Breast cancer tissue anomaly detection method based on multi-frequency electrical impedance dynamic analysis

Abstract

The invention provides a breast cancer tissue anomaly detection method based on multi-frequency electrical impedance dynamic analysis, and the method comprises the steps: generating a multi-frequency electrical excitation signal covering a preset low frequency to a high frequency, applying the multi-frequency electrical excitation signal to a to-be-detected breast tissue through an electrode array, and synchronously collecting a response voltage signal. Orthogonal demodulation is carried out on the collected voltage signals, real and imaginary components of each frequency point are separated, a complex impedance value is calculated, and a multi-frequency impedance spectrum is generated. Feature vectors including amplitude-frequency characteristic slope, phase-frequency characteristic extreme point position, high-frequency-band impedance attenuation rate and the like are extracted from the impedance spectrum, a pre-trained dynamic judgment model is input to calculate a tissue anomaly index, and finally whether the mammary gland tissue is abnormal or not is judged according to a comparison result of the anomaly index and a preset threshold value. The accuracy of early detection of breast cancer can be improved, and the stability and reliability of detection are enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of medical detection and diagnosis, and more specifically, the present invention relates to a method for detecting abnormal breast cancer tissues based on multi-frequency electrical impedance dynamic analysis. Background Art

[0002] In the field of medical detection, early diagnosis of breast cancer is of great significance for improving the cure rate and patient prognosis. Currently, the detection methods for breast cancer mainly include imaging examinations (such as mammography, ultrasound, and magnetic resonance imaging) and histopathological examinations. However, these methods have certain limitations. For example, mammography may not be sensitive enough for young women or patients with high breast density, while although ultrasound examination is radiation-free, it has a strong dependence on the operator's technology. Magnetic resonance imaging has high sensitivity, but it is costly and has a long examination time. In addition, most of these methods are invasive or semi-invasive detections, which may cause discomfort to patients or have certain risks.

[0003] In recent years, bioelectrical impedance technology, as a non-invasive detection method, has gradually received attention. Bioelectrical impedance technology is based on the electrical properties of biological tissues and measures the impedance of tissues to reflect their physiological or pathological states. There are differences in the electrical properties between normal breast tissues and diseased tissues. Therefore, in theory, early screening of breast cancer can be achieved through impedance measurement. However, most of the existing impedance detection methods use single-frequency or limited-frequency band excitation signals, which are difficult to comprehensively reflect the complex electrical properties of breast tissues, resulting in insufficient sensitivity and specificity of detection. In addition, the existing technologies lack dynamic monitoring and real-time feedback calibration mechanisms and cannot effectively cope with problems such as poor electrode contact and environmental interference that may occur during the detection process.

[0004] In the process of implementing the embodiments of the present invention, the inventors found that there are at least the following problems or defects in the prior art: The impedance detection method with single frequency or limited frequency band cannot accurately distinguish the subtle electrical differences between normal tissues and diseased tissues, resulting in insufficient detection accuracy; the lack of dynamic monitoring and feedback calibration mechanisms makes the detection results vulnerable to environmental factors and equipment stability, and it is difficult to ensure the reliability and repeatability of detection. Summary of the Invention

[0005] The present invention provides a method for detecting abnormal breast cancer tissues based on multi-frequency electrical impedance dynamic analysis, including: S1. Generating a multi-frequency electrical excitation signal including a plurality of discrete frequency points, and the frequency range of the multi-frequency electrical excitation signal covers a preset low-frequency threshold to a preset high-frequency threshold; S2. Applying the multi-frequency electrical excitation signal generated in step S1 to the breast tissue to be measured through an electrode array, and synchronously collecting the response voltage signal of the electrode array; S3. Orthogonally demodulate the response voltage signal collected in step S2 to separate the real and imaginary components at each frequency point; S4. Based on the real and imaginary components obtained in step S3, calculate the complex impedance value at each frequency point to generate a multi-frequency impedance spectrum; S5. Extract a feature vector from the multi-frequency impedance spectrum generated in step S4, which at least includes the slope of the amplitude-frequency characteristic, the position of the extreme point of the phase-frequency characteristic, and the impedance attenuation rate in the high-frequency band; S6. Input the feature vector extracted in step S5 into a pre-trained dynamic decision model, and calculate the tissue abnormality index through the iterative weighting algorithm in the dynamic decision model; S7. Determine whether there is an abnormality in the breast tissue according to the comparison result between the tissue abnormality index obtained in step S6 and a preset threshold.

[0006] Further, step S1 includes: S11. Set a preset low-frequency threshold and a preset step value , and generate a frequency sequence and based on the , , where is the preset high-frequency threshold; S12. Modulate the frequency sequence in a time-division multiplexing manner to generate a time-domain superposition signal , where is the amplitude of the nth frequency, is the preset random phase, N is the preset number of discrete points and N≥5, ; S13. Perform pre-emphasis filtering on to compensate for high-frequency signal attenuation.

[0007] Further, the amplitude in step S12 satisfies: , where is the preset maximum amplitude and satisfies the safety current limit, is the preset attenuation coefficient, is the preset low-frequency threshold, is the preset step value.

[0008] Further, step S3 includes: S31. Orthogonally demodulate the response voltage signal collected in step S2 and the reference current signal generated in step S1 to obtain the in-phase signal and the quadrature signal , where t is the time variable, is the frequency value at the nth frequency point; S32. For and perform low-pass filtering respectively to extract the baseband components at each frequency point and , where n is the frequency point serial number and n = 0, 1, …, N; S33. Based on and calculate the complex impedance at the nth frequency point , where is the excitation voltage amplitude at the nth frequency in step S1, and j is the imaginary unit.

[0009] Furthermore, step S4 includes: S41. Convert the complex impedance at each frequency point to the frequency-domain admittance ; S42. Construct an equivalent circuit model , where is the conductance component, is the capacitance component, ; S43. Optimize the frequency-varying curve of the model parameter through the non-linear least squares fitting algorithm.

