Dentin caries depth electrical impedance detection method and system

Abstract

The invention discloses a dentin caries depth electrical impedance detection method and system, and belongs to the technical field of nondestructive detection of oral diseases, and the method comprises four steps of multi-frequency impedance spectrum acquisition, impedance spectrum feature extraction, equivalent circuit model fitting and caries depth inversion. The physical property that different levels of dentin cause different conductivity due to mineralization degree difference is utilized, quantitative detection of the caries depth is realized by establishing a three-layer series impedance equivalent circuit model and nonlinear fitting inversion, the influence of the environment temperature is eliminated by adopting a temperature compensation algorithm, and the detection precision reaches 0.2 mm.

Description

Technical Field

[0001] This invention relates to the field of non-destructive testing technology for oral diseases, and in particular to a method and system for detecting dentin caries depth based on multi-frequency impedance spectroscopy analysis. Background Technology

[0002] Dentin caries is one of the most common diseases in clinical dentistry, and its accurate assessment is directly related to the formulation of clinical treatment plans, especially in the decision-making process regarding cavity preparation depth. Currently, the clinical assessment of dentin caries depth mainly relies on two methods: X-ray imaging and palpation. However, X-ray imaging methods are not sensitive enough in detecting superficial caries, especially for early proximal caries lesions. Due to the mild demineralization of dentin and the small size of the lesion, the attenuation difference of X-rays is insufficient to produce obvious density changes on the image, resulting in a false negative rate as high as 30% to 40%. At the same time, X-ray examinations inevitably expose patients to ionizing radiation. Although the effective dose of a single examination is low, in clinical scenarios such as caries progression monitoring that require repeated examinations, the cumulative effect of radiation raises significant safety concerns, especially for radiation-sensitive populations such as children and pregnant women. Although probe palpation is simple to operate and does not require special equipment, it relies heavily on the examiner's subjective experience and tactile feedback. The consistency of test results between different operators is poor, and the mechanical contact of the sharp probe may cause irreversible mechanical damage to the remineralization potential of early demineralized areas, accelerating the progression of caries.

[0003] In recent years, electrical detection methods have gradually attracted attention in the field of oral diagnosis due to their non-invasive, rapid, quantifiable, and radiation-free characteristics. Existing studies have shown that healthy dentin exhibits high electrical resistance due to its high mineralization, while carious dentin shows a significant increase in conductivity due to mineral dissolution and open tubular channels. This electrical difference provides a physical basis for the application of electrical impedance tomography (EIT) methods. Chinese invention document CN115436440A discloses an experimental device and method for detecting dentin permeability. This scheme employs a three-electrode system, including a platinum electrode, an Ag / AgCl electrode, and a glassy carbon electrode, combined with a dentin electrochemical pool, an electrochemical workstation, and a frequency response analyzer. The long-term effect of desensitizing materials on sealing dentinal tubules is evaluated by measuring the electrochemical impedance values ​​of dentin slices. This scheme has some value in achieving non-destructive and repeatable testing of dentin permeability; however, its technical approach has the following limitations: First, this scheme is an in vitro testing method, requiring the removal of dentin slices from the oral cavity and immersion in potassium chloride solution for measurement. It cannot be directly performed in vivo within the patient's oral cavity, making it unsuitable for clinical in vivo diagnostic scenarios. Second, the test frequency range of this scheme is only 0.1Hz to 60kHz, with limited frequency coverage, making it impossible to obtain dielectric properties of dentin tissue in the high-frequency range, thus limiting the ability to perform layered analysis. Third, this scheme can only obtain the overall impedance value of the dentin slices and does not establish a layered equivalent circuit model. Therefore, it cannot distinguish the electrical properties of healthy dentin layers and carious dentin layers, let alone achieve layered quantitative detection of caries depth. Fourth, this scheme does not consider the impact of temperature changes on impedance measurements and lacks a temperature compensation mechanism.

[0004] Furthermore, existing electrical impedance tomography (EIT) techniques in dental applications suffer from the following common shortcomings: First, methods based on single-frequency or narrow-band detection acquire limited information, failing to fully reveal the electrical differences in dentin structure across different layers. They can only provide a qualitative assessment of the presence or absence of caries, not quantitative information on caries depth. Second, there is a lack of miniaturized probe designs suitable for intraoral manipulation. Traditional flat or needle-shaped electrode structures are difficult to precisely position in the confined space of the oral cavity, and the detection area is uncontrollable and easily affected by interference from adjacent gingival tissue and saliva. Third, a dentin equivalent circuit layering model suitable for clinical in vivo conditions has not been established, making it impossible to invert the thickness information of each layer using impedance spectroscopy data. Finally, the systematic influence of intraoral temperature fluctuations on electrical measurements is ignored. Oral temperature fluctuates significantly due to factors such as breathing and eating; the lack of a temperature compensation mechanism directly affects the reliability and repeatability of the test results. These limitations restrict the widespread application of EIT methods in the clinical assessment of caries depth. Summary of the Invention

[0005] The purpose of this invention is to provide a method and system for detecting the depth of dentin caries, in order to solve the technical problems of existing technologies that rely on imaging methods for caries depth assessment, resulting in a high rate of missed diagnoses of superficial caries and safety hazards due to radiation exposure, and existing electrical detection methods that can only obtain overall impedance values ​​and cannot achieve layered quantitative detection of caries depth.

[0006] To achieve the above objectives, the first aspect of the present invention provides a method for detecting the electrical impedance tomography depth of dentin caries, comprising the following steps: a multi-frequency impedance spectrum acquisition step, wherein a coaxial dual-electrode detection probe is attached to the surface of the dentin to be tested, and a sinusoidal sweep excitation signal with a frequency range of 10Hz to 1MHz is applied to the detection probe through an excitation signal module, a local current loop is formed between the central electrode and the outer ring electrode of the coaxial dual-electrode detection probe, and the impedance response signals of the dentin to be tested at multiple excitation frequencies are acquired to obtain raw impedance spectrum data containing impedance amplitude and phase information; an impedance spectrum feature extraction step, wherein the raw impedance spectrum data is subjected to lock-in amplification and demodulation processing, the real and imaginary components corresponding to each excitation frequency are separated, a complex impedance spectrum is constructed, and the signal quality of the complex impedance spectrum is evaluated, and the excitation signal is adaptively adjusted according to the signal quality evaluation results. The amplitude parameters are obtained to acquire quality-verified impedance spectrum characteristic data; the equivalent circuit model fitting step establishes a dentin multilayer impedance equivalent circuit model, abstracting the dentin to be tested into a series impedance network of healthy dentin layer, carious dentin layer and pulp cavity layer. Each layer contains a parallel combination of resistive elements and constant phase angle elements. The complex impedance nonlinear least square fitting algorithm is used to fit and solve the parameters of each layer of the equivalent circuit model to obtain the resistance value and constant phase angle element parameters of each layer; the caries depth inversion step, based on the pre-established quantitative mapping relationship between impedance characteristics and caries depth, converts the resistance value and constant phase angle element parameters of the carious dentin layer obtained by fitting the equivalent circuit model into caries depth value, and eliminates the influence of oral environment temperature on impedance measurement value through temperature compensation algorithm, outputting the caries depth detection result after temperature correction. In this method, the detection results of the caries depth inversion step are fed back to the multi-frequency impedance spectrum acquisition step. The frequency distribution density of the sweep frequency signal is adaptively adjusted according to the current caries depth estimate, so that the sampling density of the low-frequency band and the high-frequency band matches the caries depth range, thereby improving the detection accuracy within the target depth range without increasing the total acquisition time.

