A signal processing method based on a matrix flexible B-ultrasound patch and an electric stimulator

Abstract

The application provides a signal processing method based on a matrix flexible B-ultrasound patch and an electric stimulator, in the method, a three-dimensional displacement vector of the whole relative to the skin is calculated based on skin relative displacement estimation values of each channel; when a preset displacement threshold is exceeded, a current spatial displacement field feature formed by the skin relative displacement estimation values of each channel is used to switch a target element from a blood flow detection mode to a motion reference mode to collect a motion reference signal; the motion reference signal is used as a reference input, an adaptive filter is used to perform adaptive noise cancellation processing on ultrasonic echo signals of the remaining elements, and a blood flow ultrasonic signal after motion artifacts are eliminated is obtained; by using array spatial difference of the matrix flexible B-ultrasound patch, precise modeling and adaptive elimination of the motion artifacts are realized without an additional motion sensor, and the strategy accuracy and robustness of the blood velocity are improved.

Description

Technical Field

[0001] This invention belongs to the field of medical ultrasound detection technology, specifically relating to a signal processing method and an electrical stimulator based on a matrix-type flexible B-mode ultrasound patch. Background Technology

[0002] Venous thromboembolism (VTE) is a common clinical complication caused by abnormal blood clotting in veins, leading to vascular obstruction. It is particularly prevalent among patients who are bedridden for extended periods, recovering from surgery, or have limited mobility. In recent years, continuous monitoring of lower extremity venous blood flow using wearable medical devices has become an important means of preventing VTE.

[0003] However, current solutions primarily assume an ideal static fit between the device and the patient's skin, failing to account for relative displacement that occurs during patient activities such as walking, turning over, or muscle contractions. In practical use, this relative displacement leads to significant motion artifacts superimposed on the ultrasound echo signal, severely interfering with the accurate measurement of blood flow velocity. Summary of the Invention

[0004] To overcome the shortcomings of the prior art, this invention proposes a signal processing method based on a matrix-type flexible ultrasound patch, wherein the matrix-type flexible ultrasound patch comprises multiple flexible ultrasound transducer array elements arranged in an M-row × N-column configuration, and the method includes: Based on the multi-channel ultrasound echo signals received through multiple array elements, the relative skin displacement estimate of each channel is calculated using the cross-correlation function between adjacent acquisition frames. Based on the skin relative displacement estimates of each channel, the three-dimensional displacement vector of the matrix flexible B-mode ultrasound patch relative to the skin is calculated using a weighted fusion algorithm of array element spatial coordinates. When the magnitude of the three-dimensional displacement vector exceeds the preset displacement threshold, a target element is selected in the matrix flexible B-mode ultrasound patch based on the current spatial displacement field characteristics formed by the skin relative displacement estimates of each channel; the target element is switched from blood flow detection mode to motion reference mode, so that the target element collects skin interface reflection signals as motion reference signals with a fixed emission frequency and a fixed focusing depth; Using the motion reference signal as a reference input, an adaptive filter is used to perform adaptive noise cancellation processing on the ultrasound echo signals of the remaining array elements that maintain the blood flow detection mode, so as to obtain the blood flow ultrasound signal after motion artifact elimination; the weight vector of the adaptive filter is determined based on the weight vector that matches the current spatial displacement field characteristics.

[0005] Preferably, the step of calculating the skin relative displacement estimate for each channel based on the multi-channel ultrasound echo signals received through multiple array elements and utilizing the cross-correlation function between adjacent acquisition frames includes: Based on each of the channels, select the radio frequency echo signal segment of the adjacent acquisition frame in the ultrasonic echo signal of the channel; based on the radio frequency echo signal segment of the adjacent acquisition frame, calculate the normalized cross-correlation function value within a preset time delay search range to generate a normalized cross-correlation sequence; use the time delay value corresponding to the peak position of the normalized cross-correlation sequence as the estimated value of the inter-frame axial displacement of the channel. Based on the inter-frame axial displacement estimates of each channel, the skin relative displacement estimates of each channel are determined.

[0006] Preferably, the step of calculating the three-dimensional displacement vector of the matrix-type flexible ultrasound patch relative to the skin based on the skin relative displacement estimates of each of the channels and using a weighted fusion algorithm of array element spatial coordinates includes: Using the two-dimensional coordinates of the corresponding array elements of each channel in the plane of the matrix-type flexible B-mode ultrasound patch as position input, and the estimated skin relative displacement of each channel as observation value, a fusion observation equation set is established. The weighted least squares method is used to solve the fusion observation equations to obtain the in-plane translation components and rotational components about the plane normal of the matrix flexible B-mode ultrasound patch as a whole. Based on the translation and rotation components in the plane, and combined with the spatial coordinates of the corresponding array elements of each channel, the three-dimensional displacement components at the position of each array element are calculated; the three-dimensional displacement components at the position of each array element are weighted and averaged to obtain the three-dimensional displacement vector of the matrix flexible B-mode ultrasound patch relative to the skin.

[0007] Preferably, the current spatial displacement field features include the divergence features and curl features of the current spatial displacement field; wherein, the divergence features reflect the degree of stretching or lifting between the patch and the skin, and the curl features reflect the degree of twisting of the patch relative to the skin; The current spatial displacement field features, based on the estimated skin relative displacement values ​​of each channel, are used to select target array elements in the matrix-type flexible B-mode ultrasound patch, including: Based on the divergence and curl characteristics of the current spatial displacement field, the array element in the channel region with the largest curl characteristic and / or divergence characteristic value is selected as the target array element. The channel region is composed of several adjacent array elements.

[0008] Preferably, the process of determining the weight vector of the adaptive filter includes: Obtain a historical optimal vector library, which is stored according to the category of spatial displacement field features, and each category of spatial displacement field features corresponds to a set of historical optimal weight vectors. Calculate the similarity between the current spatial displacement field feature and the spatial displacement field features of each category in the historical optimal vector library; take the historical optimal weight vector corresponding to the spatial displacement field of the category with the highest similarity as the initial weight vector of the adaptive filter; Based on the current operating status parameters, the weight vector update strategy of the adaptive filter is determined; the current operating status parameters include remaining battery power and / or operating temperature. The initial weight vector is iteratively updated using the weight vector update strategy to obtain the final weight vector of the adaptive filter.

[0009] Preferably, determining the weight vector update strategy of the adaptive filter based on the current operating state parameters includes: When the remaining power is lower than the first power threshold or the operating temperature is higher than the first temperature threshold, the weight vector update strategy of the adaptive filter includes iterative updating of the weight vector with a preset first update step size and / or a downsampled motion reference signal. When the remaining power is higher than the second power threshold and the operating temperature is lower than the second temperature threshold, the weight vector update strategy of the adaptive filter includes iterative updating of the weight vector with a motion reference signal of a second update step size and an original sampling rate, wherein the second power threshold is higher than the first power threshold and the second temperature threshold is lower than the first temperature threshold. When the remaining power is lower than the third power threshold, the weight vector update strategy of the adaptive filter includes freezing some weight vector elements in the initial weight vector whose correlation with the current spatial displacement field feature is lower than a preset correlation threshold, and iteratively updating the remaining weight vector elements in the initial weight vector that have not been frozen.

[0010] Preferably, after switching the target array element from blood flow detection mode to motion reference mode, the method further includes: While the other array elements besides the target array element maintain the blood flow detection mode, the expected displacement position of the target blood vessel is determined based on the three-dimensional displacement vector. The expected displacement position is tracked by the focal point of the transmitted beam as a reference, and the deflection correction of the transmitted beam is performed on the other array elements.

[0011] Preferably, after calculating the three-dimensional displacement vector of the matrix flexible ultrasound patch relative to the skin based on the skin relative displacement estimates of each of the channels using a weighted fusion algorithm of array element spatial positions, the method further includes: When the magnitude of the three-dimensional displacement vector does not exceed the preset displacement threshold, each array element is kept in blood flow detection mode, and conventional beamforming is performed on the ultrasound echo signal of each channel to obtain the blood flow velocity spectrum.

[0012] Based on the same inventive concept, the present invention also provides a thrombosis prevention electrical stimulator, comprising: an electrical stimulator assembly, an electrode assembly, and an ultrasound assembly; The electrical stimulator assembly includes a control module and an electrode connector; one end of the electrode connector is connected to the control module, and the other end is connected to the electrode assembly. The ultrasound component includes a matrix-type flexible ultrasound patch, which includes multiple flexible ultrasound transducer elements arranged in M ​​rows × N columns; the multiple flexible ultrasound transducer elements are connected to the electrode assembly. The electrode pad assembly is used to conform to the user's skin when the thrombosis prevention electrical stimulator is worn; The control module is used to implement the signal processing method based on matrix flexible ultrasound patch described above.