[0010] Furthermore, the eigenvector in step S5 includes: Amplitude-frequency characteristic slope , where , is the adjacent frequency point interval; Phase-frequency characteristic extreme point frequency , where is the frequency corresponding to the maximum value in the phase spectrum; High-frequency attenuation factor , where is the preset high-frequency threshold, is the preset reference frequency.

[0011] Furthermore, the dynamic decision model in step S6 performs: Calculate the anomaly index where is the calibration value of the amplitude-frequency characteristic slope of healthy tissue, is the calibration value of the phase-frequency extreme point frequency of healthy tissue, is the calibration value of the high-frequency attenuation factor of healthy tissue, , , is a preset weighting coefficient and satisfies .

[0012] Furthermore, it also includes a dynamic analysis step: S81. Repeat steps S1 - S6 in a continuous time series to obtain an abnormal index sequence ; S82. Calculate the dynamic factor , where is the time stamp of the k - th detection, k = 1, 2, …, n; S83. When , trigger a re - detection of the high - frequency band impedance, where the high - frequency band is a band greater than the preset re - detection frequency.

[0013] Furthermore, it also includes a feedback calibration step: S91. When the electrode contact impedance exceeds the preset contact impedance threshold , automatically switch the excitation electrode pair and recalculate the compensation matrix; S92. When the environmental temperature change exceeds the preset temperature change threshold, update the eigenvector threshold in step S5 according to the pre - stored temperature - impedance correction curve.

[0014] Furthermore, step S1 also includes an adaptive frequency adjustment: When the high - frequency attenuation factor extracted in step S5 is less than the preset attenuation threshold, expand the preset high - frequency threshold to an extended frequency threshold, and reduce the frequency step value to a preset reduced step value to generate a new multi - frequency electrical excitation signal.

[0015] According to the above - mentioned embodiments of the present invention, it has at least the following beneficial effects: First of all, through the multi - frequency electrical impedance dynamic analysis technology, the present invention can generate electrical excitation signals covering a wide frequency band and accurately measure the response voltage signals, so as to comprehensively capture the changes in the electrical characteristics of breast tissue at different frequencies. Combining with the generation of orthogonal demodulation and complex impedance spectra, it can effectively separate the real and imaginary components and extract eigenvectors, including the slope of the amplitude - frequency characteristic, the position of the extreme point of the phase - frequency characteristic, and the impedance attenuation rate in the high - frequency band, etc. These eigenvectors can more accurately reflect the physiological and pathological states of the tissue, thereby improving the sensitivity and specificity of abnormal detection of breast cancer tissue.

[0016] In addition, the present invention introduces a dynamic decision-making model and a feedback calibration mechanism, which can calculate the tissue abnormality index in real time and dynamically monitor the changes during the detection process. When detecting abnormal changes, poor electrode contact, environmental temperature changes and other problems, it can automatically trigger the re-detection or calibration process to ensure the stability and reliability of the detection results. This design of dynamic monitoring and feedback calibration can not only improve the repeatability and accuracy of the detection, but also reduce the misdiagnosis rate, providing an efficient and reliable non-invasive detection method for the early screening and diagnosis of breast cancer. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the drawings, several embodiments of the present invention are shown by way of example and not limitation, wherein: Figure 1 FIG. is a schematic flowchart of a method for detecting breast cancer tissue abnormality based on multi-frequency electrical impedance dynamic analysis provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0018] The principles and spirit of the present invention 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 then implement the present invention, and are not intended to limit the scope of the present invention in any way. On the contrary, these embodiments are provided to make the present invention more thorough and complete, and to convey the scope of the present invention fully to those skilled in the art.

[0019] Those skilled in the art know that the embodiments of the present invention can be implemented as a system, a device, an equipment, a method or a computer program product. Therefore, the present invention can be specifically implemented in the following forms, namely: completely hardware, completely software (including firmware, resident software, microcode, etc.), or a combination of hardware and software.

[0020] It should be noted that any number of elements in the drawings is for illustration rather than limitation, and any naming is only for distinction and does not have any limiting meaning.

[0021] The following reference is made to Figure 1 , Figure 1 FIG. is a schematic flowchart of a method for detecting breast cancer tissue abnormality based on multi-frequency electrical impedance dynamic analysis provided by an embodiment of the present invention. As Figure 1 shown, a method 100 for detecting breast cancer tissue abnormality based on multi-frequency electrical impedance dynamic analysis includes: S1. Generate a multi-frequency electrical excitation signal including a plurality of discrete frequency points, and the frequency range of the multi-frequency electrical excitation signal covers a preset low-frequency threshold to a preset high-frequency threshold; S2. Apply the multi-frequency electrical excitation signal generated in step S1 to the breast tissue to be measured through the electrode array, and synchronously collect the response voltage signal of the electrode array; S3. Perform quadrature demodulation on the response voltage signal collected in step S2 to separate the real and imaginary components at each frequency point; S4. Based on the real and imaginary components obtained in step S3, calculate the complex impedance value at each frequency point to generate a multi-frequency impedance spectrum; S5. Extract a feature vector from the multi-frequency impedance spectrum generated in step S4, which at least includes the slope of the amplitude-frequency characteristic, the position of the extreme point of the phase-frequency characteristic, and the impedance attenuation rate in the high-frequency band; S6. Input the feature vector extracted in step S5 into a pre-trained dynamic decision model, and calculate the tissue abnormality index through the iterative weighting algorithm in the dynamic decision model; S7. Determine whether there is an abnormality in the breast tissue according to the comparison result between the tissue abnormality index obtained in step S6 and the preset threshold.

