyography and muscle oxygenation topography maps.

[0035] In one embodiment, the feature processing module 120 obtains feature parameters to achieve the best lumbar and back classification effect based on the electromyography and muscle oxygenation dual-mode physiological information map. For example, feature extraction is performed on the electromyography topography map and muscle oxygenation topography map based on image processing technology.

[0036] Specifically, see Figure 4 As shown, a threshold-based morphological feature extraction method can be used, including the following process:

[0037] Step S1: Segment the image based on the set grayscale threshold.

[0038] For example, a grayscale threshold T is set to distinguish between the target and the background; values ​​above threshold T are considered the target, and values ​​below threshold T are considered the background. By setting this threshold T, the pixels of the original color images of electromyography and muscle oxygenation topography are converted into black-and-white binary images to achieve image segmentation. Let the input image be f(x,y), and the segmented image be g(x,y), then:

[0039]

[0040] The threshold can be selected manually or determined through simulation. For example, an experienced physician can use prior knowledge to observe the original images, conduct multiple trials, and finally select an appropriate threshold to apply to all original images to separate regions of high activity from regions of low activity.

[0041] Step S2: Perform morphological processing on the segmented image to obtain a morphologically processed topographic map.

[0042] For example, morphological processing (including but not limited to erosion, dilation, opening and closing operations) can be performed on the segmented high-activity region image to remove image noise while enhancing the edge details of the image, thereby smoothing the edge of the target image.

[0043] Step S3: Extract morphological features of highly active areas from the morphologically processed topographic map.

[0044] Specifically, morphological features of highly active regions, such as size and location distribution, are extracted. The size of highly active regions can be obtained by calculating their area, width, and height, while location distribution information is more difficult to determine. Considering human anatomy, the back muscles of a healthy person are symmetrically distributed on both sides of the spine. By analyzing the topographic map separately from the central spine to the left and right sides and calculating the differences in features between the two sides, the location distribution information of highly active regions on the complete topographic map can be indirectly reflected.

[0045] In one embodiment, the classification and diagnosis module 130 establishes a neural network analysis model to train and learn the feature data of the dual-mode physiological information map of low back pain electromyography and muscle oxygenation, so as to obtain the mutual mapping relationship between the neuromuscular electrophysiological-blood oxygenation information features of the low back and the category of low back pain, and finally provide an objective and accurate multi-classification method for low back pain to achieve accurate identification of low back pain category.

[0046] Neural network models can be of various types, including traditional neural network models and deep neural network models. Furthermore, to determine the optimal multi-classification method for lower back pain suitable for an individual, K-fold cross-validation is used to compare the performance (such as classification accuracy and computational efficiency) of the two models mentioned above, selecting the optimal model to achieve accurate multi-class identification of lower back pain.

[0047] For example, traditional neural network models can be trained using feature data based on electromyography and muscle oxygenation physiological information maps to identify the category of lower back pain.

[0048] Specifically, based on the characteristic parameters of the high-intensity activity areas obtained from electromyography and muscle oxygenation topography, a multi-layer feedforward neural network is constructed. The sigmoid activation function is adopted, and an error backpropagation algorithm based on gradient descent strategy is used to iteratively update the connection weights and thresholds of each neuron, minimizing the cumulative error of the training set and the learning time, so as to obtain the optimal artificial neural network model and achieve accurate identification of multiple categories of low back pain. Figure 5 This is a framework diagram of a multi-layer feedforward neural network, which includes an input layer, multiple hidden layers, and an output layer. The number of hidden layers can be set according to the requirements for classification accuracy and computational efficiency.

[0049] For deep neural network models, a supervised deep convolutional neural network model is trained using raw pixel data based on electromyography and muscle oxygenation physiological information maps to achieve the discrimination of low back pain categories.

[0050] Specifically, based on the raw pixel data from electromyography and muscle oxygenation topography (temporal and frequency features), a supervised deep convolutional neural network model is trained to fuse the two types of pixel data and establish relationships between different categories of low back pain, thereby achieving accurate identification of multiple types of low back pain. For example... Figure 6 This is a block diagram of a supervised deep convolutional neural network, consisting of 9 layers: 3 convolutional layers, 3 pooling layers, 1 flattening layer, 1 fully connected layer, and 1 output layer.

