Glycometabolism dynamic monitoring method and system based on multispectral imaging

Abstract

The invention discloses a glucose metabolism dynamic monitoring method and system based on multispectral imaging, and relates to the technical field of glucose metabolism dynamic monitoring, and the method comprises the following steps: obtaining polarization state distribution data and interferometric phase data of a target tissue, and constructing a Jones matrix; performing inverse polarization transformation compensation on the Jones matrix to obtain a polarization interference signal and extract a phase distribution diagram; dividing the phase distribution diagram into a plurality of sub-regions, extracting a phase gradient feature of each sub-region, obtaining a plurality of phase resolving threshold candidate sets, and performing ambiguity suppression to obtain a plurality of first phase mapping features; constructing a three-dimensional glucose metabolism activity model by adopting a feature fusion algorithm, and carrying out glucose metabolism abnormality identification; and data analysis is performed on an identification result to obtain an optimal phase solution threshold value, and the optimal phase solution threshold value is applied to the glucose metabolism monitoring process, so that the problem that the accuracy of noninvasive, dynamic and spatial resolution and high-precision monitoring of the tissue glucose metabolism process is reduced due to the fact that the precision of phase solution ambiguity of a traditional method is unstable is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of dynamic monitoring of sugar metabolism, and more specifically, to a method and system for dynamic monitoring of sugar metabolism based on multispectral imaging. Background Art

[0002] Diabetes, particularly type 2 diabetes, has become a serious public health problem worldwide, with its prevalence continuing to rise and becoming younger. While traditional blood glucose monitoring methods, such as fingertip and venous sampling, can provide blood glucose data, they are limited by their invasiveness, low frequency, and inability to reflect dynamic changes. These limitations make them unable to meet the demand for continuous, accurate, and noninvasive monitoring of glucose metabolism. Furthermore, glucose metabolism is not simply a single change in blood glucose levels; it involves complex physiological processes such as glucose uptake, transport, oxidation, and local metabolic activity in tissues. Consequently, researchers have begun exploring new monitoring technologies that can reflect the spatial distribution and temporal dynamics of glucose metabolism, leveraging tissue optical properties, multispectral imaging, and biosignal analysis. Driven by precision medicine, smart wearable devices, and artificial intelligence algorithms, the development of noninvasive, dynamic, and spatially resolved glucose metabolism monitoring systems has become a cutting-edge area. These systems can be used for early screening and intervention for diabetes, as well as for monitoring the immediate impact of factors such as exercise and diet on glucose metabolism, enabling truly personalized metabolic health management.

[0003] However, the traditional approach based on a single solution model or a fixed threshold is difficult to accurately handle phase jumps in high-noise environments or complex tissue structures, and is prone to problems such as phase discontinuity and boundary error accumulation, resulting in the phase map not being truly interpretable. Secondly, the traditional method lacks detailed modeling of the spatial phase gradient structure, making it difficult to effectively suppress local phase ambiguity and modal degradation, especially in high-throughput dynamic monitoring of glucose metabolism, making it difficult to maintain the stability of phase accuracy over time. In response to the above problems, the present invention proposes a solution. Summary of the Invention

[0004] To overcome the above-mentioned deficiencies of the prior art, embodiments of the present invention provide a method and system for dynamic monitoring of glucose metabolism based on multispectral imaging. By integrating multispectral polarization imaging with phase solution technology, a three-dimensional glucose metabolism activity model is constructed to address the unstable accuracy of phase solution ambiguity in traditional methods, which leads to a decrease in the accuracy of non-invasive, dynamic, spatially resolved, and high-precision monitoring of tissue glucose metabolism processes.