[0022] It should be noted that when implementing the abnormal detection method for breast cancer tissue based on multi-frequency electrical impedance dynamic analysis, it is first necessary to generate a multi-frequency electrical excitation signal containing multiple discrete frequency points, and its frequency range covers the preset low-frequency threshold to the preset high-frequency threshold. Here, the multi-frequency electrical excitation signal refers to an electrical signal containing multiple different frequency components, which is used to stimulate the breast tissue to be measured to generate a response. By covering a wide frequency band of signals, the electrical property changes of the tissue at different frequencies can be comprehensively captured, thus providing a rich information basis for subsequent detection. The core of this method lies in using the differences in the electrical properties of tissues to distinguish normal tissues from abnormal tissues, because there are significant differences in parameters such as conductivity and permittivity between normal breast tissues and diseased tissues.

[0023] Specifically, the generation of the multi-frequency electrical excitation signal can be achieved by setting the preset low-frequency threshold and the step value. For example, the preset low-frequency threshold can be set to 1 kHz, the step value is 10 kHz, and the preset high-frequency threshold can be set to 1 MHz. The generation of the frequency sequence is based on these parameters to form an equally spaced frequency distribution. In actual operation, these parameters can be adjusted according to the physiological characteristics of the detection object and the performance of the detection device. For example, for breast tissue, since its electrical properties show different change trends in the low-frequency band and the high-frequency band, choosing a signal covering a wider frequency band can more comprehensively reflect the electrical properties of the tissue. In addition, the setting of the amplitude needs to consider the safe current limit to ensure that the detection process is harmless to the human body.

[0024] Preferably, in order to improve the quality and stability of the detection signal, the generated multi-frequency electrical excitation signal can be pre-emphasized and filtered to compensate for the attenuation of high-frequency signals during transmission. Pre-emphasis filtering is a signal processing technique that increases a certain gain in the high-frequency band of the signal, so that the high-frequency signal can still maintain a high signal-to-noise ratio after transmission and measurement. In addition, in order to further optimize the signal generation process, the time-division multiplexing method can be used to modulate the frequency sequence to generate a time-domain superposition signal. This method can not only improve the utilization rate of the signal, but also reduce the mutual interference between signals, thereby improving the accuracy and reliability of detection.

[0025] In some embodiments, step S1 includes: S11. Set a preset low-frequency threshold and a preset step value , based on the and generate a frequency sequence , , where is the preset high-frequency threshold; S12. Modulate the frequency sequence by time-division multiplexing to generate a time-domain superposition signal , where, is the amplitude of the nth frequency, is the preset random phase, N is the preset number of discrete points and N≥5, ; S13. Perform pre-emphasis filtering on to compensate for high-frequency signal attenuation.

[0026] It should be noted that during the process of generating the multi-frequency electrical excitation signal, it is necessary to set a preset low-frequency threshold and a step value, and generate a frequency sequence based on these parameters. Here, the frequency sequence refers to a set of multiple discrete frequency points arranged according to a certain rule, which is used as the basis for constructing the multi-frequency electrical excitation signal. By modulating the frequency sequence by time-division multiplexing, a time-domain superposition signal is generated. This signal can integrate multiple frequency components into a time-domain signal, thereby improving the transmission efficiency and detection accuracy of the signal. In addition, performing pre-emphasis filtering on the generated time-domain superposition signal aims to compensate for the possible attenuation of high-frequency signals during transmission, ensure the uniformity and stability of the signal at different frequencies, and provide a high-quality excitation signal for subsequent detection.

[0027] Specifically, the preset low-frequency threshold can be set to 1 kHz, the step value is 10 kHz, and the preset high-frequency threshold is 1 MHz, so as to generate a frequency sequence from 1 kHz to 1 MHz. The generation formula of the frequency sequence is: , where is the low - frequency threshold, is the step value, is the high - frequency threshold. For example, if , the generated frequency sequence includes 1 kHz, 11 kHz, 21 kHz, etc., up to 1 MHz. During the modulation process, the amplitude of each frequency can be set according to the formula , where is the preset maximum amplitude, is the attenuation coefficient, is the current frequency. This amplitude setting method can ensure that the attenuation of the signal in the high - frequency band is compensated, while meeting the requirements of the safety current limit. The pre - emphasis filtering process can be achieved by designing a high - pass filter with a cut - off frequency slightly lower than the high - frequency threshold to enhance the amplitude of the high - frequency signal.

[0028] Preferably, in order to further optimize the signal generation process, a more complex modulation method can be adopted, such as introducing random phase modulation on the basis of time - division multiplexing. Random phase modulation can be achieved by assigning a random phase value to each frequency point, for example, using a uniform distribution or a Gaussian distribution to generate random phases. This modulation method can effectively reduce the coherent interference between signals and improve the noise resistance of the signal. In addition, the design of the pre - emphasis filter can be adjusted according to the actual detection environment, for example, by using an adaptive filtering algorithm to dynamically adjust the parameters of the filter to better compensate for the attenuation of the high - frequency signal. At the same time, in order to improve the stability and reliability of the signal, multiple samplings and averaging processes can be performed after the signal is generated to reduce the influence of random noise.

[0029] In some embodiments, the amplitude in step S12 satisfies: , where, is the preset maximum amplitude and meets the safety current limit, is the preset attenuation coefficient, is the preset low - frequency threshold, is the preset step value.

[0030] In the process of generating a multi - frequency electrical excitation signal, the setting of the signal amplitude is a key link. Specifically, the formula for the amplitude is , where is the preset maximum amplitude, is the attenuation coefficient, is the preset low - frequency threshold, is the current frequency, is a preset step value. The design of this formula aims to dynamically adjust the amplitude according to the change of frequency to ensure the uniformity and safety of the signal at different frequencies. In this way, a multi-frequency electrical excitation signal that not only meets the safety current limit but also compensates for attenuation in the high-frequency band can be generated, thus providing a high-quality excitation source for subsequent breast tissue detection.

