A method for detecting abnormality of heart function index of heart failure animal model

Abstract

This invention relates to the field of biomedical signal processing, and more specifically, to a method for detecting abnormal cardiac function indicators in an animal model of heart failure. The method includes: acquiring an original sampling point sequence containing potential values, timestamps, and channel indices; locking a feature interval where the composite feature value exceeds a threshold; generating control points within the interval and constructing a cubic B-spline reconstruction curve; adjusting the control point potential by minimizing the energy value of the evaluation function to obtain the simulated potential of the sampling points within the feature interval; connecting the feature interval reconstruction curve with the original sequence outside the interval to extract the real-time heart rate; calculating the phase correction offset based on the difference between the real-time heart rate and the baseline heart rate; correcting the timestamps of non-ECG signal channels and completing the multi-source signal timestamp alignment, thereby improving the accuracy of detecting abnormal cardiac function indicators.

Description

Technical Field

[0001] This invention relates to the field of biomedical signal processing. More specifically, this invention relates to a method for detecting abnormalities in cardiac function indicators in an animal model of heart failure. Background Technology

[0002] Animal models of heart failure are experimental subjects that artificially simulate the failure of the heart's pumping function. They play an irreplaceable role in the study of the pathological evolution of cardiovascular diseases and the evaluation of the efficacy of new drugs. In such studies, accurate detection of cardiac function indicators in experimental animals is crucial for determining the success of model construction and assessing the effectiveness of drug interventions. Typically, researchers need to use specialized electrical data acquisition equipment to simultaneously monitor and record multi-source physiological data such as electrocardiogram signals, left ventricular pressure, and aortic pressure in the experimental animals to analyze the heart's contractile capacity and valvular status.

[0003] However, in practical experimental monitoring scenarios, signal acquisition from such tiny organisms often faces a trade-off between hardware performance and the reproduction of minute features. Due to the unique cardiac physiology of experimental animals (especially small rodents), the QRS complex representing ventricular depolarization has an extremely short duration and a very high instantaneous rate of change. Existing general-purpose experimental acquisition equipment is limited by hardware sampling frequency, and when discretizing such sub-millisecond high-frequency signals, aliasing often occurs due to insufficient sampling density. This results in the acquired electrocardiogram waveform exhibiting severely reduced peak amplitude and distorted waveform slope in the time domain, leading to the loss of true feature points reflecting the intensity of myocardial depolarization. Furthermore, different types of sensors have different physical responses; for example, pressure signals transmitted through fluid conduits produce an inherent physical delay, causing a phase drift on the time axis compared to electrical signals transmitted at the speed of light. This waveform distortion and temporal disconnect between multiple signal sources make it difficult for researchers to accurately locate the instantaneous physical points of cardiac contraction and valve opening and closing based on raw data, easily leading to misdiagnosis of cardiac dysfunction and failing to meet the needs of high-precision pathological research.

[0004] Therefore, how to accurately detect abnormalities in cardiac function indicators during the construction of animal models of heart failure is an urgent problem to be solved. Summary of the Invention

[0005] To address the technical problem of accurately detecting abnormalities in cardiac function indicators during the construction of animal models of heart failure, this invention proposes a method for detecting abnormalities in cardiac function indicators in animal models of heart failure. This method includes the following steps:

[0006] The process involves: acquiring a sequence of original sampling points from multiple sources, including potential values, timestamps, and source channel indices; identifying a feature interval within the original sampling point sequence, where the composite feature value of multiple consecutive sampling points within the feature interval exceeds a preset threshold. This composite feature value is positively correlated with the voltage change rate of the corresponding sampling point and the local variance of the sampling points within the window, and negatively correlated with the local mean. Multiple control points are generated within the feature interval, and a cubic B-spline reconstruction curve is constructed based on these control points. The potential values ​​of the control points are adjusted by minimizing the energy value of the evaluation function to obtain the simulated potential of each sampling point in the final reconstruction curve. The energy value of the evaluation function is positively correlated with the second derivative of the reconstruction curve and with the difference between the simulated potential and the actual potential value of the sampling points in the reconstruction curve. The final reconstruction curve of the feature interval is connected to the original sampling point sequence outside the feature interval to obtain the real-time heart rate in the reconstruction sequence. The phase correction offset is calculated based on the difference between the real-time heart rate and the baseline heart rate, and the timestamps of the sampling points in the non-ECG signal channels of the reconstruction sequence are corrected. The timestamps of the multiple source signals are aligned to achieve abnormal detection of cardiac function indicators.