[0007] A second aspect of this invention provides a dentin caries depth electrical impedance detection system, comprising a multi-frequency impedance spectroscopy acquisition module, an impedance spectroscopy feature extraction module, an equivalent circuit model fitting module, and a caries depth inversion module. The multi-frequency impedance spectroscopy acquisition module applies a swept-frequency excitation signal through a coaxial dual-electrode detection probe and acquires the impedance response to obtain raw impedance spectroscopy data. The impedance spectroscopy feature extraction module is used for lock-in amplification, demodulation, and signal quality assessment, outputting quality-verified impedance spectroscopy feature data. The equivalent circuit model fitting module is used to establish a multi-layer equivalent circuit model and fit and solve for the parameters of each layer. The caries depth inversion module converts the model parameters into caries depth values ​​and performs temperature compensation. The four modules form a data-driven closed-loop collaborative architecture. The output of the caries depth inversion module is fed back to the impedance spectroscopy acquisition module to achieve adaptive optimization of the frequency distribution; the parameter information from the equivalent circuit model fitting module is fed back to the impedance spectroscopy feature extraction module to optimize the quality assessment criteria; and the quality assessment results from the impedance spectroscopy feature extraction module are fed back to the acquisition module to adjust the excitation parameters.

[0008] The beneficial effects of this invention are as follows: First, by combining wide-band multi-frequency impedance spectroscopy acquisition with hierarchical equivalent circuit model fitting, a non-invasive hierarchical quantitative detection of dentin caries depth is achieved for the first time, with a detection accuracy of 0.2 mm and a root mean square error of 0.19 mm for caries depth estimation. Second, the use of a coaxial dual-electrode structure to form a local current loop effectively limits the detection area to a range of approximately 3 mm in diameter, avoiding interference from adjacent tissues. Third, the closed-loop feedback mechanism enables the acquisition frequency distribution to adaptively adjust according to caries depth, significantly improving the detection efficiency and accuracy under different caries degrees. Fourth, the temperature compensation algorithm based on the Arrhenius temperature-conductivity relationship eliminates the influence of oral environmental temperature fluctuations, reducing the temperature sensitivity coefficient from 0.08 mm / ℃ to 0.01 mm / ℃, ensuring the stability and repeatability of the detection results. The overall scheme achieves a caries detection sensitivity of 92% and a specificity of 89%, completely avoiding ionizing radiation exposure, and can be used as a non-invasive auxiliary diagnostic tool for clinical caries depth assessment. Attached Figure Description

[0009] Figure 1 This is a schematic flowchart of the method for detecting the depth of dentin caries according to the present invention.

[0010] Figure 2 This is a schematic diagram of the architecture of the dentin caries depth electrical impedance detection system of the present invention. Detailed Implementation

[0011] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are for illustrative and explanatory purposes only and are not intended to limit the invention. Those skilled in the art can make various modifications and variations to the present invention without departing from its spirit and scope, and all such modifications and variations fall within the protection scope defined by the claims of the present invention.

[0012] Reference Figure 1 This invention provides a method for detecting the depth of dentin caries using electrical impedance tomography. This method utilizes the physical property that healthy dentin and carious dentin have significantly different electrical conductivities due to differences in mineralization levels, and achieves precise quantification of caries depth through multi-frequency impedance spectroscopy analysis. In one embodiment of this invention, the method specifically includes a multi-frequency impedance spectroscopy acquisition step S1, an impedance spectroscopy feature extraction step S2, an equivalent circuit model fitting step S3, and a caries depth inversion step S4. These four steps form a closed-loop architecture with step-by-step data transfer and feedback adjustment.

[0013] Step S1: Multi-frequency impedance spectroscopy acquisition. In this step, the operator first precisely positions and attaches the coaxial dual-electrode detection probe to the surface of the dentin to be tested. In one embodiment of the present invention, the coaxial dual-electrode detection probe adopts a coaxial structure design, the core of which includes a central electrode located at the center of the probe and an outer ring electrode arranged around the central electrode. The diameter of the central electrode is set to 0.5 mm to 1 mm, preferably 0.8 mm. This size range is adapted to the distribution density and spacing scale of dentinal tubules, and can cover a statistically representative number of dentinal tubules in a single measurement. The inner diameter of the outer ring electrode is set to 1.5 mm to 3 mm, preferably 2 mm, thereby forming a local current loop between the central electrode and the outer ring electrode, effectively limiting the current diffusion range and confining the detection area to a circular area with a diameter of about 3 mm directly below the probe. An insulating layer is filled between the central electrode and the outer ring electrode. This insulating layer is made of biocompatible ceramic material and has a thickness of 0.25 mm to 0.75 mm to ensure that the electrical insulation performance between the two electrodes meets the measurement requirements.

[0014] Furthermore, in a preferred embodiment of the present invention, the surfaces of the central electrode and the outer ring electrode are coated with a biocompatible conductive coating. This conductive coating can be a titanium nitride coating or a platinum coating, with a coating thickness of 0.1 μm to 0.5 μm. The titanium nitride coating exhibits excellent chemical stability and biocompatibility, effectively preventing corrosion of the electrode surface in the oral saliva environment; the platinum coating has lower electrode-tissue interface impedance, which is beneficial for improving measurement sensitivity in the low-frequency range. In one embodiment of the present invention, a platinum coating is preferably used, with its surface roughness Ra controlled below 0.05 μm to ensure uniform contact between the electrode and the dentin surface.

[0015] After the probe is in place, the excitation signal module outputs a sinusoidal sweep excitation signal to the detection probe. In one embodiment of the present invention, the frequency range of the sweep signal is set to 10Hz to 1MHz. This wide frequency band coverage is based on the physical mechanism that different layers of dentin tissue exhibit differentiated impedance responses at different frequency bands. Specifically, in the low-frequency range of 10Hz to 1kHz, the excitation current mainly flows through the ion conduction path of the fluid within the dentinal tubules. At this time, the impedance response mainly reflects the conductivity characteristics of the fluid within the tubules and the changes in ion concentration caused by the dissolution of minerals in the carious area. In the mid-frequency range of 1kHz to 100kHz, the current simultaneously flows through both the conduction path within the tubules and the conduction path through the intertubules. The impedance response reflects the overall electrical characteristics of dentin. In the high-frequency range of 100kHz to 1MHz, the current's ability to penetrate the intertubules is enhanced, and the impedance response reflects more the dielectric properties and mineralization degree of the intertubular dentin matrix.

