Gastrointestinal three-dimensional dynamic imaging analysis method based on conductivity distribution modeling

Abstract

The invention provides a gastrointestinal three-dimensional dynamic imaging analysis method based on conductivity distribution modeling, and the method comprises the steps: applying at least three composite electric signals with different frequencies to a gastrointestinal region, collecting voltage response signals, and constructing a dynamic conductivity parameter matrix based on frequency domain feature separation. And then, spatial discretization processing is carried out by using a nonlinear interpolation algorithm, a three-dimensional conductivity distribution map is generated, dynamic modulation and demodulation processing is carried out on the map, and a frequency band weight coefficient is adjusted. And mapping the modulated data to a dynamic imaging coordinate system through an inverse projection algorithm to generate a three-dimensional dynamic imaging result. According to the space coordinates and the change rate of the conductivity gradient mutation area, the position and the activity state of the gastrointestinal abnormal area are judged, and a self-adaptive feedback mechanism is adopted to update boundary condition parameters of the initial conductivity distribution model. The accuracy and efficiency of gastrointestinal disease diagnosis can be improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medical imaging, and more particularly, to a gastrointestinal three-dimensional dynamic imaging analysis method based on conductivity distribution modeling. BACKGROUND

[0002] In the field of medical diagnosis, gastrointestinal function detection has always been an important research direction. Traditional gastrointestinal detection methods mainly include X-ray contrast, endoscopy, and some physiological signal monitoring-based technologies. Although X-ray contrast can provide certain anatomical structure information, it has radiation risks and is difficult to monitor gastrointestinal function dynamic changes in real time. Endoscopy is intuitive, but it is an invasive operation, and patients have a poor experience, and it cannot fully reflect the overall gastrointestinal function state. In recent years, bioelectrical impedance imaging (EIT) technology has gradually emerged, which reflects the internal physiological function state by measuring the conductivity changes of human tissues, and has the advantages of no radiation, real-time monitoring, etc. However, the existing EIT technology still faces some challenges in gastrointestinal imaging, such as limited imaging accuracy, insufficient dynamic change capture capability, and difficulty in accurately identifying abnormal gastrointestinal regions.

[0003] In the implementation process of the embodiments of the present application, the inventors found that at least the following problems or defects exist in the prior art: When the existing EIT technology is used for gastrointestinal imaging, due to the lack of comprehensive application of multi-frequency excitation signals and dynamic modulation and demodulation processing, the imaging accuracy and resolution are low, and it is difficult to accurately reflect the dynamic changes of the gastrointestinal tract. In addition, the existing method has deficiencies in abnormal region identification, and lacks an effective adaptive feedback mechanism to optimize the imaging model, thereby affecting the accuracy and reliability of the diagnosis. SUMMARY

[0004] The present application provides a gastrointestinal three-dimensional dynamic imaging analysis method based on conductivity distribution modeling, comprising: S1. Applying a composite electrical signal containing at least three different frequencies to the gastrointestinal region through a multi-frequency current excitation device, and synchronously collecting the voltage response signals of the gastrointestinal surface electrode array; S2. Constructing a dynamic conductivity parameter matrix based on frequency domain feature separation according to the voltage response signals and a preset initial conductivity distribution model; S3. Spatially discretizing the dynamic conductivity parameter matrix using a nonlinear interpolation algorithm to generate a three-dimensional conductivity distribution map containing multiple layers of grid nodes; S4. Performing dynamic modulation and demodulation processing on the three-dimensional conductivity distribution map, including judging the modulation type according to the difference degree of the maps of adjacent time windows, adjusting the frequency band weight coefficient based on the difference degree threshold, and generating the modulated conductivity distribution data; S5. mapping the modulated conductivity distribution data to a dynamic imaging coordinate system by an inverse projection algorithm to generate a three-dimensional dynamic imaging result with time sequence labels; S6. judging the location and activity state of the gastrointestinal abnormal region according to the spatial coordinates and variation rate of the conductivity gradient mutation region in the three-dimensional dynamic imaging result; S7. updating the boundary condition parameters of the initial conductivity distribution model based on the judgment result of the abnormal region by using an adaptive feedback mechanism.

[0005] Further, the dynamic modulation processing in step S4 includes: S41. calculating the difference degree of the atlas of the adjacent time window and ; : wherein, is the three-dimensional conductivity distribution matrix of the current time window, is the Frobenius norm, and ∇ is the spatial gradient operator, represents the Hadamard product, , is the normalization coefficient; S42. when the difference degree exceeds a first threshold , activating a high-frequency modulation mode, and increasing the weight of the frequency band higher than in the composite electrical signal to times (1 > 1) of the original value; S43. when the difference degree is lower than a second threshold , enabling a low-frequency compensation mode, and applying a phase shift to the frequency band lower than .

[0006] Further, the specific implementation of the high-frequency modulation mode in step S42 includes: S421. defining an update formula of the dynamic threshold : wherein, is the initial threshold, is the time decay coefficient, is the current time window, is the initial time, is the weight factor of the maximum value of the historical difference degree; S422. when increasing the high-frequency weight, synchronously adjusting the sampling interval of the electrode array to satisfy the relationship formula: wherein, is the currently activated highest frequency, is the adjustment coefficient; S423. A band-pass filter is applied to the modulated high-frequency signal, and the passband range is set to , and the stopband attenuation is greater than 40 dB.

[0007] Further, the nonlinear interpolation algorithm in step S3 satisfies: For any grid node , the conductivity value is calculated by the following formula: wherein, N is the number of adjacent measurement points, is the weight factor and , is the conductivity value of the i-th measurement point, is the Euclidean distance, is the anisotropic adjustment radius.

[0008] Further, the step S6 of judging the gastrointestinal abnormal area includes: S61. Calculate the modulus mutation coefficient of the conductivity gradient vector: wherein, is the modulus mutation coefficient, t is the time variable, is the three-dimensional conductivity distribution matrix, is the regional average conductivity, is the anti-zero constant; S62. When exceeds the preset threshold, it is marked as a potential abnormal area, and the spatial distribution entropy thereof is calculated: wherein, H is the spatial distribution entropy, is the conductivity proportion of the i-th sub-region; S63. According to the product of the entropy value H and the time change rate , the abnormal activity level is judged, and when the product exceeds the critical value, an alarm signal is generated.