[0013] Based on the same inventive concept, this invention also provides a signal processing system based on a matrix-type flexible ultrasound patch, wherein the matrix-type flexible ultrasound patch includes multiple flexible ultrasound transducer array elements arranged in an M row × N column configuration, comprising: The displacement estimation module is used to calculate the estimated skin relative displacement of each channel based on the multi-channel ultrasound echo signals received through multiple array elements and by using the cross-correlation function between adjacent acquisition frames. The displacement fusion module is used to calculate the three-dimensional displacement vector of the matrix flexible B-mode ultrasound patch relative to the skin based on the skin relative displacement estimates of each channel and using a weighted fusion algorithm of array element spatial coordinates. The mode switching module is used to select a target element in the matrix flexible B-ultrasound patch when the magnitude of the three-dimensional displacement vector exceeds a preset displacement threshold, based on the current spatial displacement field characteristics composed of the skin relative displacement estimates of each channel; and to switch the target element from blood flow detection mode to motion reference mode, so that the target element collects skin interface reflection signals as motion reference signals with a fixed emission frequency and a fixed focusing depth. An adaptive cancellation module is used to perform adaptive noise cancellation processing on the ultrasound echo signals of the remaining array elements that maintain the blood flow detection mode using the motion reference signal as a reference input and an adaptive filter, so as to obtain a blood flow ultrasound signal after motion artifact elimination; the weight vector of the adaptive filter is determined based on a weight vector that matches the characteristics of the current spatial displacement field.

[0014] Preferably, the displacement estimation module is specifically used for: Based on each of the channels, select the radio frequency echo signal segment of the adjacent acquisition frame in the ultrasonic echo signal of the channel; based on the radio frequency echo signal segment of the adjacent acquisition frame, calculate the normalized cross-correlation function value within a preset time delay search range to generate a normalized cross-correlation sequence; use the time delay value corresponding to the peak position of the normalized cross-correlation sequence as the estimated value of the inter-frame axial displacement of the channel. Based on the inter-frame axial displacement estimates of each channel, the skin relative displacement estimates of each channel are determined.

[0015] Preferably, the displacement fusion module is specifically used for: Using the two-dimensional coordinates of the corresponding array elements of each channel in the plane of the matrix-type flexible B-mode ultrasound patch as position input, and the estimated skin relative displacement of each channel as observation value, a fusion observation equation set is established. The weighted least squares method is used to solve the fusion observation equations to obtain the in-plane translation components and rotational components about the plane normal of the matrix flexible B-mode ultrasound patch as a whole. Based on the translation and rotation components in the plane, and combined with the spatial coordinates of the corresponding array elements of each channel, the three-dimensional displacement components at the position of each array element are calculated; the three-dimensional displacement components at the position of each array element are weighted and averaged to obtain the three-dimensional displacement vector of the matrix flexible B-mode ultrasound patch relative to the skin.

[0016] Preferably, the current spatial displacement field features include the divergence features and curl features of the current spatial displacement field; wherein, the divergence features reflect the degree of stretching or lifting between the patch and the skin, and the curl features reflect the degree of twisting of the patch relative to the skin; The mode switching module is specifically used for: Based on the divergence and curl characteristics of the current spatial displacement field, the array element in the channel region with the largest curl characteristic and / or divergence characteristic value is selected as the target array element. The channel region is composed of several adjacent array elements.

[0017] Preferably, the adaptive cancellation module is also used for: Obtain a historical optimal vector library, which is stored according to the category of spatial displacement field features, and each category of spatial displacement field features corresponds to a set of historical optimal weight vectors. Calculate the similarity between the current spatial displacement field feature and the spatial displacement field features of each category in the historical optimal vector library; take the historical optimal weight vector corresponding to the spatial displacement field of the category with the highest similarity as the initial weight vector of the adaptive filter; Based on the current operating status parameters, the weight vector update strategy of the adaptive filter is determined; the current operating status parameters include remaining battery power and / or operating temperature. The initial weight vector is iteratively updated using the weight vector update strategy to obtain the final weight vector of the adaptive filter.

[0018] Preferably, the adaptive cancellation module is specifically used for: When the remaining power is lower than the first power threshold or the operating temperature is higher than the first temperature threshold, the weight vector update strategy of the adaptive filter includes iterative updating of the weight vector with a preset first update step size and / or a downsampled motion reference signal. When the remaining power is higher than the second power threshold and the operating temperature is lower than the second temperature threshold, the weight vector update strategy of the adaptive filter includes iterative updating of the weight vector with a motion reference signal of a second update step size and an original sampling rate, wherein the second power threshold is higher than the first power threshold and the second temperature threshold is lower than the first temperature threshold. When the remaining power is lower than the third power threshold, the weight vector update strategy of the adaptive filter includes freezing some weight vector elements in the initial weight vector whose correlation with the current spatial displacement field feature is lower than a preset correlation threshold, and iteratively updating the remaining weight vector elements in the initial weight vector that have not been frozen.

[0019] Preferably, the mode switching module is further configured to: While the other array elements besides the target array element maintain the blood flow detection mode, the expected displacement position of the target blood vessel is determined based on the three-dimensional displacement vector. The expected displacement position is tracked by the focal point of the transmitted beam as a reference, and the deflection correction of the transmitted beam is performed on the other array elements.

[0020] Preferably, it also includes a beamforming module for: When the magnitude of the three-dimensional displacement vector does not exceed the preset displacement threshold, each array element is kept in blood flow detection mode, and conventional beamforming is performed on the ultrasound echo signal of each channel to obtain the blood flow velocity spectrum.

[0021] Based on the same inventive concept, the present invention also provides a computer device, comprising: one or more processors; Memory, used to store one or more programs; When the one or more programs are executed by the one or more processors, a signal processing method based on a matrix-type flexible ultrasound patch, as described above, is implemented.

[0022] Based on the same inventive concept, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed, it implements the signal processing method based on a matrix-type flexible B-mode ultrasound patch as described above.

[0023] Compared with the closest existing technology, the present invention has the following beneficial effects: This invention provides a signal processing method and system based on a matrix-type flexible ultrasound patch. The method involves calculating the relative skin displacement estimate for each channel based on multi-channel ultrasound echo signals received through multiple array elements, using the cross-correlation function between adjacent acquisition frames. Based on these estimates, a weighted fusion algorithm of array element spatial coordinates is used to calculate the overall three-dimensional displacement vector of the matrix-type flexible ultrasound patch relative to the skin. When the magnitude of the three-dimensional displacement vector exceeds a preset displacement threshold, a target array element is selected within the matrix-type flexible ultrasound patch based on the current spatial displacement field characteristics formed by the relative skin displacement estimates of each channel. The target array element is switched from blood flow detection mode to motion reference mode, allowing it to acquire skin interface reflection signals at a fixed transmission frequency and fixed focusing depth as motion reference signals. Using the motion reference signal as a reference input, an adaptive filter is used to adaptively cancel the ultrasound echo signals of the remaining array elements maintaining the blood flow detection mode, obtaining a blood flow ultrasound signal with motion artifacts eliminated. The weight vector of the adaptive filter is determined based on a weight vector matching the current spatial displacement field characteristics. This method utilizes the spatial differences in the array of a matrix-type flexible ultrasound patch to achieve real-time estimation of the three-dimensional relative displacement between the patch and the skin, without the need for additional motion sensors, solely based on the cross-correlation analysis of the multi-channel ultrasound echo signals. When significant displacement is detected, some array elements are switched from blood flow detection mode to motion reference mode. The skin interface reflection signal is collected with fixed parameters as the reference input for adaptive filtering, achieving accurate modeling and adaptive elimination of motion artifacts and improving the accuracy of blood velocity measurement.

[0024] This invention also provides a thrombosis prevention electrical stimulator, comprising: an electrical stimulator assembly, an electrode pad assembly, and an ultrasound assembly; the electrical stimulator assembly includes a control module and an electrode connector; one end of the electrode connector is connected to the control module, and the other end is connected to the electrode pad assembly; the ultrasound assembly includes a matrix-type flexible ultrasound patch, which includes multiple flexible ultrasound transducer elements arranged in M ​​rows × N columns; the multiple flexible ultrasound transducer elements are connected to the electrode pad assembly; the electrode pad assembly is used to conform to the user's skin when the thrombosis prevention electrical stimulator is worn; the control module is used to implement the above-mentioned signal processing method based on the matrix-type flexible ultrasound patch. This invention deeply integrates the matrix-type flexible ultrasound patch with the thrombosis prevention electrical stimulator, achieving continuous monitoring of blood flow parameters while performing electrical stimulation treatment. This solves the technical problem of motion artifacts interfering with blood flow detection due to patient turning over or walking during wear, effectively improving the accuracy and personalization of thrombosis prevention. Attached Figure Description

[0025] Figure 1A general flowchart of a signal processing method based on a matrix-type flexible ultrasound patch provided by the present invention; Figure 2 A schematic diagram of the array element arrangement of a matrix-type flexible B-mode ultrasound patch provided by the present invention; Figure 3 A schematic diagram of a signal processing system based on a matrix-type flexible ultrasound patch provided by the present invention; Figure 4 This is a schematic diagram of an electronic device structure provided by the present invention. Detailed Implementation

[0026] The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0027] Example 1: This invention provides a signal processing method based on a matrix-type flexible ultrasound patch. The matrix-type flexible ultrasound patch includes multiple flexible ultrasound transducer elements arranged in an M-row × N-column configuration. A flowchart of the method is shown below. Figure 1 As shown, it includes: Step 101: Based on the multi-channel ultrasound echo signals received through multiple array elements, calculate the estimated relative skin displacement of each channel using the cross-correlation function between adjacent acquisition frames.