[0016] In one embodiment of the present invention, the excitation amplitude of the frequency sweep signal is set to 10mV to 100mV, preferably, the initial excitation amplitude is 50mV. The selection of this amplitude range requires a balance between measurement sensitivity and biosafety: too low an amplitude will cause the response signal to be submerged in noise, resulting in insufficient signal-to-noise ratio; too high an amplitude may produce nonlinear electrochemical effects at the electrode-tissue interface, or even induce local tissue electrostimulation responses. At an excitation amplitude of 50mV, the current density flowing through dentin is approximately 0.1mA / cm². 2 Up to 1mA / cm 2 This is far below the safe current density threshold of 5 mA / cm² for oral tissues. 2 .

[0017] The number of sweep points is set to 50 to 200, preferably 100. The frequency points are uniformly distributed on a logarithmic coordinate system, meaning the frequency ratio between adjacent frequency points is constant. For example, using 100 frequency points covering five decibel bands from 10Hz to 1MHz, each decibel contains 20 frequency points, and the frequency ratio between adjacent frequency points is approximately 1.122. This logarithmically uniform distribution ensures balanced impedance information within each frequency band, especially maintaining sufficient sampling density in the low-frequency range to capture the characteristic response of the carious layer.

[0018] During a single frequency sweep acquisition, the excitation signal module sequentially outputs a sinusoidal excitation signal at each frequency point. The duration of the signal at each frequency point is no less than 10 times the period of that frequency signal to ensure the system reaches a steady-state response. In one embodiment of the present invention, for the lowest frequency point of 10Hz, the single-point acquisition time is approximately 1 second; for the highest frequency point of 1MHz, the single-point acquisition time is approximately 0.01ms. The total time required to complete the frequency sweep acquisition of all 100 frequency points is approximately 15 to 30 seconds, which is suitable for the time constraints of clinical examination operations. The acquired raw impedance spectrum data includes the impedance amplitude corresponding to each frequency point. and phase angle ,in Indicates the first Frequency values ​​at each frequency point Values ​​range from 1 to , This represents the total number of frequency points.

[0019] It is worth noting that the geometric parameters of the coaxial dual-electrode detection probe have a significant impact on detection performance. A central electrode diameter that is too small will result in insufficient effective detection area, too few dentinal tubules covered in a single measurement, and increased statistical volatility in impedance measurements. Conversely, a central electrode diameter that is too large may exceed the range of a single carious area, leading to impedance signal aliasing between healthy and carious areas, reducing the accuracy of caries depth estimation. In one embodiment of this invention, a central electrode diameter of 0.8 mm can cover approximately 200 to 500 dentinal tubules in a single measurement, a number sufficient to produce a statistically stable impedance response. The inner diameter of the outer ring electrode determines the spatial constraint range of the current loop, thus affecting the depth penetration capability of the detection area. In one embodiment of this invention, a 2 mm inner diameter of the outer ring electrode enables an effective detection depth of approximately 4 mm to 5 mm, covering the entire clinical detection range from superficial to deep caries.

[0020] In addition, before the probe is placed against the dentin surface, it is recommended to moisten the detection area with a small amount of physiological saline to reduce the contact resistance between the electrode and the dentin surface and improve the stability of low-frequency measurements. In one embodiment of the present invention, approximately 5 μL of 0.9% sodium chloride solution is added to the detection area using a microsyringe, and after standing for 3 to 5 seconds, probe positioning and signal acquisition are performed.

[0021] Step S2: Impedance Spectrum Feature Extraction. After obtaining the original impedance spectrum data, this step performs signal processing and feature extraction. In one embodiment of the present invention, lock-in amplification and demodulation technology is first used to process the original impedance response signal. The core principle of lock-in amplification and demodulation is: using a reference signal with the same frequency and phase as the excitation signal as the demodulation carrier, multiplying the response signal with the reference signal and performing low-pass filtering, thereby accurately extracting the response component corresponding to the excitation frequency under extremely low signal-to-noise ratio conditions.

[0022] Specifically, for frequencies of The excitation signal, the lock-in amplifier simultaneously outputs two demodulation results: in-phase component and orthogonal components . In-phase component The component in the response signal that is in phase with the excitation signal corresponds to the real part of the complex impedance. Orthogonal components The component in the response signal that is 90° out of phase with the excitation signal corresponds to the imaginary part of the complex impedance. In one embodiment of the present invention, the equivalent noise bandwidth of the lock-in amplifier is set to 0.1 Hz to 10 Hz, preferably 1 Hz. This bandwidth selection enables the lock-in amplifier to effectively suppress broadband noise interference outside the excitation frequency, achieving a noise suppression ratio of at least 60 dB.

[0023] Through the above demodulation process, the complex impedance value is obtained at each frequency point. All The complex impedance values ​​at each frequency point constitute the complex impedance spectrum dataset. This dataset is typically represented by a Nyquist plot ( The horizontal axis is... (with frequency as the horizontal axis) and Bode plot (frequency as the vertical axis). and The Nyquist plot of dentin tissue is represented in two forms (vertical axis and horizontal axis, respectively). Ideally, the Nyquist plot of dentin tissue presents two or three semicircular arcs, corresponding to the impedance contributions of the healthy layer, the carious layer, and the pulp cavity layer, respectively. The difference in the time constant of different semicircular arcs reflects the difference in dielectric properties of each layer.

[0024] After constructing the complex impedance spectrum, this invention further evaluates the signal quality of the spectral data. In one embodiment of this invention, the signal quality evaluation includes verification in the following three dimensions:

[0025] The first dimension is the signal-to-noise ratio (SNR) test. For each frequency point, the SNR of the impedance measurement is calculated. The calculation method is as follows:

[0026] ,

[0027] in, For the first The impedance amplitude at each frequency point, in units of ; For the first The standard deviation of impedance measurements at each frequency point is calculated by repeatedly sampling at the same frequency point 3 to 5 times, and the unit is . ; This represents common logarithmic operations with base 10. When Below the preset threshold When this occurs, it indicates that the measurement accuracy at that frequency point is insufficient. In one embodiment of the present invention, The setting is 40dB, which means that the impedance signal amplitude must be at least 100 times the noise level.

[0028] The second dimension is the Kramers-Kronig consistency test. This test, based on the causality principle of linear time-invariant systems, verifies whether the real and imaginary parts of the complex impedance satisfy the Kramers-Kronig relationship. If the measured data deviates from the theoretical relationship by more than a preset residual limit, it indicates that there may be nonlinear effects or system drift in the measurement process. In one embodiment of the present invention, the Kramers-Kronig residual limit is set to 5%, that is, the deviation of the real and imaginary parts at each frequency point does not exceed 5%.

[0029] The third dimension is the trend continuity test. This verifies the monotonicity and smoothness of the impedance change trend between adjacent frequency points, identifying and marking abnormal jump points caused by external interference or poor contact. In one embodiment of the invention, when the rate of change of impedance amplitude between adjacent frequency points exceeds three times the average rate of change of the previous five frequency points, that point is marked as a suspected anomaly.