[0009] Further, the adaptive feedback mechanism in step S7 includes: S71. Define the exponential relationship between the model parameter update amount and the area A of the abnormal area: wherein, is the parameter update amount, A is the abnormal area, is the reference area, is the reference update amount, is the difference degree influence factor; S72. Apply constraints to boundary condition parameters: wherein, is the previous time parameter value, is the updated parameter value, , is the preset boundary; S73. When the change rate of in the last three iterations is less than 1%, trigger the parameter freezing mechanism and switch to the static model.

[0010] Further, the low-frequency compensation mode in step S43 further includes: S431. Establish the phase shift and the correlation function of the conductivity change rate: wherein, is the phase shift, erf is the error function, t is the time variable, k is the proportionality constant, is the basic conductivity; S432. Perform harmonic analysis on the phase-adjusted signal, and filter out frequency components higher than the third harmonic; S433. In the compensation mode, reduce the excitation current amplitude of the electrode array to of the original value.

[0011] Further, the determination method of the anisotropy adjustment radius includes: S81. According to the gastrointestinal anatomy, divide the reference radius of different partitions; S82. Adjust the actual radius based on the local conductivity gradient direction: wherein, is the adjusted anisotropy adjustment radius, is the reference radius, is the included angle between the conductivity gradient direction and the tissue fiber direction; S83. Apply a radius attenuation factor to the boundary area: wherein, d is the distance to the nearest anatomical boundary, is the characteristic attenuation length. ​

[0012] Further, the alarm signal generation logic in step S63 includes: S91. Define the comprehensive anomaly index: wherein, is the comprehensive anomaly index, is the modulus mutation coefficient, H is the spatial distribution entropy, is the weighted coefficient and ; S92. When continuously exceeds the threshold value T for time, classify according to the following rules: If : generate a first-level alarm; If : generate a second-level alarm and start data storage; If : generate a third-level alarm and trigger real-time transmission; wherein, T is a preset threshold value, is the duration of continuously exceeding the threshold value; S93. Record the time-frequency feature matrix for subsequent analysis.

[0013] Further, the inverse projection algorithm in step S5 adopts an improved Radon transform, specifically including: S101. Define the integral path in the projection domain as a dynamic curve: wherein, is the projection integral value, is the projection angle, t is the time variable, (x, y, z) is the three-dimensional space coordinate, is the modulated conductivity distribution data, is the Dirac function, s is the ray parameter, is the differential along the ray path; S102. Introduce the motion compensation factor : wherein, is the motion compensation factor, is the gastrointestinal peristalsis speed coefficient, is the rate of change of conductivity over time; S103. Fuse the projection integral value and the motion compensation factor to reconstruct the three-dimensional image, as shown in the following formula: wherein, is the three-dimensional conductivity distribution matrix at the n th iteration, is the update matrix for the n+1 th iteration, R is the Radon transform operator, is the step factor and 0 < < 1, n is the iteration number and n ≥ 0.

[0014] The above-mentioned embodiments of the present application have at least the following beneficial effects: the method of the present application can improve the accuracy and resolution of gastrointestinal three-dimensional dynamic imaging. By using multi-frequency current excitation and frequency domain feature separation technology, the conductivity changes of the gastrointestinal tissue at different frequencies can be more accurately captured, thereby providing a basis for constructing a high-precision three-dimensional conductivity distribution map. At the same time, combined with the nonlinear interpolation algorithm and dynamic modulation and demodulation processing, the ability to capture the dynamic changes of the gastrointestinal tract can be effectively enhanced, further improving the spatio-temporal resolution of imaging, and providing more reliable technical support for real-time monitoring of gastrointestinal function.

[0015] In addition, the present application can also realize rapid identification and accurate positioning of abnormal areas of the gastrointestinal tract. By analyzing the spatial coordinates and change rate of the conductivity gradient mutation area, and combining the adaptive feedback mechanism to dynamically update the conductivity distribution model, the accuracy of abnormal area detection can be effectively improved. This method not only can monitor the functional status of the gastrointestinal tract in real time, but also can early warn potential gastrointestinal diseases, providing a reference basis for clinical diagnosis and treatment. BRIEF DESCRIPTION OF DRAWINGS

[0016] The above and other objects, features and advantages of the exemplary embodiments of the present application will be more apparent from the following detailed description taken in conjunction with the accompanying drawings, in which: Figure 1 FIG. 1 is a flowchart of a gastrointestinal three-dimensional dynamic imaging analysis method based on conductivity distribution modeling according to an embodiment of the present application. DETAILED DESCRIPTION

[0017] The principles and spirits of the present application will be described below with reference to several exemplary embodiments. It should be understood that these embodiments are given only to enable those skilled in the art to better understand and implement the present application, and do not limit the scope of the present application in any way. On the contrary, these embodiments are provided to make the present application more thorough and complete, and to fully convey the scope of the present application to those skilled in the art.

[0018] ​Those skilled in the art will recognize that embodiments of the present invention can be implemented as a system, apparatus, device, method, or computer program product. Therefore, the present invention can be specifically implemented in the following forms: entirely hardware, entirely software (including firmware, resident software, microcode, etc.), or a combination of hardware and software.

[0019] It should be noted that the number of any elements in the accompanying drawings is for illustrative purposes only and not as a limitation, and any naming is for distinction only and has no limiting meaning.

[0020] The following is for reference. Figure 1 , Figure 1 This is a flowchart illustrating a gastrointestinal three-dimensional dynamic imaging analysis method based on conductivity distribution modeling, provided in an embodiment of the present invention. Figure 1 As shown, a three-dimensional dynamic imaging analysis method 100 for the gastrointestinal tract based on conductivity distribution modeling includes: S1. Apply a composite electrical signal containing at least three different frequencies to the gastrointestinal region using a multi-frequency current excitation device, and simultaneously acquire the voltage response signal of the gastrointestinal surface electrode array; S2. Based on the voltage response signal and the preset initial conductivity distribution model, construct a dynamic conductivity parameter matrix based on frequency domain feature separation; S3. The dynamic conductivity parameter matrix is ​​spatially discretized using a nonlinear interpolation algorithm to generate a three-dimensional conductivity distribution map containing multiple layers of grid nodes; S4. Perform dynamic modulation and demodulation processing on the three-dimensional conductivity distribution map, including determining the modulation type based on the difference in the map between adjacent time windows, adjusting the frequency band weight coefficient based on the difference threshold, and generating modulated conductivity distribution data; S5. The modulated conductivity distribution data is mapped to the dynamic imaging coordinate system using an inverse projection algorithm to generate a three-dimensional dynamic imaging result with time series markers; S6. Based on the spatial coordinates and rate of change of the conductivity gradient abrupt change region in the three-dimensional dynamic imaging results, determine the location and activity status of the abnormal gastrointestinal region; S7. Based on the judgment result of the abnormal region, the boundary condition parameters of the initial conductivity distribution model are updated using an adaptive feedback mechanism.