[0028] Step 102: Based on the estimated relative displacement of the skin in each channel, the three-dimensional displacement vector of the matrix flexible B-mode ultrasound patch relative to the skin is calculated using a weighted fusion algorithm of array element spatial coordinates.

[0029] Step 103: When the magnitude of the three-dimensional displacement vector exceeds the preset displacement threshold, the target array element is selected in the matrix flexible B-mode ultrasound patch based on the current spatial displacement field characteristics composed of the skin relative displacement estimates of each channel; the target array element is switched from blood flow detection mode to motion reference mode, so that the target array element collects the skin interface reflection signal as the motion reference signal with a fixed transmission frequency and a fixed focusing depth.

[0030] Step 104: Using the motion reference signal as the reference input, the ultrasound echo signal of the remaining array elements maintaining the blood flow detection mode is adaptively noise-cancelled using an adaptive filter to obtain the blood flow ultrasound signal after motion artifact elimination; the weight vector of the adaptive filter is determined based on the weight vector matching the current spatial displacement field characteristics.

[0031] In this embodiment of the invention, by utilizing the spatial differences in the array of a matrix-type flexible ultrasound patch, real-time estimation of the three-dimensional relative displacement between the patch and the skin can be achieved solely based on the cross-correlation analysis of the multi-channel ultrasound echo signals, without the need for additional motion sensors. When a significant displacement is detected, some array elements are switched from blood flow detection mode to motion reference mode. The skin interface reflection signal is collected with fixed parameters as the reference input for adaptive filtering, thereby achieving accurate modeling and adaptive elimination of motion artifacts and improving the accuracy of blood velocity measurement.

[0032] This invention provides a signal processing method based on a matrix-type flexible ultrasound patch that enables adaptive beamforming and motion artifact elimination. This method can be applied to thrombosis prevention electrical stimulators that monitor venous blood flow velocity using flexible ultrasound technology. Specific details regarding thrombosis prevention electrical stimulators can be found in subsequent embodiments.

[0033] A matrix-type flexible ultrasound patch structure as follows Figure 2 As shown, the array includes multiple flexible ultrasound transducer elements (hereinafter referred to as elements) arranged in 2 (M=2) rows × 4 (N=4) columns. In actual scenarios, the specific values ​​of M and N can be determined based on monitoring needs and the deployable volume of the thrombosis prevention electrical stimulator, among other factors. No restrictions are imposed here. Each element in the matrix-type flexible ultrasound patch can independently transmit and receive ultrasound pulses. In normal blood flow monitoring mode, all M×N elements work synchronously with preset blood flow detection parameters (e.g., but not limited to a transmission frequency of 3.0 to 7.5 MHz and a focusing depth of 2.0 to 6.0 cm). Each element continuously acquires ultrasound echo signals at a frame rate F (e.g., but not limited to 20 to 50 frames / second), forming a multi-channel radio frequency echo data stream.

[0034] In this embodiment of the invention, one array element can correspond to one channel, that is, a fully parallel transceiver architecture is adopted. At this time, the array element, channel and ultrasonic echo signal correspond one-to-one to ensure the accuracy of subsequent displacement field estimation to the greatest extent.

[0035] Conventional displacement tracking methods often use demodulated envelope signals or amplitude image data for correlation analysis. However, the phase information of the envelope signal is discarded, and the displacement estimation accuracy achievable is fundamentally limited by signal bandwidth and sampling rate. Especially in wearable medical monitoring scenarios, the relative displacement between the patch and skin caused by patient activity is often a sub-millimeter level micro-movement, and envelope-level estimation cannot meet the accuracy requirements of blood flow detection. Therefore, in step 101 of this embodiment, the displacement estimation of each array element channel directly uses the radio frequency echo signal as the processing object to retain complete phase information, thereby achieving high-precision displacement tracking. In one implementation, step 101 includes: Based on each channel, select the radio frequency echo signal segment of the adjacent acquisition frame in the ultrasonic echo signal of that channel; based on the radio frequency echo signal segment of the adjacent acquisition frame, calculate the normalized cross-correlation function value within the preset time delay search range to generate a normalized cross-correlation sequence; use the time delay value corresponding to the peak position of the normalized cross-correlation sequence as the estimated value of the inter-frame axial displacement of that channel. Based on the inter-frame axial displacement estimates of each channel, the relative skin displacement estimates of each channel are determined.

[0036] When selecting the corresponding radio frequency echo signal segment for a channel, for the array element channel numbered (i, j), the system reads the radio frequency echo signals of the k-th and (k-1)-th frames from the frame buffer. In each frame of radio frequency echo signal, not all data across the entire depth range is used for cross-correlation calculation. Instead, based on a preset target depth for blood flow detection (e.g., the target vessel is located 2.0 cm to 4.0 cm subcutaneously), a continuous signal segment is extracted from the skin surface to the lower boundary of the target vessel. In this implementation, the signal segment length corresponds to the ultrasound propagation time window from the skin surface to the target vessel depth. When the patch shifts relative to the skin, the strong reflection signals from the skin interface and superficial tissues shift over time, and the deep blood flow signal also undergoes amplitude modulation due to changes in the relative position of the beam and the vessel. Therefore, containing the entire signal window from the skin surface to the target depth captures more complete tissue displacement information, resulting in higher robustness than tracking only a single interface reflection.

[0037] Taking the above example again, based on the radio frequency echo signal segments of adjacent acquisition frames, the normalized cross-correlation function value is calculated within a preset time delay search range. When generating the normalized cross-correlation sequence, the (k-1)th frame signal segment is used as a reference template, and it is offset point by point on the time axis relative to the kth frame signal segment. The normalized similarity between the two signal segments is calculated once for each offset of a sampling interval, and recorded as a normalized cross-correlation function value. After traversing the preset time delay search range, a discrete sequence (i.e., the normalized cross-correlation sequence) is obtained with time delay as the horizontal axis and normalized similarity as the vertical axis. The peak position of this sequence is the optimal matching time delay of the echo signal of this channel between two frames, which is used as the estimated value of the inter-frame axial displacement of this channel. In a typical scenario, the sampling rate of ultrasonic echo signals is usually as high as 40 MS / s or more, and the spatial distance between adjacent sampling points is on the order of 20 μm. Therefore, the spatial resolution of cross-correlation peak localization in this implementation method is much higher than the sampling interval accuracy of subsequent signal processing stages, avoiding the loss of phase information caused by envelope demodulation, and achieving subwavelength level displacement estimation accuracy without increasing the hardware sampling rate.

[0038] In one example, the normalized cross-correlation function value satisfies the following formula:

[0039] in, For the channel numbered (i, j), search within a preset time range The normalized cross-correlation function value within, , Let be the radio frequency echo signal segments of the k-th and (k-1)-th frames at time t.

[0040] The preset latency search range is limited by the physical constraints of the wearable scenario. Considering the maximum displacement velocity of the wearable device during human activity (e.g., the displacement velocity of the skin relative to the patch during rapid walking or turning over does not exceed approximately 2 m / s), and combined with the ultrasound frame rate (e.g., 20 to 50 frames / second), the maximum possible displacement between adjacent frames is constrained within a limited numerical range. Therefore, in one example, the preset latency search range can be set to the corresponding ±3 mm to ±5 mm range. This setting method can significantly reduce the computational load of related operations to meet the real-time, low-power processing requirements of the wearable device, while also eliminating spurious peak interference caused by noise, thus improving the reliability of peak detection.

[0041] To overcome the physical limitations of sampling rate on time delay resolution and achieve a finer displacement estimation accuracy than that of the sampling interval, in one example, when using the time delay value corresponding to the peak position of the normalized cross-correlation sequence as the inter-frame axial displacement estimate for that channel, after coarsely locating the peak position of the normalized cross-correlation sequence, parabolic or Gaussian interpolation is performed on the peak and several adjacent points. The continuous peak positions after interpolation are used as sub-sampling level time delay estimates. Multiplying the sub-sampling level time delay estimate by the ultrasound propagation velocity in soft tissue and dividing by two yields the inter-frame axial displacement estimate for that element channel, which is the axial displacement of the skin relative to the patch at that element position. The displacement estimation accuracy in this example can reach the order of several micrometers to tens of micrometers, far superior to the hundreds of micrometers achieved by envelope-level methods.

[0042] This implementation can be executed in parallel on all channels. Spatially, each channel independently provides sampling at its own position. Temporally, the dynamic changes in displacement are tracked through adjacent frames. By utilizing the spatial and temporal dimensions of the array, the spatiotemporal distribution field of the relative motion between the patch body and the skin can be constructed.