[0030] When signal quality assessment results indicate that the measurement data at certain frequency points do not meet quality requirements, the system automatically executes an adaptive adjustment strategy. In one embodiment of the present invention, the adaptive adjustment strategy includes: for frequency points with insufficient signal-to-noise ratio, increasing the amplitude of the excitation signal to 1.5 to 2 times the original value and re-acquiring; for abnormal jump points, adding auxiliary frequency points near the frequency point for supplementary acquisition; for frequency bands where Kramers-Kronig consistency is not satisfied, extending the acquisition time of each frequency point in the band to twice the original time to improve the steady-state performance. After adaptive adjustment and re-acquisition, quality-verified impedance spectrum characteristic data is obtained, which serves as the input for the subsequent step S3.

[0031] Step S3: Equivalent circuit model fitting. This step establishes a multilayer impedance equivalent circuit model based on the layered structural characteristics of dentin tissue, and uses the impedance spectrum feature data output in step S2 to fit and solve the model parameters. In one embodiment of the present invention, the equivalent circuit model abstracts the dentin region to be tested into a three-layer series impedance network along the depth direction, namely, a healthy dentin layer, a carious dentin layer, and a pulp cavity layer.

[0032] For each tissue layer, the present invention employs a resistive element. The parallel combination with a constant-phase-angle element (CPE) serves as the basic equivalent unit. The reason for choosing a constant-phase-angle element instead of an ideal capacitor is that dentin tissue is a heterogeneous complex composed of hydroxyapatite mineral crystals, collagen fibers, and tubular fluid, exhibiting significant frequency dispersion characteristics in its dielectric response, meaning its capacitive behavior deviates from that of an ideal capacitor. The constant-phase-angle element can introduce a frequency dispersion index... To accurately describe this non-ideal capacitance behavior.

[0033] In one embodiment of the present invention, the impedance expression of the constant phase angle element is as follows: ,in, The complex impedance value of a constant phase angle element, in units of ; This is the pseudo-capacitance parameter for a constant phase angle element, and its unit is... This reflects the organization's energy storage capacity; The imaginary unit; The excitation frequency is expressed in Hz. The frequency dispersion index is dimensionless and ranges from 0 to 1. It degenerates into an ideal capacitor when It degenerates into pure resistance. The physical significance of this parameter lies in characterizing the degree of non-uniformity of the tissue's microstructure: the microstructure of well-mineralized healthy dentin is relatively ordered, and its... The value is relatively high; the demineralized dentin has a disordered microstructure due to demineralization. The value is low. In one embodiment of the present invention, the frequency diffusion index of the healthy dentin layer... The value ranges from 0.85 to 0.95, preferably 0.90; the frequency diffusion index of the decayed dentin layer. The value ranges from 0.60 to 0.80, preferably 0.70; the frequency diffusion index of the pulp cavity layer. The value ranges from 0.90 to 0.98. Since the pulp cavity is filled with blood vessels and nerve tissue, its ionic liquid properties are closer to those of an ideal electrolyte.

[0034] Based on the above equivalent unit, the total impedance expression of the three-layer series impedance network model is:

[0035] ,

[0036] in, For frequency The total complex impedance value of the model at the location, in units of ; , , These are the electrical resistance values ​​of the healthy dentin layer, the decayed dentin layer, and the pulp cavity layer, respectively, in units of [unit missing]. This reflects the magnitude of ion conduction resistance in each layer of tissue; , , These are the pseudo-capacitor parameters for the three layers; , , These are the frequency dispersion indices for the three layers; The series impedance is the contact resistance and wire resistance, expressed in ohms, with units of Ω. The value is usually 10. Up to 100 The model contains a total of 10 parameters to be fitted, namely... .

[0037] In one embodiment of the present invention, a complex impedance nonlinear least squares fitting algorithm is used to solve for the above 10 parameters. Specifically, the fitting objective function is defined as:

[0038] ,

[0039] in, The sum of squares of the weighted residuals is dimensionless. This represents the total number of frequency points. and The first The measured real and imaginary parts of the complex impedance at each frequency point, in units of . ; and The model is in the th The real and imaginary parts of the complex impedance calculated at each frequency point, in units of... ; For the first The impedance amplitude at each frequency point is used for normalization to ensure that the contribution weights of high-frequency and low-frequency data points to the fitting results are balanced. The weighting factor is proportional to the reciprocal of the impedance amplitude at each frequency point, i.e. The unit is This weighting design allows high-frequency data points with low impedance values ​​to receive higher fitting weights, compensating for the problem of relatively large measurement errors caused by the small impedance amplitude in the high-frequency band.

[0040] Preferably, the nonlinear least squares fitting employs the Levenberg-Marquardt optimization strategy, which combines the global convergence of gradient descent with the fast local convergence of the Gauss-Newton method. In one embodiment of the invention, the initial value of the damping factor in the Levenberg-Marquardt algorithm is set to 0.001. After each iteration, the damping factor is adaptively adjusted according to the direction of change of the objective function value: if the objective function value decreases, the damping factor is reduced to 0.1 times its original value to accelerate convergence; if the objective function value increases, the damping factor is increased to 10 times its original value to enhance stability. The iteration termination condition is set as follows: the relative change in the objective function value is less than... Or the relative change in the parameter vector is less than Or the number of iterations reaches the maximum value of 1000.

[0041] To improve the fitting convergence and the uniqueness of parameter identification, one embodiment of the present invention also includes parameter constraints. Resistance values ​​of each layer. The constraint range is set based on the typical electrical parameters of dentin tissue: The range is to , The range is to , The range is to The electrical resistance of the decayed dentin layer is significantly lower than that of the healthy layer. This is because mineral dissolution during caries leads to the opening of tubular channels and an increase in ion concentration, thus significantly increasing conductivity. (Pseudo-capacitance parameter) The constraint range is to Frequency dispersion index The constraint range is 0.5 to 1.0. Series resistor. The constraint range is 1 Up to 1000 .

[0042] After fitting is complete, the system outputs the fitting parameter values ​​and goodness-of-fit indices for each layer. In one embodiment of the present invention, the root mean square (RMSE) value of the fitting residuals and the coefficient of determination are used. As a goodness-of-fit evaluation index. When When the value is greater than 0.98, the fitting result is considered reliable, and the fitting parameters can be used for subsequent caries depth inversion. When the value is below 0.95, the system issues a fitting quality warning, prompting the operator to check the probe contact status or reacquire data.

[0043] Preferably, an intelligent selection strategy for initial parameter values ​​can be employed during the fitting process. In one embodiment of the present invention, the system first performs a coarse estimation based on the geometric features of the Nyquist plot: by identifying the diameter of the semicircular arcs in the plot, an initial estimate of the resistance value of each layer is obtained approximately; by identifying the characteristic frequency corresponding to each semicircular arc, an initial estimate of the time constant is obtained approximately. This provides a high-quality starting point for the Levenberg-Marquardt iteration, significantly improving the fitting convergence speed and success rate. Practice shows that after adopting the above-mentioned intelligent initialization strategy, the average number of iterations required for fitting convergence is reduced from 200 to about 50, and the time taken for a single fitting is shortened from 1.5s to 0.4s.