[0021] It should be noted that the method applies a composite electrical signal containing at least three different frequencies to the gastrointestinal region through a multi-frequency current excitation device, and synchronously collects the voltage response signals of the gastrointestinal surface electrode array. The multi-frequency current excitation device here refers to a device capable of generating and applying multiple frequency current signals to stimulate the electrical response of the gastrointestinal tissue, while the gastrointestinal surface electrode array refers to a plurality of electrodes distributed on the surface of the gastrointestinal or corresponding positions on the body surface for detecting the propagation of electrical signals in the tissue. In this way, the electrical properties of the gastrointestinal region under different frequencies can be obtained, providing basic data for subsequent conductivity distribution modeling.

[0022] Specifically, the frequency selection of the composite electrical signal can be optimized according to the electrical properties of the gastrointestinal tissue. For example, low-frequency signals (such as 1-10 kHz) can better reflect the bulk conductivity of the tissue, while high-frequency signals (such as above 100 kHz) are more sensitive to the boundary and cell membrane properties of the tissue. The collection of voltage response signals can be achieved through a high-precision electrode array, which can be arranged in a regular grid or optimized according to the gastrointestinal anatomy. The collected voltage data will be used as input for subsequent dynamic conductivity parameter matrix construction to reflect the electrical state of the gastrointestinal region at different time points. In addition, the sampling rate and signal processing method of the electrode array will directly affect the quality of the data and the accuracy of the subsequent imaging.

[0023] Preferably, when applying the composite electrical signal, a combination of signals with a frequency range of 1-200 kHz can be used, such as 1 kHz, 10 kHz, and 100 kHz, to cover the electrical property changes from the inside to the surface of the tissue. The collection frequency of the voltage response signal can be set to 10-50 times per second to ensure that dynamic changes such as gastrointestinal peristalsis can be captured. In addition, in order to improve the signal-to-noise ratio, filtering techniques can be introduced during the collection process to remove high-frequency noise and power frequency interference. As an alternative, a higher density electrode array can also be used, such as increasing the number of electrodes or optimizing the electrode spacing, to improve the spatial resolution and more accurately reflect the conductivity distribution of the gastrointestinal region.

[0024] In some embodiments, the dynamic modulation processing in step S4 includes: S41. Calculate adjacent time windows The difference between the maps of : where, is the three-dimensional conductivity distribution matrix of the current time window, is the Frobenius norm, and ∇ is the spatial gradient operator, represents the Hadamard product, , ​a normalization coefficient; S42. When the first threshold is exceeded , the high-frequency modulation mode is activated, and the weight of the frequency band higher than in the composite signal is increased to times of the original value (>1); S43. When the second threshold is exceeded , the low-frequency compensation mode is enabled, and a phase offset is applied to the frequency band lower than . .

[0025] It should be noted that the core of the dynamic modulation and demodulation processing in this method is to calculate the atlas difference degree of adjacent time windows and adjust the frequency band weight coefficient according to the difference degree. The atlas difference degree is calculated by combining the Frobenius norm and the spatial gradient operator, which is used to quantify the degree of change of the three-dimensional conductivity distribution atlas between different time windows. The Frobenius norm is a matrix norm used to measure the size of a matrix, while the spatial gradient operator is used to detect the spatial variation of the conductivity distribution. The Hadamard product is the point-by-point product of corresponding elements of matrices, which is used to consider the local changes of the conductivity distribution. In this way, the signal processing strategy can be dynamically adjusted to adapt to the complexity of the dynamic changes of the gastrointestinal tract.

[0026] Specifically, in the calculation formula of the atlas difference degree, the normalization coefficients and are used to balance the contributions of the Frobenius norm and the spatial gradient operator, and their values usually range from 0 to 1, and the specific values can be adjusted according to the actual application scenario and the characteristics of the conductivity distribution. For example, when the gastrointestinal peristalsis is more intense, the value of can be increased to enhance the sensitivity to spatial gradient changes. The first threshold and the second threshold are key parameters for determining whether to activate the high-frequency modulation mode or enable the low-frequency compensation mode, and are usually set according to experimental data or experience. For example, can be set to between 0.1 and 0.3, can be set to between 0.01 and 0.05. When the atlas difference degree exceeds , the high-frequency modulation mode is activated, and the weight of the high-frequency signal is increased to better capture the rapidly changing conductivity distribution; when the atlas difference degree is lower than , the low-frequency compensation mode is started, and a phase offset is applied to the low-frequency signal to enhance the stability of the signal.

[0027] Preferably, in the high-frequency modulation mode, the frequency threshold can be dynamically adjusted according to the electrical properties of the gastrointestinal tissue, for example, set between 50 kHz and 100 kHz. When the atlas difference exceeds , the high-frequency weight multiplication factor k can be set to 1.5 to 2.0 to significantly enhance the contribution of high-frequency signals. At the same time, in order to further optimize the modulation effect, adaptive filtering technology can be introduced to dynamically adjust the filter parameters according to the real-time signal characteristics. As an alternative, the phase shift φ in the low-frequency compensation mode can be obtained by fitting experimental data, for example, using a linear or nonlinear function to describe the relationship between φ and the atlas difference , so as to more accurately compensate for the phase change of low-frequency signals.

[0028] In some embodiments, the specific implementation of the high-frequency modulation mode in step S42 includes: S421. Define the update formula of the dynamic threshold wherein, is the initial threshold, is the time decay coefficient, is the current time window, is the initial time, is the weight factor of the maximum historical difference; S422. When the high-frequency weight is increased, the sampling interval of the electrode array is adjusted to satisfy the relationship: wherein, is the currently activated highest frequency, is the adjustment coefficient; S423. Apply band-pass filtering to the modulated high-frequency signal, with the passband range set to and the stopband attenuation greater than 40 dB.

