Deep tissue damage detection method and system based on dielectric spectrum analysis

Abstract

The invention relates to the technical field of biomedical detection, in particular to a deep tissue damage detection method and system based on dielectric spectrum analysis. The method comprises the following steps: acquiring a complex dielectric constant frequency spectrum of a pelvic tissue to be detected; an improved bipolar Cole-Cole model is established; establishing a mapping relation between tissue dielectric parameters and damage degrees; establishing a damage judgment function; obtaining judgment results of tissue damage grades and damage types; and the deep tissue damage detection is realized. According to the method, the time-frequency joint sparse reconstruction algorithm and the deep generative adversarial network are utilized to realize high-precision inversion of the deep organization complex dielectric constant spectrum; by constructing an improved bipolar Cole-Cole model, the layered dielectric characteristics of the tissue are accurately represented; by establishing a damage judgment function and combining damage characteristic function interaction analysis, precise classification of single damage and composite damage is achieved. According to the method, the sensitivity, specificity and clinical applicability of deep tissue damage detection are effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of biomedical detection, and particularly relates to a deep tissue injury detection method and system based on dielectric spectroscopy analysis. Background Art

[0002] Early detection of deep tissue injury is a major challenge in clinical medicine. Especially for pelvic tissues, due to their deep location and complex structure, it is difficult for traditional imaging methods to achieve high-sensitivity and specific injury diagnosis. Tissue injury will cause the destruction of cell membrane integrity, changes in ion concentration and abnormal water distribution, which will in turn cause changes in dielectric properties, such as dielectric constant and conductivity. Dielectric spectroscopy analysis technology provides a new idea for non-invasive detection of deep tissue injury by detecting the electromagnetic response characteristics of tissues in a wide frequency range.

[0003] In the prior art, single-frequency impedance measurement or the classical Cole-Cole model is mainly used for tissue dielectric analysis, which has the following defects: First, it is difficult to effectively separate the signal aliasing between deep tissues and surface tissues, resulting in large errors in the extraction of dielectric parameters; Second, the existing Cole-Cole model has insufficient fitting accuracy in a wide frequency range and is difficult to characterize the non-Debye relaxation characteristics of deep tissues with depth changes; Third, there is a lack of a multi-layer mapping model between dielectric parameters and injury degree, resulting in low accuracy in determining injury grades and types; Fourth, the ability to analyze the interaction effects of complex injuries is insufficient, and it cannot meet the needs of clinical precise diagnosis.

[0004] At present, there is not enough research work on deep tissue injury detection, and there is no specific deep tissue injury detection method based on dielectric spectroscopy analysis. Summary of the Invention

[0005] Aiming at the defects in the prior art, the present invention provides a deep tissue injury detection method and system based on dielectric spectroscopy analysis.

[0006] In a first aspect, a deep tissue damage detection method based on dielectric spectrum analysis provided by the present invention includes the following steps: obtaining the complex dielectric constant spectrum of the pelvic tissue to be measured; establishing an improved bipolar Cole-Cole model based on the complex dielectric constant spectrum; using the improved bipolar Cole-Cole model to establish a mapping relationship between tissue dielectric parameters and the degree of damage; based on the mapping relationship, establishing a damage determination function for the pelvic tissue to be measured; using the damage determination function to obtain the determination results of the damage level and damage type of the pelvic tissue to be measured; and realizing the damage detection of deep tissues through the determination results. By obtaining the complex dielectric constant spectrum of the pelvic tissue to be measured, the present invention realizes the high-precision quantitative characterization of the dielectric properties of deep tissues, providing a reliable data basis for improving the sensitivity and specificity of damage detection; by establishing an improved bipolar Cole-Cole model, it overcomes the defect of insufficient fitting accuracy of the existing model in the wide frequency range, providing a more accurate mathematical description of the dielectric behavior of pelvic tissues; by establishing a mapping relationship between tissue dielectric parameters and the degree of damage, it reveals the correlation law between dielectric property changes and tissue pathological states, creating conditions for the quantitative assessment of damage; by constructing a damage determination function, it realizes the intelligent damage classification of multi-parameter fusion, providing an objective basis for distinguishing different damage levels and types; and through the determination results of the damage level and damage type, it realizes the non-invasive and rapid detection of deep tissue damage, providing a new technical means for clinical pelvic disease diagnosis and treatment effect monitoring.

[0007] Optionally, the obtaining of the complex dielectric constant spectrum of the pelvic tissue to be measured includes: transmitting a swept-frequency electromagnetic wave signal to the pelvic tissue to be measured to obtain a transmitted signal and a reflected signal; designing a time-frequency joint sparse reconstruction algorithm based on the transmitted signal and the reflected signal; using the time-frequency joint sparse reconstruction algorithm to obtain the signal separation result of the pelvic tissue to be measured and the surface tissue; establishing a deep generative adversarial network dielectric inversion model based on the signal separation result; and obtaining the complex dielectric constant spectrum of the pelvic tissue to be measured at different depths through the deep generative adversarial network dielectric inversion model. By transmitting a swept-frequency electromagnetic wave signal to the pelvic tissue to be measured and collecting the transmitted signal and the reflected signal, and combining with the time-frequency joint sparse reconstruction algorithm, the present invention realizes the efficient separation of deep tissue signals and surface tissue signals, laying a foundation for improving the accuracy of deep tissue dielectric parameter extraction; by designing a time-frequency joint sparse reconstruction algorithm for signal separation, it effectively suppresses the interference caused by the signal aliasing of multiple layers of tissues, providing an important technical support for the accurate dielectric analysis of pelvic tissue structures; and by establishing a deep generative adversarial network dielectric inversion model, it realizes the high-precision inversion of the complex dielectric constant spectrum at different depths, providing a basis for the tomographic detection of tissue damage.