[0044] Furthermore, in one embodiment of the present invention, to address the degradation of parameter identification in the three-layer model caused by severe caries, the system also implements an adaptive switching mechanism for the model order. When caries has penetrated deep into the pulp cavity, the electrical difference between the caries layer and the pulp cavity layer may become insignificant, and the parameter identification of the three-layer model may encounter pathological conditions. The system determines the suitability of the model order by comparing the Akaike information criterion values ​​of the two-layer model and the three-layer model: when the Akaike information criterion value of the three-layer model is not significantly better than that of the two-layer model, it degenerates into a two-layer model for fitting, and a mapping function specifically designed for deep caries scenarios is used in the depth inversion step.

[0045] Step S4: Caries Depth Inversion. This step, based on the equivalent circuit model fitting parameters output in Step S3, achieves numerical output of caries depth through a pre-established quantitative mapping relationship. In one embodiment of the present invention, this step includes two sub-processes: caries depth mapping and temperature compensation.

[0046] In the caries depth mapping process, it is first necessary to establish a quantitative mapping relationship between impedance characteristic parameters and caries depth based on extracted tooth samples. In one embodiment of the present invention, this calibration process is carried out as follows: 30 to 50 extracted tooth samples with different degrees of caries are collected, and steps S1 to S3 are performed on each sample to obtain equivalent circuit fitting parameters. Then, the samples are sectioned along the caries direction, and the actual caries depth is measured under an optical microscope. Based on the resistance value of the caries layer and caries layer frequency diffusion index The independent variable is the actual depth of caries. Using the variable as the dependent variable, a quantitative mapping model is established through multiple regression analysis.

[0047] In one embodiment of the present invention, the quantitative mapping model takes the following form:

[0048] ,

[0049] in, The value is the estimated depth of caries, in mm; The electrical resistance of the decayed dentin layer, in units of . ; The frequency diffusion index of the decayed dentin layer is dimensionless. For natural logarithm operations; This is the regression intercept term, in mm; The regression coefficient for the logarithmic term of resistance is given in mm. The regression coefficient for the frequency dispersion index term is expressed in mm. The regression coefficient for the interaction term, in mm, reflects the synergistic effect of electrical resistance and frequency dispersion index on caries depth. The reason for this is... The natural logarithm is used because the resistance of the caries layer varies across multiple orders of magnitude; logarithmic transformation compresses the dynamic range of the data and improves the linearity of the regression model. Interaction term. The introduction of this concept is based on physical considerations: the greater the depth of caries, the greater the decrease in resistance value, and the lower the degree of mineralization, the lower the frequency dispersion index, and there is a coupling effect between the two.

[0050] In one embodiment of the present invention, the regression coefficients obtained based on 40 extracted tooth samples are as follows: mm, mm, mm, mm. The root mean square error of the cross-validation of this calibration model is 0.18 mm, and the coefficient of determination is... The value is 0.94. Preferably, when the calibrated samples cover a caries depth range of 0 mm to 4 mm, the model's applicability corresponds to the entire detection scenario from superficial caries to deep caries.

[0051] In the temperature compensation process, this invention introduces a temperature correction mechanism to eliminate the systematic bias of oral environment temperature fluctuations on impedance measurement results. Oral temperature typically fluctuates between 32°C and 37°C, but can change within a range of 20°C to 40°C for a short period after consuming hot or cold foods. Temperature changes primarily alter impedance measurements by affecting the ion mobility of the fluid within the tubules and the dielectric constant of the tissue matrix.

[0052] In one embodiment of the present invention, the temperature compensation algorithm is established based on the Arrhenius temperature-conductivity relationship:

[0053] ,

[0054] in, To correct the resistance value of the caries layer to the reference temperature, the unit is... ; The resistance value of the caries layer obtained at the actual measurement temperature, in units of... ; The activation energy for ion conduction, expressed in J, is obtained by Arrhenius fitting of impedance measurement data of standard dentin samples at different temperatures. This is described in one embodiment of the present invention. Values J; Here is the Boltzmann constant, which takes the value of J / K; For reference temperature, it is set to 310.15K, or 37℃, based on the normal temperature of the human oral cavity; The temperature measured is the local oral cavity temperature in K, obtained by measuring a miniature thermistor integrated on the probe, with a measurement accuracy better than 0.1℃.

[0055] After temperature compensation, the corrected resistance value of the caries layer is... and frequency diffusion index Substituting the values ​​into the quantitative mapping model, the final caries depth detection result is obtained. In one embodiment of the present invention, the system also simultaneously outputs the confidence interval of the detection result: when the goodness of fit... And temperature deviation When K is reached, a high confidence level is output; otherwise, a low confidence level is output and the operator is advised to retest.

[0056] It is worth noting that the calibration accuracy of the quantitative mapping model directly determines the final accuracy of caries depth inversion; therefore, the selection of calibration samples and the standardization of the calibration process are crucial. In a preferred embodiment of the present invention, the calibration samples should cover different age groups (18 to 65 years old) and different tooth positions (premolars and molars) to fully cover individual differences in the microstructure of dentin tissue. During the calibration process, tissue sections should be cut using a hard tissue microtome at low speed, with the section thickness controlled between 100 μm and 200 μm to ensure that the identification accuracy of caries boundaries under an optical microscope is better than 0.05 mm. Furthermore, after the calibration model is established, leave-one-out cross-validation or ten-fold cross-validation should be used to evaluate the model's generalization ability to prevent overfitting and subsequent decrease in extrapolation accuracy.

[0057] Furthermore, in one embodiment of the present invention, the system also includes a clinical grading output mechanism for the caries depth detection results. Based on the relationship between the caries depth value and the remaining dentin thickness, the system automatically generates a suggested range of cavity preparation depths to assist clinicians in developing precise treatment plans. When the caries depth is detected to be close to the pulp chamber, the system automatically issues a pulp proximity alarm, prompting the operator to exercise extra caution during cavity preparation to protect pulp vitality.

[0058] Furthermore, this invention constructs a closed-loop feedback path from step S4 to step S1. Specifically, after obtaining a preliminary estimate of the caries depth in step S4, the system adaptively adjusts the frequency distribution density of the sweep signal in step S1 based on this estimate. In one embodiment of this invention, when the estimated caries depth is shallow, i.e. When the depth of caries is mm, the impedance difference between the carious layer and the healthy layer is relatively small. In this case, increasing the sampling density in the mid-frequency range of 1kHz to 100kHz improves the discrimination. When the estimated caries depth is deeper, i.e. When the thickness reaches 1 mm, the low-frequency impedance characteristics of the carious layer become more pronounced. At this point, the sampling density is increased in the low-frequency range of 10 Hz to 1 kHz. This closed-loop feedback mechanism enables adaptive configuration of detection resources, improving detection accuracy within the target depth range without significantly increasing the total acquisition time.

[0059] Furthermore, the model parameter fitting results from step S3 are also fed back to step S2, affecting the threshold for signal quality assessment. In one embodiment of the present invention, when the model fitting finds that the time constants of each layer differ significantly, i.e., the time constant of the healthy layer... Time constant of caries layer When the signal-to-noise ratio exceeds 100, the system automatically increases the Kramers-Kronig consistency check accuracy of the corresponding frequency band to 3% to ensure that the data quality of this key frequency band meets the model identification requirements. The signal quality evaluation result of step S2 is also fed back to step S1. When the signal-to-noise ratio at a certain frequency point continues to be substandard, the system automatically increases the excitation amplitude at that frequency point to the maximum value within the safe range. Through the above multi-level closed-loop feedback mechanism, a deeply coupled collaborative working architecture is formed among the four steps, making the overall system performance significantly better than the linear processing method of simply connecting the steps.