[0029] It should be noted that the implementation of the high-frequency modulation mode in this method involves the update of the dynamic threshold, the adjustment of the sampling interval, and the application of band-pass filtering, which aims to more accurately capture the changes in high-frequency signals and optimize the imaging effect. The update formula of the dynamic threshold comprehensively considers the initial threshold , the time decay coefficient , and the weight factor of the maximum historical difference , and is used to dynamically adjust the threshold according to the current signal state. The sampling interval ​The adjustment of the sampling interval is to meet the sampling requirement of high-frequency signals and avoid aliasing. The setting of the band-pass filter is to filter out noise and retain the effective signal frequency band, thereby improving the imaging quality.

[0030] Specifically, the dynamic threshold is updated according to the following formula: where the initial threshold can be set according to experimental data or experience, for example, between 0.1 and 0.5. The time decay coefficient is used to control the decay rate of the threshold over time, and is usually in the range of 0.01 to 0.1. The weight factor of the maximum historical difference degree is used to adjust the influence of historical data on the current threshold, and is generally in the range of 0.1 to 0.5. In the adjustment of the sampling interval , the highest frequency can be determined according to the signal frequency range in actual application, for example, between 50 kHz and 200 kHz. The adjustment coefficient is used to fine-tune the sampling interval, and is in the range of 0.1 to 1. The passband range of the band-pass filter is set to , 1.5 , and the stopband attenuation is greater than 40 dB to ensure effective filtering of high-frequency noise and retention of key signal components.

[0031] Preferably, the update of the dynamic threshold can introduce an adaptive mechanism, such as dynamically adjusting the time decay coefficient according to the signal intensity of the current time window, so that it decreases when the signal changes dramatically and increases when the signal is stable. For the adjustment of the sampling interval , a signal quality evaluation mechanism can be further introduced, which automatically reduces the sampling interval to improve signal quality when the signal signal-to-noise ratio is below a certain threshold. In addition, the design of the band-pass filter can use an adaptive filtering algorithm, such as a Kalman filter, to dynamically adjust the filtering parameters according to real-time signal characteristics. As an alternative, a multi-stage filtering structure can be introduced, which first removes high-frequency noise through a low-pass filter and then extracts target frequency band signals through a band-pass filter, thereby further optimizing the processing effect of high-frequency signals.

[0032] In some embodiments, the non-linear interpolation algorithm in step S3 satisfies: for any grid node , its conductivity value is calculated by the following formula: where N is the number of adjacent measurement points, is the weight factor and , is the conductivity value of the i-th measurement point. is the Euclidean distance, is the anisotropic adjustment radius.

[0033] It should be noted that the implementation of the nonlinear interpolation algorithm in this method is to perform spatial discretization processing on the dynamic conductivity parameter matrix, thereby generating a three-dimensional conductivity distribution map containing multiple grid nodes. The algorithm calculates the conductivity value of each grid node, taking into account the conductivity values of adjacent measurement points and their distance relationship with the grid node. Among them, the weight factor is used to adjust the influence degree of adjacent measurement points on the conductivity value of the grid node, and the anisotropic adjustment radius is used to consider the anisotropic characteristics of the conductivity distribution, that is, the conductivity variation characteristics in different directions may be different. This method can effectively improve the accuracy and reliability of the three-dimensional conductivity distribution map, and better reflect the electrical characteristics of the gastrointestinal region.

[0034] Specifically, in the nonlinear interpolation algorithm, the number N of adjacent measurement points depends on the layout and density of the electrode array, and the value range is usually 3 to 10. The calculation of the weight factor needs to satisfy the normalization condition, that is, the sum of the weight factors of all adjacent measurement points is 1. The selection of the anisotropic adjustment radius is related to the conductivity gradient direction and local structure of the tissue, and the value range is usually 1 to 10 cm. In calculating the conductivity value of the grid node , the Euclidean distance is used to measure the spatial distance between the grid node and the adjacent measurement points, and the smaller the value, the greater the influence of the adjacent measurement points on the grid node. By reasonably setting these parameters, the relationship between the calculation complexity and the interpolation accuracy can be effectively balanced, thereby generating a high-precision three-dimensional conductivity distribution map.

[0035] Preferably, to further improve the interpolation accuracy, a local conductivity gradient direction weight adjustment mechanism can be introduced when calculating the weight factor , so that the influence of the adjacent measurement points consistent with the conductivity gradient direction on the grid node is greater. For example, the weight factor can be adjusted by introducing a directional weight factor , so that its weight in the conductivity gradient direction increases. In addition, the anisotropic adjustment radius can be dynamically adjusted according to the strength of the local conductivity gradient, and the greater the gradient, the smaller the adjustment radius, thereby better reflecting the detailed features of the local conductivity distribution. As an alternative, a machine learning-based interpolation algorithm such as neural network interpolation can be used to automatically learn the relationship between adjacent measurement points and grid nodes by training the model, thereby further improving the accuracy and adaptability of interpolation.

[0036] In some embodiments, the step S6 of determining the gastrointestinal abnormal region comprises: S61. Calculate the modulus mutation coefficient of the conductivity gradient vector: wherein, is the modulus mutation coefficient, t is the time variable, is the three-dimensional conductivity distribution matrix, is the regional average conductivity, is the anti-zero constant; S62. When exceeds the preset threshold, mark it as a potential abnormal region, and calculate its spatial distribution entropy: wherein, H is the spatial distribution entropy, is the conductivity proportion of the i-th sub-region; S63. Determine the abnormal activity level according to the product of the entropy value H and the time change rate When the product exceeds the critical value, generate an alarm signal.

[0037] It should be noted that the present method determines the gastrointestinal abnormal region by calculating the modulus mutation coefficient of the conductivity gradient vector and the spatial distribution entropy. The modulus mutation coefficient of the conductivity gradient vector is used to quantify the sharp change of conductivity in space, reflecting the non-uniformity of the electrical properties of the tissue, while the spatial distribution entropy is used to measure the complexity and non-uniformity of the conductivity distribution of the abnormal region. By considering these two parameters comprehensively, potential gastrointestinal abnormal regions can be effectively identified, and their activity state can be evaluated. This method can further provide quantitative analysis of gastrointestinal dysfunction based on three-dimensional dynamic imaging results, providing strong support for clinical diagnosis.