[0005] To achieve the above object, the present invention provides the following technical solutions: The method for dynamic monitoring of glucose metabolism based on multispectral imaging includes the following steps: obtaining polarization state distribution data and interference phase data of the target tissue and constructing a Jones matrix; performing inverse polarization transformation compensation on the Jones matrix to obtain a polarization interference signal and extract a phase distribution map; dividing the phase distribution map into several sub-regions, and extracting the phase gradient characteristics of each sub-region to obtain several phase solution threshold candidate sets; performing ambiguity suppression on the several phase solution threshold candidate sets to obtain several first phase mapping features; constructing a three-dimensional glucose metabolism activity model based on the first phase mapping features using a feature fusion algorithm to identify glucose metabolism abnormalities; performing data analysis on the identification results to obtain the optimal phase solution threshold, and applying it to the glucose metabolism monitoring process.

[0006] In a preferred embodiment, the polarization state distribution data and interference phase data of the target tissue are obtained and the Jones matrix is constructed, specifically: the polarization state distribution data and interference phase data of the target tissue are obtained, and the Mueller matrix is constructed; based on the Mueller matrix, the polarization state direction angle and ellipticity of each pixel point of the target tissue are calculated; the polarization state direction angle and ellipticity of each pixel point are superimposed by the Jones vector to obtain the first Jones matrix of each pixel point; and the first Jones matrix is normalized to obtain the Jones matrix.

[0007] In a preferred embodiment, the Jones matrix is subjected to inverse polarization transformation compensation to obtain a polarization interference signal and extract a phase distribution diagram, specifically: based on the preset birefringence characteristics of the target tissue, an inverse polarization transformation compensation model is constructed; the Jones matrix is input into the inverse polarization transformation compensation model, and a polarization interference signal is obtained through matrix decomposition; and the polarization interference signal is subjected to Fourier transformation to obtain a phase distribution diagram.

[0008] In a preferred embodiment, the phase distribution diagram is divided into several sub-regions, and the phase gradient characteristics of each sub-region are extracted to obtain several phase solution threshold candidate sets, specifically: the phase distribution diagram is divided into several sub-regions, and the horizontal phase gradient and vertical phase gradient of each sub-region are calculated to obtain a two-dimensional gradient vector sequence; based on the DTW algorithm, the two-dimensional gradient vector sequences of adjacent sub-regions are path matched, and the direction difference between the path pairs is calculated; the phase jump variable is calculated based on the direction difference between the path pairs, and the corrected gradient direction is determined by the main direction extraction method; according to the path matching result and based on the corrected gradient direction, the phase gradient jump error at the sub-region boundary is corrected; the corrected maximum gradient direction is extracted as the phase gradient feature of each path; and the weighted average method is used to fuse the gradient features of each path to obtain a phase solution threshold candidate set for each sub-region.

[0009] In a preferred embodiment, the ambiguity suppression is performed on several phase solution threshold candidate sets respectively to obtain several first phase mapping features, specifically: the local gradient vector in each phase solution threshold candidate set is extracted and the main gradient direction angle is obtained; a two-dimensional rotation matrix is constructed based on the main gradient direction angle, and all local gradient vectors are aligned and transformed to obtain several first local gradient vectors; and several first local gradient vectors are weightedly fused according to preset weights to obtain the first phase mapping feature.

[0010] In a preferred embodiment, the three-dimensional glucose metabolism activity model is constructed based on the first phase mapping feature using a feature fusion algorithm to identify abnormal glucose metabolism. Specifically, the first phase mapping features of each sub-region are spliced according to the spatial position to obtain a first phase mapping image; based on the light absorption coefficient of the preset multispectral band, the glucose metabolism activity parameters of the target tissue are calculated, and the glucose metabolism activity parameters include the glucose uptake rate and metabolic heat production distribution; the first phase mapping image is fused with the glucose substitute metabolism activity parameters using a voxelization algorithm to construct a three-dimensional glucose metabolism activity model; and normal and abnormal metabolic areas are divided based on the color coding output by the three-dimensional glucose metabolism activity model.