[0043] The inter-frame axial displacement estimates for each channel can reflect the one-dimensional displacement of the skin relative to the patch along the ultrasound propagation direction at each element position. However, the relative motion between the patch and the skin is essentially a three-dimensional spatial motion, including translation within the patch plane, rotation around the plane normal, and translation along the axial direction. Therefore, the axial displacement observations of a single element in step 101 cannot distinguish these motion components. For example, when the patch as a whole translates along a certain direction in the plane, elements at different positions will produce different projection components in the axial observation due to their different spatial coordinates; while when the patch rotates around the plane normal, the axial displacement amplitude of the edge elements is significantly greater than that of the center element. Therefore, in step 102, in order to accurately describe the overall motion state of the patch, a method of inferring three-dimensional motion from one-dimensional projection of spatial distribution is used. The spatial distribution information of all elements and their respective axial displacement observations are jointly solved to decouple the three-dimensional motion vector from the one-dimensional observation. In one implementation, step 102 includes: Using the two-dimensional coordinates of the corresponding array elements of each channel in the matrix-type flexible B-mode ultrasound patch as the position input, and the estimated relative skin displacement of each channel as the observation value, a set of fusion observation equations is established. The weighted least squares method was used to solve the fusion observation equations to obtain the in-plane translational components and the rotational components about the plane normal of the matrix flexible B-mode ultrasound patch. Based on the translation and rotation components in the plane, and combined with the spatial coordinates of the corresponding array elements in each channel, the three-dimensional displacement components at the position of each array element are calculated; the three-dimensional displacement components at the position of each array element are weighted and averaged to obtain the three-dimensional displacement vector of the matrix flexible B-mode ultrasound patch relative to the skin.

[0044] In one example, when establishing the fusion observation equations using the two-dimensional coordinates of each channel's corresponding array element within the matrix-type flexible ultrasound patch plane as position input and the estimated relative skin displacement of each channel as observation value, the following steps can be included: the two-dimensional coordinates of each array element within the patch plane are... Where i=1,…,M, j=1,…,N, the two-dimensional coordinate system in the patch plane is a local coordinate system established with the geometric center of the patch as the origin and the two orthogonal directions of the patch plane as axes; the estimated skin relative displacement values ​​of each channel are... As observed values, the two-dimensional coordinates of each channel's corresponding array element in the plane of the matrix-type flexible ultrasound patch refer to the inherent array arrangement coordinates of each array element when the matrix-type flexible ultrasound patch is in its unfolded state. When the patch deforms while flexibly adhering to the skin surface, the spatial coordinates of each array element change accordingly. Specifically, when the patch as a whole translates in the plane relative to the skin or rotates around the plane's normal, array elements at different spatial positions will produce different projection components in the axial observation values ​​due to their different coordinates. For example, when the patch translates along the x-direction in the plane, array elements located on the positive x-axis and those located on the negative x-axis show opposite displacement trends in the axial observation; when the patch rotates around the normal, the axial displacement amplitude of the edge array elements is significantly greater than that of the center array elements. Therefore, in this embodiment of the invention, this array spatial difference can be utilized to decouple four motion components—in-plane translation, rotation, and axial translation—from the estimated relative skin displacement values ​​(mainly axial displacement values) of each channel, thereby reconstructing a complete three-dimensional displacement vector.

[0045] In this embodiment of the invention, the motion of the patch relative to the skin is parameterized into four degrees of freedom, namely, the motion parameter vector includes the following four vectors: the translation component of the array element along the x-direction in the plane. Translation component along the y-direction in the plane The rotation component ω about the plane normal (this rotation causes a change in the position of the array element in the plane, thus changing its relative geometric relationship with the target tissue), and the axial translation component. For the (i, j)th element, its axial displacement observation value This can be represented as a linear combination of the above motion components: in-plane translation component. , The axial projection at the position of the array element is superimposed with the axial projection of the position change of the array element caused by the rotation component ω, and then superimposed with a uniform axial translation component. Finally, observation noise is added. Example The linear relationship between each motion vector satisfies the formula (hereinafter referred to as the observation equation): ,in, Respectively, two-dimensional coordinates The geometric coefficients related to the ultrasonic propagation direction represent the contribution coefficients of each motion component to the observed axial displacement, and are a known coefficient matrix. Specifically, This represents the rate of change of the axial displacement observation when the array element undergoes a unit translation in the x-direction. The ultrasonic propagation direction is approximately along the normal direction of the patch plane. The value of can be related to the deflection angle of the ultrasonic beam in the x-direction. Similarly, This represents the rate of change of the axial displacement observation value when the array element undergoes a unit translation in the y direction. The value can be related to the deflection angle of the ultrasonic beam in the y direction. This represents the rate of change of the axial observation value of the array element when the patch rotates one unit around the plane normal. The coefficient is 1, indicating that the axial translation vector has a uniform contribution to the axial observations of each array element.

[0046] By simultaneously solving the observation equations for all M×N array elements, the resulting fused observation equations satisfy the formula... Where d is the fusion observation equation set, and A represents the set of geometric coefficients. The set representing the motion components. Let the set of observed noise satisfy the following formula:

[0047] As can be seen from the above formula, there are three unknowns in the fused observation equations (actually...). , , The four unknowns ω, axial observations and Since the sensitivity depends on the angle between the ultrasonic propagation direction and the normal of the patch plane, appropriate parameterization is required in the actual solution based on the array geometry. The number of equations is M×N, which is usually much larger than the number of unknowns. Therefore, the fused observation equation set can be regarded as a redundant observation system.

[0048] When solving the fusion observation equations using the weighted least squares method to obtain the in-plane translational components and rotational components about the plane normal of the matrix-type flexible ultrasound patch, the principle of minimizing the weighted residual peak is followed, i.e., finding a set of motion parameters. This makes the observation values ​​of each array element... The goal is to minimize the weighted sum of squared errors between the estimated values ​​and the theoretical values ​​predicted by the motion parameters. Since this optimization problem has an analytical solution, the optimal estimates of the four motion components can be obtained in one step by constructing a weighted pseudo-inverse matrix, without iteration. This results in extremely high computational efficiency, making it suitable for the real-time processing requirements of wearable devices and also taking into account the different reliability levels of the displacement estimates from each channel. Specifically, in the solution process, the objective function of the weighted least squares method satisfies the following formula: ,in Here, represents the channel coefficients. The optimal estimates of the four motion components constructed using this objective function are: ,in This is the optimal estimate of the translation component along the x-direction. This represents the optimal estimate of the translation component along the y-direction. This is the optimal estimate of the rotation component about the plane normal. This represents the optimal estimate of the axial translation component. The weighted least squares fusion process fully utilizes the spatial redundancy information of the matrix array, adaptively adjusting the contribution weights of each channel in both signal-to-noise ratio and geometric sensitivity dimensions. Even in scenarios where the signal quality of some channels degrades, it can still maintain robust displacement estimation through high-quality channels, demonstrating good fault tolerance. In one example, the channel coefficients... The peak value of the cross-correlation of the channels is determined. For example, for the (i, j)th channel, the peak value of its normalized cross-correlation sequence is extracted. and the mean value of the side lobes Define the reliability of this channel. ;like If the reliability is less than the preset reliability threshold, the weight coefficient of that channel is set to zero; otherwise... The average reliability of all channels is taken. This determination process can ensure that the contribution of channels with lower displacement estimation reliability to the fusion result is suppressed, thereby improving the robustness of the three-dimensional displacement vector to motion artifact estimation.

[0049] Based on the translation and rotation components in the plane, and combined with the spatial coordinates of the corresponding array elements in each channel, when calculating the three-dimensional displacement components at the position of each array element, for each array element, the spatial coordinates of the (i, j)th array element are compared with the previously solved... Substituting the geometric relationships, the displacement components of the array element in three-dimensional space can be calculated. , , ), satisfying the following formula: This example considers that the spatial positions of each array element are different, and the direction and magnitude of the displacement vector caused by the rotation component are different at different array elements. Therefore, by using the differentiated responses of multiple array elements with known spatial positions, the axial displacement effects caused by translation and rotation can be effectively distinguished. This allows for the differentiation between the two motion modes of translation and rotation, and the overall displacement vector can still be accurately estimated when the patch is twisted, providing accurate motion state input for the initial weight vector matching of the subsequent adaptive filter.

[0050] It is worth noting that since the surface shape of the patch is relatively fixed when worn by the user, in this embodiment of the invention, a standard curve model of the user's wearing part in the patch bonding area can be pre-constructed. Therefore, after obtaining the two-dimensional coordinates of each element, the two-dimensional coordinates of each element can be mapped to the standard curve model to obtain the spatial coordinates of each element.

[0051] The weighted average of the three-dimensional displacement components at each element position yields the overall three-dimensional displacement vector of the matrix-type flexible ultrasound patch relative to the skin, satisfying the following formula: .

[0052] This step condenses the high-density displacement information at the element level into a concise overall motion state representation. Although this three-dimensional displacement vector represents the overall average, it already encompasses the complete motion state of the patch in space because the spatial position information and global optimization solutions for motion parameters of each element were preserved in previous steps. That is, it includes not only translational components but also rotational information through the differences in the displacement components of each element. In subsequent steps, when switching some elements to motion reference mode, this overall displacement vector can be used to determine the main direction and amplitude of the motion, thereby selecting the optimal element switching scheme.