[0031] Specifically, the preset maximum amplitude is set according to the safety standards of the detection device and the physiological tolerance range of the human body. Usually, it does not exceed the milliampere-level current to ensure that the detection process is harmless to the human body. The attenuation coefficient is a parameter used to control the change of amplitude with frequency, and its value can be adjusted according to actual detection requirements. For example, when the detection target is breast tissue, considering the rapid change of its electrical properties in the high-frequency band, the attenuation coefficient can be set to a relatively small value (such as 0.01) to ensure that the amplitude of the high-frequency signal will not be too low. The low-frequency threshold and the step value are the basic parameters for generating the frequency sequence. The low-frequency threshold is usually set to about 1 kHz, and the step value can be set to 10 kHz or smaller according to the detection accuracy requirements. By reasonably setting these parameters, a multi-frequency electrical excitation signal covering a wide frequency band with a reasonable amplitude distribution can be generated.

[0032] Preferably, in order to further optimize the setting of the amplitude, an adaptive adjustment mechanism can be introduced. For example, according to the signal quality feedback in real time during the detection process, dynamically adjust the attenuation coefficient and the maximum amplitude . If the signal-to-noise ratio of the high-frequency signal is detected to be low, the attenuation coefficient can be appropriately reduced to enhance the amplitude of the high-frequency signal; if the detected current exceeds the safety range, the maximum amplitude can be reduced. In addition, a segmented attenuation method can be adopted, that is, different attenuation coefficients are set in the low-frequency band and the high-frequency band respectively to better adapt to the signal characteristics of different frequency bands. This flexible amplitude adjustment strategy can not only improve the reliability of the detection but also further optimize the quality of the detection signal.

[0033] In some embodiments, step S3 includes: S31. Orthogonally demodulate the response voltage signal collected in step S2 and the reference current signal generated in step S1 to obtain the in-phase signal and the quadrature signal , where t is the time variable, is the frequency value of the nth frequency point; S32. Perform low-pass filtering on and respectively to extract the baseband components of each frequency point and , where n is the frequency point serial number and n = 0, 1, …, N; S33. Based on and calculate the complex impedance at the nth frequency point , where is the amplitude of the excitation voltage at the nth frequency in step S1, and j is the imaginary unit.

[0034] It should be noted that during the process of processing the acquired response voltage signal, it is necessary to perform quadrature demodulation on it with the reference current signal to separate the real and imaginary components of each frequency point. Quadrature demodulation is a signal processing technique that can separate the amplitude and phase information of a signal by performing orthogonal decomposition with a phase difference of 90° between the signal and the reference signal. The real and imaginary components here correspond to the in-phase and quadrature components of the signal respectively, and they are the basis for calculating complex impedance. In this way, the impedance information of each frequency point can be accurately extracted, providing accurate data support for subsequent feature extraction and tissue abnormality judgment.

[0035] Specifically, the quadrature demodulation process includes three key steps: First, multiply the acquired response voltage signal by the reference current signal to obtain the in-phase signal and the quadrature signal; Second, perform low-pass filtering on these two signals respectively to extract the baseband components; Finally, calculate the complex impedance value based on the baseband components. In actual operation, the frequency of the reference current signal is the same as that of the excitation signal, and the design of the low-pass filter needs to be optimized according to the frequency range of the signal. For example, if the frequency range of the excitation signal is from 1 kHz to 1 MHz, the cut-off frequency of the low-pass filter can be set slightly higher than half of the highest frequency (such as 500 kHz) to ensure that the baseband components of the signal can be completely extracted. In addition, the calculation formula for complex impedance is , where and are the real and imaginary parts of the baseband component respectively, is the amplitude of the excitation voltage, and j is the imaginary unit.

[0036] Preferably, in order to improve the accuracy and stability of quadrature demodulation, digital signal processing techniques can be adopted. For example, a digital filter can be used to replace the analog filter for low-pass filtering to reduce the impact of hardware errors on signal processing. At the same time, the frequency resolution can be improved by increasing the signal sampling rate, so as to extract the baseband component more accurately. In addition, in order to cope with possible signal interference, the signal can be preprocessed before demodulation, such as removing the DC bias or suppressing high-frequency noise. During the complex impedance calculation process, an error correction mechanism can also be introduced. For example, the system error in the calculation model can be corrected by calibrating a standard sample with a known impedance, thereby further improving the accuracy of the detection results.

[0037] In some embodiments, step S4 includes: S41. Convert the complex impedance at each frequency point to the frequency-domain admittance ; S42. Construct an equivalent circuit model , where is the conductance component, is the capacitance component, ; S43. Optimize the frequency-varying curve of the model parameters by the non-linear least squares fitting algorithm.

[0038] It should be noted that in the process of extracting feature vectors from the multi-frequency impedance spectrum, parameters such as the slope of the amplitude-frequency characteristic, the position of the extreme point of the phase-frequency characteristic, and the impedance attenuation rate in the high-frequency band need to be calculated. These feature vectors can effectively characterize the changes in the electrical properties of breast tissue, thereby providing key information for the detection of tissue abnormalities. The slope of the amplitude-frequency characteristic reflects the trend of the impedance amplitude changing with frequency, the position of the extreme point of the phase-frequency characteristic indicates the frequency point where the impedance phase changes most significantly, and the impedance attenuation rate in the high-frequency band is used to measure the decline speed of the impedance in the high-frequency region. By extracting these features, the complex impedance spectrum information can be simplified into a set of representative parameters, which is convenient for subsequent analysis and judgment.

