Multi-probe radio frequency distributed lung water monitoring system and method

Abstract

The invention relates to the technical field of data processing, and discloses a multi-probe radio frequency distributed lung water monitoring system and method.The method comprises the steps that low, high and frequency protection transmission parameters are collected and converted into a logarithm domain by deploying a thoracic radio frequency array; fitting the rib quasi-periodic model based on frequency protection scanning to obtain a rib arch trend and a synthetic pitch; estimating effective wavelengths in tissues corresponding to low and high frequencies; obtaining a window consistency weight by combining geometric and wavelength parameters, and weighting and normalizing logarithmic parameters to generate a standard complex response; carrying out self-spectrum analysis on the complex response to extract respiratory and cardiac angular frequencies, and carrying out orthogonal projection on the basis of sine, cosine and direct current to obtain a modulation slow-varying complex response; taking the initial breath-holding as a baseline, constructing a high-low frequency complex response complex number difference, and performing two-dimensional vector projection on real and imaginary parts of the high-low frequency complex response complex number difference to generate a lung water index.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, more particularly, it relates to a multi-probe radio frequency distributed lung water monitoring system and method. BACKGROUND

[0002] In recent years, non-invasive monitoring methods based on radio frequency electromagnetic waves have been gradually applied to the dynamic monitoring of lung liquid content. Compared with traditional imaging methods, radio frequency measurement has the advantages of real-time, wearable and no radiation, and therefore has wide application prospects in intensive care, management of cardiogenic edema and prevention and control of postoperative complications. In order to improve the spatial resolution and stability, researchers usually arrange multiple radio frequency probes around the chest, and estimate the lung water level through the amplitude and phase changes of the cross-probe signals.

[0003] However, in practical application, the measurement signal is not only related to the lung water content, but also significantly affected by the chest wall tissue structure. The bony components (ribs) of the chest wall are greatly different from soft tissues, muscles and lung parenchyma in terms of dielectric constant and conductivity. When the radio frequency signal passes through the chest wall, it will encounter a strong discontinuous boundary, resulting in scattering, reflection and attenuation effects. These effects are not uniformly distributed in space, but are limited to the periodic arrangement of ribs and intercostal spaces. Further, at certain frequencies, radio frequency waves are more likely to penetrate intercostal spaces along a specific angle, while being greatly attenuated when directly opposite the ribs. This causes the sensitivity between probe channels to have a directional difference.

[0004] When the probe array is fixed to the body surface, the relationship between its direction and the arrangement of the ribs directly determines whether the radio frequency signal can smoothly penetrate to the lung tissue. If some channels are aligned with the intercostal space, a significant electromagnetic response can be generated when the lung water changes slightly; otherwise, if the channel is directly opposite the ribs, the perception of lung water changes will be severely weakened. More complexly, when the patient's body position is adjusted or the chest wall moves, causing a slight shift in the relative direction of the probe and the ribs, the originally sensitive channel may suddenly become insensitive, and the originally insensitive channel may suddenly become enhanced. This makes it difficult to maintain consistency in monitoring results at different time points, and weakens the reliability of the radio frequency distributed method in clinical monitoring. SUMMARY

[0005] The present application provides a multi-probe radio frequency distributed lung water monitoring system and method, which solves the technical problems raised in the background art.

[0006] In a first aspect, the present application provides a multi-probe radio frequency distributed lung water monitoring method, comprising: arranging a radio frequency array around the chest, acquiring transmission parameters in a low frequency working frequency band, a high frequency working frequency band and a guard frequency working frequency band, and converting the transmission parameters into a logarithmic domain to form logarithmic transmission parameters; Based on the guard band, shallow high-frequency spatial spectrum scanning is performed, a quasi-periodic lattice model of the rib and the intercostal space is fitted, a rib arch direction unit vector and a combined pitch of the rib and the intercostal space are obtained; The effective wavelength in the tissue corresponding to the low-frequency working frequency band and the high-frequency working frequency band is determined; According to the rib arch direction unit vector, the combined pitch of the rib and the intercostal space, and the effective wavelength in the tissue, a window consistency weight is determined; and the logarithmic transmission parameters are weighted and summed and normalized with the window consistency weight to obtain a standard complex response; The standard complex response is subjected to a self-spectrum analysis, and a respiratory angular frequency and a cardiac angular frequency are extracted; a linear orthogonal projection is performed on the respiratory angular frequency and the cardiac angular frequency, and a direct current component as a basis vector to obtain a demodulation slow-varying complex response; Taking the initial breath-holding time as a baseline, a complex difference of the demodulation slow-varying complex responses of the low-frequency working frequency band and the high-frequency working frequency band is constructed, the real part and the imaginary part of the complex difference are taken as a two-dimensional vector, and a lung water index is generated by projecting the two-dimensional vector.

[0007] Further, the transmission parameters are collected in the low-frequency working frequency band, the high-frequency working frequency band and the guard band, and the transmission parameters are converted into the logarithmic domain to form logarithmic transmission parameters, including: The low-frequency working frequency band is 0.6 GHz to 0.8 GHz, the high-frequency working frequency band is 1.2 GHz to 1.5 GHz, and the guard band is 2.0 GHz to 2.5 GHz; The transmission parameters of the time t in the frequency band are collected at a fixed time interval to obtain the amplitude and the phase of the transmission parameters, and the frequency band represents the low-frequency working frequency band, the high-frequency working frequency band or the guard band; The transmission parameters are converted into the logarithmic domain according to the amplitude and the phase of the transmission parameters, as follows: wherein, represents the logarithmic transmission parameter, represents an imaginary unit, .

[0008] Further, based on the guard band, shallow high-frequency spatial spectrum scanning is performed, a quasi-periodic lattice model of the rib and the intercostal space is fitted, a rib arch direction unit vector and a combined pitch of the rib and the intercostal space are obtained, including: In a first preset time period, the logarithmic transmission parameters of each frequency in the guard band are determined at a fixed time interval, and the logarithmic transmission parameter variance of the corresponding frequency is calculated; the frequency corresponding to the maximum logarithmic transmission parameter variance is taken as a scanning frequency point; In the second preset time period, the M scanning probes are controlled to collect the logarithmic transmission parameters according to the scanning frequencies, and the average value of each logarithmic transmission parameter is calculated ; A one-dimensional mapping is set to determine the one-dimensional coordinate of the mth scanning probe ; wherein, ; wherein, represents the origin coordinate of the one-dimensional coordinate system, represents the coordinate of the mth scanning probe, represents the initial rib arch direction unit vector, represents the included angle of the one-dimensional mapping with the horizontal direction; Based on the average value of the logarithmic transmission parameter of each scanning probe and the one-dimensional coordinate, fitting is performed as follows: wherein, represents the constant term of the quasi-periodic lattice model, represents the amplitude coefficient of the quasi-periodic lattice model, represents the phase coefficient of the quasi-periodic lattice model, represents the combined pitch of the initial rib and intercostal space; The combined pitch of the initial rib and intercostal space and the explained variance of the quasi-periodic lattice model corresponding to the initial rib arch direction unit vector are calculated as follows: wherein, represents the fitting value of the quasi-periodic lattice model to the average value of the logarithmic transmission parameter of the mth scanning probe, represents the average value of the logarithmic transmission parameter of all scanning probes in the second preset time period; Along the direction that maximizes the explained variance, the corresponding initial rib and intercostal combined pitch and the initial rib arch direction unit vector are taken as the rib and intercostal combined pitch and the rib arch direction unit vector .