[0008] Optionally, establishing an improved Debye - Cole - Cole model based on the complex permittivity spectrum includes: obtaining the depth - dependent relaxation time distribution characteristics according to the complex permittivity spectrum; and establishing an improved Debye - Cole - Cole model based on the relaxation time distribution characteristics. Through the complex permittivity spectrum, the present invention extracts the depth - dependent relaxation time distribution characteristics, providing important parameters for more accurately characterizing the complex dielectric relaxation behavior of deep tissues; by analyzing the correlation between the relaxation time distribution characteristics and tissue depth, it reveals the dielectric response differences of the microstructures of tissues at different depths, laying a theoretical foundation for establishing a dielectric model with depth - resolution ability; by constructing an improved Debye - Cole - Cole model, it realizes the accurate fitting of the dielectric spectra of multi - layer tissues, providing an important method for analyzing the multi - scale dielectric properties of deep tissues; by integrating the depth - dependent relaxation characteristics into the Debye - Cole - Cole model, it significantly improves the sensitivity of the model to pathological changes in deep tissues, providing an important technical approach for the detection and differential diagnosis of early minor injuries.

[0009] Optionally, the improved Debye - Cole - Cole model satisfies the following expression: , where is the complex permittivity at different depths and different frequencies , is the high - frequency limit permittivity, , are the DC permittivities related to depth , , are the depth -dependent relaxation times, , are the distribution coefficients, is the static ionic conductivity, is the vacuum permittivity. By introducing the depth - dependent DC permittivity, the present invention can accurately characterize the differences in static polarization characteristics of tissues at different depths, providing an important parameter basis for analyzing the layered dielectric properties of deep tissues; by setting the relaxation time as a depth - related variable, it effectively reflects the dynamic relaxation behavior of tissue microstructures changing with depth, creating conditions for revealing the dielectric response mechanism of pathological changes in deep tissues; by setting the distribution coefficients, it significantly improves the fitting ability of the model to the non - Debye relaxation characteristics of deep tissues, providing a mathematical basis for the accurate analysis of complex permittivity spectra; by integrating the static ionic conductivity and the high - frequency limit permittivity, it constructs a complete depth - resolved dielectric model system, laying a theoretical model support for developing a quantitative detection technology for deep tissue injuries with clinical application value.

[0010] Optionally, establishing the mapping relationship between the tissue dielectric parameters and the degree of damage by using the improved bipolar Cole-Cole model includes: establishing a mapping model between the tissue dielectric parameters and the degree of damage by using the improved bipolar Cole-Cole model; and obtaining the mapping relationship between the tissue dielectric parameters and the degree of damage through the mapping model. By establishing a mapping model between the deep tissue dielectric parameters and the degree of damage, the present invention realizes the accurate quantification of the tissue damage degree, and has higher accuracy, adaptability and real-time performance; by obtaining the dynamic mapping relationship between the tissue dielectric parameters and the degree of damage, the intelligent conversion from dielectric measurement to clinical evaluation is realized, providing an important technical means for the real-time monitoring and prognosis evaluation of deep tissue damage.

[0011] Optionally, the mapping model between the tissue dielectric parameters and the degree of damage satisfies the following expression: , where, is the tissue damage degree at depth , is the complex dielectric constant at depth and frequency , is the high-frequency limit dielectric constant, , is the DC dielectric constant related to depth , , are the relaxation times dependent on depth , , are the distribution coefficients, is the static ionic conductivity, is the vacuum dielectric constant, , , are the weight coefficients, is the offset constant. By introducing a multi-band weighted fusion mechanism, the present invention realizes the differential regulation of different dielectric characteristics, such as polarization, relaxation, and conductivity, significantly improving the model's ability to identify the degree of damage; by establishing a mapping model between the tissue dielectric parameters and the degree of damage, the amplitude-frequency characteristics of the complex dielectric spectrum are directly correlated with the degree of damage; by combining the depth-dependent relaxation time with the distribution coefficient, the dielectric behavior changes caused by the variation of the tissue microstructure are effectively captured, providing a sensitive index for the quantitative detection of early minor damage.

[0012] Optionally, establishing the damage determination function of the pelvic tissue to be measured based on the mapping relationship includes: establishing the damage determination function of the pelvic tissue to be measured based on the mapping relationship between the tissue dielectric parameters and the degree of damage at different depths, and the damage determination function is as follows: , wherein, is the damage determination value of the damage determination function, represents at the th characteristic frequency and the multi-damage state variable the dielectric constant, represents at the th characteristic frequency and the multi-damage state variable the conductivity, represents at the th characteristic frequency and the multi-damage state variable the loss factor, , , represent the frequency and damage-dependent weight coefficients, , represent the multi-damage compensation coefficients, represents the tissue density, represents the moisture content, , represent the frequency-dependent quadratic coefficients, represents the damage-related offset term, represents the number of characteristic frequency points, , are the non-linear correction exponents, is the damage interaction term coefficient, is the characteristic function of the th type of damage, is the characteristic function of the th type of damage. The present invention realizes the accurate quantitative assessment of pelvic tissue damage by fusing the relationship between multi-band dielectric parameters and dynamic damage variables; by introducing a compensation mechanism and an offset term adaptive to the damage state, the adaptability of the model to different pathological stages and individual differences is significantly improved, and the clinical applicability is enhanced; by constructing a frequency-related quadratic term and a non-linear correction term, the interference of high-frequency measurement noise is effectively suppressed, and the stability and reliability of damage determination are improved; by designing the collaborative calculation of the damage interaction term and the multi-damage characteristic function, the collaborative diagnosis of complex pelvic damage is realized, and the problem that the existing methods are insensitive to the detection of concurrent pathologies is solved.