[0060] Reference Figure 2 This invention also provides a dentin caries depth electrical impedance detection system, which corresponds one-to-one with the steps in the above method embodiments. The technical solution of this invention is described from the perspective of hardware architecture and functional implementation. In one embodiment of this invention, the system includes a multi-frequency impedance spectrum acquisition module, an impedance spectrum feature extraction module, an equivalent circuit model fitting module, and a caries depth inversion module.

[0061] The multi-frequency impedance spectroscopy acquisition module is the front-end sensing unit of the system. Its core hardware includes a coaxial dual-electrode detection probe and an excitation signal generator. As detailed in the previous method embodiments, the coaxial dual-electrode detection probe adopts a coaxial structure design with a central electrode diameter of 0.5mm to 1mm and an outer ring electrode inner diameter of 1.5mm to 3mm. An insulating layer is provided between the central electrode and the outer ring electrode, and both surfaces are coated with a biocompatible conductive coating. The probe tip is designed with an arcuate configuration and a radius of curvature of 5mm to 10mm to adapt to the curved surface shape of the dentin occlusal surface and proximal surface, ensuring that the probe can achieve uniform contact at different tooth positions. The excitation signal generator adopts direct digital frequency synthesis technology, with a frequency control word width of 32 bits, and can output a sinusoidal sweep signal with a frequency range of 10Hz to 1MHz, an amplitude of 10mV to 100mV, and a frequency resolution better than 0.01Hz. In one embodiment of the present invention, the total harmonic distortion of the excitation signal generator is better than -80dB, ensuring that the spectral purity of the excitation signal meets the requirements of precision impedance measurement. The probe also integrates a miniature NTC thermistor sensor, with a package size of 0.5mm × 0.5mm × 0.3mm. This sensor is used to acquire the local oral temperature at the probe contact point in real time. The sensor sensitivity is 10mV / ℃, the temperature measurement range is 15℃ to 45℃, and the response time is better than 0.5s. This module corresponds to step S1 in the method embodiment, responsible for generating the excitation signal, acquiring the impedance response, and outputting the raw impedance spectrum data.

[0062] The impedance spectrum feature extraction module is the signal conditioning unit of the system, and its core hardware includes a lock-in amplifier and an analog-to-digital converter (ADC). The lock-in amplifier adopts a digital lock-in detection architecture, and its equivalent noise bandwidth can be programmably adjusted within the range of 0.1Hz to 10Hz by the cutoff frequency of the digital low-pass filter, with a dynamic reserve better than 80dB and a common-mode rejection ratio better than 100dB. The ADC adopts a Σ-Δ architecture, with a sampling accuracy of 24 bits, a sampling rate of 1MSPS, and an effective number of bits (ENOB) of not less than 21 bits. This module receives the raw impedance spectrum data output from the multi-frequency impedance spectrum acquisition module, performs lock-in amplification and demodulation processing to separate the real and imaginary components at each frequency point, constructs a complex impedance spectrum, and performs data quality verification according to the three-dimensional signal quality assessment mechanism described in step S2 of the method embodiment. When the data quality is substandard, the excitation parameters are adjusted by feeding back a control signal to the acquisition module. Preferably, the module further includes a programmable gain amplifier with a gain range of 0dB to 60dB and a step accuracy of 0.1dB, used to adaptively adjust the front-end gain according to the dynamic range of impedance amplitude in different frequency bands, avoiding signal saturation or undersampling. In one embodiment of the present invention, the gain switching time of the programmable gain amplifier is less than 1µs, ensuring that gain switching does not introduce additional transition delay during rapid frequency sweeping.

[0063] The equivalent circuit model fitting module is the core computing unit of the system, implemented using an embedded digital signal processor. In one embodiment of the invention, the processor has a main frequency of no less than 200MHz, is equipped with at least 256MB of computing memory, and supports single-precision and double-precision floating-point operations. This module incorporates the dentin three-layer series impedance network equivalent circuit model described in step S3 of the method embodiment, as well as the Levenberg-Marquardt nonlinear least squares fitting algorithm. It receives the quality-verified data output by the impedance spectrum feature extraction module and automatically completes the fitting solution for 10 model parameters. Preferably, this module pre-stores the initial value range and constraint boundaries of each parameter obtained from a large amount of clinical data statistics, and provides differentiated initial parameter lookup tables for different tooth positions and different patient age groups to accelerate the fitting convergence process and reduce the risk of getting trapped in local optima. The single execution time of the fitting operation is better than 2 seconds, meeting the timeliness requirements of real-time clinical diagnosis. In one embodiment of the invention, this module also implements a multi-starting-point strategy, that is, simultaneously starting the fitting operation from 5 different initial points in the parameter space, selecting the fitting result with the smallest objective function value as the final output, to improve the reliability of the global optimal solution search.

[0064] The caries depth inversion module is the system's decision output unit. It receives the parameter values ​​of each layer output by the equivalent circuit model fitting module and the local oral temperature values ​​collected by the temperature sensor. It performs the temperature compensation calculation and caries depth mapping calculation described in step S4 of the method embodiment, and finally outputs the temperature-corrected caries depth detection result. In one embodiment of the present invention, the module presents the detection result to the operator through an LCD display interface. The displayed content includes the caries depth value and its 95% confidence interval, the caries severity grading indicator, and a real-time visualization interface of the Nyquist and Bode plots. The caries severity grading standard is as follows: a depth of 0mm to 0.5mm is marked in green as a superficial caries indication, a depth of 0.5mm to 2mm is marked in yellow as a moderate caries warning, and a depth greater than 2mm is marked in red as a deep caries alarm. Preferably, the module also stores a comparison table of caries depth and cavity preparation depth recommendations, which can provide clinical operators with reference recommendations for cavity preparation depth.

[0065] The closed-loop feedback connections between the system modules are as follows: The caries depth inversion module feeds back the depth estimate to the multi-frequency impedance spectrum acquisition module for adaptive adjustment of the sweep frequency distribution density, forming the main feedback loop. The equivalent circuit model fitting module feeds back the time constant information of each layer to the impedance spectrum feature extraction module for optimizing the signal quality assessment criterion threshold. The impedance spectrum feature extraction module feeds back the signal quality assessment results to the acquisition module for adjusting the excitation amplitude and the number of acquisitions. The multi-level closed-loop architecture formed by the three feedback paths enables the system to progressively optimize the acquisition parameters and processing strategies during a single detection, achieving a gradual improvement in detection accuracy. This closed-loop collaborative architecture produces a significant 1+1>2 nonlinear synergistic gain effect: using multi-frequency acquisition technology alone can only obtain coarse spectral information, and using the equivalent circuit model alone can only perform theoretical parameter estimation. However, the deep coupling formed by the closed-loop feedback allows the acquisition strategy to be accurately optimized for model identification needs, and the model fitting can obtain more reliable parameter estimates based on the optimized high-quality data. The overall detection accuracy of the system far exceeds the level that can be achieved by simply superimposing the various technical links.