[0009] Further, the effective wavelength in the tissue corresponding to the low-frequency working frequency band and the high-frequency working frequency band is determined, including: The interval distance of the transmission channel from the mth scanning probe to the transmitting probe is determined; The patient to be monitored is controlled to suspend breathing for at least 10 seconds, and the phase corresponding to the transmission channel is collected at a fixed time interval to calculate the average phase, and the average phase is phase unfolded to obtain the corresponding unfolded phase ; The interval distance the angle between the rib arch direction unit vector and the transmission channel , and the angle between the rib arch direction unit vector and the transmission channel ; The unwrapped phase and baseline length are fitted by weighted least squares as follows: wherein, represents the slope to be fitted; The corresponding slope to be fitted is taken as the target slope along the direction that minimizes the weighted least squares fitting value; The effective wavelength in the tissue is calculated according to the target slope, including: wherein, represents the effective wavelength in the tissue, belongs to the low-frequency working frequency band or the high-frequency working frequency band, represents the target slope.

[0010] Further, the window consistency weight is determined according to the rib arch direction unit vector, the combined pitch of the rib and the intercostal space, and the effective wavelength in the tissue, including: The combined pitch of the rib and the intercostal space is projected to the angle between the transmission channel and the rib arch direction unit vector, and the equivalent pitch of the transmission channel is calculated ; The corresponding matching residual is calculated based on the equivalent pitch and the effective wavelength in the tissue as follows: wherein, represents the projection multiple, represents the matching residual of the transmission channel; The matching residual of each transmission channel is extracted to determine the median of the matching residual; the product of the median of the matching residual and a normal distribution constant is taken as the frequency The corresponding robust scale ; The window consistency weight of the transmission channel at the frequency f is calculated according to the robust scale and the matching residual: wherein, represents the window consistency weight of the transmission channel at the frequency f.

[0011] Further, the log transmission parameters are weighted and summed and normalized with the window consistency weight to obtain the standard complex response, including: The window consistency weight of each log transmission parameter is determined, and the product of the log transmission parameter and the corresponding window consistency weight is taken as the weighted composite complex quantity; The window consistency weights of all transmission channels are accumulated to obtain a window composite weight; The ratio of the weighted composite complex to the window composite weight is taken as a standard complex response.

[0012] Further, the standard complex response is subjected to a spectral analysis to extract a respiratory angular frequency and a cardiac angular frequency, including: The standard complex response is subjected to a discrete Fourier transform to obtain an estimated spectrum; The combination composite weight of the window composite weight corresponding to a low-frequency working frequency band and the window composite weight corresponding to a high-frequency working frequency band is calculated; the working frequency band corresponding to the standard complex response is determined, and the ratio of the window composite weight corresponding to the working frequency band to the combination composite weight is taken as a frequency band combination weight coefficient; The product of the frequency band combination weight coefficient and the estimated spectrum is taken as a joint spectrum; The maximum value and the minimum value in the joint spectrum are extracted, and the respiratory angular frequency and the cardiac angular frequency are calculated.

[0013] Further, the respiratory angular frequency and the cardiac angular frequency are subjected to a linear orthogonal projection with respect to a direct current component and sine and cosine functions as basis vectors to obtain a demodulated slow-varying complex response, including: A negative response sequence is constructed from the standard complex responses at N time points ; A basis matrix is constructed , as follows: A projection sequence is constructed from the basis matrix ; The projection sequence is mapped to a demodulated slow-varying complex response sequence ; wherein, .

[0014] Further, an initial breath-holding time is taken as a baseline to construct a complex difference of the demodulated slow-varying complex responses of the low-frequency working frequency band and the high-frequency working frequency band, and the real part and the imaginary part of the complex difference are taken as a two-dimensional vector, and a lung water index is generated by projecting the two-dimensional vector, including: The demodulated slow-varying complex response sequence of the low-frequency working frequency band and the demodulated slow-varying complex response sequence of the high-frequency working frequency band are determined respectively, and a starting time at which the patient to be monitored suspends breathing is determined ; A complex difference at the tth time is constructed , including: wherein, represents the time a demodulated slow-varying complex response of the high-frequency operating frequency band, denotes a starting time a demodulated slow-varying complex response of the low-frequency operating frequency band, denotes a starting time a demodulated slow-varying complex response of the high-frequency operating frequency band, denotes a starting time a demodulated slow-varying complex response of the low-frequency operating frequency band; performing complex decomposition on the complex difference to obtain a two-dimensional vector ; wherein, denotes a real part of the complex difference , denotes an imaginary part of the complex difference ; performing mean value processing on each two-dimensional vector to obtain a zero-mean sequence according to a mean vector of the two-dimensional vectors at N time points , and estimating a two-dimensional covariance matrix through the zero-mean sequence; performing eigenvalue decomposition processing on the two-dimensional covariance matrix to obtain a maximum eigenvalue, and determining a unit eigenvector corresponding to the maximum eigenvalue ; calculating a lung water index at time t.

[0015] In a second aspect, a multi-probe radio frequency distributed lung water monitoring system is applied to any one of the multi-probe radio frequency distributed lung water monitoring methods, and comprises: a logarithmic parameter acquisition module, configured to arrange a radio frequency array around a thoracic cage, acquire transmission parameters in a low-frequency operating frequency band, a high-frequency operating frequency band and a guard frequency operating frequency band, and convert the transmission parameters into a logarithmic domain to form logarithmic transmission parameters; a rib arch scanning module, configured to perform a shallow high-frequency spatial spectrum scan based on the guard frequency operating frequency band, and obtain a rib arch trend unit vector and a combined pitch of ribs and intercostal spaces by fitting a quasi-periodic lattice model of ribs and intercostal spaces; a wavelength determination module, configured to determine effective wavelengths in tissues corresponding to the low-frequency operating frequency band and the high-frequency operating frequency band; a standard complex response module, configured to determine window consistency weights according to the rib arch trend unit vector, the combined pitch of ribs and intercostal spaces and the effective wavelengths in tissues, and perform weighted summation and normalization processing on the logarithmic transmission parameters with the window consistency weights to obtain a standard complex response; A demodulation response module is configured to perform a power spectrum analysis on the standard complex response, extract a respiratory angular frequency and a cardiac angular frequency, and perform a linear orthogonal projection on the standard complex response with the positive and negative sine functions of the respiratory angular frequency and the cardiac angular frequency and a direct current component as base vectors to obtain a demodulation slow-varying complex response; A lung water index module is configured to take an initial breath-holding time as a baseline, construct a complex difference of the demodulation slow-varying complex responses in the low-frequency working frequency band and the high-frequency working frequency band, take a real part and an imaginary part of the complex difference as a two-dimensional vector, and generate a lung water index according to the projection of the two-dimensional vector.