[0011] In a preferred embodiment, the data analysis of the recognition results is performed to obtain the optimal phase resolution threshold, and the optimal phase resolution threshold is applied to the glucose metabolism monitoring process, specifically: based on the abnormal recognition results, the false detection rate and missed detection rate corresponding to different phase resolution thresholds are statistically analyzed; with the weighted sum of the false detection rate and missed detection rate being minimized as the optimization goal, the optimal phase resolution threshold is searched through a particle swarm algorithm; the optimal threshold is embedded in the real-time processing module of the multispectral imaging system for dynamically adjusting the phase resolution accuracy in subsequent monitoring; based on the adjusted phase resolution accuracy, the target tissue is continuously scanned and a metabolic activity heat map is generated; When the confidence level of the abnormal area in the heat map exceeds the preset alarm threshold, a visual early warning signal is triggered and a diagnostic report is generated.

[0012] The technical effects and advantages of the method and system for dynamic monitoring of glucose metabolism based on multispectral imaging of the present invention are as follows: 1. The present invention uses multispectral polarization imaging technology, combined with Jones matrix construction and inverse polarization transformation compensation, to effectively suppress tissue optical scattering interference and improve imaging contrast and phase extraction accuracy; secondly, by dividing the phase distribution map into sub-regions and using the DTW algorithm and gradient direction correction method to extract phase gradient features, the spatial resolution and error correction capability of the phase solution are significantly improved; further, the ambiguity suppression and main direction alignment method enhance the consistency of local features, effectively construct a robust first phase mapping feature, and provide a solid foundation for the subsequent construction of a three-dimensional glucose metabolism activity model; the model integrates the light absorption coefficient and voxel algorithm, not only realizing the spatial visualization of glucose uptake rate and metabolic heat production, but also accurately identifying metabolic abnormality areas through color coding, with extremely high clinical auxiliary diagnosis value. In addition, the system also combines the particle swarm optimization algorithm to adaptively extract the optimal phase solution threshold from the recognition results, and use it to dynamically adjust subsequent monitoring parameters in real time to improve the adaptive ability and diagnostic accuracy of the monitoring system. In summary, the present invention integrates polarization analysis, image processing, feature fusion and intelligent optimization technologies based on multispectral imaging. It has the advantages of high imaging accuracy, strong recognition robustness, excellent real-time performance and high degree of automation. It is suitable for multiple medical fields such as diabetes and tumor metabolism research, showing broad application prospects and industrialization potential. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 Schematic diagram of the process of the method for dynamic monitoring of glucose metabolism based on multispectral imaging of the present invention.

[0014] Figure 2 Schematic diagram of the structure of the sugar metabolism dynamic monitoring system based on multispectral imaging of the present invention. DETAILED DESCRIPTION

[0015] The following will provide a clear and complete description of the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0016] Example 1, Figure 1 The present invention provides a method for dynamic monitoring of glucose metabolism based on multispectral imaging, which includes the following steps: S1, obtain the polarization state distribution data and interference phase data of the target tissue and construct the Jones matrix; In this example, the polarization state distribution data and interference phase data of the target tissue are obtained, and the Jones matrix is constructed, specifically: Obtain polarization state distribution data and interference phase data of the target tissue and construct the Mueller matrix; Based on the Mueller matrix, the polarization direction angle and ellipticity of each pixel of the target tissue are calculated; The polarization direction angle and ellipticity of each pixel are superimposed by Jones vectors to obtain the first Jones matrix of each pixel; The first Jones matrix is normalized to obtain the Jones matrix.

[0017] It should be noted that a multi-channel polarization interferometer imaging device performs multispectral scanning of the target tissue at multiple polarization states (e.g., 0°, 45°, 90°, 135°, and circular polarization), acquiring corresponding polarization reflection intensity images and interference fringe images. Using a Stokes parameter measurement module, the Mueller matrix for each pixel of the target tissue is calculated by combining the incident polarization information of the light source with the received polarization configuration. The Mueller matrix is a 4×4 real matrix used in polarization optics to describe the effect of an object or medium on changes in the polarization state of light.