[0053] The preset displacement threshold in step 103 is used to determine whether significant relative movement has occurred between the patch and the skin. For example, but not limited to, the threshold displacement value ranges from 0.5 to 0.2 mm. In a preferred embodiment, the preset displacement threshold is 1 mm, because this value is a small proportion compared to the typical diameter of veins in the lower limbs of the human body (e.g., approximately 8 to 12 mm for the femoral vein and approximately 5 to 8 mm for the popliteal vein). When the displacement exceeds this value, it indicates that the relative positional deviation between the ultrasound beam and the target blood vessel is sufficient to cause a significant error in the blood flow velocity measurement.

[0054] Unlike selecting edge elements or elements distributed along the displacement direction, this step uses the spatial displacement field characteristics formed by the relative displacement estimates of the skin across all channels to select target elements. This is because the relative motion between the patch and the skin is not rigid body motion, but rather an elastic or non-rigid motion mode with a complex spatial distribution. For example, when the patch as a whole translates along a certain direction, the distribution of the displacement field at each element position exhibits an approximately uniform characteristic, with the displacement estimates of each element being similar in magnitude and direction. When the patch twists around the plane normal, the amplitude of the displacement field increases radially from the center of the patch to the edge, with the displacement gradient in the edge region being significantly greater than that in the center region. When the patch locally lifts or the skin undergoes stretching deformation in a certain area, the displacement field near that area exhibits obvious convergence or divergence characteristics, and the rate of change of the displacement vector in space reaches a local extremum in that area. It can be seen that different motion modes exhibit drastically different distribution patterns in the spatial derivative characteristics of the displacement field. Therefore, by analyzing the spatial derivative characteristics of the displacement field in this step, the spatial regions with the most intense motion or the most sensitive to specific motion modes can be accurately identified.

[0055] For example, the current spatial displacement field features include the divergence and curl features of the current spatial displacement field; wherein, the divergence feature reflects the degree of stretching or lifting between the patch and the skin (e.g., a positive divergence indicates that the displacement diverges outward from the center, corresponding to the stretching or separation trend between the skin and the patch; a negative divergence indicates that the displacement converges towards the center, corresponding to the compression or adhesion trend between the skin and the patch), and the curl feature reflects the degree of torsion of the patch relative to the skin (e.g., a region with non-zero curl indicates that the patch has undergone local torsion or shearing in that region, corresponding to the non-uniform rotational motion of the patch relative to the skin).

[0056] The current spatial displacement field characteristics can be achieved through the following steps: Estimating the axial displacement values ​​of all M×N channels. According to its two-dimensional spatial coordinates in the patch plane The data is organized to form a scalar field d(x, y) defined on a discrete grid. This scalar field reflects the spatial distribution of the axial displacement of the skin relative to the patch at various locations. Although this displacement field contains only the axial component, since the axial displacement is the projection of the relative motion between the patch and the skin in the direction of ultrasound propagation, its spatial gradient characteristics have a definite geometric mapping relationship with the divergence and curl of the in-plane motion. Therefore, by analyzing the spatial rate of change of this scalar field, the divergence and curl distribution of the in-plane motion can be inferred.

[0057] The process of inferring the divergence and curl distribution of in-plane motion by analyzing the spatial rate of change of the scalar field can include: calculating the spatial gradient characteristics of the displacement field at each point (the gradient describes the rate and direction of change of displacement at that point in adjacent spatial locations, i.e., the magnitude of strain at that point). Since the calculation of divergence and curl involves the partial derivatives of the displacement field in two orthogonal directions in space, and the displacement estimates of each array element are distributed on a discrete grid, the finite difference method is used here to estimate the partial derivatives of the spatial gradient characteristics at each grid node. For the partial derivatives of the spatial gradient characteristics at the boundary array elements, a one-sided difference scheme is used, and for the partial derivatives of the spatial gradient characteristics at the internal array elements, a central difference scheme is used. A pair of eigenvalues ​​can be obtained at each array element location: divergence eigenvalue and curl eigenvalue.

[0058] In one example, step 103 above, which selects target array elements in a matrix-type flexible ultrasound patch based on the current spatial displacement field features derived from the skin relative displacement estimates of each channel, may include the following steps: selecting array elements within the channel region with the largest curl and / or divergence values ​​as target array elements, based on the divergence and curl features of the current spatial displacement field. The channel region consists of several adjacent array elements. For example, in a matrix-type flexible ultrasound patch, with each array element as the center, several array elements adjacent to that center are selected as a channel region (also called an array element region) according to the local neighborhood window selection principle. The window size in the local neighborhood window selection principle can be, but is not limited to, 3×3 or 5×5 array elements, or the window size can be determined based on the number of array elements within a channel region accounting for 10% to 30% of the total number of array elements, to balance the quality of the motion reference signal and the maintenance of blood flow detection performance. In this example, compared to simply selecting edge elements or elements distributed along the displacement direction, the element selection based on the spatial derivative characteristics of the displacement field can dynamically adapt the spatial distribution of the reference elements according to the actual type of motion. For example, in torsional motion, edge region elements with the largest tangential displacement gradient are selected first, and in tensile motion, deformation region elements with the largest divergence are selected first, thereby achieving on-demand optimization of the motion reference signal quality.

[0059] In some scenarios, the current motion is primarily rotational, with rotational features dominating the displacement field. In such cases, the element region with the largest absolute value of the curl feature can be prioritized as the target element. In other scenarios, the current motion is primarily stretching or lifting, with divergence features dominating the displacement field. In this case, the element region with the largest absolute value of the divergence feature can be prioritized as the target element. In yet another scenario, when both divergence and curl are significant, it indicates the existence of a composite motion pattern. In this case, the element region with the largest weighted combination of the absolute values ​​of the divergence and curl features can be selected as the target element.

[0060] When the target array elements selected in this manner acquire skin interface reflection signals at a fixed transmission frequency and fixed focusing depth, they can capture motion information at different spatial locations in the displacement field. This results in a certain degree of spatial non-correlation between the acquired motion reference signals, further improving the accuracy of motion artifact estimation. Specifically, the target array elements acquiring skin interface reflection signals at a fixed transmission frequency and fixed focusing depth in motion reference mode include amplitude and phase modulation caused by the relative motion between the patch and the skin, but are almost unaffected by deep blood flow signals; therefore, they can serve as reference signals for motion artifacts.

[0061] When the magnitude of the three-dimensional displacement vector exceeds a preset displacement threshold, in addition to switching the motion reference mode for the target element, this embodiment of the invention also proposes another level of motion compensation operation. That is, at the physical level, the direction of the emitted beams of the remaining elements maintaining the blood flow detection mode is actively adjusted so that the focusing center of the ultrasound emitted beam can follow the relative position change of the target blood vessel caused by the sliding of the patch. In one implementation, after switching the target element from the blood flow detection mode to the motion reference mode in step 103, while the remaining elements other than the target element maintain the blood flow detection mode, the expected displacement position of the target blood vessel can be determined based on the three-dimensional displacement vector. Using the focus point of the emitted beam tracking the expected displacement position as a reference, the deflection correction of the emitted beam is performed on the remaining elements.

[0062] In one example, when determining the expected displacement position of the target blood vessel based on the three-dimensional displacement vector, the displacement of the patch relative to the skin is numerically equal to but opposite in direction to the displacement of the target blood vessel relative to the patch. That is, if the initial reference position of the target blood vessel in the patch coordinate system is P0=(x0, y0, z0), and the three-dimensional displacement vector is D=(Δx, Δy, Δz), then the expected displacement position of the target blood vessel at the current moment is P0. Pred =P0-D=(x0-Δx, y0-Δy, z0-Δz). To ensure the real-time performance of the correction, the three-dimensional displacement vector on which this correction process is based can be the three-dimensional displacement vector estimated in the previous frame. In this way, in continuous motion scenarios where all three-dimensional displacement vectors exceed the preset displacement threshold, the cumulative effect of the wave deflection correction can enable the transmitted beam to smoothly track the continuous displacement trajectory of the target blood vessel.

[0063] Accordingly, when performing transmit beam deflection correction on the remaining array elements based on the target displacement position tracked by the focal point of the transmit beam, the transmit beam deflection correction is achieved by adjusting the transmit delay parameters of each array element. For the (i, j)th array element, the adjustment amount Δτ of its transmit delay relative to the reference delay is... ij The distance from the element to the expected displacement position is proportional to the difference between the distance from the element to the original reference position. In this method, the transmission delay parameters of each element are recalculated based on the expected displacement position, so that the focal point shifts from the original reference position to the expected displacement position, ensuring that the adjusted beam achieves synchronous changes in pointing deflection and focusing depth in space. The deflection correction result of the transmitted beam can be sent to the adaptive filter in step 104 for processing together with the motion reference signal. The deflection correction result provides energy assurance, ensuring that the ultrasound energy can reach the target blood vessel, while the motion reference signal can be used to interfere with motion artifacts. The combination of the two further improves the reliability of blood flow detection in motion scenarios.