[0013] Optionally, the obtaining of the determination results of the damage level and damage type of the pelvic tissue to be measured by using the damage determination function includes: using the damage determination function to obtain a damage determination value; determining the damage level based on the damage determination value; based on the damage level, determining a damage characteristic function, where the damage characteristic function includes a temperature damage characteristic function, a pressure damage characteristic function, and an ischemic damage characteristic function; and obtaining the determination results of the single damage type and the composite damage type of the pelvic tissue to be measured according to the damage characteristic function. By dynamically matching the damage determination value with multi-dimensional damage level criteria, the present invention realizes the objective grading of the damage degree of pelvic tissue; by establishing specific damage characteristic functions for temperature, pressure, and ischemia, the recognition accuracy of single-type damage is significantly improved; through the non-linear combination analysis of the damage characteristic functions, the precise analysis of pelvic composite damage is realized, solving the technical problem of difficult differentiation of mixed pathologies.

[0014] Optionally, the obtaining of the determination results of the single damage type and the composite damage type of the pelvic tissue to be measured according to the damage characteristic function includes: determining the interaction contributions of temperature and pressure, temperature and ischemia, and pressure and ischemia according to the damage characteristic function; determining the dominant damage type based on the interaction contributions; and obtaining the determination results of the single damage type and the composite damage type of the pelvic tissue to be measured through the dominant damage type. By analyzing the interaction contribution degrees of the temperature, pressure, and ischemic damage characteristic functions, the present invention realizes the precise quantification of each damage component in composite damage, breaking through the limitation of the existing methods in difficult analysis of mixed damage; by identifying the dominant damage type, the discrimination accuracy of single damage and composite damage is significantly improved.

[0015] Optionally, the obtaining of the determination results of the single damage type and the composite damage type of the pelvic tissue to be measured according to the damage characteristic function includes: determining the interaction contributions of temperature and pressure, temperature and ischemia, and pressure and ischemia according to the damage characteristic function; determining the dominant damage type based on the interaction contributions; and obtaining the determination results of the single damage type and the composite damage type of the pelvic tissue to be measured through the dominant damage type. By establishing a dynamic interaction model of temperature, pressure, and ischemic damage, the present invention realizes the precise quantification of the contribution degrees of each damage component in pelvic composite damage; by analyzing the interaction contributions of each damage characteristic, identifying the dominant damage factors, the discrimination accuracy of single damage and composite damage is greatly improved; by constructing a dynamic weight allocation mechanism for damage interaction contributions, the adaptive discrimination of different damage types is realized, significantly improving the clinical applicability of damage diagnosis; by integrating a collaborative analysis algorithm for multi-damage characteristics, a diagnostic chain from single damage recognition to composite damage analysis is formed, providing a reliable basis for the formulation of precise medical treatment plans.

[0016] Second aspect, a deep tissue injury detection system based on dielectric spectrum analysis provided by the present invention includes an input device, a processor, an output device, and a memory. The input device, the processor, the output device, and the memory are interconnected. Among them, the memory is used to store a computer program, the computer program includes program instructions, the processor is configured to call the program instructions, and the system uses the deep tissue injury detection method based on dielectric spectrum analysis. The system provided by the present invention has a high degree of integration, and the information transmission between components is smooth. Through the deep coupling modeling of multi-band dielectric parameters and tissue injury characteristics, the accurate and non-invasive detection of deep tissue injury is realized, breaking through the limitations of invasive biopsy and imaging examination in the prior art; by establishing a dielectric parameter analysis model adaptive to the dynamic injury state, the recognition sensitivity and specificity of the system for tissue injuries of different depths and different types are significantly improved; through the intelligent analysis algorithm that fuses dielectric spectrum features and multi-physical field parameters, the quantitative evaluation of deep tissue complex injuries is realized, providing an objective basis for clinical diagnosis in a new dimension; by correlating dielectric parameters with tissue pathology, a full-automatic analysis process from data collection to injury determination is formed, greatly improving the clinical detection efficiency and diagnostic consistency. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a flowchart of a deep tissue injury detection method based on dielectric spectrum analysis according to an embodiment of the present invention; Figure 2 It is a flowchart for determining the injury level and injury type according to an embodiment of the present invention; Figure 3 It is a schematic structural diagram of a deep tissue injury detection system based on dielectric spectrum analysis according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0018] Specific embodiments of the present invention will be described in detail below. It should be noted that the embodiments described here are only for illustrative purposes and are not used to limit the present invention. In the following description, in order to provide a thorough understanding of the present invention, a large number of specific details are set forth. However, it is obvious to those of ordinary skill in the art that the present invention does not have to employ these specific details. In other instances, well-known circuits, software, or methods have not been described in detail in order to avoid obscuring the present invention.