[0039] Specifically, the slope of the amplitude-frequency characteristic can be obtained by performing linear regression calculation on the impedance amplitude in a preset middle frequency band (for example, from 100 kHz to 500 kHz), and its calculation formula is , where is the difference in impedance amplitude between adjacent frequency points, is the frequency interval. The position of the extreme point of the phase-frequency characteristic refers to the frequency corresponding to the maximum value of the impedance phase angle, which can be obtained by solving the extreme point of the phase spectrum, that is . The impedance attenuation rate in the high-frequency band is calculated by comparing the impedance amplitude at the high-frequency threshold (such as 1 MHz) with the impedance amplitude at the reference frequency (such as 500 kHz), and the formula is 。In actual operation, the calculation of these parameters needs to be based on accurate impedance measurement and data processing to ensure that the eigenvector can accurately reflect the differences in the electrical properties of tissues.

[0040] Preferably, to improve the accuracy and reliability of feature extraction, the calculation process of the eigenvector can be optimized. For example, when calculating the slope of the amplitude-frequency characteristic, the accuracy of linear regression can be improved by increasing the number of frequency points in the middle frequency band, or a weighted linear regression method can be adopted to assign different weights to different frequency points to highlight the influence of key frequency bands. For the extraction of the extreme point position of the phase-frequency characteristic, the phase spectrum can be preprocessed by combining smoothing algorithms (such as moving average or Gaussian filtering) to reduce the influence of noise interference on the detection of extreme points. In addition, a dynamic reference frequency can be introduced in the calculation of the impedance attenuation rate in the high-frequency band, and the position of the reference frequency can be adjusted according to the actual detection situation to better adapt to the characteristic changes of different tissue types. These optimization measures can further improve the quality of the eigenvector, thereby improving the sensitivity and specificity of breast tissue abnormality detection.

[0041] In some embodiments, the eigenvector in step S5 includes: Slope of the amplitude-frequency characteristic Linear regression coefficient in the preset middle frequency band, where , is the interval between adjacent frequency points; Frequency of the extreme point of the phase-frequency characteristic , where is the frequency corresponding to the maximum value in the phase spectrum; High-frequency attenuation factor , where is the preset high-frequency threshold, is the preset reference frequency.

[0042] It should be noted that in the process of inputting the extracted eigenvector into the dynamic decision model to calculate the tissue abnormality index, the dynamic decision model analyzes and processes the eigenvector through an iterative weighting algorithm. Here, the dynamic decision model is a pre-trained mathematical model that can calculate the tissue abnormality index based on the input eigenvector, and then determine whether there is an abnormality in the breast tissue. The iterative weighting algorithm is an optimization algorithm that assigns different weights to each feature in the eigenvector and continuously iteratively adjusts these weights to improve the model's ability to identify abnormal tissues. The core of this process is to use the model to perform quantitative analysis on the eigenvector, so as to achieve accurate judgment of breast tissue abnormalities.

[0043] Specifically, the input of the dynamic decision model is a feature vector, including parameters such as the slope of the amplitude-frequency characteristic, the position of the extreme point of the phase-frequency characteristic, and the impedance attenuation rate in the high-frequency band. These parameters respectively reflect the change trends of the electrical properties of tissues in different frequency bands. In the model, through preset weighting coefficients (such as ), each feature is weighted, and the sum of the weighting coefficients is 1, that is . The calculation formula of the tissue abnormality index can be expressed as where S0, and are the calibration values of the slope of the amplitude-frequency characteristic, the frequency of the extreme point of the phase-frequency characteristic, and the high-frequency attenuation factor of healthy tissues respectively. These calibration values are obtained by measuring and statistically analyzing a large number of healthy tissue samples and are used as a benchmark for judging abnormal tissues.

[0044] Preferably, in order to further improve the accuracy and adaptability of the dynamic decision model, the training process of the model can be optimized. For example, machine learning algorithms (such as support vector machines or neural networks) are used to train the model. By increasing the number and diversity of training samples, the recognition ability of the model for different tissue types is improved. In addition, the weighting coefficients can be dynamically adjusted. According to the change trend of the feature vector in the actual detection process, the weight allocation is updated in real time to better reflect the importance of each feature in abnormal detection. For example, when the impedance attenuation rate in the high-frequency band makes a greater contribution to abnormal detection, its corresponding weighting coefficient can be increased. At the same time, in order to improve the robustness of the model, a threshold judgment mechanism can be introduced when calculating the tissue abnormality index. When the abnormality index exceeds the preset threshold, it is determined that the tissue is abnormal, so as to achieve rapid and accurate detection of breast tissue abnormalities.

[0045] In some embodiments, the dynamic decision model in step S6 performs: Calculate the abnormality index where is the calibration value of the slope of the amplitude-frequency characteristic of healthy tissues, is the calibration value of the frequency of the extreme point of the phase-frequency of healthy tissues, is the calibration value of the high-frequency attenuation factor of healthy tissues, , , are preset weighting coefficients and satisfy .

[0046] It should be noted that in the dynamic decision model, the calculation formula of the tissue abnormality index is , where is a preset weighting coefficient and satisfies . This formula quantifies the difference between the eigenvector and the calibration value of healthy tissue, and comprehensively considers the influence of the slope of the amplitude-frequency characteristic, the extreme point frequency of the phase-frequency characteristic, and the high-frequency attenuation factor on tissue abnormality. This calculation method can effectively reflect the changes in the electrical properties of breast tissue, thus providing a quantitative index for the judgment of tissue abnormality. By reasonably setting the weighting coefficient and calibration value, the model can flexibly adapt to the detection requirements of different tissue types, improving the accuracy and reliability of detection.