[0064] In one possible scenario, if the magnitude of the three-dimensional displacement vector does not exceed a preset displacement threshold, it indicates that the current situation is quasi-static or micro-motion, and the relative motion amplitude between the patch and the skin is insufficient to significantly affect the blood flow detection accuracy. In this case, no additional compensation process is required. In one implementation, after step 102 above, if the magnitude of the three-dimensional displacement vector does not exceed the preset displacement threshold, each array element can be kept in blood flow detection mode, and conventional beamforming can be performed on the ultrasound echo signals of each channel to obtain the blood flow velocity spectrum. For example, during the conventional beamforming process, the ultrasound echo signals of each channel can be aligned and superimposed according to the time delay parameters of the reference focusing position to form a high signal-to-noise ratio synthetic signal. The synthetic signal is then subjected to Doppler processing, and low-frequency components from blood vessel walls and tissue motion are filtered out by a wall filter, while high-frequency Doppler components from red blood cell scattering are retained. The retained Doppler signal is then subjected to spectral analysis to calculate the power spectral density. The peak frequency and average frequency are extracted from the power spectrum, and then the blood flow velocity is calculated according to the Doppler equation to finally obtain the blood flow velocity spectrum. By calculating hemodynamic parameters such as peak velocity, average velocity, and resistance index based on the blood flow velocity spectrum, and monitoring the changing trends of these parameters over a continuous time series, it is possible to determine whether there is a risk of abnormal blood flow or thrombosis.

[0065] The motion artifact components doped in the motion reference signal and the echo signal from the blood flow detection channel share the same physical origin—both stem from the same relative motion between the patch and the skin, thus exhibiting a high correlation in both the frequency and time domains. However, due to their different spatial positions on the patch, their response amplitude and phase to the same motion may differ. Therefore, in step 104, an adaptive filter is used for compensation and matching to address this difference, enabling the filter to generate the optimal estimate of the motion artifact components in the blood flow channel from the reference signal.

[0066] Step 104 above can be the acquisition of motion reference signals using motion reference array elements. As a reference input, the ultrasound echo signal of the blood flow detection array element is used. The desired signal contains the actual blood flow signal. and motion artifact noise ,Right now Noise cancellation is achieved using an adaptive filter.

[0067] Adaptive filters can employ transverse filter structures based on the minimum mean square error (LMS) criterion, satisfying the formula... ,in The output of the filter is the motion artifact component at the current moment. Linear estimate, The motion reference signal (historical sample vector). This is the current weight vector. In one implementation, step 104 above can involve obtaining a linear estimate of the adaptive filter output. From the expected signal Subtracting from the middle, we obtain the error signal. This error signal is the hemoglobinometry signal after motion artifact removal. (The sentence about motion artifact removal and hemoglobinometry is incomplete and lacks context.) Doppler spectral analysis is performed to obtain the corrected blood flow velocity spectrum. Based on this, hemodynamic parameters such as peak flow velocity and average flow velocity are calculated, which can determine whether thrombosis has occurred.

[0068] In adaptive filters, the iteration of weight vectors follows the principle of minimum mean square error, and the vector in each iteration is related to the weight vector matched with the current spatial displacement field characteristics. The update process of weight vectors can be related to various factors, such as, but not limited to, the current operating status (remaining power and / or operating temperature), and / or the magnitude of the three-dimensional displacement vector.

[0069] In one example, the weight vector update process is related to the current running state. For instance, the process of determining the weight vector of an adaptive filter includes the following steps: Obtain the historical best vector library, which is stored according to the category of spatial displacement field features, and each category of spatial displacement field features corresponds to a set of historical best weight vectors. Calculate the similarity between the current spatial displacement field features and the spatial displacement field features of each category in the historical best vector library; use the historical best weight vector corresponding to the spatial displacement field of the category with the highest similarity as the initial weight vector of the adaptive filter; Based on the current operating status parameters, determine the weight vector update strategy for the adaptive filter; the current operating status parameters include remaining battery power and / or operating temperature; The initial weight vector is iteratively updated using a weight vector update strategy to obtain the final weight vector of the adaptive filter.

[0070] This example starts iterating based on the historical best vector library, which can shorten the convergence transient process and is more suitable for power consumption sensitive scenarios of the device; and the hierarchical update strategy based on the remaining power and temperature enables the filtering accuracy to automatically adapt to the physical state of the device, extending the effective monitoring time after a single charge, while preventing device damage or wearing discomfort caused by overheating.

[0071] The historical optimal weight vector library can be established during the offline training phase before the device leaves the factory. For example, the establishment process includes: collecting multi-channel ultrasonic echo data under different motion modes, labeling the spatial displacement field feature category of each motion mode, running adaptive filters until convergence, and storing the converged weight vector as the historical optimal weight vector for that category in the library. In this example, the iterative update results of the initial weight vector can also be used to update and continuously optimize the historical optimal weight vector library online.

[0072] One approach involves determining the weight vector update strategy for the adaptive filter based on the current running state parameters, including: When the remaining battery power is lower than the first battery power threshold or the operating temperature is higher than the first temperature threshold, the adaptive filter's weight vector update strategy includes iteratively updating the weight vector with a preset first update step size and / or a downsampled motion reference signal. In this case, when resources are limited, the computational load is reduced by reducing the precision step size or downsampling, which can balance performance and battery life. When the remaining power is higher than the second power threshold and the operating temperature is lower than the second temperature threshold, the adaptive filter's weight vector update strategy includes iteratively updating the weight vector using a motion reference signal with a second update step size and the original sampling rate. The second power threshold is higher than the first power threshold, and the second temperature threshold is lower than the first temperature threshold. In this case, when resources are sufficient, iterative updates are performed with a high-precision step size and the original sampling rate, which can ensure optimal filtering performance. When the remaining power is below the third power threshold, the adaptive filter's weight vector update strategy includes freezing some weight vector elements in the initial weight vector whose correlation with the current spatial displacement field features is below a preset correlation threshold, and iteratively updating the remaining weight vector elements in the initial weight vector that are not frozen; the third power threshold is less than the first power threshold; in this case, when the power is almost exhausted, some weight vector elements that contribute little to the current motion artifact estimation are frozen, and only the key weight vector is updated to ensure that the basic filtering function is maintained and does not completely fail under extreme resource constraints.

[0073] In this example, the final weight vector after each iteration converges can be written back to the historical optimal weight vector library to update the optimal weight vector for the corresponding feature category. This allows the knowledge in the library to be continuously enriched and optimized as the device is used, forming an online learning closed loop of training as it is used.

[0074] In another example, the update process of the weight vector is related to the magnitude of the three-dimensional displacement vector. For instance, the determination process of the weight vector of an adaptive filter includes the following steps: Based on the magnitude of the three-dimensional displacement vector, the updated step size is obtained by adaptively adjusting the basic step size using a dynamic step size strategy. The current weight vector is updated based on the update step size combined with the error signal and the motion reference signal.

[0075] The dynamic step size strategy includes: when the displacement magnitude is large, it indicates that the non-stationarity of motion artifacts is enhanced, and the filter needs a faster tracking speed to capture the rapidly changing statistical characteristics of artifacts. At this time, the step size is increased to speed up convergence; when the displacement magnitude is close to or less than the threshold, the statistical characteristics of motion artifacts are relatively stable, and the step size returns to the base value to reduce steady-state misalignment error.

[0076] In this example, the update step size is dynamically associated with the magnitude of the displacement vector, which ensures faster convergence in turbulent motion scenarios and reduces steady-state error in stable scenarios, thus achieving an adaptive balance between tracking accuracy and stability.

[0077] Example 2: Based on the same inventive concept, the present invention also provides a thrombosis prevention electrical stimulator, comprising: an electrical stimulator assembly, an electrode assembly, an ultrasound assembly, and a control module (not shown in the figure). The electrical stimulator assembly includes a control module and an electrode connector; one end of the electrode connector is connected to the control module, and the other end is connected to the electrode assembly. The ultrasound component includes a matrix-type flexible ultrasound patch, which includes multiple flexible ultrasound transducer elements arranged in M ​​rows × N columns; the multiple flexible ultrasound transducer elements are connected to the electrode assembly. The electrode pad assembly is used to conform to the user's skin when the thrombosis prevention electrical stimulator is worn; The control module is used to implement the signal processing method based on matrix-type flexible ultrasound patch in the above embodiments.