[0019] Throughout the specification, references to "an embodiment", "embodiments", "an example" or "examples" mean that the specific features, structures, or characteristics described in connection with the embodiment or example are included in at least one embodiment of the present invention. Thus, the phrases "in an embodiment", "in embodiments", "an example" or "examples" that appear throughout the specification do not necessarily refer to the same embodiment or example. In addition, the specific features, structures, or characteristics may be combined in any suitable combination and / or sub-combination in one or more embodiments or examples. Further, those of ordinary skill in the art should understand that the diagrams provided herein are for illustrative purposes only and are not necessarily drawn to scale.

[0020] Please refer to Figure 1 , embodiments of the present invention provide a deep tissue damage detection method based on dielectric spectroscopy analysis, and the method includes the following steps: S1. Obtain the complex dielectric constant spectrum of the pelvic tissue to be measured.

[0021] In one embodiment, first use a vector network analyzer as a signal source to generate a swept electromagnetic wave signal with a frequency range of 1 MHz - 10 GHz; at the same time, place the transmitting antenna on one side of the surface of the pelvic tissue to be measured, ensuring good coupling between the antenna and the tissue surface to reduce signal loss. It should be noted that the pelvic tissue belongs to deep tissue, located in the lower part of the abdominal cavity, containing reproductive, urinary, and part of the digestive organs, and is relatively deep.

[0022] Further, place a receiving antenna on the other side of the pelvic tissue to be measured for receiving the transmitted signal; at the same time, set up a reflected signal receiving device near the transmitting antenna to obtain the reflected signal.

[0023] It should be noted that when the electromagnetic wave signal is emitted from the transmitting antenna and passes through the pelvic tissue to be measured, it will inevitably pass through the surface tissue first, and the surface tissue will reflect, scatter, and absorb the signal, so that the received transmitted signal and reflected signal contain the information of the surface tissue. If the signal of the surface tissue is not separated, the signal characteristics of the pelvic tissue to be measured itself cannot be accurately obtained.

[0024] Further, based on the transmitted signal and the reflected signal, design a time-frequency joint sparse reconstruction algorithm. The time-frequency joint sparse reconstruction algorithm combines the information in the time domain and the frequency domain and uses the sparsity of the signal to reconstruct the original signal. In the present invention, this algorithm separates the pelvic tissue signal to be measured and the surface tissue signal from the mixed signal of the transmitted signal and the reflected signal by constructing and solving an optimization problem. For this algorithm, construct the following optimization problem: , where Denote the received signal vector, which contains the transmitted signal and the reflected signal; is the observation matrix, which describes the observation method of the signal during the measurement process and maps the original signal space to the observation signal space. is the sparse transformation matrix, which transforms the signal from the time domain to the frequency domain, making the signal exhibit sparsity in the frequency domain. 、 Denote the regularization parameter, which is used to control the performance of the reconstructed signal in terms of fidelity, sparsity, and smoothness. is the sparse signal vector to be reconstructed. is an element in the sparse signal vector. 、 Denote the index numbers of the row vector and the column vector respectively. is a very small positive number, which is used to avoid the denominator being zero. It should be noted that the core of this algorithm is to utilize the sparsity of the signal in the time-frequency domain, construct an optimization problem with composite regularization, and achieve signal separation by solving this problem.

[0025] Furthermore, by using the time-frequency joint sparse reconstruction algorithm, the sparse signal vector is solved. Then, using the time-frequency mask, the pelvic tissue signal to be measured and the surface tissue signal and are separated. The time-frequency mask is a key separation technique in signal processing, which is used to extract the target component from the mixed signal. In the problem of separating pelvic signals and surface tissue signals, the role of the mask is to distinguish the different distribution characteristics of the two types of signals in the time-frequency domain.

[0026] Furthermore, perform the inverse sparse transformation to obtain the separated pelvic tissue signal to be measured and surface tissue signal.

[0027] Specifically, let the separated pelvic tissue signal to be measured be , and the surface tissue signal be , which can be expressed as: , , where denotes the inverse transformation of the sparse transformation matrix , and denote the signals corresponding to the pelvic tissue to be measured and the surface tissue in the sparse signal vector respectively, denotes time.

[0028] The advantage of this method lies in accurately distinguishing the time-frequency characteristics of pelvic tissue and surface tissue signals through joint time-frequency sparse reconstruction and mask separation, achieving high-precision signal separation. Among them, the inverse sparse transform restores the original signal, effectively improving the signal-to-noise ratio and resolution, providing purer tissue characteristic information for medical diagnosis, and having significant clinical value.

[0029] Furthermore, a deep generative adversarial network dielectric inversion model is established.

[0030] Specifically, the time-domain signal and the frequency-domain signal obtained through Fourier transform are concatenated into a joint feature vector: , where is the joint feature vector, is the frequency-domain amplitude spectrum, is the frequency.

[0031] Furthermore, a deep generative adversarial network structure is designed, and the deep generative adversarial network structure includes a generator and a discriminator.

[0032] Specifically, the generator takes the joint feature vector and the random noise as inputs and outputs the complex permittivity spectrum :

[0033]

[0034] where is the real part of the complex permittivity, representing the ability to store electrical energy, is the imaginary part of the complex permittivity, representing the ability to consume electrical energy, represents the tissue depth, is the frequency, are the generator parameters, is the generator, and the random noise follows a Gaussian distribution .