[0016] The present application has the beneficial effects that: by arranging a multi-probe radio frequency array around a thoracic cavity, combining a guard frequency scanning to fit a quasi-periodic model of ribs and intercostal spaces, effective wavelength estimation in tissues and window consistency weighting processing, directional interference caused by bone structures of a chest wall can be significantly weakened, and unified representation of cross-band complex responses can be realized; on this basis, respiratory and cardiac components are effectively separated through power spectrum analysis and orthogonal projection, and a lung water index is constructed through a difference between low-frequency and high-frequency complex responses, so that the stability and comparability of monitoring results at different body positions and different time points are improved under the premise of non-invasiveness, real-time and non-radiation, and the sensitivity and reliability of lung water monitoring are significantly enhanced. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 is a flow chart of a multi-probe radio frequency distributed lung water monitoring method of the present application; Figure 2 is a module diagram of a multi-probe radio frequency distributed lung water monitoring system of the present application. DETAILED DESCRIPTION

[0018] The subject matter described herein will now be discussed with reference to example implementations. It should be understood that the discussion of these implementations is merely meant to provide a better understanding of the subject matter described herein and can be changed in function and arrangement without departing from the scope of the present description. Various processes or components can be omitted, substituted, or added according to desired implementations. Additionally, features described in relation to some examples can be combined in other examples.

[0019] Example One:

[0020] As shown in Figure 1 A multi-probe radio frequency distributed lung water monitoring method includes: Arranging a radio frequency array around a thoracic cavity, acquiring transmission parameters in a low-frequency working frequency band, a high-frequency working frequency band and a guard frequency working frequency band, and converting the transmission parameters into a logarithmic domain to form logarithmic transmission parameters; Based on the shallow high-frequency spatial spectrum scanning of the guard frequency working band, the rib arch direction unit vector and the synthetic pitch of the rib and intercostal space are obtained by fitting the quasi-periodic lattice model of the rib and intercostal space. The effective wavelength in the tissue corresponding to the low-frequency working band and the high-frequency working band is determined. According to the rib arch direction unit vector, the synthetic pitch of the rib and intercostal space and the effective wavelength in the tissue, the window consistency weight is determined; and the logarithmic transmission parameters are weighted and summed and normalized to obtain the standard complex response. The respiratory angular frequency and the heart angular frequency are extracted by performing self-spectrum analysis on the standard complex response, and the linear orthogonal projection is performed on the sine and cosine functions of the respiratory angular frequency and the heart angular frequency and the direct current component to obtain the demodulation slow-varying complex response. The complex difference of the demodulation slow-varying complex responses of the low-frequency working band and the high-frequency working band is constructed with the initial breath-holding time as the baseline, and the real part and the imaginary part of the complex difference are taken as a two-dimensional vector to generate the lung water index by two-dimensional vector projection.

[0021] It should be noted that the present application aims to solve the problem of signal sensitivity fluctuation caused by the quasi-periodic structure of the chest wall rib and intercostal space in traditional radio frequency monitoring, and to realize reliable lung water dynamic monitoring across time periods and body positions, as follows: A radio frequency probe array is arranged around the thorax, and the transmission parameters (amplitude and phase) of the radio frequency signal are collected in the low frequency (0.6 to 0.8 GHz), high frequency (1.2 to 1.5 GHz) and guard frequency (2.0 to 2.5 GHz) three frequency bands, and are converted into logarithmic domain complex number form.

[0022] The quasi-periodic lattice model of the rib and intercostal space is fitted by shallow scanning with the guard frequency, and the rib arch direction and the synthetic pitch of the rib and intercostal space are accurately obtained.

[0023] For the low-frequency and high-frequency working frequency bands, the effective wavelength of the radio frequency signal in the human tissue is calculated in situ, and the effective wavelength and the periodic structure of the rib and intercostal space together determine the penetration characteristics of the signal.

[0024] Based on the rib arch direction, the synthetic pitch and the wavelength in the tissue, the window consistency weight is calculated, that is, higher weight is given to those channels that match well with the periodic structure of the thorax, the standard complex response is obtained by weighted normalization and aggregation of the signal, and the channel sensitivity mutation problem caused by body position change is eliminated.

[0025] The feature frequencies of respiration and heartbeat are extracted by performing frequency spectrum analysis on the standard complex response, the interference of the two high-frequency physiological motions is removed by linear orthogonal projection, and the demodulation slow-varying complex response related to the slow-varying lung water is retained.

[0026] Taking the initial breath-holding moment as a baseline, the complex difference of the low-frequency and high-frequency demodulation complex response is calculated, the real part and the imaginary part of the difference are taken as a two-dimensional vector, the main change direction is extracted through projection, and a dimensionless lung water index is generated to quantitatively reflect the dynamic change of lung water.

[0027] To sum up, the present application can realize stable and reliable monitoring of lung water change by perceiving the periodic structure of thorax, dynamically correcting signal weight, and stripping physiological interference, thereby facilitating doctors to conduct inquiry.

[0028] It should be noted that the transmission parameter represents the physical quantity related to the signal propagation characteristics detected by the receiving probe after the radio frequency signal passes through the thoracic tissue, and specifically includes amplitude and phase.

[0029] The amplitude represents the intensity of the radio frequency signal after passing through the thorax (including ribs, intercostal space, lung tissue, etc.) in the transmission channel from the transmitting probe to the receiving probe at time t and frequency f, reflecting the attenuation degree of the signal in the propagation process (affected by tissue water content, density, etc.).

[0030] The phase represents the phase shift of the radio frequency signal after passing through the thorax at time t and frequency f, reflecting the path length and medium property (dielectric constant) change of the signal in the tissue.

[0031] In an embodiment of the present application, the transmission parameters are collected in the low-frequency working frequency band, the high-frequency working frequency band and the guard frequency working frequency band, and the transmission parameters are converted into logarithmic domain to form logarithmic transmission parameters, including: The low-frequency working frequency band is 0.6GHz to 0.8GHz, the high-frequency working frequency band is 1.2GHz to 1.5GHz, and the guard frequency working frequency band is 2.0GHz to 2.5GHz; It should be noted that the low-frequency working frequency band is 0.6GHz to 0.8GHz, the high-frequency working frequency band is 1.2GHz to 1.5GHz, and the two frequency bands are closely related to the propagation characteristics of the radio frequency signal in the human tissue and the change of lung water content, which are the core frequency bands reflecting the lung water state. The guard frequency working frequency band is 2.0GHz to 2.5GHz, and the signal penetration depth of this frequency band is shallow, and the sensitivity to the bony structure of the chest wall is higher, which is the core frequency band reflecting the thoracic structure.

[0032] The transmission parameters of time t in the frequency band are collected at a fixed time interval, the amplitude and the phase of the transmission parameters, and the frequency band represents the low-frequency working frequency band, the high-frequency working frequency band or the guard frequency working frequency band. In detail, the system continuously collects signals at fixed time intervals to obtain the transmission parameters of each transmission channel in the three frequency bands at each time t. The transmission parameters include the amplitude, which reflects the intensity of the microwave signal after passing through the chest tissue (including ribs, intercostal space, lung tissue, etc.), and the change is related to the attenuation degree of the signal in the tissue; and the phase, which reflects the phase shift in the signal propagation process, and the change is related to the propagation path of the signal in the tissue and the medium characteristics (dielectric constant).

[0033] The transmission parameters are converted into the logarithmic domain according to the amplitude and phase of the transmission parameters, as follows: wherein, represents the logarithmic transmission parameter, represents the imaginary unit, .