[0078] Specifically, the basic structure of the thrombosis prevention electrical stimulator is as shown in application number 202410981582.X, which describes a thrombosis prevention electrical stimulator and electrodes based on flexible ultrasound technology for monitoring lower limb venous blood flow velocity. It includes an electrical stimulator assembly, an electrode assembly, and an ultrasound assembly. The electrical stimulator assembly includes a screen protector, a screen, a top cover, a switch, a battery, a middle frame, a PCB circuit board (in this embodiment, the control module can be integrated into the PCB circuit board), a bottom cover, conductive electrode needles, electrode connectors, a magnet, a data storage module, and an information transmission module. The top cover, middle frame, and bottom cover are connected by snap-fit ​​connections. A battery is installed between the top cover and the middle frame, and a PCB circuit board is installed between the middle frame and the bottom cover. The battery and the PCB circuit board are connected by a circuit. The switch is a mechanical switch button installed on both sides of the middle frame, a physical button electrically connected to the PCB circuit board, used to adjust the stimulation level. The screen is an LCD display installed on the top cover, electrically connected to the PCB circuit board, used to display data from the data storage module. A screen protector, i.e., an LCD screen protective cover, is installed on the outside of the top cover. The lower cover has a through hole for installing electrode connectors. Electrode conductive pins are mounted on the electrode connectors and soldered to the PCB circuit board. A magnet is located at the bottom of the electrode connector for connecting the electrode pad assembly. A data storage module is mounted on the PCB circuit board to record working time, stimulation intensity, and ultrasound component detection results, and uploads this data to a cloud server via an information transmission module. The electrode pad assembly includes an electrode pad body, an elastic band, a magic hook, an electrode pad fixing upper cover, a surface impedance detection electrode, an output electrode, a reference electrode, a gel, a flexible ultrasound electrode connection interface, a magnet, and an electrode pad fixing lower cover. The surface impedance detection electrode, output electrode, and reference electrode are mounted on the underside of the electrode pad body. The surface impedance detection electrode is used to determine whether the electrode pad assembly adheres to the skin through impedance detection. The electrode pad fixing upper and lower covers are concentrically aligned and connected by snap-fit ​​mechanisms to press and secure the electrode pad body and the connectors of the three electrodes. The magnet is installed inside the electrode pad fixing lower cover for alignment and attraction with the magnet in the electrical stimulator assembly. A first hydrogel is mounted on the underside of the surface impedance detection electrode, and a second hydrogel is mounted on the underside of the output electrode and reference electrode, for conforming to the patient's lower limb skin. Elastic straps and magic hooks are mounted on both sides of the electrode pad fixing cover on the upper part of the electrode pad body. The back of the elastic strap has magic hairs for cooperating with the magic hooks to secure the electrode pad assembly to the patient's lower limb. A flexible ultrasound electrode connection interface connects to the three electrodes and is located under the three electrodes. The ultrasound assembly includes a flexible ultrasound electrode connection plug, a flexible ultrasound transducer, flexible ultrasound electrode pads, and hydrogels.The flexible ultrasound electrode pad is equipped with an array of multiple flexible ultrasound transducer elements to acquire vascular images, calculate blood flow velocity and volume, detect peak blood flow and velocity, and identify any thrombus formation. Each flexible ultrasound transducer element is connected to a flexible ultrasound electrode connector on the ultrasound assembly, which in turn connects to the flexible ultrasound electrode interface on the electrode pad assembly. The lower part of the flexible ultrasound electrode pad contains hydrogel for conforming to the patient's lower limb skin.

[0079] Example 3: Based on the same inventive concept, this invention also provides a signal processing system based on a matrix-type flexible ultrasound patch, such as... Figure 3 As shown, the matrix-type flexible ultrasound patch includes multiple flexible ultrasound transducer elements arranged in an M-row × N-column configuration, including: The displacement estimation module is used to calculate the estimated skin relative displacement of each channel based on the multi-channel ultrasound echo signals received through multiple array elements and by using the cross-correlation function between adjacent acquisition frames. The displacement fusion module is used to calculate the three-dimensional displacement vector of the matrix-type flexible B-mode ultrasound patch relative to the skin based on the estimated skin displacement values ​​of each channel and using a weighted fusion algorithm of array element spatial coordinates. The mode switching module is used to select target array elements in the matrix flexible B-ultrasound patch when the magnitude of the three-dimensional displacement vector exceeds the preset displacement threshold, based on the current spatial displacement field characteristics composed of the skin relative displacement estimates of each channel; and to switch the target array element from blood flow detection mode to motion reference mode, so that the target array element collects skin interface reflection signals as motion reference signals with a fixed transmission frequency and a fixed focusing depth. The adaptive cancellation module is used to adaptively cancel the ultrasound echo signals of the remaining array elements that maintain the blood flow detection mode using an adaptive filter, with the motion reference signal as the reference input, to obtain the blood flow ultrasound signal after motion artifact elimination; the weight vector of the adaptive filter is determined based on the weight vector that matches the characteristics of the current spatial displacement field.

[0080] Preferably, the displacement estimation module is specifically used for: Based on each channel, select the radio frequency echo signal segment of the adjacent acquisition frame in the ultrasonic echo signal of the channel; based on the radio frequency echo signal segment of the adjacent acquisition frame, calculate the normalized cross-correlation function value within the preset time delay search range to generate a normalized cross-correlation sequence; use the time delay value corresponding to the peak position of the normalized cross-correlation sequence as the estimated value of the inter-frame axial displacement of the channel. Based on the inter-frame axial displacement estimates of each channel, the relative skin displacement estimates of each channel are determined.

[0081] Preferably, the displacement fusion module is specifically used for: Using the two-dimensional coordinates of the corresponding array elements of each channel in the matrix-type flexible B-mode ultrasound patch as the position input, and the estimated relative skin displacement of each channel as the observation value, a set of fusion observation equations is established. The weighted least squares method was used to solve the fusion observation equations to obtain the in-plane translational components and the rotational components about the plane normal of the matrix flexible B-mode ultrasound patch. Based on the translation and rotation components in the plane, and combined with the spatial coordinates of the corresponding array elements in each channel, the three-dimensional displacement components at the position of each array element are calculated; the three-dimensional displacement components at the position of each array element are weighted and averaged to obtain the three-dimensional displacement vector of the matrix flexible B-mode ultrasound patch relative to the skin.

[0082] Preferably, the current spatial displacement field features include the divergence feature and the curl feature of the current spatial displacement field; wherein, the divergence feature reflects the degree of stretching or lifting between the patch and the skin, and the curl feature reflects the degree of twisting of the patch relative to the skin; The mode switching module is specifically used for: Based on the divergence and curl characteristics of the current spatial displacement field, the array element in the channel region with the largest curl and / or divergence characteristics is selected as the target array element. The channel region consists of several adjacent array elements.

[0083] Preferably, the adaptive cancellation module is also used for: Obtain the historical best vector library, which is stored according to the category of spatial displacement field features, and each category of spatial displacement field features corresponds to a set of historical best weight vectors. Calculate the similarity between the current spatial displacement field features and the spatial displacement field features of each category in the historical best vector library; use the historical best weight vector corresponding to the spatial displacement field of the category with the highest similarity as the initial weight vector of the adaptive filter; Based on the current operating status parameters, determine the weight vector update strategy for the adaptive filter; the current operating status parameters include remaining battery power and / or operating temperature; The initial weight vector is iteratively updated using a weight vector update strategy to obtain the final weight vector of the adaptive filter.

[0084] Preferably, the adaptive cancellation module is specifically used for: When the remaining power is lower than the first power threshold or the operating temperature is higher than the first temperature threshold, the weight vector update strategy of the adaptive filter includes iterative updating of the weight vector with a preset first update step size and / or a downsampled motion reference signal. When the remaining power is higher than the second power threshold and the operating temperature is lower than the second temperature threshold, the weight vector update strategy of the adaptive filter includes iterative updating of the weight vector with the motion reference signal of the second update step size and the original sampling rate, wherein the second power threshold is higher than the first power threshold and the second temperature threshold is lower than the first temperature threshold. When the remaining power is lower than the third power threshold, the adaptive filter's weight vector update strategy includes freezing some weight vector elements in the initial weight vector whose correlation with the current spatial displacement field features is lower than a preset correlation threshold, and iteratively updating the remaining weight vector elements in the initial weight vector that have not been frozen.

[0085] Preferably, the mode switching module is also used for: While maintaining the blood flow detection mode for the array elements other than the target element, the expected displacement position of the target blood vessel is determined based on the three-dimensional displacement vector. The deflection correction of the transmitted beam is performed on the remaining array elements with the focus point of the transmitted beam tracking the expected displacement position as a reference.

[0086] Preferably, it also includes a beamforming module for: When the magnitude of the three-dimensional displacement vector does not exceed the preset displacement threshold, each array element is kept in blood flow detection mode, and conventional beamforming is performed on the ultrasound echo signals of each channel to obtain the blood flow velocity spectrum.

[0087] Example 4: like Figure 4 As shown, the present invention also provides an electronic device, which may be a computer device, a microcontroller device, a smart mobile device, etc. The electronic device in this embodiment may include a processor, a memory, a transceiver component, etc. The memory, processor, and transceiver component are connected via a bus; the memory can be used to store executable programs, and an exemplary executable program may include instructions; the processor is used to execute the instructions stored in the memory. The memory can also be used to store data, which can be accessed and / or modified when instructions are executed.

[0088] The processor may be a Central Processing Unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, and it is suitable for implementing one or more instructions. Specifically, it is suitable for loading and executing one or more instructions in the storage medium to realize the corresponding method flow or corresponding function, so as to realize the steps of the signal processing method based on a matrix flexible B-mode ultrasound patch in the above embodiments.