[0007] This invention provides a method for detecting abnormalities in cardiac function indicators in animal models of heart failure. It improves the quality of electrocardiogram (ECG) and stress signals, thereby assisting in the detection of abnormal cardiac function indicators and effectively enhancing the accuracy of detection results. In acquiring the detection results, this invention considers the problems of aliasing, waveform distortion, and phase drift caused by differences in the physical responses of different sensors due to hardware sampling rate limitations. Based on this, this invention constructs a high-dimensional feature space with physical consistency, uses composite features to lock key waveform intervals, and, based on B-spline curve reconstruction technology that minimizes the energy function, accurately restores the bioelectrical spike features lost by discrete sampling at the sub-millisecond level. Simultaneously, by combining real-time heart rate with dynamic time axis alignment of multi-source signals (ECG and stress), errors caused by physical delays are effectively eliminated, thus significantly improving the sensitivity and accuracy of detecting abnormalities in minute waveform changes in animal models of heart failure.

[0008] According to the present invention, a method for detecting abnormal cardiac function indicators in a heart failure animal model includes obtaining a multi-source signal raw sampling point sequence containing potential values, timestamps, and source channel indices. This includes: simultaneously polling and acquiring the electrocardiogram (ECG) potential value and pressure signal potential value of the experimental target using an ADC (Analog-Digital Converter), wherein the pressure signal includes left ventricular pressure signal and aortic pressure signal; marking the ECG and pressure signals according to their source channel indices to finally obtain the raw sampling point sequence, wherein the source channels include ECG signal channels and non-ECG signal channels; each sampling point in the raw sampling point sequence has a corresponding potential value, timestamp, and source channel index.

[0009] This invention establishes a traceable physical data foundation through a hardware-level polling acquisition and source channel indexing mechanism. This ensures that mixed signal streams from different sensors maintain strict temporal linearity and identification during discretization, providing reliable data structure support for subsequent data processing.

[0010] According to the present invention, a method for detecting abnormal cardiac function indicators in a heart failure animal model includes obtaining the composite feature value by: acquiring sampling points in the original sampling point sequence whose source channel index is an electrocardiogram signal channel; constructing a window centered on the sampling point; and calculating the composite feature value of the sampling point.

[0011] ;

[0012] , The first Composite feature value and potential value of each sampling point , The first Local variance and local mean of the sampling points within a sampling window For the first The potential value at each sampling point The sampling period is As a preset physical correction factor, To prevent constants with a denominator of zero, It is an absolute value function.

[0013] This invention introduces a composite feature extraction algorithm that includes an energy perturbation term. This algorithm can effectively suppress isolated high-frequency noise on a stable baseline, while being extremely sensitive to energy abrupt changes generated by real myocardial depolarization signals (QRS complexes). Thus, it can accurately locate the feature intervals of ECG signals in complex noise environments, improving the robustness of feature recognition.

[0014] According to the present invention, a method for detecting abnormalities in cardiac function indicators in an animal model of heart failure includes, wherein locking the feature interval in the original sampling point sequence comprises: if continuous The composite feature values ​​of all sampling points exceed the threshold. The sequence of sampling points is used as the feature interval, where the threshold is... , The preset number.

[0015] According to the present invention, a method for detecting abnormalities in cardiac function indicators in an animal model of heart failure includes generating multiple control points within a feature interval, comprising: using the starting index of the feature interval... Starting from the sampling step size, interpolation is performed between adjacent sampling points to generate... The coordinates of control points that provide physical guidance, wherein the number of control points... satisfy , This is the terminating index of the feature interval. This is the starting index of the feature interval.

[0016] This invention generates control points within a locked feature range using high-magnification interpolation, providing a high-density geometric skeleton for subsequent curve reconstruction. This reduces the impact of sparse original low-frequency sampling points, which prevents the description of steep changes in high-frequency ECG waveforms.

[0017] According to the present invention, a method for detecting abnormalities in cardiac function indicators in an animal model of heart failure includes obtaining the energy value of the evaluation function in the following manner:

[0018] ;

[0019] The energy value of the evaluation function for reconstructing the curve. To control the point potential coefficient, The potential value of the characteristic interval changes with time The acceleration of change, i.e., the second derivative of the reconstructed curve, To preset smoothing weights, To control the number of points, , These are the simulated potential and potential value at the sampling points in the reconstructed curve, respectively. As the credibility weight of the location, This is the terminating index of the feature interval. This is the starting index of the feature interval. This is a minimization optimization algorithm.