[0034] In detail, the conversion formula is .

[0035] wherein is the converted logarithmic transmission parameter, which is a complex number. The real part is obtained by converting the amplitude of the transmission parameter by the natural logarithm, and this processing can convert the nonlinear attenuation characteristics of the signal into a linear form, which is convenient for subsequent superposition and weighting operations; the imaginary part is directly composed of the phase of the transmission parameter , is the imaginary unit and , used to distinguish the real part and the imaginary part. Through the conversion, the originally separated amplitude and phase information are integrated into a unified complex form.

[0036] In an embodiment of the present application, shallow high-frequency spatial spectrum scanning is performed based on the guard band working frequency band, and a quasi-periodic lattice model of the ribs and the intercostal space is fitted to obtain a rib arch direction unit vector and a combined pitch of the ribs and the intercostal space, including: In the first preset time period, the logarithmic transmission parameters of each frequency in the guard band working frequency band are determined at fixed time intervals, and the logarithmic transmission parameter variance of the corresponding frequency is calculated; the frequency corresponding to the maximum logarithmic transmission parameter variance is selected as the scanning frequency point; In detail, in the first preset time period, the system collects the logarithmic transmission parameters of each frequency in the guard band working frequency band at fixed time intervals. Since the guard band working frequency band is highly sensitive to the chest wall bony structure, and the quasi-periodic arrangement of the ribs and the intercostal space will cause the logarithmic transmission parameters at different frequencies to present differential fluctuations, the greater the fluctuation, the more the frequency can reflect the periodic structure characteristics. By calculating the logarithmic transmission parameter variance corresponding to each frequency, the frequency with the maximum variance means that its signal is most significantly affected by the periodicity of the ribs and the intercostal space, so the frequency is selected as the scanning frequency point.

[0037] During the second preset time period, M scanning probes are controlled to collect logarithmic transmission parameters according to the scanning frequency, and the average value of each logarithmic transmission parameter is calculated. ; In detail, within the second preset time period, the system controls M scanning probes to continuously acquire logarithmic transmission parameters at selected scanning frequencies. Since a single acquisition may be affected by noise interference, by calculating the average value of the logarithmic transmission parameters of each scanning probe, the influence of random noise can be reduced, resulting in a more stable signal basis characterizing the thoracic structure, namely the average value of the logarithmic transmission parameters of each scanning probe.

[0038] Define a one-dimensional mapping to determine the one-dimensional coordinates of the m-th scanning probe. ;in, ;in, This represents the coordinates of the origin of a one-dimensional coordinate system. This represents the coordinates of the m-th scanning probe. This represents the initial rib arch orientation unit vector. This represents the angle between the one-dimensional mapping and the horizontal direction; In detail, to simplify the analysis of the periodicity of the ribs and intercostal spaces, it is necessary to map the two-dimensional probe coordinates to a one-dimensional space. The origin coordinates of the one-dimensional coordinate system are set as follows: The coordinates of the m-th scanning probe are The initial rib arch orientation unit vector is ( (The angle between the vector and the horizontal direction). This is achieved through vector dot product. The one-dimensional coordinates of the m-th scanning probe in the initial rib arch orientation direction are obtained. This mapping transforms the periodic distribution along the rib arch in two-dimensional space into a periodic distribution on a one-dimensional line.

[0039] The fitting is performed based on the average value of the logarithmic transmission parameters of each scanning probe and the one-dimensional coordinates, as follows: in, This represents the constant term of the quasi-periodic lattice model. This represents the amplitude coefficient of the quasi-periodic lattice model. This represents the phase coefficient of the quasi-periodic lattice model. Indicates the initial rib and the combined intercostal distance; Detailed, based on the logarithmic transmission parameter average A for each scanning probe. m and its corresponding one-dimensional coordinates s m A periodic function is used for fitting, and the fitting model is as follows: .in This is a model constant term, reflecting the overall baseline level of the signal; and are the coefficients of cosine term and sine term respectively, which reflect the amplitude and phase of periodic fluctuation together; is the initial combined pitch of rib and intercostal, which is the periodic length of the sum of rib width and intercostal gap width. The model simulates the modulation effect of quasi-periodic structure formed by the alternating arrangement of rib and intercostal gap on the signal through trigonometric function.

[0040] Calculate the explained variance of quasi-periodic lattice model corresponding to the initial combined pitch of rib and intercostal and the initial rib arch direction unit vector, as follows: wherein, represents the fitting value of the quasi-periodic lattice model to the average value of the logarithmic transmission parameters of the mth scanning probe, represents the average value of the logarithmic transmission parameters of all scanning probes in the second preset time period; In detail, to evaluate the model fitting effect, the explained variance needs to be calculated , the calculation formula of which is . Among them, the numerator is the square sum of the difference between the actual average value of the logarithmic transmission parameters of each scanning probe and the model fitting value , which reflects the unexplained signal fluctuation of the model; the denominator is the square sum of the difference between each and the average value of all probes , which reflects the total fluctuation of the signal. The closer the explained variance is to 1, the stronger the model's ability to explain the signal fluctuation, that is, the closer the assumed initial combined pitch p and the initial rib arch direction unit vector to the actual thoracic structure.

[0041] Along the direction that maximizes the explained variance, the corresponding initial combined pitch of rib and intercostal and the initial rib arch direction unit vector are taken as the combined pitch of rib and intercostal and the rib arch direction unit vector .

[0042] It should be noted that by traversing different initial combined pitch p and initial rib arch direction unit vector (i.e. changing the value of ), the corresponding explained variance is calculated, and the and that maximize the explained variance are selected as the combined pitch of rib and intercostal and the rib arch direction unit vector . Thus, through data-driven optimization, the obtained parameters can most accurately reflect the quasi-periodic arrangement characteristics of rib and intercostal gap.

[0043] In one embodiment of the present application, the effective wavelength in the tissue corresponding to the low-frequency working frequency band and the high-frequency working frequency band is determined, comprising: determining the interval distance of the transmission channel from the mth scanning probe to the transmitting probe ; controlling the patient to be monitored to suspend breathing for at least 10 seconds, and acquiring the phase corresponding to the transmission channel at fixed time intervals to calculate the average phase and perform phase unwrapping processing on the average phase to obtain the unwrapped phase ; In detail, the patient to be monitored is controlled to suspend breathing for at least 10 seconds, so as to eliminate the interference of the chest fluctuation caused by the respiratory motion on the signal propagation path and ensure the stability of the phase measurement. During this period, the phase corresponding to the transmission channel is acquired at fixed time intervals, and the average phase is calculated to reduce the influence of random noise. Since the phase measurement is usually limited to the range of -π to π, the average phase needs to be unwrapped to obtain the continuous unwrapped phase, so as to truly reflect the cumulative phase shift in the signal propagation process.

[0044] determining the angle between the transmission channel and the unit vector of the rib arch direction , and calculating the corresponding direction smoothing weight based on the angle ; In detail, the angle between the transmission channel and the unit vector of the rib arch direction is determined, which reflects the relative relationship between the channel direction and the arrangement direction of the ribs. The corresponding direction smoothing weight is calculated based on the angle, and the weight formula is When the angle between the channel and the rib arch direction is large, the value is small, and the weight is large, which means that the channel in this direction is less affected by the ribs, and the phase data is more stable; otherwise, the weight is small, so as to realize the differentiated weighting of the channel data in different directions.