[0047] Specifically, S0 is the calibration value of the slope of the amplitude-frequency characteristic of healthy tissue, which is the average value obtained by statistically analyzing the slopes of the amplitude-frequency characteristics of a large number of healthy breast tissue samples; is the calibration value of the extreme point frequency of the phase-frequency characteristic of healthy tissue, representing the typical value of the extreme point frequency of the phase-frequency characteristic of healthy tissue; is the calibration value of the high-frequency attenuation factor of healthy tissue, reflecting the impedance attenuation characteristics of healthy tissue in the high-frequency band. The setting of the weighting coefficient needs to be adjusted according to the importance of each feature in tissue abnormality detection. For example, if the slope of the amplitude-frequency characteristic is more sensitive to tissue abnormality, the weight of can be appropriately increased. In practical applications, these calibration values and weighting coefficients can be obtained by fitting experimental data or optimizing machine learning algorithms to ensure that the model can accurately distinguish normal tissue from abnormal tissue.

[0048] Preferably, in order to further improve the accuracy of calculating the tissue abnormality index, a dynamic adjustment mechanism can be introduced. For example, according to the signal quality or tissue type fed back in real time during the detection process, dynamically adjust the ratio of the weighting coefficient . In addition, a multi-model fusion method can be adopted, combining different types of dynamic decision models (such as models based on statistical analysis and models based on machine learning), and calculating the final tissue abnormality index by integrating the output results of multiple models. This fusion method can make full use of the advantages of different models, improving the robustness and adaptability of detection. At the same time, in order to better reflect the dynamic changes of tissue, time series analysis can be introduced during the calculation process to dynamically monitor the continuously detected abnormality indexes, so as to more accurately capture the early signs of tissue abnormality.

[0049] In some embodiments, it further includes a dynamic analysis step: S81. Repeat steps S1 - S6 under the continuous time series to obtain the abnormality index sequence ; S82. Calculate the dynamic factor , where is the time stamp of the k-th detection, k = 1, 2,..., n; S83. When this occurs, high-frequency band impedance re-detection is triggered, and the high-frequency band is a band greater than a preset re-detection frequency.

[0050] It should be noted that during the dynamic analysis process, by repeatedly executing the detection steps in a continuous time series and obtaining an abnormal index sequence, a dynamic factor is further calculated to evaluate the change in tissue state. Here, the dynamic factor is an index that measures the dynamic change of tissue characteristics by analyzing the change rate of the abnormal index over time. Its calculation formula is , where represents the abnormal index of the k-th detection, is the corresponding detection timestamp. When the dynamic factor exceeds the preset threshold, high-frequency band impedance re-detection is triggered to ensure the reliability of the detection results in the high-frequency band. This dynamic analysis mechanism can monitor the change of tissue state in real time, detect abnormal changes in a timely manner, and improve the sensitivity and reliability of detection.

[0051] Specifically, the calculation of the dynamic factor involves successive differences of the abnormal index sequence and normalization processing of the time interval. In practical applications, the detection time series can be set at a fixed interval according to the detection requirements, such as detecting once every 10 seconds or every 30 seconds. The preset threshold is set according to the dynamic change range of normal tissues and is usually obtained through experimental data statistics. For example, if the fluctuation range of the dynamic factor of normal tissues is within 0.01, the threshold can be set to 0.02. When the dynamic factor D exceeds this threshold, it indicates that the tissue characteristics may have changed significantly, and re-detection of the high-frequency band impedance is required. The definition of the high-frequency band can be set as a band greater than a certain frequency value (such as 500 kHz) according to the detection target. The purpose of re-detection is to further confirm the existence of tissue abnormalities by increasing the detection accuracy in the high-frequency band.

[0052] Preferably, in order to improve the accuracy and efficiency of dynamic analysis, the calculation process of the dynamic factor can be optimized. For example, a sliding window mechanism is introduced to segment the continuous abnormal index sequence and calculate the dynamic factor within each window, thereby reducing the influence of noise on the calculation of the dynamic factor. In addition, according to the characteristics of the dynamic change of tissue characteristics, the detection time interval can be dynamically adjusted. For example, when significant changes in tissue characteristics are detected, the detection interval is shortened to improve the monitoring accuracy; when the tissue characteristics are stable, the detection interval is extended to save detection resources. At the same time, in order to further improve the reliability of high-frequency band re-detection, the number of detection frequency points can be increased or a higher-precision detection device can be used during the re-detection process, so as to more accurately capture the impedance change of the tissue in the high-frequency band.

[0053] In some embodiments, a feedback calibration step is further included: S91. When the electrode contact impedance exceeds a preset contact impedance threshold automatically switch the excitation electrode pair and recalculate the compensation matrix; S92. When the environmental temperature change exceeds a preset temperature change threshold, update the eigenvector threshold in step S5 according to the pre-stored temperature-impedance correction curve.

[0054] It should be noted that during the detection process, the feedback calibration mechanism is used to ensure the accuracy and reliability of the detection results. When the electrode contact impedance exceeds the preset threshold automatically switch the excitation electrode pair and recalculate the compensation matrix to correct the measurement error caused by poor electrode contact. In addition, when the environmental temperature change exceeds the preset temperature change threshold, update the threshold of the eigenvector according to the pre-stored temperature-impedance correction curve. This feedback calibration mechanism can effectively cope with external interferences that may occur during the detection process and ensure the stability and repeatability of the detection results.

[0055] Specifically, the electrode contact impedance refers to the impedance generated when the electrode contacts the skin, and its magnitude directly affects the quality of the measurement signal. The preset threshold is set according to the impedance range measured under the condition of good electrode contact, usually determined through experiments. For example, if the impedance range when the electrode contact is good is 100 Ω, then can be set to 150 Ω. When it is detected that the electrode contact impedance exceeds this threshold, the system will automatically switch the excitation electrode pair and recalculate the compensation matrix to correct the measurement error. The compensation matrix is pre-calculated based on the characteristics of the electrode contact impedance and is used to compensate for the signal deviation caused by poor electrode contact. The influence of environmental temperature change on the detection results cannot be ignored, and the preset temperature change threshold can be set according to the temperature stability of the device and the requirements for detection accuracy. For example, if the device can still maintain the detection accuracy within a temperature change of ±2 °C, then the temperature change threshold can be set to ±3 °C. When the temperature change exceeds this threshold, the system will update the threshold of the eigenvector according to the pre-stored temperature-impedance correction curve to correct the influence of temperature change on impedance measurement.