[0038] Specifically, in the calculation formula of the modulus mutation coefficient, represents the modulus of the conductivity gradient vector, reflecting the rate of change of conductivity in space; is the regional average conductivity, used for normalization processing; is the anti-zero constant, to avoid numerical instability in the calculation process. In the calculation of the spatial distribution entropy H, is the conductivity proportion of the i-th sub-region, reflecting the distribution of conductivity in different sub-regions. In practical applications, the threshold value of the modulus mutation coefficient can be set according to experimental data or clinical experience, for example, between 1.5 and 3. The calculation of the spatial distribution entropy requires the gastrointestinal region to be divided into several sub-regions, and the conductivity proportion of each sub-region is calculated by integral or discrete summation. By reasonably setting these parameters, normal and abnormal tissues can be effectively distinguished, thereby achieving precise positioning of gastrointestinal abnormal regions.

[0039] Preferably, to improve the accuracy of abnormal area judgment, the time dimension can be considered when calculating the modulus mutation coefficient, such as calculating its rate of change over time, so as to better reflect the dynamic changes of the abnormal area. At the same time, the calculation of spatial distribution entropy can combine the direction information of local conductivity gradient to further refine the division of sub-regions and improve the accuracy of entropy value calculation. As an alternative, machine learning algorithms such as support vector machine (SVM) or deep learning models can be introduced to extract features and classify the modulus mutation coefficient and spatial distribution entropy of the conductivity gradient vector, thereby realizing automatic identification and classification of gastrointestinal abnormal areas. In addition, other physiological parameters (such as gastrointestinal peristalsis speed) can also be combined to further optimize the abnormal area judgment logic and improve the reliability of diagnosis.

[0040] In some embodiments, the adaptive feedback mechanism in step S7 includes: S71. Define the model parameter update amount and the exponential relationship between the abnormal area A: wherein, is the parameter update amount, A is the abnormal area, is the reference area, is the baseline update amount, is the difference degree influence factor; S72. Apply constraints to the boundary condition parameters: wherein, is the previous time parameter value, is the updated parameter value, , is the preset boundary; S73. When the change rate of is less than 1% for three consecutive iterations, trigger the parameter freezing mechanism and switch to the static model.

[0041] It should be noted that the adaptive feedback mechanism in this method is to update the boundary condition parameters of the initial conductivity distribution model based on the judgment results of the abnormal area. The core of this mechanism is to dynamically adjust the model parameters so that the imaging system can better adapt to the dynamic changes of the gastrointestinal tissue, thereby improving the imaging accuracy and the reliability of abnormal detection. The parameter update amount The exponential relationship between and the abnormal area A reflects the influence of the size of the abnormal area on the adjustment of the model parameters, and the constraint conditions of the boundary condition parameters ensure the stability and reasonableness of the model update. Through this feedback mechanism, the imaging system can continuously optimize its performance in continuous monitoring.

[0042] Specifically, the parameter update amount is calculated by the formula: wherein is the baseline update amount, reflecting the basic adjustment range of the model parameters in the absence of abnormal areas; A is the area of the abnormal area, used to quantify the size of the abnormal area; is the difference influence factor, used to adjust the influence of the abnormal area size on the parameter update. In practical applications, the baseline update amount can be set according to experimental data or experience, for example, between 0.01 and 0.1; the reference area A0 is used for normalization processing, and is usually a certain proportion of the area of the normal gastrointestinal region; the difference influence factor is adjusted according to the significance of the abnormal area, and the value range is generally 0.1 to 1. In the constraint conditions of the boundary condition parameters, and are preset boundary values, used to limit the range of parameter update and ensure the stability of the model. By reasonably setting these parameters, dynamic adjustment of the model parameters can be achieved, thereby improving the adaptability and accuracy of the imaging system.

[0043] Preferably, in order to further optimize the adaptive feedback mechanism, a time decay factor can be introduced in the parameter update process, so that the adjustment of the model parameters gradually tends to be stable over time, avoiding the drastic fluctuations of the model parameters caused by the temporary changes of the abnormal area. For example, the time decay factor can be set to 0.05 to 0.2, and dynamically adjusted according to the length of continuous monitoring. In addition, when the change rate of is less than 1% in three consecutive iterations, the parameter freezing mechanism is triggered and switched to a static model, which can effectively reduce the consumption of computing resources while maintaining the stability and reliability of the imaging system. As an alternative, a data-driven method such as a machine learning algorithm can be used to automatically learn the relationship between the abnormal area and the model parameters based on historical monitoring data, thereby achieving more accurate adaptive adjustment.

[0044] In some embodiments, the low-frequency compensation mode in step S43 further includes: S431. Establishing a correlation function between the phase shift amount and the conductivity change rate: wherein is the phase shift amount, erf is the error function, t is the time variable, k is the proportionality constant, is the base conductivity; S432. Performing harmonic analysis on the phase-adjusted signal to filter out frequency components higher than the third harmonic; S433. In the compensation mode, the excitation current amplitude of the electrode array is reduced to of the original value.

[0045] It should be noted that the low-frequency compensation mode in the method is to optimize the signal processing process by establishing a correlation function between the phase shift and the rate of change of conductivity. The core of this mode is to compensate for the phase delay of low-frequency signals through phase adjustment, thereby improving the integrity of the signal and the accuracy of the imaging. The relationship between the phase shift and the rate of change of conductivity is described by the error function (erf), and this nonlinear relationship can better adapt to the propagation characteristics of low-frequency signals in gastrointestinal tissues. Harmonic analysis is used to filter out frequency components higher than the third harmonic to reduce the interference of high-frequency noise and further optimize signal quality. At the same time, the adjustment of the excitation current amplitude can dynamically change according to the needs of the compensation mode to improve the stability of the signal and the adaptability of the imaging system.

[0046] Specifically, in the calculation formula of the phase shift, the proportional constant k is used to adjust the sensitivity between the phase shift and the rate of change of conductivity, and its value range is usually 0.1 to 1. The basic conductivity is the conductivity reference value of the gastrointestinal tissue in the normal state, which is used for normalization processing. In harmonic analysis, the signal is decomposed into different frequency components through Fourier transform, and frequency components higher than the third harmonic are filtered out to retain the main characteristics of the signal. The adjustment coefficient (1- ) of the excitation current amplitude can be dynamically set according to the needs of the compensation mode, where The value range of is generally 0 to 0.5. By reasonably setting these parameters, the phase delay of low-frequency signals can be effectively compensated, and the stability of the signal and the imaging quality can be improved.