[0066] In addition, in one embodiment of the present invention, the system further includes a power management unit and a wireless communication unit. The power management unit is powered by a rechargeable lithium polymer battery with a capacity of not less than 2000mAh, an operating voltage of 3.3V to 4.2V, and supports continuous operation for at least 4 hours. The wireless communication unit supports Bluetooth 5.0 Low Energy protocol with a data transmission rate of not less than 2Mbps, used to transmit test results in real time to external terminal devices such as tablets or smartphones for data storage, historical comparison analysis, and remote consultation. Preferably, the overall size of the system is controlled within a handheld pen-shaped shell with a length not exceeding 200mm and a maximum diameter not exceeding 15mm, and a weight not exceeding 80g, facilitating flexible operation and precise positioning by the operator within the oral cavity. The shell material is medical-grade stainless steel or polycarbonate, with an antibacterial surface treatment, allowing for high-temperature and high-pressure sterilization using standard medical sterilization methods, meeting the hygiene and safety requirements for oral clinical instruments.

[0067] To verify the detection performance of the method and system of the present invention, systematic experimental tests were conducted under the following conditions.

[0068] For the experimental samples, 60 premolars and third molars that needed to be extracted for orthodontic treatment were collected. After independent confirmation by two dental specialists, they were divided into four groups according to the degree of caries: 15 teeth without caries, 15 teeth with superficial caries (0mm to 0.5mm caries depth), 15 teeth with moderate caries (0.5mm to 2mm caries depth), and 15 teeth with deep caries (caries depth greater than 2mm). All samples underwent impedance testing within 24 hours after extraction and were stored in artificial saliva at 37°C to maintain tissue hydration during the testing period. After the testing of each sample, sections were prepared along the direction of caries, and the actual caries depth was independently measured by two pathologists under an optical microscope as the gold standard.

[0069] Regarding environmental conditions, the experiment was conducted in a temperature- and humidity-controlled laboratory, with the room temperature maintained at 25±1℃ and the humidity at 50±5%RH. During testing, samples were fixed in custom-made tooth retainers, and the probe contact surface was immersed in artificial saliva at 37℃ to simulate the intraoral testing environment. The artificial saliva formula was prepared according to a standard prescription and contained NaCl, KCl, and CaCl₂. NaH PO It contains ingredients such as [list of ingredients], with the pH value adjusted to 6.8 to 7.2, and its conductivity is comparable to that of human saliva.

[0070] In terms of detection operation, impedance spectrum was acquired at three different detection sites for each sample. The acquisition parameters were set as follows: 100 frequency points, frequency range of 10Hz to 1MHz, and excitation amplitude of 50mV. Each site was measured three times consecutively, and the average value was taken as the measurement result for that site. A single full-band scan took approximately 20 seconds, and the complete detection process per unit point, including data processing, took approximately 25 seconds.

[0071] Regarding the accuracy of caries depth detection, the root mean square error (RMSE) between the caries depth estimate obtained by the method of this invention and the gold standard for tissue sections is 0.19 mm, and the coefficient of determination is [missing information]. The RMSE was 0.93, and the mean absolute error (MAE) was 0.15 mm. Looking at the groups, the RMSE was 0.14 mm for the superficial caries group, 0.18 mm for the moderate caries group, and 0.24 mm for the deep caries group, reflecting a reasonable trend that the inversion error slightly increases with the depth of caries. This level of accuracy is better than the 0.2 mm detection accuracy target set in the user's requirements.

[0072] In the binary classification task of caries detection (healthy vs. decayed), using tissue section results as the gold standard, the sensitivity was 92%, the specificity was 89%, the positive predictive value was 93%, the negative predictive value was 87%, and the area under the curve (AUC) was 0.95. In the caries depth grading task (superficial caries vs. moderate caries vs. deep caries), an ordered logistic regression model was used for grading, with an overall grading accuracy of 87%. The grading accuracy was 85% for the superficial caries group, 89% for the moderate caries group, and 88% for the deep caries group.

[0073] In comparison with existing technologies, parallel detection was performed using the same sample set with both X-ray digital imaging and traditional single-frequency impedance spectroscopy. The X-ray digital imaging method achieved a caries detection sensitivity of 78%, a specificity of 91%, and a caries depth estimation error (RMSE) of 0.45 mm. Compared to the X-ray method, this invention improves sensitivity by 14 percentage points and depth estimation accuracy by approximately 58%, while completely avoiding ionizing radiation exposure. Compared to the traditional single-frequency impedance spectroscopy method, this invention improves the accuracy of caries depth grading by 12 percentage points. This improvement is mainly attributed to the ability of multi-frequency impedance spectroscopy acquisition combined with a layered equivalent circuit model to extract richer tissue electrical characteristics, while the single-frequency method can only obtain the impedance amplitude at a single frequency point, resulting in limited information dimensions.

[0074] Regarding the effect of temperature compensation, to quantify the role of the temperature compensation algorithm, impedance spectra were repeatedly measured on 10 moderate caries samples at five different temperatures: 20℃, 25℃, 30℃, 35℃, and 40℃. Without temperature compensation, the temperature sensitivity coefficient of the caries depth estimate was 0.08 mm / ℃, meaning that a 1℃ change in temperature caused a 0.08mm shift in the depth estimate. Within a 20℃ temperature range (20℃ to 40℃), the maximum deviation in the depth estimate reached 1.6mm, severely impacting the clinical reliability of the detection. After correction using the temperature compensation algorithm of this invention, the temperature sensitivity coefficient decreased to 0.01 mm / ℃, the maximum deviation decreased to 0.2mm, and the compensation effect reached 87.5%, ensuring stable detection results within the normal oral temperature fluctuation range.

[0075] Regarding the closed-loop feedback effect, comparative tests were conducted on all 60 samples in both open-loop mode (fixed frequency distribution) and closed-loop mode (adaptive frequency distribution). In closed-loop mode, the depth estimation RMSE of the deep caries sample group decreased from 0.28 mm in open-loop mode to 0.24 mm, an improvement of approximately 15%; the detection sensitivity of the superficial caries sample group increased from 85% in open-loop mode to 92%, an improvement of approximately 8%. Closed-loop feedback enables the system to concentrate limited sampling resources on the most diagnostically valuable frequency bands based on preliminary diagnostic results, achieving synergistic optimization of detection efficiency and accuracy without changing the total acquisition time.

[0076] Regarding repeatability and reproducibility, the standard deviation of caries depth measured 10 times consecutively for the same sample was 0.05 mm, and the coefficient of variation was 3.2%. The standard deviation between groups of 15 samples from the same group, measured independently by three different operators at different time periods, was 0.08 mm, and the intra-group correlation coefficient (ICC) was 0.95, indicating that the system of the present invention has excellent measurement repeatability and inter-operator reproducibility.

[0077] Based on the above experimental results, the method and system of this invention are significantly superior to existing technologies in key performance indicators such as accuracy, sensitivity, specificity, and resistance to environmental interference in caries depth detection. They can serve as a non-invasive auxiliary diagnostic tool for clinical caries depth assessment, providing objective and quantitative basis for accurate decision-making on the depth of clinical cavity preparation.