[0035] Furthermore, the discriminator takes the real permittivity data or the generated data as inputs and outputs the discrimination probability: , where is the discriminator, is the input data, including the real permittivity data and the generated data, are the discriminator parameters, is the discrimination probability.

[0036] Furthermore, a loss function is designed, and the losses involved in the loss function include the generator loss and the discriminator loss.

[0037] Specifically, the generator loss includes the adversarial loss and the physical constraint loss; The generator loss includes the adversarial loss and the physical constraint loss; , Among them, represents the generator loss, represents the adversarial loss, represents the physical constraint loss, 、 are weight coefficients. The physical constraint loss is as follows: , Among them, is the electric field strength, is the frequency, is the depth, satisfies the following expression:

[0038] Among them, is the wave number, is the speed of light, is the complex permittivity spectrum. The adversarial loss is expressed as the negative log-likelihood that the generated sample is judged by the discriminator to be a real sample, that is:

[0039] Among them, is the noise vector, is the noise distribution, is the sample generated by the generator, is the discriminator's discrimination result for the generated sample, is the mathematical expectation.

[0040] Furthermore, the discriminator loss is as follows: , Among them, is the discriminator's loss, is the dielectric constant data of the real sample, is the discriminator's discrimination result for the real sample data, is the noise vector, is the additional conditional input information, is the probability that the discriminator believes the sample is real, is the generator according to the conditional input and the noise vector The generated forged samples are used as the input of the discriminator to train the discriminator's ability to recognize forged samples.

[0041] Furthermore, based on the above deep generative adversarial network structure, considering the losses of the generator and the discriminator, a deep generative adversarial network dielectric inversion model is established. The deep generative adversarial network dielectric inversion model includes a generator, a discriminator, a generator loss, and a discriminator loss. The deep generative adversarial network dielectric inversion model is a non-linear inversion method based on adversarial learning, which is used to reconstruct the vertical distribution of dielectric constants from observed data. Its core idea is to achieve high-precision and high-resolution dielectric parameter inversion through the adversarial training of the generator and the discriminator, combined with physical constraints.

[0042] Furthermore, the depth is discretized layer by layer into , and the generator outputs a multi-dimensional tensor: , where each layer corresponds to an independent physical constraint loss , is the frequency, is the depth of the th layer, is the additional conditional input information, is the noise vector, is the generator parameter, is the complex dielectric constant spectrum corresponding to different layer depths.

[0043] Furthermore, model training is carried out, including the following steps: Randomly initialize and ; Iterative training: Fix , update , and minimize ; Fix , update , and minimize ; Judgment of termination: When the error between the generated data and the real data is less than , the iteration stops. Here,

[0044] Furthermore, after training is completed, the generator can directly predict the dielectric constant of any input signal : , where the noise Reset, is the optimal parameter.

[0045] The advantage of this method is that through the adversarial training of the generator and discriminator, combined with hierarchical physical constraints, high-precision and high-resolution dielectric parameter inversion is achieved, overcoming the problems of existing methods relying on linear approximation and low computational efficiency. At the same time, through end-to-end prediction and hierarchical adaptive optimization, the accuracy and generalization ability of the vertical distribution reconstruction of pelvic tissues are significantly improved, and the dependence on a large amount of labeled data is reduced.

[0046] S2. Establish an improved Debye model based on the complex dielectric constant spectrum.

[0047] In one embodiment, first, for each depth of the complex dielectric constant spectrum is decomposed to identify the existing relaxation processes, including low-frequency relaxation and high-frequency relaxation .

[0048] Furthermore, through the particle swarm optimization algorithm, the relaxation time and and their distribution coefficients and are extracted.

[0049] Furthermore, analyze the variation law of the relaxation time with depth to establish the and depth-dependent models.

[0050] Specifically, the depth-dependent model of is as follows: the depth-dependent model of is as follows: where , are scale parameters for controlling the initial amplitude of the relaxation time, , are attenuation rates describing the variation of the relaxation time with depth, , represent the offset, indicating the baseline value of the relaxation time when the depth approaches infinity.

[0051] Furthermore, through the and depth-dependent models, the depth-dependent relaxation time distribution characteristics are obtained.

[0052] Further, according to the extracted relaxation time distribution characteristics, an improved bipolar Cole-Cole model is established, and the improved bipolar Cole-Cole model satisfies the following expression: , where is the complex permittivity at different depths and different frequencies , is the high-frequency limit permittivity, , are the DC permittivities related to the depth , , are the relaxation times dependent on the depth , , are the distribution coefficients, is the static ionic conductivity, is the vacuum permittivity.

[0053] Further, the model parameters are determined by fitting experimental data.

[0054] First, for deep tissue detection, corresponding samples are selected. A vector network analyzer equipped with a minimally invasive dielectric probe is used to collect time-domain voltage signals and frequency-domain amplitude spectra at 1 mm depth intervals within the target frequency band of 1 MHz - 10 GHz, constructing a multi-depth and multi-frequency experimental dataset. Based on the particle swarm optimization algorithm, first, according to the prior dielectric characteristics of normal / lesion tissues, the initial search ranges of model parameters ( , , , , etc.) are set. The candidate parameter values are substituted into the improved bipolar Cole-Cole model, and the mean square error between the calculated output complex permittivity and the measured data is used as the fitness. The particles iteratively adjust their positions until the mean square error is less than 0.01 or the iteration reaches 1000 times. Finally, 20% of the independently collected sample data is used for verification. If the correlation coefficient between the predicted complex permittivity of the model and the measured value is greater than 0.95, the parameters are determined for use as input feature constraints for the depth generative adversarial network (such as the initial conditions of the generator and the real sample labels of the discriminator), supporting the inversion of the complex permittivity. If not satisfied, the sample stratification strategy or algorithm parameters are adjusted for re-fitting.