[0045] The unwrapped phase and the baseline length are fitted by the weighted least squares method as follows: wherein, represents the slope to be fitted; In detail, the relationship between the unwrapped phase and the baseline length is fitted by the weighted least squares method, and the fitting target is to minimize , wherein is the unwrapped phase, is the interval distance of the channel, is the slope to be fitted. The slope to be fitted reflects the phase change rate corresponding to a unit distance, and the weighted processing makes the fitting process more focused on the channel data with high weight (i.e. less affected by interference), so as to obtain a more accurate slope.​​

[0046] a target slope along a direction that minimizes the weighted least square fitting value; calculating the effective wavelength in tissue according to the target slope, including: wherein, represents the effective wavelength in tissue, belongs to a low-frequency working frequency band or a high-frequency working frequency band, represents the target slope.

[0047] It should be noted that the corresponding slope to be fitted is taken as the target slope along the direction that minimizes the weighted least square fitting value. According to the physical relationship between the wave number and the wavelength, the wave number (phase change per unit length) and the wavelength are inversely related (wave number = 2π / λ), so the effective wavelength in tissue can be calculated by . Wherein belongs to a low-frequency working frequency band or a high-frequency working frequency band, that is, the frequency of the downlink radio frequency signal in the human tissue.

[0048] In an embodiment of the present application, the window consistency weight is determined according to the rib arch direction unit vector, the combined pitch of the rib and the intercostal space, and the effective wavelength in tissue, including: projecting the combined pitch of the rib and the intercostal space to the angle between the transmission channel and the rib arch direction unit vector to calculate the equivalent pitch of the transmission channel. In detail, the combined pitch of the rib and the intercostal space is used to reflect the periodicity of the thoracic cavity, but the direction of the transmission channel may have an angle with the rib arch direction, and the combined pitch needs to be projected to the channel direction to obtain the effective periodicity length. Specifically, the equivalent pitch is the product of the combined pitch and the cosine value of the angle between the transmission channel and the rib arch direction unit vector, that is, the equivalent pitch reflects the actual periodicity length of the rib and the intercostal space arranged alternately in the channel direction.

[0049] Based on the equivalent pitch and the effective wavelength in tissue, the corresponding matching residual is calculated as follows: wherein, represents the projection multiple, represents the matching residual of the transmission channel; and In detail, the matching residual is used to measure the matching degree of the effective wavelength and the equivalent pitch in the organization, and the calculation formula is the minimum value of the absolute value of the difference between the effective wavelength and the n times equivalent pitch when the projection multiple n is 1 and 2. Because when the wavelength of the radio frequency signal is close to the 1st or 2nd order period (i.e. the pitch itself or twice the pitch) of the periodic structure of the thorax, a significant periodic modulation effect will be produced, and the smaller the residual is, the higher the matching degree is, and the more stable the sensitivity of the channel to the lung water change is.

[0050] The matching residual of each transmission channel is extracted to determine the median of the matching residual; and the product of the median of the matching residual and a normal distribution constant is taken as the frequency The corresponding robust scale ; In detail, the robust scale is used to standardize the matching residual to eliminate the influence of individual differences or noise on the weight calculation. Specifically, by extracting the matching residual of all transmission channels, the median is calculated, and then the median is multiplied by a normal distribution constant (usually 1.4826, which is used to convert the absolute deviation of the median to a robust scale estimate), and the result is the robust scale at the frequency. The robust scale reflects the overall distribution characteristics of the residual.

[0051] According to the robust scale and the matching residual, the window consistency weight of the transmission channel at the frequency f is calculated: Wherein, Indicates the transmission channel The window consistency weight at the frequency f.

[0052] It should be noted that the window consistency weight is determined by the relationship between the matching residual and the robust scale, and the formula is When the matching residual is smaller (i.e. the wavelength and the equivalent pitch match better), the weight is closer to 1, indicating that the channel is stably modulated by the periodic structure of the thorax, and the response to the lung water change is more reliable; otherwise, the smaller the weight is, the lower the matching degree of the channel and the periodic structure is, and the response stability is poor. Through the weight, the signals of different channels can be differentially weighted.

[0053] In an embodiment of the present application, the window consistency weight is used to weight and sum the logarithmic transmission parameters and normalize to obtain a standard complex response, including: Determine the window consistency weight of each logarithmic transmission parameter, and calculate the product of the logarithmic transmission parameter and the corresponding window consistency weight as the weighted composite complex quantity; In detail, the logarithmic transmission parameter is a complex number containing amplitude and phase information, reflecting the propagation characteristics of the radio frequency signal after passing through the chest tissue. The window consistency weight quantifies the matching degree of each transmission channel with the periodic structure of the chest, and the higher the weight, the more stable the channel is affected by the periodicity of the ribs and intercostal space, and the more reliable the response to lung water changes. By multiplying each logarithmic transmission parameter with its corresponding window consistency weight, the signal contribution of reliable channels can be enhanced, and the signal contribution of unreliable channels can be weakened, and the product obtained is the weighted complex composite quantity of the channel, which reflects the effective signal of the channel after reliability adjustment.

[0054] Cumulative window consistency weights of all transmission channels to obtain window composite weights; In detail, the window composite weight is the sum of the window consistency weights of all transmission channels. This can eliminate the influence of the number of channels or the absolute value difference of individual weights by aggregating the weights of all channels, ensuring that the signals at different body positions or different times have a unified reference basis.

[0055] The ratio of the weighted complex composite quantity to the window composite weight is taken as the standard complex response.

[0056] It should be noted that the sum of the weighted complex composite quantities of all channels divided by the window composite weight is the standard complex response. This ratio operation realizes normalization, so that the result is not affected by the absolute value of the weight, and only reflects the comprehensive effect of the signals of each channel after reliability weighting. The standard complex response, as the integrated complex signal, not only retains the amplitude and phase information in the logarithmic transmission parameter, but also filters the interference of unstable channels through the window consistency weight, and the normalization processing ensures its comparability under different conditions, providing a stable and reliable basis signal for subsequent stripping of respiratory and cardiac interference and calculation of lung water index.

[0057] In an embodiment of the present application, the standard complex response is subjected to spectral analysis to extract the respiratory angular frequency and the cardiac angular frequency, comprising: The standard complex response is subjected to discrete Fourier transform to obtain an estimated spectrum; In detail, the standard complex response is a complex signal after weighted normalization, containing slow-varying information related to lung water and high-frequency interference of physiological movements such as respiration and heartbeats. Discrete Fourier transform of the standard complex response can convert the time-domain signal into the estimated spectrum in the frequency domain, which can intuitively present the energy distribution of different frequency components in the signal, laying a foundation for identifying the characteristic frequencies of respiration and heartbeats.