[0051] The input layer is used to feed the raw image pixels of electromyography and muscle oxygenation topography into the convolutional neural network structure and produce the input of the first convolutional layer.

[0052] Convolutional layers are responsible for feature learning and fusion. This layer extracts features by convolutionally calculating the outputs of neurons connected to local regions in the input layer or the previous layer. Each neuron is sparsely connected to the region in the previous layer. The depth of the convolutional layer is consistent with the number of feature maps in the previous layer to achieve feature fusion. A learnable convolutional kernel convolves with several feature maps from the previous layer, summing all elements and adding a bias before passing the result to the ReLU activation function.

[0053] Pooling layers reduce the dimensionality of feature maps by decreasing the number of similar feature points, and can also reduce noise and expand the receptive field. Pooling layers can employ either max-pooling or mean-pooling strategies to achieve feature dimensionality reduction.

[0054] Flatten layers are used to "flatten" the input, that is, to reduce the multidimensional input to one dimension, for the transition with fully connected layers.

[0055] Each neuron in a fully connected layer is fully connected to the layer above, resulting in a multidimensional feature vector.

[0056] The output layer can use a softmax classifier to quantize the probability of distributed features and calculate the probability of each category of lower back pain.

[0057] During the training of the deep neural network model, based on the softmax loss function, the backpropagation (BP) algorithm is used to iteratively train and optimize the convolutional network parameters to obtain the optimal convolutional neural network model, thereby achieving accurate identification of multiple categories of lower back pain.

[0058] Accordingly, the present invention also provides a method for identifying low back pain based on a combined electromyography and muscle oxygenation array. See also Figure 7 As shown, the method includes: step S110, generating an electromyography (EMG) topographic map and a muscle oxygenation (MOO) topographic map based on EMG and MOO information of the lower back; step S120, extracting features from the EMG and MOO topographic maps to obtain EMG topographic map feature data and MOO topographic map feature data, wherein the EMG topographic map feature data and MOO topographic map feature data are morphological features of areas of high activity; step S130, inputting the EMG topographic map feature data and MOO topographic map feature data into a trained neural network model to obtain a lower back pain category recognition result.

[0059] It should be understood that in this paper, the type of neural network model, number of layers, activation function, loss function, etc., can all be selected or set according to actual needs.

[0060] In summary, accurate diagnosis of the pathology and etiology is necessary for the effective treatment and rehabilitation of low back pain. However, currently, 85% of low back pain is nonspecific (i.e., low back pain with unknown etiology), leading to diagnoses that are more like symptom descriptions than pathological diagnoses. To improve the accuracy of low back pain identification, this invention obtains a dual-mode physiological information map reflecting neuromuscular activity through the fusion analysis of neuromuscular electrophysiology and blood oxygenation information, and uses a neural network model to obtain the mapping relationship of pain categories. Verification has shown that this invention can objectively and accurately detect pain characteristic information and quantitatively classify pain types, providing a diagnostic reference for the precise and objective classification of low back pain, thereby improving the treatment and rehabilitation effects.

[0061] This invention can be a system, method, and / or computer program product. A computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for causing a processor to implement various aspects of the invention.

[0062] Computer-readable storage media can be tangible devices capable of holding and storing instructions for use by an instruction execution device. Computer-readable storage media can be, for example, but not limited to, electrical storage devices, magnetic storage devices, optical storage devices, electromagnetic storage devices, semiconductor storage devices, or any suitable combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), static random access memory (SRAM), portable compact disc read-only memory (CD-ROM), digital multifunction disc (DVD), memory sticks, floppy disks, mechanical encoding devices, such as punch cards or recessed protrusions storing instructions thereon, and any suitable combination thereof. The computer-readable storage media used herein are not to be construed as transient signals themselves, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through waveguides or other transmission media (e.g., light pulses through fiber optic cables), or electrical signals transmitted through wires.