[0047] Preferably, to further optimize the low-frequency compensation mode, an adaptive mechanism can be introduced in the calculation of the phase shift, such as dynamically adjusting the proportional constant k according to the real-time monitored rate of change of conductivity, so that it increases when the signal changes dramatically and decreases when the signal is stable. In addition, harmonic analysis can be combined with multi-resolution analysis methods such as wavelet transform to more accurately identify and filter out high-frequency noise components. As an alternative, the adjustment of the excitation current amplitude can use a piecewise linear function or a nonlinear function instead of a simple linear adjustment, thereby better adapting to different compensation needs. For example, when the intensity of low-frequency signals is weak, the compensation proportion of the current amplitude can be appropriately increased to improve the signal-to-noise ratio of the signal.

[0048] In some embodiments, the determination method of the anisotropy adjustment radius includes: S81. Dividing the reference radius of different partitions according to the gastrointestinal anatomy ; S82. Adjusting the actual radius based on the local conductivity gradient direction: wherein, is the adjusted anisotropic adjustment radius, is the reference radius, is the angle between the conductivity gradient direction and the tissue fiber orientation; S83. Apply a radius decay factor to the boundary region: wherein d is the distance to the nearest anatomical boundary, is the characteristic decay length.

[0049] It should be noted that the determination of the anisotropic adjustment radius in this method is based on the gastrointestinal anatomy and the local conductivity gradient direction to optimize the accuracy of the three-dimensional conductivity distribution map. The anisotropic adjustment radius refers to the weight radius adjusted in different directions according to the change of the conductivity gradient, which is used to better reflect the anisotropic characteristics of the tissue. The reference radius is the initial radius divided according to the gastrointestinal anatomy, which is used to determine the basic value of the anisotropic adjustment radius. By considering the angle between the conductivity gradient direction and the tissue fiber orientation, as well as the decay characteristics of the boundary region, the anisotropic adjustment radius can be further adjusted to improve the accuracy and adaptability of the imaging.

[0050] Specifically, the selection of the reference radius can be divided according to the characteristics of the gastrointestinal anatomy, for example, different reference radius values can be set for different regions of the stomach and intestines, usually between 1 to 5 centimeters. The angle refers to the angle between the conductivity gradient direction and the tissue fiber orientation, with a value range of 0 to π / 2 radians. When the angle is 0, it means that the conductivity gradient direction is consistent with the tissue fiber orientation, and the anisotropic adjustment radius reaches the maximum value; when the angle is π / 2, the adjustment radius is the smallest. The characteristic decay length is used to describe the decay characteristics of the boundary region, with a value range of generally 0.5 to 2 centimeters. By reasonably setting these parameters, dynamic adjustment of the anisotropic adjustment radius can be achieved, thereby better adapting to the complex structure of the gastrointestinal tissue.

[0051] Preferably, to further optimize the calculation of the anisotropy adjustment radius, an adaptive adjustment mechanism based on the tissue conductivity distribution can be introduced in the selection of the reference radius. For example, for regions with large conductivity variations, the reference radius can be appropriately reduced to increase the spatial resolution; for regions with small conductivity variations, the reference radius can be increased to reduce the computational complexity. In addition, for the decay factor of the boundary region, a nonlinear decay function such as a Gaussian function or an exponential function can be introduced to more accurately simulate the boundary effect. As an alternative, a machine learning-based method can be used to automatically learn the relationship between the conductivity gradient direction and the tissue fiber orientation through training the model, thereby dynamically adjusting the anisotropy adjustment radius to further improve the accuracy and adaptability of the imaging.

[0052] In some embodiments, the alarm signal generation logic in step S63 includes: S91. Define the comprehensive anomaly index: wherein, is the comprehensive anomaly index, is the modulus mutation coefficient, H is the spatial distribution entropy, is the weighting coefficient and ; S92. When continuously exceeds the threshold value T for time, the following rules are classified: If : generate a first-level alarm; If : generate a second-level alarm and start data storage; If : generate a third-level alarm and trigger real-time transmission; wherein T is a preset threshold value, is the duration of continuous threshold exceedance; S93. All alarm events are recorded in the time-frequency feature matrix for subsequent analysis.

[0053] It should be noted that the alarm signal generation logic in this method is achieved by calculating the comprehensive anomaly index and judging the threshold value to grade the early warning of gastrointestinal abnormal activity. The comprehensive anomaly index is a quantitative index for comprehensive evaluation of the contribution of the conductivity gradient mutation coefficient, the spatial distribution entropy and the conductivity change rate to abnormal activity. By defining different alarm levels and corresponding threshold values, real-time monitoring and classification warning of gastrointestinal abnormalities can be achieved, thereby providing timely feedback for clinical diagnosis. This method can effectively distinguish between normal and abnormal states in a complex and dynamic gastrointestinal imaging environment, and take different alarm measures according to the degree of abnormality.

[0054] Specifically, in the calculation formula of the comprehensive anomaly index, the modulus mutation coefficient, the spatial distribution entropy, and the conductivity change rate respectively reflect the severity of the conductivity change, the complexity of the conductivity distribution, and the dynamic change rate of the abnormal region. The weighting coefficient and is used to balance the contributions of the three parameters, usually ranging from 0 to 1, and satisfying . The preset threshold T is used to determine whether the comprehensive anomaly index exceeds the normal range, and its value can be set according to experimental data or clinical experience, for example, between 1 and 3. The duration is used to determine whether the abnormal state lasts long enough to trigger an alarm, usually ranging from a few seconds to tens of seconds. By reasonably setting these parameters, precise grading and early warning of gastrointestinal abnormal activity can be achieved.

[0055] Preferably, to further optimize the alarm signal generation logic, a time weighting factor can be introduced in the calculation of the comprehensive anomaly index to consider the persistence and dynamic changes of the abnormal state. For example, for long-lasting abnormal states, their weight in the comprehensive anomaly index can be increased. In addition, the classification of alarm levels can be further refined according to clinical needs, such as adding more intermediate levels to provide more detailed warning information. As an alternative, a machine learning-based method can be used to automatically learn the characteristics of abnormal activity by training a model, thereby dynamically adjusting the alarm threshold and generation logic. This method can automatically optimize the performance of the alarm system according to actual monitoring data, improving its adaptability and accuracy.