[0018] Furthermore, by first constructing the Mueller matrix and extracting the polarization state angle and ellipticity, and then superimposing them into Jones vectors and normalizing them to generate the Jones matrix, high-precision modeling of the polarization information of the target tissue is achieved, which has significant advantages. Compared with the method of directly constructing the Jones matrix, this process fully utilizes the Mueller matrix's complete expression ability of the composite polarization characteristics, improving the accuracy and stability of the acquisition of polarization state data. At the same time, by extracting the polarization state angle and ellipticity pixel by pixel and superimposing them into Jones vectors, the information of tiny polarization changes in the tissue structure is effectively retained, and the spatial resolution of the polarization characteristics is enhanced. The normalization process further eliminates the interference of intensity differences in the measurement and improves the accuracy and consistency of subsequent interferometric phase calculations. The overall process combines the strong robustness of the Mueller matrix with the high resolution characteristics of the Jones matrix, achieving accurate extraction of polarization features at multiple scales, laying a solid foundation for subsequent phase compensation, feature mapping and three-dimensional modeling, and is particularly suitable for high-sensitivity and high-resolution application scenarios in tissue optical imaging.

[0019] S2, perform inverse polarization transformation compensation on the Jones matrix to obtain the polarization interference signal and extract the phase distribution map; In this example, the Jones matrix is subjected to inverse polarization transformation compensation to obtain the polarization interference signal and extract the phase distribution diagram, specifically: Based on the preset birefringence characteristics of the target tissue, an inverse polarization transformation compensation model is constructed; The Jones matrix is input into the inverse polarization transformation compensation model, and the polarization interference signal is obtained through matrix decomposition; Perform Fourier transform on the polarization interference signal to obtain a phase distribution diagram.

[0020] It should be noted that the construction process of the inverse polarization transformation compensation model is as follows: Based on the birefringence characteristic parameters of the target tissue, including the slow axis direction angle and phase delay. First, these parameters are used to establish the equivalent birefringence Jones matrix model of the tissue. This matrix describes the polarization state changes caused by the propagation of light in a birefringent medium. Then, in order to achieve reverse compensation for the birefringence effect in the tissue, the inverse matrix of the Jones matrix is calculated, usually by solving its adjoint matrix and dividing it by the determinant. On this basis, the original Jones matrix obtained by actual observation is input into the compensation model to obtain a compensated correction matrix, which effectively removes the polarization perturbation caused by the optical anisotropy of the tissue. Finally, the main diagonal complex elements are extracted from the corrected Jones matrix as polarization interference signals and input into the Fourier transform module. The main carrier frequency component of the frequency domain signal is extracted and the phase distribution map is reconstructed to achieve high-precision measurement of the phase change of the optical path inside the tissue, providing key input for subsequent glucose metabolism modeling.

[0021] Furthermore, the inverse polarization transformation compensation model, constructed based on the birefringence characteristics of the preset target tissue, can accurately compensate for the polarization distortion caused by the optical anisotropy within the tissue by effectively correcting the Jones matrix, significantly improving the purity of the polarization interference signal and the accuracy of phase extraction. This method separates the true polarization interference signal through matrix decomposition and then extracts a high-resolution phase distribution map using Fourier transform, achieving sensitive capture of tiny optical path differences within the tissue and improving the spatial consistency and signal-to-noise ratio of the imaging. The advantage of this technology is not only that it improves the resolution and robustness of tissue optical information, but also provides key technical support for non-invasive blood glucose monitoring. Optical changes in glucose metabolism are often accompanied by birefringence and scattering effects. Traditional optical measurements are easily affected by the complex structure of the tissue and produce errors. This inverse transformation compensation model effectively overcomes this problem, enabling multispectral polarization imaging to more accurately and in real time reflect changes in blood glucose levels and their metabolic dynamics, promoting the development of non-invasive blood glucose testing towards higher sensitivity and stability, and has broad clinical application potential.