[0055] The improvements of this method are as follows: First, by introducing the depth functions of relaxation time and permittivity, the existing uniform medium assumption is broken through, and the dielectric gradient change of pelvic biological tissues is accurately characterized; second, by separating high-frequency molecular polarization and low-frequency interfacial polarization through the double relaxation terms and combining the dynamic distribution coefficients, the fitting accuracy of complex relaxation spectra is significantly improved; third, by considering the static conductivity, the distortion problem of the existing model in the low-frequency region is solved, especially suitable for the detection of high-water-content tissues.

[0056] S3. Use the improved bipolar Cole-Cole model to establish the mapping relationship between tissue dielectric parameters and the degree of damage.

[0057] In one embodiment, use the improved bipolar Cole-Cole model to establish a mapping model between tissue dielectric parameters and the degree of damage, and the mapping model between the tissue dielectric parameters and the degree of damage satisfies the following expression: , where, is the degree of tissue damage at depth , is the complex dielectric constant at different depths and different frequencies , is the high-frequency limit dielectric constant, , is the DC dielectric constant related to depth , , is the relaxation time dependent on depth , , are distribution coefficients, is the static ionic conductivity, is the vacuum permittivity, , , are weighting coefficients, is an offset constant used to adjust the reference level of the degree of damage to ensure that the model output is consistent with the observed data.

[0058] Furthermore, through the mapping model, obtain the mapping relationship between the tissue dielectric parameters and the degree of damage, that is, the one-to-one correspondence between the tissue dielectric parameters and the degree of damage.

[0059] It should be noted that although the above mapping model is beneficial for quickly screening tissues with damage, it does not solve the problems of why the damage occurs and how to intervene.

[0060] The advantage of this method is that through the mapping model between tissue dielectric parameters and the degree of damage, the accurate quantification of the degree of tissue damage is realized, with higher accuracy, adaptability and real-time performance.

[0061] S4. Based on the mapping relationship, establish the damage determination function of the pelvic tissue to be measured.

[0062] To solve the defect problem of the mapping model in step S3, in one embodiment, based on the mapping relationship between tissue dielectric parameters and the degree of damage at different depths, establish the damage determination function of the pelvic tissue to be measured, and the damage determination function is as follows: , Among them, is the damage determination value of the hierarchical damage determination function, represents at the th characteristic frequency and the multi-damage state quantity the dielectric constant under, represents at the th characteristic frequency and the multi-damage state variable the conductivity under, represents at the th characteristic frequency and the multi-damage state variable the loss factor under, , , represent the frequency and damage-dependent weight coefficients, , represent the multi-damage compensation coefficients, represents the tissue density, represents the moisture content, , represent the frequency-dependent quadratic coefficients, represents the damage-related offset term, represents the number of characteristic frequency points, , is the non-linear correction exponent, is the damage interaction term coefficient, is the characteristic function of the th kind of damage, is the characteristic function of the th kind of damage, the damage state variable includes temperature, pressure and mechanical strain.

[0063] The advantage of this method is that by establishing a multi-parameter and multi-frequency damage determination function, integrating dielectric properties, tissue parameters and multi-physical field variables, and introducing non-linear correction and damage interaction terms, the accuracy and dynamic adaptability of pelvic tissue damage assessment are significantly improved, overcoming the defect of insufficient analysis of complex damage states by existing models.

[0064] S5. Use the damage determination function to obtain the determination results of the damage level and damage type of the pelvic tissue to be measured.

[0065] Please refer to Figure 2 , in one embodiment, first use the damage determination function to obtain the damage determination value .

[0066] Furthermore, the damage level is determined by using the damage determination value; Specifically, the thresholds of the damage determination value are set, including a low threshold , a medium threshold , and a high threshold ; Furthermore, the damage determination value is compared with the thresholds to obtain the following comparison results: When , it is in the normal state level, that is, the pelvic tissue remains basically intact without obvious signs of damage; When , it is in the mild damage level, that is, the pelvic tissue has minor damage, but the tissue function can still be maintained and time is needed for maintenance; When , it is in the moderate damage level, that is, the pelvic tissue damage is relatively obvious and the tissue function is affected to a certain extent, and timely repair is required.

[0067] When , it is in the severe damage level, that is, the pelvic tissue is severely damaged and large-scale repair is needed.

[0068] Furthermore, the damage feature functions are extracted , including the temperature damage feature function , the pressure damage feature function , and the ischemic damage feature function . They satisfy the following expressions: , , , where is the current tissue temperature, is the normal physiological temperature, is the pressure applied to the current tissue, is the critical pressure tolerated by the tissue, is the current blood perfusion volume, is the normal blood perfusion volume.