[0056] In some embodiments, the back-projection algorithm in step S5 employs an improved Radon transform, specifically including: S101. Define the integral path in the projection domain as a dynamic curve: where, is the projection integral value, is the projection angle, t is the time variable, and (x, y, z) is the three-dimensional space coordinate, is the modulated conductivity distribution data, is the Dirac function, s is the ray parameter, is the differential along the ray path; S102. Introduce a motion compensation factor : where, is the motion compensation factor, is the gastrointestinal peristalsis speed coefficient, is the rate of change of conductivity over time; S103. fusing the projection integral values based on the formula with the motion compensation factor reconstructing a three-dimensional image, as shown in the following formula: wherein, is the three-dimensional conductivity distribution matrix at the nth iteration, is the update matrix for the (n+1)th iteration, R is the Radon transform operator, is the step factor and 0 < a < 1, n is the iteration number and n ≥ 0.

[0057] It should be noted that the inverse projection algorithm in this method uses an improved Radon transform to map the modulated conductivity distribution data into the dynamic imaging coordinate system, thereby generating a three-dimensional dynamic imaging result with time sequence markers. The improved Radon transform defines a dynamic curve as the integral path in the projection domain and introduces a motion compensation factor to adapt to dynamic changes such as gastrointestinal peristalsis, thereby improving the accuracy and stability of image reconstruction. This method can effectively reduce motion artifacts in gastrointestinal dynamic imaging and improve imaging quality, providing more reliable image information for clinical diagnosis.

[0058] Specifically, Radon transform is a mathematical tool for projecting a two-dimensional image into one-dimensional space, widely used in medical imaging. In this method, the projection integral value is calculated by integrating the modulated conductivity distribution data on the dynamic curve, where the projection angle and the time variable t are used to describe the changes of the dynamic curve. The introduction of the motion compensation factor is to compensate for the conductivity changes caused by gastrointestinal peristalsis. In its calculation formula, the gastrointestinal peristalsis velocity coefficient is used to describe the motion rate of the gastrointestinal tissue, while the rate of change of conductivity with time reflects the dynamic changes of conductivity. The step factor is used to control the convergence speed of the iteration process, and its value range is usually between 0 and 1. By reasonably setting these parameters, accurate reconstruction of dynamic conductivity distribution data can be achieved, generating high-quality three-dimensional dynamic imaging results.

[0059] Preferably, to further optimize the performance of the inverse projection algorithm, more physiological parameters can be introduced in the definition of the dynamic curve, such as the thickness change of the gastrointestinal wall or the direction of peristalsis, to more accurately describe the dynamic characteristics of the gastrointestinal tissue. In addition, the calculation of the motion compensation factor can be combined with real-time monitoring of gastrointestinal motion data, and the gastrointestinal peristalsis velocity coefficient , thereby improving the compensation accuracy. As an alternative, a multi-scale Radon transform can be employed to project and reconstruct the conductivity distribution data at different scales to improve the detail resolution and overall quality of the imaging. This method can better capture the subtle changes in the gastrointestinal tissue in complex dynamic environments, enhancing the adaptability and reliability of the imaging system.

[0060] The above-mentioned various embodiments of the present application have the following beneficial effects: The present application can achieve high-precision three-dimensional imaging of the gastrointestinal region by multi-frequency current excitation and construction of a dynamic conductivity parameter matrix. The use of nonlinear interpolation algorithms and frequency domain feature separation techniques can effectively improve the spatial resolution and dynamic capture capability of the imaging. The dynamic modulation and demodulation processing can adjust the frequency band weight coefficients according to the atlas difference degree, further optimizing the imaging effect. In addition, the introduction of the inverse projection algorithm can accurately map the modulated conductivity distribution data to the dynamic imaging coordinate system, generating a three-dimensional dynamic imaging result with time sequence labels, providing strong support for real-time monitoring of gastrointestinal function. Through the analysis of the conductivity gradient mutation region, the position and activity state of the abnormal gastrointestinal region can be quickly identified, and the boundary condition parameters of the conductivity distribution model are dynamically updated in combination with the adaptive feedback mechanism, thereby improving the accuracy and reliability of abnormal detection. The introduction of low-frequency compensation mode and high-frequency modulation mode can further optimize the signal processing process, improving the adaptability and stability of the imaging system. The dynamic adjustment of the anisotropy adjustment radius can better adapt to the gastrointestinal anatomical structure, further improving the imaging accuracy. The introduction of improved Radon transform and motion compensation factor can effectively reduce the influence of gastrointestinal peristalsis on imaging, improving the quality and efficiency of image reconstruction.

[0061] In addition, the present application can also realize precise positioning and graded early warning of the gastrointestinal abnormal region. By calculating the modulus mutation coefficient and spatial distribution entropy of the conductivity gradient vector, potential abnormal regions can be effectively identified, and the abnormal activity level can be judged according to the product of the entropy value and the time change rate, thereby generating alarm signals of different levels. The time-frequency feature matrix record of the alarm event can provide important data support for subsequent analysis. This method not only can monitor the gastrointestinal function state in real time, but also can early warn potential gastrointestinal diseases, providing important reference for clinical diagnosis and treatment, and has important medical application value.

[0062] Furthermore, the storage medium in the embodiments of this application stores program instructions capable of implementing all the above methods. These program instructions can be stored in the storage medium in the form of a software product, including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks, or terminal devices such as computers, servers, mobile phones, and tablets.

[0063] The above description is merely an explanation of some preferred embodiments of the present invention and the technical principles employed. Those skilled in the art should understand that the scope of the invention as described in the embodiments of the present invention is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features with similar functions disclosed in the embodiments of the present invention.