[0078] The embodiments of the present invention are not limited to the specific embodiments described above. Those skilled in the art can make various equivalent changes or substitutions based on the technical solutions of the present invention, and all such changes or substitutions should be included within the protection scope of the present invention.

Claims

1. A method for detecting the depth of dentin caries using electrical impedance to measure its properties, characterized in that, Includes the following steps: Multi-frequency impedance spectrum acquisition steps: The coaxial dual-electrode detection probe is attached to the surface of the dentin to be tested. A sinusoidal sweep excitation signal with a frequency range of 10Hz to 1MHz is applied to the detection probe through the excitation signal module. A local current loop is formed between the central electrode and the outer ring electrode of the coaxial dual-electrode detection probe. The impedance response signals of the dentin to be tested at multiple excitation frequencies are acquired to obtain the raw impedance spectrum data containing impedance amplitude and phase information. Impedance spectrum feature extraction steps: The original impedance spectrum data is subjected to lock-in amplification and demodulation processing, the real and imaginary components corresponding to each excitation frequency are separated, a complex impedance spectrum is constructed, and the signal quality of the complex impedance spectrum is evaluated. Based on the signal quality evaluation results, the amplitude parameters of the excitation signal are adaptively adjusted to obtain quality-verified impedance spectrum feature data. Equivalent circuit model fitting steps: Establish a dentin multilayer impedance equivalent circuit model. This equivalent circuit model abstracts the dentin to be tested into a series impedance network of healthy dentin layer, carious dentin layer and pulp cavity layer. Each layer contains a parallel combination of resistive elements and constant phase angle elements. Using impedance spectrum characteristic data as input, the parameters of each layer of the equivalent circuit model are fitted and solved using a complex impedance nonlinear least square fitting algorithm to obtain the resistance value and constant phase angle element parameters of each layer. Caries depth inversion steps: Based on the pre-established quantitative mapping relationship between impedance characteristics and caries depth, the resistance value of the carious dentin layer and the constant phase angle element parameters obtained by fitting the equivalent circuit model are converted into caries depth values. The influence of oral environment temperature on impedance measurement values ​​is eliminated by temperature compensation algorithm, and the temperature-corrected caries depth detection results are output.

2. The method for detecting dentin caries depth using electrical impedance to determine the depth of dentin caries according to claim 1, characterized in that, In the multi-frequency impedance spectrum acquisition step, the amplitude of the sinusoidal sweep excitation signal is 10mV to 100mV, the number of sweep points is 50 to 200, and the frequency points are evenly distributed on the logarithmic coordinate.

3. The method for detecting dentin caries depth using electrical impedance to determine the depth of dentin caries according to claim 1, characterized in that, In the impedance spectrum feature extraction step, signal quality assessment includes calculating the signal-to-noise ratio (SNR) of the impedance measurement at each frequency point. When the SNR is lower than a preset threshold, the amplitude of the excitation signal at the corresponding frequency point is increased and the signal is re-acquired.

4. The method for detecting dentin caries depth using electrical impedance to determine the depth of dentin caries according to claim 1, characterized in that, In the caries depth inversion step, the temperature acquisition range of the temperature compensation algorithm is 20℃ to 40℃, and the temperature compensation coefficient is obtained by calibration of standard dentin samples under known temperature conditions.

5. The method for detecting dentin caries depth using electrical impedance to determine the depth of dentin caries according to claim 1, characterized in that, In the equivalent circuit model fitting step, the complex impedance nonlinear least squares fitting algorithm adopts the Levenberg-Marquardt optimization strategy, with the objective function being to minimize the weighted sum of squares of the fitting residuals. The weighting factor is proportional to the reciprocal of the impedance amplitude at each frequency point.

6. The method for detecting dentin caries depth using electrical impedance to determine the depth of dentin caries according to claim 1, characterized in that, In the equivalent circuit model fitting step, the impedance expression of the constant phase angle element includes the frequency diffusion index. The frequency diffusion index of the healthy dentin layer ranges from 0.85 to 0.95, while the frequency diffusion index of the carious dentin layer ranges from 0.60 to 0.

80.

7. The method for detecting dentin caries depth using electrical impedance to determine the depth of dentin caries according to claim 1, characterized in that, The detection results of the caries depth inversion step are fed back to the multi-frequency impedance spectroscopy acquisition step. The frequency distribution density of the sweep signal is adaptively adjusted according to the current caries depth estimate so that the sampling density of the low-frequency and high-frequency bands matches the caries depth range. In the caries depth inversion step, the quantitative mapping relationship is established in the following way: multiple extracted tooth samples with known caries depths are obtained, the impedance spectra of each extracted tooth sample are collected and fitted to obtain the caries layer parameters, the caries layer resistance value and frequency diffusion index are used as independent variables, and the actual caries depth verified by tissue sections is used as the dependent variable. A quantitative mapping model is established through multiple regression analysis.

8. The method for detecting dentin caries depth using electrical impedance to determine the depth of dentin caries according to claim 1, characterized in that, The diameter of the central electrode of the coaxial dual-electrode detection probe is 0.5 mm to 1 mm, and the inner diameter of the outer ring electrode is 1.5 mm to 3 mm. An insulating layer is provided between the central electrode and the outer ring electrode.

9. The method for detecting dentin caries depth using electrical impedance to determine the depth of dentin caries according to claim 1, characterized in that, The surfaces of the central electrode and the outer ring electrode of the coaxial dual-electrode detection probe are coated with a biocompatible conductive coating, which is either a titanium nitride coating or a platinum coating.

10. A dentin caries depth electrical impedance detection system, used to implement the dentin caries depth electrical impedance detection method according to any one of claims 1-9, characterized in that, include: The multi-frequency impedance spectrum acquisition module is used to apply a sinusoidal sweep excitation signal with a frequency range of 10Hz to 1MHz to the dentin under test through a coaxial dual-electrode detection probe, acquire the impedance response signal of the dentin under test at multiple excitation frequencies, and obtain the raw impedance spectrum data. The impedance spectrum feature extraction module is used to perform lock-in amplification and demodulation processing on the original impedance spectrum data, separate the real and imaginary components corresponding to each excitation frequency to construct the complex impedance spectrum, and adaptively adjust the amplitude parameters of the excitation signal according to the signal quality assessment results to obtain impedance spectrum feature data. The equivalent circuit model fitting module is used to establish an equivalent circuit model of a series impedance network containing a healthy dentin layer, a carious dentin layer, and a pulp cavity layer. The model parameters are fitted and solved using a complex impedance nonlinear least squares fitting algorithm to obtain the resistance values ​​of each layer and the parameters of constant phase angle elements. The caries depth inversion module is used to convert the resistance value of the carious dentin layer and the constant phase angle element parameters into caries depth values ​​based on the pre-established quantitative mapping relationship between impedance characteristics and caries depth. It also eliminates the influence of oral environment temperature through a temperature compensation algorithm and outputs the caries depth detection results. The output of the caries depth inversion module is fed back to the multi-frequency impedance spectrum acquisition module to adaptively adjust the frequency distribution density of the sweep frequency signal.

Images