[0058] Calculate the combined composite weight of the window composite weight corresponding to the low-frequency working frequency band and the window composite weight corresponding to the high-frequency working frequency band; determine the working frequency band corresponding to the standard complex response, and take the ratio of the window composite weight corresponding to the working frequency band to the combined composite weight as the inter-frequency combination weight coefficient; In detail, the combined composite weight is the sum of the window composite weight corresponding to the low-frequency operating band and the window composite weight corresponding to the high-frequency operating band, used to integrate the weight information of the two frequency bands. The inter-band weighting coefficient is the ratio of the window composite weight to the combined composite weight of the standard complex response's operating band (low-frequency or high-frequency). This coefficient, by balancing the signal reliability of different frequency bands (the window composite weight reflects the overall matching degree of the channel in that frequency band), makes subsequent spectrum analysis focus more on the frequency bands with higher weights and more stable signals.

[0059] The product of the inter-band weighting coefficients and the estimated spectrum is used as the joint autospectrum; In detail, the inter-band weighting coefficients are multiplied by the estimated spectrum to obtain the joint autospectrum. This weighting coefficients are then used to weight and integrate the spectra of different frequency bands, enhancing the frequency characteristics of physiological interference in reliable bands while weakening the noise effects of unreliable bands. This makes the frequency peaks of respiration and heartbeat more prominent, facilitating accurate identification.

[0060] Extracting the maximum value from the joint autospectrum and minimum value And calculate the respiratory angular frequency. and cardiac angular frequency .

[0061] It should be noted that respiration and cardiac activity are periodic physiological activities, and their corresponding frequency components appear as distinct peaks in the joint autospectrum. Generally, cardiac frequency is higher than respiratory frequency; therefore, the frequency corresponding to the maximum value in the joint autospectrum is the cardiac frequency, and the frequency corresponding to the minimum value is the respiratory frequency. Using the physical relationship between angular frequency and frequency (angular frequency = 2π × frequency), the respiratory angular frequency and cardiac angular frequency can be calculated separately.

[0062] In one embodiment of the present invention, a linear orthogonal projection is performed using the sine and cosine functions of the respiratory angular frequency and the cardiac angular frequency, along with their DC components, as basis vectors to obtain a demodulated slowly varying complex response, including: Construct a negative response sequence from the standard complex responses at N time points ; In detail, standard complex responses at N time points are selected and arranged in chronological order to form a response sequence. This sequence contains slowly varying information related to pulmonary edema, while also being mixed with periodic interferences caused by respiration and heartbeat. The purpose of constructing the sequence is to integrate the time-domain dispersed signals into a unified processing object, laying the foundation for subsequent batch interference removal.

[0063] Constructing the basis matrix ,as follows: In detail, each row of the basis matrix corresponds to a time point, and each column corresponds to a basis vector, and a total of five basis vectors are included. The first column is all 1, representing a direct current component, used to capture a constant baseline in the signal; the second column and the third column are respectively a cosine function value and a sine function value of a respiratory angular frequency, used to represent a periodic characteristic of respiratory motion; and the fourth column and the fifth column are respectively a cosine function value and a sine function value of a cardiac angular frequency, used to represent a periodic characteristic of heartbeat motion. The construction of the basis matrix is essentially to establish a subspace of respiratory and cardiac interference signals, which covers all possible periodic components of the two physiological activities.

[0064] Projecting the response sequence by the basis matrix ; In detail, the projection sequence is obtained by multiplying the response sequence by the formula . Wherein, I is a unit matrix, B is a basis matrix, , and B T is the transpose of the basis matrix. The core of the formula is to realize orthogonal projection: subtract the projection of the response sequence on the subspace spanned by the basis matrix from the response sequence, and the remaining part is the signal component orthogonal to the respiratory and cardiac interference. This operation can effectively filter out the periodic fluctuations in the response sequence caused by respiration and heartbeat, and retain the slowly varying signal related to lung water changes.

[0065] Map the projection sequence into a demodulated slowly varying complex response sequence ; wherein, .

[0066] It should be noted that each element in the projection sequence corresponds to a signal after interference removal at a time point, and mapping these elements in time sequence into a demodulated slowly varying complex response sequence. The demodulated slowly varying complex response at each time point is equal to the element at the corresponding position in the projection sequence, that is, the conversion from the projection result to the slowly varying signal with clear physical meaning is completed. The sequence has stripped off high-frequency physiological interference and only retains the slow fluctuation component related to lung water changes.

[0067] In an embodiment of the present application, taking the initial breath-holding time point as the baseline, a complex difference of the demodulated slowly varying complex response of the low-frequency working frequency band and the high-frequency working frequency band is constructed, the real part and the imaginary part of the complex difference are taken as a two-dimensional vector, and a lung water index is generated by two-dimensional vector projection, including: respectively determining the demodulated slowly varying complex response sequence of the low-frequency working frequency band and the demodulated slowly varying complex response sequence of the high-frequency working frequency band, and determining the starting time point of the patient to suspend breathing ; In detail, it is necessary to determine the demodulated slow-varying complex response sequences for the low-frequency and high-frequency operating bands respectively. These two sequences have been stripped of respiratory and cardiac interferences and contain only slow-varying signals related to pulmonary edema. At the same time, the onset time of respiratory arrest in the patient to be monitored needs to be determined. At this time, since respiratory movements have stopped and the chest wall is stable, it can be used as a reference point to eliminate fixed signal biases caused by individual physiological structures.

[0068] Construct the complex difference at time t ,include: in, Indicates time The demodulation slow-varying complex response of the high-frequency operating band. Indicates time The demodulation of the slow-varying complex response in the low-frequency operating band. Indicates the start time The demodulation slow-varying complex response of the high-frequency operating band. Indicates the start time The demodulation of the slow-varying complex response in the low-frequency operating band; In detail, complex difference is used to highlight the differences in high and low frequency signals caused by changes in lung water. The calculation formula is as follows: This operation, through baseline calibration, eliminates the influence of inherent individual differences (such as chest wall thickness, initial lung condition, etc.), retaining only the lung water-related signal components that change over time, so that the difference results directly reflect the differences in electromagnetic properties caused by changes in lung water.

[0069] For complex differences Perform complex factorization to obtain a two-dimensional vector. ;in, Representing complex difference The real part, Representing complex difference The imaginary part; In detail, the complex difference is decomposed into real and imaginary parts, forming a two-dimensional vector. The real part reflects the change in signal amplitude (attenuation), and the imaginary part reflects the change in signal phase. Together, they constitute the multidimensional characteristics of lung water changes. A mean vector is calculated from the two-dimensional vectors at N time points. Then, the mean vector is subtracted from the two-dimensional vector at each time point to obtain a zero-mean sequence. Mean removal eliminates the overall data bias, causing the sequence to fluctuate around zero, thus more accurately reflecting the relative changes in lung water.

[0070] Based on the mean vector of the two-dimensional vectors at N time points The mean is removed from the two-dimensional vector at each time step to obtain the zero-mean sequence. The two-dimensional covariance matrix is ​​estimated by zero-meaning sequence. ; In detail, the two-dimensional covariance matrix of the zero-mean sequence is obtained by accumulating the product of each zero-mean vector and its transpose and taking the average, which reflects the correlation of the real and imaginary parts and the respective fluctuation intensity.

[0071] The eigenvalue decomposition process is performed on the two-dimensional covariance matrix to obtain the maximum eigenvalue and determine the unit eigenvector corresponding to the maximum eigenvalue In detail, the eigenvalue decomposition is performed on the covariance matrix, and the unit eigenvector corresponding to the maximum eigenvalue represents the direction of the most significant change in the two-dimensional vector, which concentrates the main signal fluctuation information caused by the change of lung water.