[0022] S3, dividing the phase distribution map into several sub-regions, and extracting the phase gradient features of each sub-region to obtain several phase solution threshold candidate sets; In this example, the phase distribution map is divided into several sub-regions, and the phase gradient features of each sub-region are extracted to obtain several phase solution threshold candidate sets, specifically: The phase distribution map is divided into several sub-regions, and the horizontal phase gradient and vertical phase gradient of each sub-region are calculated to obtain a two-dimensional gradient vector sequence; Based on the DTW algorithm, path matching is performed on the two-dimensional gradient vector sequences of adjacent sub-regions, and the direction difference between the path pairs is calculated; The phase jump variable is calculated based on the direction difference between the path pairs, and the correction gradient direction is determined by the main direction extraction method; According to the path matching result and based on the correction gradient direction, the phase gradient jump error at the sub-region boundary is corrected; Extract the corrected maximum gradient direction as the phase gradient feature of each path; The weighted average method is used to fuse the gradient features of each path to obtain the candidate set of phase solution thresholds for each sub-region.

[0023] It should be noted that the phase distribution map is divided into several sub-regions and the phase gradient characteristics of each sub-region are extracted. By calculating the horizontal and vertical phase gradients to form a two-dimensional gradient vector sequence, the gradient sequences of adjacent sub-regions are then path-matched using the dynamic time warping (DTW) algorithm. This effectively addresses the errors caused by local discontinuities and noise interference in the phase map. By calculating the directional difference between path pairs and introducing phase jump variables, combined with the main direction extraction method, the gradient direction is accurately corrected, significantly reducing the phase gradient jump error at the sub-region boundaries and improving the continuity and accuracy of the gradient characteristics.

[0024] Finally, the weighted average method is used to fuse the phase gradient features of each path to ensure the robustness and representativeness of the sub-region phase solution threshold candidate set. The advantage of this method is that it not only enhances the accuracy and robustness of the phase solution, but also can effectively deal with local abnormalities caused by complex tissue structures, greatly improving the applicability and sensitivity of the multispectral imaging system in non-invasive blood glucose monitoring. Small changes in blood glucose metabolism often cause subtle adjustments in the optical properties of local tissues. This technology achieves efficient identification of areas with abnormal glucose metabolism by accurately capturing and correcting phase gradient changes, providing a stable and reliable quantitative basis for dynamic monitoring of non-invasive blood glucose, and promoting the development of non-invasive blood glucose detection towards higher accuracy and real-time performance. S4, performing ambiguity suppression on a plurality of phase resolution threshold candidate sets to obtain a plurality of first phase mapping features; In this example, ambiguity suppression is performed on several phase resolution threshold candidate sets to obtain several first phase mapping features, specifically: Extract the local gradient vector in each phase solution threshold candidate set and obtain the main gradient direction angle; Construct a two-dimensional rotation matrix based on the main gradient direction angle, and align all local gradient vectors to obtain several first local gradient vectors; A plurality of first local gradient vectors are weightedly fused according to preset weights to obtain a first phase mapping feature.

[0025] It should be noted that for the phase solution threshold candidate sets extracted from multiple sub-regions, a sequence of local phase gradient vectors is first selected from each candidate set. The orientation angles of these gradient vectors are calculated to determine the main gradient orientation angle θ for the local region. Subsequently, based on the main gradient orientation angle θ, a two-dimensional rotation matrix R(θ) is constructed. This rotation matrix is used to rotate and align all local gradient vectors, eliminating directional differences between different candidate sets. Specifically, for each gradient vector, a first local gradient vector is obtained through matrix multiplication, so that all gradient vectors are compared and fused under a unified orientation reference.

[0026] Subsequently, each first local gradient vector is weighted and fused according to pre-set weight coefficients. The weights can be adjusted based on local gradient strength, noise level, or regional importance to highlight key features and suppress the influence of noise. The fusion result is the first phase map feature, which reflects the comprehensive phase gradient information of multiple candidate sets after ambiguity suppression and direction alignment. This method effectively improves the accuracy and stability of phase solution and provides high-quality input for the subsequent construction of a three-dimensional glucose metabolism activity model.