[0063] The computer-readable program instructions described herein can be downloaded from computer-readable storage media to various computing / processing devices, or downloaded via a network, such as the Internet, local area network, wide area network, and / or wireless network, to an external computer or external storage device. The network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards them to the computer-readable storage media in the respective computing / processing device.

[0064] The computer program instructions used to perform the operations of this invention may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk, C++, Python, etc., and conventional procedural programming languages ​​such as "C" or similar languages. The computer-readable program instructions may be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or may be connected to an external computer (e.g., via the Internet using an Internet service provider). In some embodiments, electronic circuitry, such as programmable logic circuitry, field-programmable gate arrays (FPGAs), or programmable logic arrays (PLAs), is personalized by utilizing state information from the computer-readable program instructions. This electronic circuitry can execute the computer-readable program instructions to implement various aspects of the invention.

[0065] Various aspects of the present invention are described herein with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer-readable program instructions.

[0066] These computer-readable program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that, when executed by the processor of the computer or other programmable data processing apparatus, they create means for implementing the functions / actions specified in one or more blocks of the flowchart and / or block diagram. These computer-readable program instructions can also be stored in a computer-readable storage medium that causes a computer, programmable data processing apparatus, and / or other device to operate in a particular manner; thus, the computer-readable medium storing the instructions comprises an article of manufacture that includes instructions for implementing aspects of the functions / actions specified in one or more blocks of the flowchart and / or block diagram.

[0067] Computer-readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, thereby causing the instructions executed on the computer, other programmable data processing apparatus, or other device to perform the functions / actions specified in one or more boxes of a flowchart and / or block diagram.

[0068] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of an instruction containing one or more executable instructions for implementing a specified logical function. In some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions. It will be known to those skilled in the art that implementation in hardware, implementation in software, and implementation using a combination of software and hardware are equivalent.

[0069] The various embodiments of the present invention have been described above. These descriptions are exemplary and not exhaustive, and are not limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen to best explain the principles, practical application, or technical improvements to the embodiments in the market, or to enable others skilled in the art to understand the embodiments disclosed herein. The scope of the invention is defined by the appended claims.

Claims

1. A low back pain recognition system based on electromyography and muscle oxygenation combined array information, comprising: Data generation module: used to generate electromyography topography maps and muscle oxygenation topography maps based on electromyography and muscle oxygenation information of the lower back; Feature processing module: used to extract features from the electromyography topography map and the oxygenation topography map to obtain electromyography topography map feature data and oxygenation topography map feature data, wherein the electromyography topography map feature data and the oxygenation topography map feature data are morphological features of strong activity areas; Analysis and Diagnosis Module: Used to input the electromyography topography feature data and the muscle oxygenation topography feature data into a trained neural network model to obtain the low back pain category recognition result; The electromyography topography includes root mean square value topography, autoregressive model coefficient topography, average frequency value topography, and median frequency value topography; the muscle oxygenation topography includes deoxyhemoglobin topography, oxygenated hemoglobin topography, and total hemoglobin topography. The electromyography topographic feature data and the muscle oxygenation topographic feature data are obtained according to the following steps: Using a set grayscale threshold T, the pixels of the original color images of electromyography and muscle oxygenation topography are converted into black-and-white binary images to obtain segmented images, wherein the segmented images contain strong activity regions and weak activity regions. With the goal of removing image noise and enhancing image edge details, the segmented image is subjected to morphological processing to obtain a morphologically processed topographic map; Morphological features of highly active areas are extracted from the morphologically processed topographic map and used as the electromyography (EMG) topographic map feature data and the muscle oxygenation (MOO) topographic map feature data. The morphological features of the highly active areas include size features and location distribution features. The location distribution features are obtained by separating the EMG topographic map and the MOO topographic map from the middle spine to the left and right sides and calculating the difference between the topographic map features on the left and right sides.