[0072] The lung water index at time t is calculated .

[0073] It should be noted that the lung water index at time t is obtained by taking the inner product of the zero-mean two-dimensional vector at that time and the unit eigenvector corresponding to the maximum eigenvalue. Thus, the two-dimensional feature is condensed into a one-dimensional index, which not only retains the main information of the change of lung water, but also realizes quantitative representation. When the lung water increases or decreases, the index will present a corresponding monotonic change, thereby directly reflecting the dynamic change trend of the lung water.

[0074] It should be noted that the lung water index is only used to assist the doctor in judging the patient's condition, and is not directly used to diagnose the patient's condition.

[0075] Embodiment two:

[0076] As shown in Figure 2 , a multi-probe radio frequency distributed lung water monitoring system applied to any one of the multi-probe radio frequency distributed lung water monitoring methods, comprising: A logarithmic parameter acquisition module is configured to arrange a radio frequency array around the thoracic cage, acquire transmission parameters in a low-frequency working frequency band, a high-frequency working frequency band and a guard frequency working frequency band, and convert the transmission parameters into a logarithmic domain to form logarithmic transmission parameters. A rib arch scanning module is configured to perform a shallow high-frequency spatial spectrum scan based on the guard frequency working frequency band, and obtain a rib arch direction unit vector and a combined pitch of the ribs and the intercostal space by fitting a quasi-periodic lattice model of the ribs and the intercostal space. A wavelength determination module is configured to determine the effective wavelength of the tissue corresponding to the low-frequency working frequency band and the high-frequency working frequency band. A standard complex response module is configured to determine a window consistency weight according to the rib arch direction unit vector, the combined pitch of the ribs and the intercostal space, and the effective wavelength of the tissue, and perform weighted summation and normalization processing on the logarithmic transmission parameters with the window consistency weight to obtain a standard complex response. ​The demodulation response module is configured to perform a power spectrum analysis on the standard complex response, extract a respiratory angular frequency and a cardiac angular frequency, and perform a linear orthogonal projection on the standard complex response with a direct current component and sine and cosine functions of the respiratory angular frequency and the cardiac angular frequency as base vectors to obtain a demodulated slow-varying complex response; The lung water index module is configured to take an initial breath-holding time as a baseline, construct a complex difference of the demodulated slow-varying complex responses in the low-frequency working frequency band and the high-frequency working frequency band, take real and imaginary parts of the complex difference as a two-dimensional vector, and generate the lung water index according to projection of the two-dimensional vector.

[0077] The above describes the embodiments of the present embodiment, but the present embodiment is not limited to the specific embodiments described above, and the specific embodiments described above are only illustrative but not restrictive. Those skilled in the art can make many forms under the inspiration of the present embodiment, which are all within the protection scope of the present embodiment.

Claims

1. A multi-probe radio frequency distributed lung water monitoring method, characterized in that, The method comprises the following steps: An RF array is arranged around the thoracic cage, transmission parameters are collected in a low-frequency operating frequency band, a high-frequency operating frequency band and a guard operating frequency band, and the transmission parameters are converted into a logarithmic domain to form logarithmic transmission parameters; A shallow high-frequency spatial spectrum scan is performed based on the guard operating frequency band, a quasi-periodic lattice model of the ribs and intercostal spaces is fitted, a rib arch direction unit vector and a combined pitch of the ribs and intercostal spaces are obtained; Effective wavelengths in tissues corresponding to the low-frequency operating frequency band and the high-frequency operating frequency band are determined; A window consistency weight is determined according to the rib arch direction unit vector, the combined pitch of the ribs and intercostal spaces and the effective wavelengths in tissues, and the logarithmic transmission parameters are weighted and summed and normalized with the window consistency weight to obtain a standard complex response; Self-spectrum analysis is performed on the standard complex response, a respiratory angular frequency and a heart motion angular frequency are extracted, and a demodulation slow-varying complex response is obtained by linear orthogonal projection with the sine and cosine functions of the respiratory angular frequency and the heart motion angular frequency and a direct current component as base vectors; A complex difference of the demodulation slow-varying complex responses of the low-frequency operating frequency band and the high-frequency operating frequency band is constructed with an initial breath-holding time as a baseline, the real part and the imaginary part of the complex difference are taken as a two-dimensional vector, and a lung water index is generated by two-dimensional vector projection.

2. A multi-probe RF distributed lung water monitoring method according to claim 1, wherein, Transmission parameters are collected in a low-frequency operating frequency band, a high-frequency operating frequency band and a guard operating frequency band, and the transmission parameters are converted into a logarithmic domain to form logarithmic transmission parameters, which comprises the following steps: The low-frequency operating frequency band is 0.6 GHz to 0.8 GHz, the high-frequency operating frequency band is 1.2 GHz to 1.5 GHz, and the guard operating frequency band is 2.0 GHz to 2.5 GHz; The amplitude and phase of the transmission parameter at time t are collected at fixed time intervals The frequency band represents a low-frequency operating frequency band, a high-frequency operating frequency band, or a guard-band operating frequency band.​​​ The transmission parameters are converted into a logarithmic domain according to the amplitudes and phases of the transmission parameters, as follows: wherein represents a logarithmic transmission parameter, represents the imaginary unit, .

3. A multi-probe RF distributed lung water monitoring method according to claim 2, wherein, A shallow high-frequency spatial spectrum scan is performed based on the guard operating frequency band, a quasi-periodic lattice model of the ribs and intercostal spaces is fitted, a rib arch direction unit vector and a combined pitch of the ribs and intercostal spaces are obtained, which comprises the following steps: In a first preset time period, logarithmic transmission parameters of each frequency in the guard operating frequency band are determined at a fixed time interval, and a logarithmic transmission parameter variance corresponding to the frequency is calculated; the frequency corresponding to the maximum logarithmic transmission parameter variance is taken as a scanning frequency point; In the second preset time period, the M scanning probes are controlled to collect the logarithmic transmission parameters according to the scanning frequency points, and the average value of each logarithmic transmission parameter is calculated ; A one-dimensional mapping is set to determine a one-dimensional coordinate of the mth scanning probe ; wherein, ; wherein, represents an origin coordinate of a one-dimensional coordinate system, represents a coordinate of the mth scanning probe, represents an initial rib arch direction unit vector, represents an included angle of the one-dimensional mapping with a horizontal direction; The logarithmic transmission parameters of each scanning probe are fitted based on the average value and one-dimensional coordinates, as follows: wherein, represents a constant term of the quasi-periodic lattice model, represents an amplitude coefficient of the quasi-periodic lattice model, represents a phase coefficient of the quasi-periodic lattice model, represents a combined pitch of the initial rib and intercostal. The explained variance of the quasi-periodic lattice model corresponding to the initial rib arch direction unit vector and the combined pitch of the ribs and intercostal spaces is calculated, as follows: wherein, represents a fitting value of the quasi-periodic lattice model to the average of the logarithmic transmission parameters of the mth scanning probe, represents the average of the logarithmic transmission parameters of all scanning probes within the second preset time period; corresponding initial rib and intercostal synthetic pitch and initial rib arch orientation unit vectors as the rib and intercostal synthetic pitch and rib arch orientation unit vectors along which to maximize explained variance.