[0027] S5, constructing a three-dimensional glucose metabolism activity model based on the first phase mapping feature using a feature fusion algorithm to identify glucose metabolism abnormalities; In this example, a feature fusion algorithm is used to construct a three-dimensional glucose metabolism activity model based on the first phase mapping feature to identify abnormal glucose metabolism. Specifically, splicing the first phase mapping features of each sub-region according to spatial position to obtain a first phase mapping image; Calculating glucose metabolism activity parameters of the target tissue based on the light absorption coefficient of the preset multispectral band, wherein the glucose metabolism activity parameters include glucose uptake rate and metabolic heat production distribution; The voxelization algorithm was used to fuse the first phase map with the glucose metabolism activity parameters to construct a three-dimensional glucose metabolism activity model. Normal and abnormal metabolic areas were divided based on the color coding output from the three-dimensional glucose metabolism activity model.

[0028] It is important to note that by spatially concatenating the first phase map features of each subregion to form an overall first phase map, spatial integration of local tissue glucose metabolism characteristics is achieved, thereby enhancing image continuity and overall expressiveness. Furthermore, by combining the tissue light absorption characteristics of each band in the multispectral imaging system, key glucose metabolism activity parameters, such as glucose uptake rate and metabolic heat production distribution, are calculated. These parameters sensitively reflect the efficiency and metabolic intensity of cellular glucose utilization, thereby revealing the state of tissue glucose metabolism. Fusion of the two-dimensional phase map with three-dimensional metabolic parameters via a voxelization algorithm not only achieves structure-function coupled modeling from two-dimensional to three-dimensional, but also enhances the spatial precision and layering of glucose metabolism activity expression. The resulting three-dimensional glucose metabolism activity model intuitively annotates normal and abnormal regions with color coding, facilitating rapid identification of the distribution of abnormal glucose metabolism. The greatest advantage of this method is that it enables high-resolution, non-invasive tissue metabolism imaging, significantly improving the accuracy and visualization of non-invasive blood glucose monitoring. It is particularly suitable for continuously monitoring individual glucose metabolism fluctuations, providing a scientific basis and technical support for early warning and personalized treatment of diabetes.

[0029] S6, performing data analysis on the recognition results to obtain the optimal phase resolution threshold, which is then applied to the glucose metabolism monitoring process.

[0030] In this example, data analysis is performed on the recognition results to obtain the optimal phase resolution threshold, which is then applied to the glucose metabolism monitoring process. Specifically: According to the anomaly recognition results, the false detection rate and missed detection rate corresponding to different phase resolution thresholds are calculated; Taking the minimum weighted sum of false detection rate and missed detection rate as the optimization goal, the particle swarm algorithm is used to search for the optimal phase solution threshold; The optimal threshold is embedded in the real-time processing module of the multispectral imaging system to dynamically adjust the phase resolution accuracy in subsequent monitoring; Based on the adjusted phase resolution accuracy, the target tissue is continuously scanned and a metabolic activity heat map is generated; When the confidence level of the abnormal area in the heat map exceeds the preset alarm threshold, a visual early warning signal is triggered and a diagnostic report is generated.

[0031] It is important to note that by analyzing the results of abnormal glucose metabolism identification, a quantitative assessment of the system's recognition performance was achieved based on the statistical analysis of false detection and missed detection rates at different phase resolution thresholds. By minimizing the weighted sum of these values as the optimization objective, a particle swarm optimization algorithm was used to efficiently search for the optimal phase resolution threshold, thereby improving overall recognition accuracy and robustness. This optimal threshold can be automatically embedded in the real-time processing module of the multispectral imaging system, enabling dynamic adaptive adjustment during subsequent detection processes, effectively addressing imaging changes in different individuals or tissue states and ensuring long-term monitoring stability. The adjusted phase resolution accuracy significantly improves the resolution of metabolic activity thermograms and the clarity of abnormal region boundaries, enabling precise capture of the evolution of abnormal metabolism during continuous scanning. When the confidence level of an abnormal region in the thermogram exceeds a preset alarm threshold, the system automatically triggers a visual warning and generates a diagnostic report, providing clinicians with real-time, quantitative support for judgment. This method significantly enhances the intelligent level of noninvasive blood glucose monitoring, not only enhancing the early identification of abnormal glucose metabolism but also providing a personalized dynamic feedback mechanism, meeting the practical needs of continuous, accurate, and noninvasive monitoring in chronic disease management.