2. The system according to claim 1, characterized in that, The electromyography topography and the muscle oxygenation topography are obtained according to the following steps: Cubic spline interpolation was performed on the electromyographic information of the lower back, and the matrix structure was formed together with the electromyographic feature values ​​of multiple parts of the lower back. The electromyographic feature values ​​include root mean square value, autoregressive model coefficient, mean frequency value and median frequency value. For each element in the matrix, a linearization method is used to transform it to the range [0,1]. The converted elements are mapped to colors in the three primary color light mode, and then displayed as a visualized electromyographic topography map; The oxygenation information of the muscles in the lower back and its deoxygenated hemoglobin, oxygenated hemoglobin and total hemoglobin were calculated by three interpolation to obtain the distribution of muscle blood oxygen concentration in the test area of ​​the lower back, and to realize near-infrared spectral imaging of blood oxygen information to obtain muscle oxygenation topography.

3. The system according to claim 1, characterized in that, The neural network model is a multi-layer feedforward neural network that uses the sigmoid activation function for non-linear processing, and includes an input layer, multiple hidden layers, and an output layer.

4. The system according to claim 1, characterized in that, The neural network model is a deep convolutional neural network model, which includes, in sequence, an input layer, multiple convolutional layers, multiple pooling layers corresponding to the convolutional layers, a flattening layer, a fully connected layer, and an output layer.

5. The system according to claim 4, characterized in that, The convolutional neural network model is trained according to the following steps: The input layer feeds the raw image pixels of the electromyography and muscle oxygenation topography into the convolutional neural network model; Multiple convolutional layers are used for feature learning and fusion; Multi-layer pooling reduces the dimensionality of learned features by reducing the number of similar feature points; The flattening layer is used to convert the multidimensional features output by the multi-layer pooling layer into one dimension. Each neuron in a fully connected layer is fully connected to the layer above, resulting in a multidimensional distributed feature vector. The output layer uses a softmax classifier to perform probability quantization on the distributed feature vectors and calculate the probability of each category of lower back pain. Based on the softmax loss function, the parameters of the convolutional neural network model are iteratively trained and optimized through the error backpropagation algorithm.

6. The system according to claim 1, characterized in that, The neural network model is the optimal performance model selected by performing K-fold cross-validation on various types of pre-trained models.

7. A method for identifying low back pain based on electromyography and muscle oxygenation combined array information, comprising the following steps: Electromyography (EMG) and oxygenation (OO) topography maps are generated based on EMG and OO data from the lower back. Feature extraction is performed on the electromyography topography map and the oxygenation topography map to obtain electromyography topography map feature data and oxygenation topography map feature data, wherein the electromyography topography map feature data and the oxygenation topography map feature data are morphological features of strong activity areas; The electromyography topography feature data and the muscle oxygenation topography feature data are input into a trained neural network model to obtain the low back pain category recognition result. The electromyography topography includes root mean square value topography, autoregressive model coefficient topography, average frequency value topography, and median frequency value topography; the muscle oxygenation topography includes deoxyhemoglobin topography, oxygenated hemoglobin topography, and total hemoglobin topography. The electromyography topographic feature data and the muscle oxygenation topographic feature data are obtained according to the following steps: Using a set grayscale threshold T, the pixels of the original color images of electromyography and muscle oxygenation topography are converted into black-and-white binary images to obtain segmented images, wherein the segmented images contain strong activity regions and weak activity regions. With the goal of removing image noise and enhancing image edge details, the segmented image is subjected to morphological processing to obtain a morphologically processed topographic map; Morphological features of highly active areas are extracted from the morphologically processed topographic map and used as the electromyography (EMG) topographic map feature data and the muscle oxygenation (MOO) topographic map feature data. The morphological features of the highly active areas include size features and location distribution features. The location distribution features are obtained by separating the EMG topographic map and the MOO topographic map from the middle spine to the left and right sides and calculating the difference between the topographic map features on the left and right sides.

8. A computer-readable storage medium having a computer program stored thereon, wherein, When the computer program is executed by the processor, it implements the steps of the method according to claim 7.

Images