[0056] Preferably, to further improve the accuracy and efficiency of the feedback calibration mechanism, real-time monitoring and dynamic adjustment strategies can be introduced. For example, during the detection process, the change in electrode contact impedance is monitored in real time. Once an abnormal increase in impedance is detected, the electrode pair is immediately switched and recalibrated instead of waiting for the impedance to exceed the threshold. In addition, an adaptive temperature compensation algorithm can be adopted to dynamically adjust the threshold of the eigenvector according to the real-time temperature change, rather than relying solely on the pre-stored calibration curve. This adaptive algorithm can be achieved through machine learning or data fitting and can better adapt to the detection requirements under different environmental conditions. At the same time, to improve the stability of electrode contact, a conductive gel can be coated on the electrode surface or more advanced electrode materials can be used to reduce the measurement error caused by poor contact.

[0057] In some embodiments, step S1 further includes adaptive frequency adjustment: When the high-frequency attenuation factor extracted in step S5 is less than the preset attenuation threshold, the preset high-frequency threshold is extended to the extended frequency threshold, and the frequency step value is reduced to the preset reduced step value to generate a new multi-frequency electrical excitation signal.

[0058] It should be noted that during the generation of the multi-frequency electrical excitation signal, an adaptive frequency adjustment mechanism is introduced. When it is detected that the high-frequency attenuation factor is less than the preset attenuation threshold, the system automatically extends the preset high-frequency threshold to the extended frequency threshold and reduces the frequency step value to the preset reduced step value, thereby generating a new multi-frequency electrical excitation signal. This adaptive frequency adjustment mechanism can dynamically optimize the detection frequency range according to the signal characteristics feedback in real time during the detection process, ensure the detection accuracy in the high-frequency band, and improve the flexibility and adaptability of the detection at the same time.

[0059] Specifically, the high-frequency attenuation factor is calculated by comparing the ratio of the impedance amplitude in the high-frequency band to the impedance amplitude at the reference frequency and is used to measure the attenuation degree of the high-frequency signal. The preset attenuation threshold is set according to the high-frequency attenuation characteristics of normal tissues. For example, if the range of the high-frequency attenuation factor of normal tissues is 0.5 - 1.0, the attenuation threshold can be set to 0.4. When it is detected that the high-frequency attenuation factor is less than this threshold, it indicates that the high-frequency band signal may be affected by tissue characteristic changes and it is necessary to expand the detection frequency range to obtain more information. At this time, the system extends the preset high-frequency threshold from the initial value (such as 1 MHz) to a higher frequency (such as 2 MHz) and reduces the frequency step value Reduce from the initial value (such as 10 kHz) to a smaller step value (such as 5 kHz) to improve the frequency resolution in the high-frequency band. This adjustment mechanism can ensure that the detected signal better covers the tissue property changes in the high-frequency band, thereby improving the detection sensitivity and accuracy.

[0060] Preferably, in order to further optimize the adaptive frequency adjustment mechanism, dynamic monitoring and real-time feedback can be introduced during the detection process. For example, the system can monitor the change trend of the high-frequency attenuation factor in real time and dynamically adjust the extended frequency threshold and the size of the reduced step value according to its change rate. In addition, multiple frequency adjustment strategies can be preset in combination with the tissue type and detection target, and flexibly selected according to the actual detection situation. For example, for the detection of breast tissue, if the high-frequency attenuation factor continuously remains below the threshold, the high-frequency threshold can be gradually extended to 3 MHz, and the frequency step value can be further reduced to 1 kHz to more finely capture the changes in the high-frequency band. At the same time, in order to improve the detection efficiency, after expanding the frequency range, the signals at the newly added frequency points can be quickly screened and analyzed, and only the frequency bands with significant changes are retained for detailed detection, thereby reducing unnecessary computational complexity.

[0061] The above-mentioned various embodiments of the present invention have the following beneficial effects: Through the multi-frequency electrical impedance dynamic analysis technology, the present invention can generate electrical excitation signals covering a wide frequency band and accurately measure the response voltage signals, thereby comprehensively capturing the electrical property changes of breast tissue at different frequencies. Combining orthogonal demodulation and complex impedance spectrum generation can effectively separate the real and imaginary components and extract eigenvectors, including the slope of the amplitude-frequency characteristic, the position of the extreme point of the phase-frequency characteristic, and the impedance attenuation rate in the high-frequency band, etc. These eigenvectors can more accurately reflect the physiological and pathological states of the tissue, improving the sensitivity and specificity of abnormal detection of breast cancer tissue. At the same time, the dynamic decision model can calculate the tissue abnormality index based on the extracted eigenvectors and compare it with a preset threshold to determine whether the tissue is abnormal, which can further enhance the accuracy and reliability of the detection.

[0062] In addition, the present invention also introduces dynamic analysis, feedback calibration, and an adaptive frequency adjustment mechanism. Dynamic analysis can monitor the change of the tissue abnormality index in real time. When an abnormal change is detected, it can trigger the re-detection of the high-frequency band impedance to ensure the timeliness and accuracy of the detection results. The feedback calibration mechanism can automatically switch the excitation electrode pair or update the eigenvector threshold when the electrode contact impedance is too high or the environmental temperature changes, thereby compensating for the influence of external factors on the detection results and improving the stability and repeatability of the detection. Adaptive frequency adjustment can dynamically adjust the detection frequency range according to the change of the high-frequency attenuation factor to further optimize the detection accuracy. The introduction of these mechanisms makes the detection process more intelligent and automated, and can provide an efficient and reliable non-invasive detection means for the early screening and diagnosis of breast cancer.