[0069] Furthermore, the interactive term contributions are calculated, including: Calculating the interactive contribution of temperature and pressure: , where is the damage interaction term coefficient of temperature and pressure, is the feature function of temperature damage, is the feature function of pressure damage; Calculating the interactive contribution of temperature and ischemia: , where is the coefficient of the damage interaction term between temperature and ischemia, is the characteristic function of temperature damage, is the characteristic function of ischemic damage; Calculate the interaction contribution between pressure and ischemia: , where is the coefficient of the damage interaction term between pressure and ischemia, is the characteristic function of pressure damage, is the characteristic function of ischemic damage; Furthermore, determine the dominant damage type, which includes temperature damage, pressure damage, and ischemic damage. The characteristics of the temperature damage include: Higher than the normal value, Changing, and Generating interaction contributions; The characteristics of the pressure damage include: Increasing, Decreasing, Generating interaction contributions; The characteristics of the ischemic damage include: Increasing, Changing, Becoming higher.

[0070] Furthermore, perform single-damage determination. Specifically as follows: Set the coefficient of the damage interaction term between temperature and pressure , the threshold of the coefficient of the damage interaction term between temperature and ischemia , the coefficient of the damage interaction term between pressure and ischemia .

[0071] If and 、 , it is determined as temperature damage; If and , it is determined as pressure damage; If and is abnormal, it is determined as ischemic damage.

[0072] Where 、 、 are respectively 、 、 's thresholds.

[0073] Furthermore, perform combined-damage determination. Specifically as follows: If 、 and If all are higher than the corresponding thresholds, it is determined as a combined injury of temperature and pressure; If , and are all higher than the corresponding thresholds, it is determined as a combined injury of temperature and ischemia; If , and are all higher than the corresponding thresholds, it is determined as a combined injury of pressure and ischemia; If , and are all higher than the corresponding thresholds, and , and are also higher than the corresponding thresholds, it is determined as a combined injury of temperature, pressure and ischemia.

[0074] The advantage of this method is that through multi-parameter hierarchical determination and interaction term analysis, it realizes the accurate quantitative evaluation of pelvic tissue injury, can distinguish single-factor and combined injury types, dynamically judges the injury level in combination with physiological thresholds, provides a scientific basis for clinical diagnosis and repair, and has comprehensiveness, sensitivity and operability.

[0075] Please refer to Figure 3 , Figure 3 which is a schematic structural diagram of the deep tissue injury detection system based on dielectric spectrum analysis in the embodiment of the present invention. The system includes an input device, a processor, an output device, and a memory. The input device, the processor, the output device, and the memory are interconnected. Among them, the memory is used to store computer programs, the computer programs include program instructions, the processor is configured to call the program instructions, and the system uses the above-mentioned deep tissue injury detection method based on dielectric spectrum analysis.

[0076] In this embodiment, the input device includes a swept-frequency electromagnetic wave transmitting module and a signal receiving module. The functions of the input device include: transmitting a swept-frequency electromagnetic wave signal to the pelvic tissue to be measured to stimulate the dielectric response of the tissue; synchronously collecting the transmitted signal and the reflected signal; converting the original analog signal into a digital signal and transmitting it to the processor for subsequent analysis.

[0077] The processor includes a signal processing unit, a modeling and mapping unit, and a damage determination unit. The signal processing unit includes a time-frequency joint sparse reconstruction algorithm module and a deep generative adversarial network dielectric inversion module. The modeling and mapping unit includes an improved bipolar Cole-Cole model solver and a mapping model of dielectric parameters and damage degree. The damage determination unit includes a multi-parameter fusion damage determination function calculation engine and a damage type classifier. The functions of the processor include: separating deep tissue and surface tissue signals using the time-frequency joint sparse reconstruction algorithm; inverting the complex dielectric constant spectra of tissues at different depths using the model; establishing an improved bipolar Cole-Cole model and extracting depth-dependent relaxation times; calculating damage determination values, classifying damage levels, and analyzing the interactive contributions of temperature, pressure, and ischemia damage characteristic functions.

[0078] The output device includes a visual interaction interface and a report generation module. The functions of the output device include: real-time displaying the damage heat map and key dielectric parameters at each depth of pelvic tissues; generating a clinically readable damage detection report, including the analysis of damage level, damage type, and damage cause.

[0079] The memory uses a high-speed solid-state drive, which features fast read and write speeds, large capacity, and high reliability. It is mainly used to store the data input by the input device and the result data processed by the processor, and can meet the needs of large data volume storage.

[0080] In summary, the present invention realizes the high-precision quantitative characterization of the dielectric properties of deep tissues by obtaining the complex dielectric constant spectra of the pelvic tissues to be measured, providing a reliable data basis for improving the sensitivity and specificity of damage detection; overcomes the defect of insufficient fitting accuracy of the existing model in the wide frequency range by establishing an improved bipolar Cole-Cole model, providing a more accurate mathematical description of the dielectric behavior of pelvic tissues; reveals the correlation law between the change of dielectric properties and the tissue pathological state by establishing the mapping relationship between tissue dielectric parameters and damage degree, creating conditions for the quantitative evaluation of damage; realizes the intelligent damage classification of multi-parameter fusion by constructing a damage determination function, providing an objective basis for distinguishing different damage levels and types; and realizes the non-invasive and rapid detection of deep tissue damage through the determination results of damage levels and types, providing a new technical means for clinical pelvic disease diagnosis and treatment effect monitoring. This method effectively improves the sensitivity, specificity, and clinical applicability of deep tissue damage detection.

[0081] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the claims and the description of the present invention.