Claims

1. A method of analysis for three-dimensional dynamic imaging of the gastrointestinal tract based on modeling of conductivity distribution, characterized in that, The method comprises the following steps: S1. Apply a composite electrical signal containing at least three different frequencies to the gastrointestinal region through a multi-frequency current excitation device, and synchronously collect the voltage response signals of the gastrointestinal surface electrode array; S2. Construct a dynamic conductivity parameter matrix based on frequency domain feature separation according to the voltage response signals and a preset initial conductivity distribution model; S3. Perform spatial discretization processing on the dynamic conductivity parameter matrix using a nonlinear interpolation algorithm to generate a three-dimensional conductivity distribution map containing multiple layers of grid nodes; S4. Perform dynamic modulation and demodulation processing on the three-dimensional conductivity distribution map, including judging the modulation type according to the difference degree of the map of adjacent time windows, adjusting the frequency band weight coefficient based on the difference degree threshold, and generating modulated conductivity distribution data; S5. Map the modulated conductivity distribution data to a dynamic imaging coordinate system through an inverse projection algorithm to generate a three-dimensional dynamic imaging result with time sequence labels; S6. Determine the position and activity state of the gastrointestinal abnormal region according to the spatial coordinates and change rate of the conductivity gradient mutation region in the three-dimensional dynamic imaging result; S7. Update the boundary condition parameters of the initial conductivity distribution model using an adaptive feedback mechanism based on the determination result of the abnormal region.

2. The method of claim 1, wherein, The dynamic modulation and demodulation processing in step S4 comprises: S41. Calculate adjacent time windows With the atlas difference degree : wherein is the three-dimensional conductivity distribution matrix of the current time window, is the Frobenius norm, ∇ is the spatial gradient operator, denotes the Hadamard product, , is a normalization coefficient; S42. When the first threshold is exceeded , a high-frequency modulation mode is activated, and the frequency band weight of the frequency band higher than in the composite electrical signal is raised to times the original value; S43. When the second threshold is exceeded , a low frequency compensation mode is enabled, applying a phase offset to frequency bands with frequencies below .​​ 3. The method of claim 2, wherein, The update formula of the dynamic threshold value in the step S42 is: wherein, is an initial threshold value, is a time decay coefficient, is a current time window, is an initial time, is a weight factor of the maximum historical difference degree, is a historical atlas difference degree; Synchronously adjusting the sampling interval of the electrode array when boosting the high frequency weight , the sampling interval satisfies the relationship wherein, is the currently activated maximum frequency, is a tuning factor.

4. The method of claim 1, wherein, The nonlinear interpolation algorithm in step S3 satisfies: For any grid node whose conductivity value is calculated from the formula: where N is the number of neighboring measurement points, is a weight factor and , is the conductivity value of the i-th measurement point, is the Euclidean distance, is the anisotropy adjustment radius.

5. The method of claim 1, wherein, The determination of the gastrointestinal abnormal region in step S6 comprises: S61. Calculate the modulus mutation coefficient of the conductivity gradient vector: wherein is the modulus jump coefficient, t is the time variable, is the three-dimensional conductivity distribution matrix, is the area-averaged conductivity, is the prevention of zero constants; S62. When When the value exceeds a preset threshold, mark it as a potential abnormal region and calculate its spatial distribution entropy: H is a spatial distribution entropy, is the conductivity proportion of the i-th sub-region; S63. The abnormal activity level is judged according to the product of the entropy value H and the time change rate and an alarm signal is generated when the product exceeds a threshold value.

6. The method of claim 1, wherein, The adaptive feedback mechanism in step S7 comprises: S71. Define model parameter update amount Exponential relationship with abnormal area A: wherein, is a parameter update amount, A is an abnormal area, and is a reference area, is a reference update amount, is a difference degree influence factor; S72. Apply a constraint condition to the boundary condition parameters: wherein, is the parameter value at the previous time instant, is the updated parameter value, , is a preset boundary; S73. When the rate of change of the parameter is less than 1% for three consecutive iterations, trigger the parameter freeze mechanism and switch to the static model. S73. When the rate of change of the parameter is less than 1% for three consecutive iterations, trigger the parameter freeze mechanism and switch to the static model.

7. The method of claim 2, wherein, The low-frequency compensation mode in step S43 comprises: S431. Establishing the phase shift amount Correlation function with the rate of change of conductivity: wherein, is the phase offset, erf is the error function, t is the time variable, k is a proportionality constant, is the base conductivity; S432. Perform harmonic analysis on the phase-adjusted signal to filter out frequency components higher than the third harmonic; S433. In the compensation mode, the amplitude of the stimulation current of the electrode array is reduced to a fraction of the original value of the original value.

8. The method of claim 4, wherein, The anisotropic adjustment radius The determination method comprises: S81. The reference radius of different partitions according to the gastrointestinal anatomy ; S82. Adjust the actual radius based on the local conductivity gradient direction: wherein, is the adjusted anisotropy adjustment radius, is the reference radius, is the angle between the conductivity gradient direction and the tissue fiber orientation; S83. Apply a radius attenuation factor to the boundary region: where d is the distance to the nearest anatomical boundary, is the characteristic attenuation length.

9. The method of claim 5, wherein, The generation logic of the alarm signal in step S63 comprises: S91. Define a comprehensive abnormality index: wherein, is a comprehensive anomaly index, is a modulus mutation coefficient, H is a spatial distribution entropy, is a weighting coefficient and ; S92. When the threshold T is exceeded for more than time, classify according to the following rules: If : generate a primary alert; If : generate secondary alert and initiate data storage; If : generate a level three alert and trigger live transmission; wherein T is a preset threshold value, is a duration of time continuously exceeding the threshold value; S93. All alarm events are logged with a time-frequency feature matrix for subsequent analysis.

10. The method of claim 1, wherein, The inverse projection algorithm in step S5 uses an improved Radon transform, specifically comprising: S101. Define the integral path in the projection domain as a dynamic curve: wherein is the projection integral value, is the projection angle, t is the time variable, (x, y, z) is the three-dimensional spatial coordinate, is the modulated conductivity distribution data, is the Dirac function, s is the ray parameter, is the differential along the ray path; S102. Introducing a motion compensation factor : wherein is a motion compensation factor, is a coefficient of gastrointestinal peristalsis speed, is a rate of change of electrical conductivity with time; S103. fusing the projection integral values based on the formula with the motion compensation factor reconstructing the three-dimensional image as shown in the following formula: wherein is a three-dimensional conductivity distribution matrix at the n-th iteration, is an update matrix for the n+1-th iteration, R is a Radon transform operator, is a step factor and 0 < a < 1, n is the iteration number and n > 0.

Images