[0032] Example 2, Figure 2 The present invention provides a dynamic monitoring system for glucose metabolism based on multispectral imaging, which includes a data acquisition module, a phase distribution module, a feature extraction module, a feature mapping module, a model building module, and a metabolic monitoring module: A data acquisition module is used to obtain polarization state distribution data and interference phase data of the target tissue and construct a Jones matrix; Phase distribution module, used to perform inverse polarization transformation compensation on the Jones matrix, obtain polarization interference signal and extract phase distribution diagram; A feature extraction module is used to divide the phase distribution map into several sub-regions and extract the phase gradient features of each sub-region to obtain several candidate sets of phase solution thresholds; A feature mapping module is used to perform ambiguity suppression on a plurality of phase resolution threshold candidate sets to obtain a plurality of first phase mapping features; A model building module is used to build a three-dimensional glucose metabolism activity model based on the first phase mapping feature using a feature fusion algorithm to identify abnormal glucose metabolism; The metabolic monitoring module is used to perform data analysis on the recognition results, obtain the optimal phase solution threshold, and apply it to the glucose metabolism monitoring process.

[0033] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formulas are set by technicians in this field according to actual conditions.

[0034] The above embodiments may be implemented in whole or in part through software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments may be implemented in whole or in part in the form of a computer program product.

[0035] Those skilled in the art will appreciate that the modules and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0036] In addition, each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0037] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

[0038] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for dynamic monitoring of glucose metabolism based on multispectral imaging, characterized in that: The following steps are involved: Obtain polarization state distribution data and interference phase data of the target tissue and construct the Jones matrix; Perform inverse polarization transformation compensation on the Jones matrix to obtain polarization interference signal and extract phase distribution map; The phase distribution map is divided into several sub-regions, and the phase gradient features of each sub-region are extracted to obtain several candidate sets of phase solution thresholds; Performing ambiguity suppression on a plurality of phase resolution threshold candidate sets respectively to obtain a plurality of first phase mapping features; Based on the first phase mapping features, a feature fusion algorithm is used to construct a three-dimensional glucose metabolism activity model to identify abnormal glucose metabolism. The recognition results were analyzed to obtain the optimal phase resolution threshold, which was then applied to the glucose metabolism monitoring process.

2. The method for dynamic monitoring of glucose metabolism based on multispectral imaging according to claim 1, characterized in that: The polarization state distribution data and interference phase data of the target tissue are obtained, and the Jones matrix is constructed, specifically: Obtain polarization state distribution data and interference phase data of the target tissue and construct the Mueller matrix; Based on the Mueller matrix, the polarization direction angle and ellipticity of each pixel of the target tissue are calculated; The polarization direction angle and ellipticity of each pixel are superimposed by Jones vectors to obtain the first Jones matrix of each pixel; The first Jones matrix is normalized to obtain the Jones matrix.

3. The method for dynamic monitoring of glucose metabolism based on multispectral imaging according to claim 2, characterized in that: The inverse polarization transformation compensation is performed on the Jones matrix to obtain the polarization interference signal and extract the phase distribution diagram, specifically: Based on the preset birefringence characteristics of the target tissue, an inverse polarization transformation compensation model is constructed; The Jones matrix is input into the inverse polarization transformation compensation model, and the polarization interference signal is obtained through matrix decomposition; Perform Fourier transform on the polarization interference signal to obtain a phase distribution diagram.