[0063] Further, the storage medium of the embodiment of the present application stores program instructions capable of implementing all the above methods. Among them, the program instructions can be stored in the above storage medium in the form of a software product, including several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor to execute all or part of the steps of the methods described in various embodiments of the present application. The aforementioned storage medium includes: various media that can store program codes such as USB flash drives, mobile hard disks, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical discs, or terminal devices such as computers, servers, mobile phones, and tablets.

[0064] The above description is only some preferred embodiments of the present invention and an explanation of the applied technical principles. Those skilled in the art should understand that the scope of the invention involved in the embodiments of the present invention is not limited to the technical solutions formed by the specific combination of the above technical features, and should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above inventive concept. For example, the technical solutions formed by mutually replacing the above features with the (but not limited to) technical features having similar functions disclosed in the embodiments of the present invention.

Claims

1. A method for detecting abnormalities in breast cancer tissue based on multi-frequency electrical impedance dynamic analysis, characterized in that: The following steps are involved: S1. Generate a multi-frequency electrical excitation signal comprising a plurality of discrete frequency points, wherein the frequency range of the multi-frequency electrical excitation signal covers a preset low frequency threshold to a preset high frequency threshold; S2. applying the multi-frequency electrical excitation signal to the breast tissue to be tested through the electrode array, and synchronously collecting the response voltage signal of the electrode array; S3. Performing orthogonal demodulation on the collected response voltage signal to separate the real and imaginary components of each frequency point; S4. Based on the obtained real and imaginary components, the complex impedance value of each frequency point is calculated to generate a multi-frequency impedance spectrum; S5. Extracting a feature vector including at least the slope of the amplitude-frequency characteristic, the position of the extreme point of the phase-frequency characteristic and the impedance attenuation rate in the high-frequency band from the multi-frequency impedance spectrum; S6. Inputting the extracted feature vector into the pre-trained dynamic decision model, and calculating the tissue abnormality index by the iterative weighted algorithm in the dynamic decision model; S7. Determine whether there is any abnormality in the breast tissue based on the comparison result between the obtained tissue abnormality index and the preset threshold value.

2. The detection method according to claim 1, characterized in that: The step S1 comprises: S11. Set the preset low frequency threshold and preset step values , based on the and Generate frequency sequence; S12. Modulate the frequency sequence using a time division multiplexing method to generate a time domain superposition signal, as shown in the following calculation formula; ,in, is the amplitude of the nth frequency, is the preset random phase, N is the preset discrete points, ; S13.Yes Perform pre-emphasis filtering to compensate for high-frequency signal attenuation.

3. The detection method according to claim 2, characterized in that: The amplitude in step S12 satisfy: ,in, To preset the maximum value and meet the safety current limit, is the preset attenuation coefficient, To preset the low frequency threshold, is the preset step value.

4. The detection method according to claim 1, characterized in that: The step S3 comprises: S31. The response voltage signal collected in step S2 The reference current signal generated in step S1 Perform orthogonal demodulation to obtain an in-phase signal and an orthogonal signal. The in-phase signal is shown in the following formula; ； The quadrature signal is shown in the following formula; ； Among them, t is the time variable, is the frequency value of the nth frequency point; S32.Yes and Perform low-pass filtering to extract the baseband components at each frequency point and , where n is the frequency point number; S33.Based on and Calculate the complex impedance at the nth frequency point as shown in the following formula; ,in, is the excitation voltage amplitude of the nth frequency, and j is an imaginary unit.

5. The detection method according to claim 1, characterized in that: The step S4 comprises: S41. The complex impedance at each frequency point Converted to frequency domain admittance as shown in the following formula; ； S42. Construct an equivalent circuit model; ,in, is the conductivity component, is the capacitance component, ; S43. Optimizing model parameters by nonlinear least squares fitting algorithm The frequency curve of .

6. The detection method according to claim 1, characterized in that: The characteristic vector in step S5 includes the slope of the amplitude-frequency characteristic, the frequency of the extreme point of the phase-frequency characteristic, and the high-frequency attenuation factor.

7. The detection method according to claim 1, characterized in that: The dynamic decision model in step S6 calculates the abnormality index based on the slope of the amplitude-frequency characteristic, the frequency of the extreme point of the phase-frequency characteristic and the high-frequency attenuation factor, as shown in the following formula; in, is the slope calibration value of the amplitude-frequency characteristic of healthy tissue, is the frequency calibration value of the phase frequency extreme point of healthy tissue, is the calibration value of the high-frequency attenuation factor of healthy tissue, is the slope of the amplitude-frequency characteristic, is the frequency of the extreme point of the phase-frequency characteristic, is the high frequency attenuation factor, is the abnormality index, , , is the preset weighting factor.

8. The detection method according to claim 7, characterized in that: The method further comprises the following steps: S81. In continuous time series Repeat steps S1-S6 to obtain the abnormal index sequence ; S82. Based on abnormal index sequence The dynamic factor is calculated as shown in the following formula; ,in, is the timestamp of the k-th detection, For the kth The timestamp of the detection. is the dynamic factor; S83. When When the high frequency band impedance re-detection is triggered, the high frequency band is a frequency band greater than the preset re-detection frequency.

9. The detection method according to claim 1, characterized in that: The following steps are also included: S91. When the electrode contact impedance exceeds the preset contact impedance threshold When the excitation electrode pair is switched, the compensation matrix is ​​recalculated. S92. When the ambient temperature changes When the temperature exceeds a preset temperature change threshold, the characteristic vector threshold in step S5 is updated according to the pre-stored temperature-impedance correction curve.

10. The detection method according to claim 1, characterized in that: The method further comprises: When the high frequency attenuation factor When the frequency step value is less than the preset attenuation threshold, the preset high frequency threshold is extended to the extended frequency threshold, and the frequency step value is increased. The multi-frequency electric excitation signal is generated by reducing the preset reduction step value.

Images