Claims

1. A method for detecting deep tissue damage based on dielectric spectroscopy analysis, characterized in that The method includes the following steps: Obtain the complex permittivity spectrum of the pelvic tissue to be measured; Based on the complex permittivity spectrum, establish an improved bipolar Cole-Cole model; Use the improved bipolar Cole-Cole model to establish the mapping relationship between tissue dielectric parameters and damage degree; Based on the mapping relationship, establish the damage determination function of the pelvic tissue to be measured; Use the damage determination function to obtain the determination results of the damage level and damage type of the pelvic tissue to be measured; Through the determination results, realize the damage detection of deep tissues.

2. The method for detecting deep tissue damage based on dielectric spectroscopy according to claim 1, wherein The obtaining of the complex permittivity spectrum of the pelvic tissue to be measured includes: Transmit a swept-frequency electromagnetic wave signal to the pelvic tissue to be measured to obtain a transmitted signal and a reflected signal; Based on the transmitted signal and the reflected signal, design a time-frequency joint sparse reconstruction algorithm; Use the time-frequency joint sparse reconstruction algorithm to obtain the signal separation result of the pelvic tissue to be measured and the surface tissue; Based on the signal separation result, establish a deep generative adversarial network dielectric inversion model; Through the deep generative adversarial network dielectric inversion model, obtain the complex permittivity spectrum of the pelvic tissue to be measured at different depths.

3. A deep tissue injury detection method based on dielectric spectroscopy analysis according to claim 1, characterized in that The establishing of the improved bipolar Cole-Cole model based on the complex permittivity spectrum includes: Based on the complex permittivity spectrum, obtain the depth-dependent relaxation time distribution characteristics; Based on the relaxation time distribution characteristics, establish an improved bipolar Cole-Cole model.

4. The deep tissue injury detection method based on dielectric spectrum analysis according to claim 3, characterized in that, The improved bipolar Cole-Cole model satisfies the following expression: , in, For different depths and different frequencies The complex dielectric constant under is the high frequency limiting dielectric constant, , For Depth The related DC dielectric constant, , For Depth Dependent on the relaxation time, , is the distribution coefficient, is the static ionic conductivity, is the dielectric constant of vacuum.

5. The method for detecting deep tissue damage based on dielectric spectroscopy according to claim 1, characterized in that The using of the improved bipolar Cole-Cole model to establish the mapping relationship between tissue dielectric parameters and damage degree includes: Use the improved bipolar Cole-Cole model to establish the mapping model between tissue dielectric parameters and damage degree; Through the mapping model, obtain the mapping relationship between the tissue dielectric parameters and the damage degree.

6. The deep tissue injury detection method based on dielectric spectroscopy analysis according to claim 5, characterized in that The mapping model between the tissue dielectric parameters and the damage degree satisfies the following expression: , Among them, is the degree of tissue damage at depth . is the complex permittivity at depth and frequency . is the high-frequency limit permittivity, , are the DC permittivities related to depth . , are the relaxation times dependent on depth . , are the distribution coefficients, is the static ionic conductivity, is the vacuum permittivity, , , are the weighting coefficients, is the offset constant.

7. A method for detecting deep tissue damage based on dielectric spectroscopy according to claim 1, characterized in that The establishing of the damage determination function of the pelvic tissue to be measured based on the mapping relationship includes: Based on the mapping relationship between tissue dielectric parameters and damage degree at different depths, establish the damage determination function of the pelvic tissue to be measured, and the damage determination function is as follows: , wherein, is the damage determination value of the damage determination function, represents at the th characteristic frequency and the multi-damage state variable the permittivity, represents at the th characteristic frequency and the multi-damage state variable the conductivity, represents at the th characteristic frequency and the multi-damage state variable the loss factor, , , represent the frequency and damage-dependent weight coefficients, , represent the multi-damage compensation coefficients, represents the tissue density, represents the moisture content, , represent the frequency-dependent quadratic coefficients, represents the damage-related offset term, represents the number of characteristic frequency points, , are the non-linear correction exponents, is the damage interaction term coefficient, is the characteristic function of the th type of damage, is the characteristic function of the th type of damage.

8. A deep tissue injury detection method based on dielectric spectroscopy analysis according to claim 1, characterized in that The using of the damage determination function to obtain the determination results of the damage level and damage type of the pelvic tissue to be measured includes: Use the damage determination function to obtain a damage determination value; Through the damage determination value, determine the damage level; Based on the damage level, determine the damage characteristic functions, and the damage characteristic functions include temperature damage characteristic function, pressure damage characteristic function, and ischemic damage characteristic function; According to the damage characteristic functions, obtain the determination results of the single damage type and the composite damage type of the pelvic tissue to be measured.

9. The method for detecting deep tissue damage based on dielectric spectroscopy according to claim 8, wherein, The obtaining of the determination results of the single damage type and the composite damage type of the pelvic tissue to be measured according to the damage characteristic functions includes: According to the damage characteristic functions, determine the interactive contributions of temperature and pressure, temperature and ischemia, and pressure and ischemia; Based on the interactive contributions, determine the dominant damage type; Obtain the determination results of single injury types and compound injury types of the pelvic tissue to be measured through the dominant injury type.

10. A deep tissue injury detection system based on dielectric spectroscopy analysis, the system using the deep tissue injury detection method based on dielectric spectroscopy analysis according to any one of claims 1 to 9, characterized in that, The system includes an input device, a processor, an output device, and a memory. The input device, the processor, the output device, and the memory are interconnected. Among them, the memory is used to store a computer program, the computer program includes program instructions, and the processor is configured to call the program instructions.

Images