[0089] Example 5: Based on the same inventive concept, this invention also provides a readable storage medium, specifically an electronic device readable storage medium (Memory). This readable storage medium is a memory device within an electronic device used to store programs and data. It is understood that the storage medium here can include both built-in storage media within the electronic device and extended storage media supported by the electronic device. The storage medium provides storage space, which stores the terminal's operating system. Furthermore, this storage space also stores one or more instructions suitable for loading and execution by a processor. These instructions can be one or more executable programs (including program code). It should be noted that the storage medium here can be high-speed RAM or non-volatile memory, such as at least one disk storage device. Loading and executing one or more instructions stored in the storage medium by the processor can implement the steps of the signal processing method based on a matrix-type flexible ultrasound patch in the above embodiments.

[0090] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0091] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0092] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0093] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0094] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit its scope of protection. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that after reading the present invention, they can still make various changes, modifications or equivalent substitutions to the specific implementation methods of the application, but these changes, modifications or equivalent substitutions are all within the scope of protection of the claims of the present invention.

Claims

1. A signal processing method based on a matrix-type flexible ultrasound patch, characterized in that, The matrix-type flexible ultrasound patch includes multiple flexible ultrasound transducer elements arranged in an M-row × N-column configuration, and the method includes: Based on the multi-channel ultrasound echo signals received through multiple array elements, the relative skin displacement estimate of each channel is calculated using the cross-correlation function between adjacent acquisition frames. Based on the skin relative displacement estimates of each channel, the three-dimensional displacement vector of the matrix flexible B-mode ultrasound patch relative to the skin is calculated using a weighted fusion algorithm of array element spatial coordinates. When the magnitude of the three-dimensional displacement vector exceeds the preset displacement threshold, a target element is selected in the matrix flexible B-mode ultrasound patch based on the current spatial displacement field characteristics formed by the skin relative displacement estimates of each channel; the target element is switched from blood flow detection mode to motion reference mode, so that the target element collects skin interface reflection signals as motion reference signals with a fixed emission frequency and a fixed focusing depth; Using the motion reference signal as a reference input, an adaptive filter is used to perform adaptive noise cancellation processing on the ultrasound echo signals of the remaining array elements that maintain the blood flow detection mode, so as to obtain the blood flow ultrasound signal after motion artifact elimination; the weight vector of the adaptive filter is determined based on the weight vector that matches the current spatial displacement field characteristics.

2. The method as described in claim 1, characterized in that, The method of calculating the skin relative displacement estimate for each channel based on the multi-channel ultrasound echo signals received through multiple array elements and utilizing the cross-correlation function between adjacent acquisition frames includes: Based on each of the channels, select the radio frequency echo signal segment of the adjacent acquisition frame in the ultrasonic echo signal of the channel; based on the radio frequency echo signal segment of the adjacent acquisition frame, calculate the normalized cross-correlation function value within a preset time delay search range to generate a normalized cross-correlation sequence; use the time delay value corresponding to the peak position of the normalized cross-correlation sequence as the estimated value of the inter-frame axial displacement of the channel. Based on the inter-frame axial displacement estimates of each channel, the skin relative displacement estimates of each channel are determined.

3. The method as described in claim 1 or 2, characterized in that, The three-dimensional displacement vector of the matrix flexible ultrasound patch relative to the skin is calculated using a weighted fusion algorithm of array element spatial coordinates based on the skin relative displacement estimates of each channel, including: Using the two-dimensional coordinates of the corresponding array elements of each channel in the plane of the matrix-type flexible B-mode ultrasound patch as position input, and the estimated skin relative displacement of each channel as observation value, a fusion observation equation set is established. The weighted least squares method is used to solve the fusion observation equations to obtain the in-plane translation components and rotational components about the plane normal of the matrix flexible B-mode ultrasound patch as a whole. Based on the translation and rotation components in the plane, and combined with the spatial coordinates of the corresponding array elements of each channel, the three-dimensional displacement components at the position of each array element are calculated; the three-dimensional displacement components at the position of each array element are weighted and averaged to obtain the three-dimensional displacement vector of the matrix flexible B-mode ultrasound patch relative to the skin.

4. The method as described in claim 1, characterized in that, The current spatial displacement field features include the divergence and curl features of the current spatial displacement field; wherein, the divergence feature reflects the degree of stretching or lifting between the patch and the skin, and the curl feature reflects the degree of twisting of the patch relative to the skin; The current spatial displacement field features, based on the estimated skin relative displacement values ​​of each channel, are used to select target array elements in the matrix-type flexible B-mode ultrasound patch, including: Based on the divergence and curl characteristics of the current spatial displacement field, the array element in the channel region with the largest curl characteristic and / or divergence characteristic value is selected as the target array element. The channel region is composed of several adjacent array elements.

5. The method as described in claim 1, characterized in that, The process of determining the weight vector of the adaptive filter includes: Obtain a historical optimal vector library, which is stored according to the category of spatial displacement field features, and each category of spatial displacement field features corresponds to a set of historical optimal weight vectors. Calculate the similarity between the current spatial displacement field feature and the spatial displacement field features of each category in the historical optimal vector library; take the historical optimal weight vector corresponding to the spatial displacement field of the category with the highest similarity as the initial weight vector of the adaptive filter; Based on the current operating status parameters, the weight vector update strategy of the adaptive filter is determined; the current operating status parameters include remaining battery power and / or operating temperature. The initial weight vector is iteratively updated using the weight vector update strategy to obtain the final weight vector of the adaptive filter.

6. The method as described in claim 5, characterized in that, The step of determining the weight vector update strategy for the adaptive filter based on the current operating state parameters includes: When the remaining power is lower than the first power threshold or the operating temperature is higher than the first temperature threshold, the weight vector update strategy of the adaptive filter includes iterative updating of the weight vector with a preset first update step size and / or a downsampled motion reference signal. When the remaining power is higher than the second power threshold and the operating temperature is lower than the second temperature threshold, the weight vector update strategy of the adaptive filter includes iterative updating of the weight vector with a motion reference signal of a second update step size and an original sampling rate, wherein the second power threshold is higher than the first power threshold and the second temperature threshold is lower than the first temperature threshold. When the remaining power is lower than the third power threshold, the weight vector update strategy of the adaptive filter includes freezing some weight vector elements in the initial weight vector whose correlation with the current spatial displacement field feature is lower than a preset correlation threshold, and iteratively updating the remaining weight vector elements in the initial weight vector that have not been frozen.

7. The method as described in claim 1, characterized in that, After switching the target array element from blood flow detection mode to motion reference mode, the method further includes: While the other array elements besides the target array element maintain the blood flow detection mode, the expected displacement position of the target blood vessel is determined based on the three-dimensional displacement vector. The expected displacement position is tracked by the focal point of the transmitted beam as a reference, and the deflection correction of the transmitted beam is performed on the other array elements.

8. The method as described in claim 1, characterized in that, After calculating the three-dimensional displacement vector of the matrix flexible ultrasound patch relative to the skin based on the skin relative displacement estimates of each of the aforementioned channels using a weighted fusion algorithm of array element spatial positions, the method further includes: When the magnitude of the three-dimensional displacement vector does not exceed the preset displacement threshold, each array element is kept in blood flow detection mode, and conventional beamforming is performed on the ultrasound echo signal of each channel to obtain the blood flow velocity spectrum.

9. A thrombosis prevention electrical stimulator, characterized in that, include: Electrical stimulator assembly, electrode assembly, and ultrasound assembly; The electrical stimulator assembly includes a control module and an electrode connector; one end of the electrode connector is connected to the control module, and the other end is connected to the electrode assembly. The ultrasound component includes a matrix-type flexible ultrasound patch, which includes multiple flexible ultrasound transducer elements arranged in M ​​rows × N columns; the multiple flexible ultrasound transducer elements are connected to the electrode assembly. The electrode pad assembly is used to conform to the user's skin when the thrombosis prevention electrical stimulator is worn; The control module is used to implement the signal processing method based on matrix flexible ultrasound patch as described in any one of claims 1-8.

10. A signal processing system based on a matrix-type flexible ultrasound patch, characterized in that, The matrix-type flexible ultrasound patch includes multiple flexible ultrasound transducer array elements arranged in M ​​rows × N columns, including: The displacement estimation module is used to calculate the estimated skin relative displacement of each channel based on the multi-channel ultrasound echo signals received through multiple array elements and by using the cross-correlation function between adjacent acquisition frames. The displacement fusion module is used to calculate the three-dimensional displacement vector of the matrix flexible B-mode ultrasound patch relative to the skin based on the skin relative displacement estimates of each channel and using a weighted fusion algorithm of array element spatial coordinates. The mode switching module is used to select a target element in the matrix flexible B-ultrasound patch when the magnitude of the three-dimensional displacement vector exceeds a preset displacement threshold, based on the current spatial displacement field characteristics composed of the skin relative displacement estimates of each channel; and to switch the target element from blood flow detection mode to motion reference mode, so that the target element collects skin interface reflection signals as motion reference signals with a fixed emission frequency and a fixed focusing depth. An adaptive cancellation module is used to perform adaptive noise cancellation processing on the ultrasound echo signals of the remaining array elements that maintain the blood flow detection mode using the motion reference signal as a reference input and an adaptive filter, so as to obtain a blood flow ultrasound signal after motion artifact elimination; the weight vector of the adaptive filter is determined based on a weight vector that matches the characteristics of the current spatial displacement field.

Images