4. The method for dynamic monitoring of glucose metabolism based on multispectral imaging according to claim 3, characterized in that: The phase distribution map is divided into several sub-regions, and the phase gradient features of each sub-region are extracted to obtain several phase solution threshold candidate sets, specifically: The phase distribution map is divided into several sub-regions, and the horizontal phase gradient and vertical phase gradient of each sub-region are calculated to obtain a two-dimensional gradient vector sequence; Based on the DTW algorithm, path matching is performed on the two-dimensional gradient vector sequences of adjacent sub-regions, and the direction difference between the path pairs is calculated; The phase jump variable is calculated based on the direction difference between the path pairs, and the correction gradient direction is determined by the main direction extraction method; According to the path matching result and based on the correction gradient direction, the phase gradient jump error at the sub-region boundary is corrected; Extract the corrected maximum gradient direction as the phase gradient feature of each path; The weighted average method is used to fuse the gradient features of each path to obtain the candidate set of phase solution thresholds for each sub-region.

5. The method for dynamic monitoring of glucose metabolism based on multispectral imaging according to claim 4, characterized in that: The ambiguity suppression is performed on a plurality of phase resolution threshold candidate sets to obtain a plurality of first phase mapping features, specifically: Extract the local gradient vector in each phase solution threshold candidate set and obtain the main gradient direction angle; Construct a two-dimensional rotation matrix based on the main gradient direction angle, and align all local gradient vectors to obtain several first local gradient vectors; A plurality of first local gradient vectors are weightedly fused according to preset weights to obtain a first phase mapping feature.

6. The method for dynamic monitoring of glucose metabolism based on multispectral imaging according to claim 5, characterized in that: The three-dimensional glucose metabolism activity model is constructed based on the first phase mapping feature using a feature fusion algorithm to identify glucose metabolism abnormalities, specifically: splicing the first phase mapping features of each sub-region according to spatial position to obtain a first phase mapping image; Calculating glucose metabolism activity parameters of the target tissue based on the light absorption coefficient of the preset multispectral band, wherein the glucose metabolism activity parameters include glucose uptake rate and metabolic heat production distribution; The voxelization algorithm was used to fuse the first phase map with the glucose metabolism activity parameters to construct a three-dimensional glucose metabolism activity model. Normal and abnormal metabolic areas were divided based on the color coding output from the three-dimensional glucose metabolism activity model.

7. The method for dynamic monitoring of glucose metabolism based on multispectral imaging according to claim 6, characterized in that: The data analysis of the recognition results is performed to obtain the optimal phase solution threshold, which is applied to the glucose metabolism monitoring process, specifically: According to the anomaly recognition results, the false detection rate and missed detection rate corresponding to different phase resolution thresholds are calculated; Taking the minimum weighted sum of false detection rate and missed detection rate as the optimization goal, the particle swarm algorithm is used to search for the optimal phase solution threshold; The optimal threshold is embedded in the real-time processing module of the multispectral imaging system to dynamically adjust the phase resolution accuracy in subsequent monitoring; Based on the adjusted phase resolution accuracy, the target tissue is continuously scanned and a metabolic activity heat map is generated; When the confidence level of the abnormal area in the heat map exceeds the preset alarm threshold, a visual early warning signal is triggered and a diagnostic report is generated.

8. A system using the method for dynamic monitoring of glucose metabolism based on multispectral imaging according to any one of claims 1 to 7, characterized in that: Including data acquisition module, phase distribution module, feature extraction module, feature mapping module, model building module, and metabolic monitoring module: A data acquisition module is used to obtain polarization state distribution data and interference phase data of the target tissue and construct a Jones matrix; Phase distribution module, used to perform inverse polarization transformation compensation on the Jones matrix, obtain polarization interference signal and extract phase distribution diagram; A feature extraction module is used to divide the phase distribution map into several sub-regions and extract the phase gradient features of each sub-region to obtain several candidate sets of phase solution thresholds; A feature mapping module is used to perform ambiguity suppression on a plurality of phase resolution threshold candidate sets to obtain a plurality of first phase mapping features; A model building module is used to build a three-dimensional glucose metabolism activity model based on the first phase mapping feature using a feature fusion algorithm to identify abnormal glucose metabolism; The metabolic monitoring module is used to perform data analysis on the recognition results, obtain the optimal phase solution threshold, and apply it to the glucose metabolism monitoring process.

Images