4. A multi-probe RF distributed lung water monitoring method according to claim 3, wherein, Effective wavelengths in tissues corresponding to the low-frequency operating frequency band and the high-frequency operating frequency band are determined, which comprises the following steps: determining a separation distance of the transmission channel of the mth scan probe to the transmission probe ;​ controlling the patient to be monitored to suspend breathing for at least 10 seconds, collecting and transmitting the channel at fixed time intervals corresponding phase, to calculate the average phase, and to perform phase unwrapping processing on the average phase to obtain the corresponding unwrapped phase ; Determining transmission channel The angle between the rib arch direction unit vector And the angle Calculate the corresponding direction smoothing weight ; The unwrapped phase and the baseline length are fitted by weighted least squares, as follows: wherein, represents the slope to be fitted; The corresponding to-be-fitted slope is taken as a target slope by minimizing the weighted least squares fitting value; The effective wavelengths in tissues are calculated according to the target slope, which comprises the following steps: wherein, represents an effective wavelength within the tissue, belongs to a low-frequency operating band or a high-frequency operating band, represents a target slope.

5. A multi-probe RF distributed lung water monitoring method according to claim 4, wherein, A window consistency weight is determined according to the rib arch direction unit vector, the combined pitch of the ribs and intercostal spaces and the effective wavelengths in tissues, which comprises the following steps: The rib and the intercostal synthetic pitch are projected to the angle between the transmission channel and the rib arch unit vector, and the equivalent pitch of the transmission channel is calculated ; Based on the equivalent pitch and the effective wavelengths in tissues, a matching residual corresponding to the window consistency weight is calculated, as follows: wherein denotes the projection magnification, denotes the matching residual of the transmission channel; extracting the matching residual of each transmission channel to determine a matching residual median; multiplying the matching residual median by a normal distribution constant to obtain a frequency corresponding robust scale ; The window consistency weight of the transmission channel at a frequency f is calculated according to the robust scale and the matching residual: wherein, represents a transmission channel window consistency weight at frequency f.

6. A multi-probe RF distributed lung water monitoring method according to claim 5, wherein, The logarithmic transmission parameters are weighted and summed and normalized with the window consistency weight to obtain a standard complex response, which comprises the following steps: determining a window consistency weight of each logarithmic transmission parameter, calculating a product of the logarithmic transmission parameter and the corresponding window consistency weight as a weighted complex composite variable; accumulating the window consistency weights of all transmission channels to obtain a window composite weight; taking a ratio of the weighted complex composite variable and the window composite weight as a standard complex response.

7. A multi-probe RF distributed lung water monitoring method according to claim 6, wherein, performing a spectral analysis on the standard complex response to extract a respiratory angular frequency and a cardiac angular frequency, including: performing a discrete Fourier transform on the standard complex response to obtain an estimated spectrum; calculating a combined composite weight of the window composite weight corresponding to the low-frequency working frequency band and the window composite weight corresponding to the high-frequency working frequency band; determining a working frequency band corresponding to the standard complex response, and taking a ratio of the window composite weight corresponding to the working frequency band and the combined composite weight as an inter-frequency combination weight coefficient; taking a product of the inter-frequency combination weight coefficient and the estimated spectrum as a joint spectrum; Extracting the maximum value from the joint autospectrum and minimum value And calculate the respiratory angular frequency. and cardiac angular frequency .

8. A multi-probe RF distributed lung water monitoring method according to claim 7, wherein, performing a linear orthogonal projection on the respiratory angular frequency and the cardiac angular frequency and a direct current component as a basis vector to obtain a demodulated slow-varying complex response, including: Constructing a negative response sequence from N standard complex responses ; Constructing a base matrix As follows: Constructing a projection sequence by a basis matrix ; The projection sequence is mapped to a sequence of de-modulated slowly varying complex impulse responses ; wherein, .

9. A multi-probe RF distributed lung water monitoring method according to claim 8, wherein, taking an initial breath-holding time as a baseline to construct a complex difference of the demodulated slow-varying complex responses of the low-frequency working frequency band and the high-frequency working frequency band, taking a real part and an imaginary part of the complex difference as a two-dimensional vector, and generating a lung water index by projecting the two-dimensional vector, including: determining a demodulated slow varying complex response sequence for a low frequency operating band and a demodulated slow varying complex response sequence for a high frequency operating band, respectively, and a start instant of a patient's breath-hold to be monitored ; constructing the complex difference at the tth time instant , comprising: in, Indicates time The demodulation of the slow-varying complex response in the high-frequency operating band. Indicates time The demodulation of the slow-varying complex response in the low-frequency operating band. Indicates the start time The demodulation of the slow-varying complex response in the high-frequency operating band. Indicates the start time The demodulation of the slow-varying complex response in the low-frequency operating band; complex difference performing complex resolution to obtain a two-dimensional vector ; wherein denotes the real part of the complex difference denotes the imaginary part of the complex difference denotes the imaginary part of the complex difference denotes the imaginary part of the complex difference According to the mean vector of the two-dimensional vectors at N time instants The two-dimensional vectors at each time instant are de-meaned to obtain a zero-mean sequence And a two-dimensional covariance matrix is estimated from the zero-mean sequence ; Eigenvalue decomposition is performed on the two-dimensional covariance matrix to obtain a maximum eigenvalue and determine a unit eigenvector corresponding to the maximum eigenvalue ; The lung water index at time t is calculated .

10. A multi-probe RF distributed lung water monitoring system, applied to the multi-probe RF distributed lung water monitoring method of any one of claims 1-9, characterized in that, including: a logarithmic parameter sampling module, configured to arrange a radio frequency array around a thorax, collect transmission parameters in a low-frequency working frequency band, a high-frequency working frequency band and a guard frequency working frequency band, and convert the transmission parameters into a logarithmic domain to form logarithmic transmission parameters; a rib arch scanning module, configured to perform a shallow high-frequency spatial spectrum scan based on the guard frequency working frequency band, fit a quasi-periodic lattice model of ribs and intercostal spaces, and obtain a rib arch trend unit vector and a combined pitch of the ribs and the intercostal spaces; a wavelength determination module, configured to determine effective wavelengths in tissues corresponding to the low-frequency working frequency band and the high-frequency working frequency band; a standard complex response module, configured to determine window consistency weights according to the rib arch trend unit vector, the combined pitch of the ribs and the intercostal spaces and the effective wavelengths in tissues, and perform weighted summation and normalization processing on the logarithmic transmission parameters with the window consistency weights to obtain a standard complex response; a demodulated response module, configured to perform a spectral analysis on the standard complex response to extract a respiratory angular frequency and a cardiac angular frequency, and perform a linear orthogonal projection on the respiratory angular frequency and the cardiac angular frequency and a direct current component as a basis vector to obtain a demodulated slow-varying complex response; a lung water index module, configured to take an initial breath-holding time as a baseline to construct a complex difference of the demodulated slow-varying complex responses of the low-frequency working frequency band and the high-frequency working frequency band, take a real part and an imaginary part of the complex difference as a two-dimensional vector, and generate a lung water index by projecting the two-dimensional vector.

Images