[0020] This invention establishes an evaluation function that integrates physical laws and data fidelity. By balancing the second derivative of the reconstructed curve with sampling point errors and introducing point reliability weights, the reconstructed curve is forced to approximate the true sampling value as closely as possible while satisfying the smooth evolution law of bioelectricity, thereby accurately obtaining the simulated potential of each sampling point.

[0021] According to the present invention, a method for detecting abnormal cardiac function indicators in a heart failure animal model includes adjusting the potential values ​​of control points by minimizing the energy value of the evaluation function to obtain the simulated potentials of each sampling point in the final reconstructed curve. This method comprises: dynamically iteratively adjusting the potential coefficients of each control point within the feature interval by executing a minimization optimization algorithm in the energy value of the evaluation function to obtain a set of optimal coefficient vectors that minimize the total energy value of the evaluation function; and generating a final reconstructed curve with continuous second derivative characteristics on the time axis by linear weighted combination of cubic B-spline basis functions based on the solved optimal coefficient vectors, thereby obtaining the simulated potentials corresponding to each sampling point within the feature interval.

[0022] According to the present invention, a method for detecting abnormal cardiac function indicators in an animal model of heart failure is provided. The method for obtaining the real-time heart rate in the reconstructed sequence includes: determining the peak time of two adjacent R waves in the reconstructed sequence by using the first derivative zero-finding method, taking the difference between the peak times as the instantaneous cardiac cycle, and obtaining the real-time heart rate in the reconstructed sequence based on the instantaneous cardiac cycle.

[0023] According to the present invention, a method for detecting abnormalities in cardiac function indicators in an animal model of heart failure includes calculating a phase correction offset based on the difference between real-time heart rate and a reference heart rate, comprising:

[0024] ;

[0025] This refers to the phase correction offset for non-ECG signal channels. This is the fixed delay constant for the non-ECG signal channel. To preset the heart rate regulation coefficient, For real-time heart rate, The average heart rate over the 24 hours prior to modeling the experimental target is the baseline heart rate.

[0026] According to the present invention, a method for detecting abnormal cardiac function indicators in an animal model of heart failure includes aligning the timestamps of multi-source signals to achieve abnormal detection of cardiac function indicators. The method comprises: extracting cardiac function feature parameters based on the aligned multi-source signal sequence to assist in the detection of abnormal cardiac function indicators.

[0027] The present invention has the following beneficial effects:

[0028] Based on the above technical solution, the present invention provides a method for detecting abnormal cardiac function indicators in animal models of heart failure. By constructing a high-dimensional feature space with physical consistency, using composite features to lock key waveform intervals, and based on the B-spline curve reconstruction technique of minimizing the energy function, the bioelectrical spike features lost by discrete sampling are accurately restored at the sub-millisecond level. At the same time, by combining real-time heart rate to dynamically align the time axis of multi-source signals (ECG and stress), the error caused by physical delay is effectively eliminated, thereby effectively improving the sensitivity and accuracy of judging abnormalities in minute waveform changes in the construction of animal models of heart failure. Attached Figure Description

[0029] Figure 1 A flowchart illustrating the steps of a method for detecting abnormal cardiac function indicators in an animal model of heart failure, as provided in an embodiment of the present invention.

[0030] Figure 2 This is a schematic diagram of feature interval locking and sub-millisecond waveform physical reconstruction provided by an embodiment of the present invention;

[0031] Figure 3 This is a schematic diagram comparing the anomaly detection performance of the present invention and the prior art, provided as an embodiment of the present invention. Detailed Implementation

[0032] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments.

[0033] Please see Figure 1 , Figure 1 This invention provides a flowchart of a method for detecting abnormal cardiac function indicators in an animal model of heart failure. The method constructs a high-dimensional feature space with physical coherence, which can reduce QRS waveform distortion caused by sampling rate limitations in the acquisition device, as well as phase shifts between multiple source signals due to physical delays. This helps improve the sensitivity and accuracy of identifying subtle pathological waveform abnormalities in the heart failure model. The method specifically includes the following steps:

[0034] S1: Obtain the sequence of raw sample points for multi-source signals, including potential values, timestamps, and source channel indices.

[0035] It should be noted that the data acquisition step, as the logical starting point of the entire detection system, has the core task of completing the raw conversion of physical signals to digital signals and establishing a unified index space. The source channel index, as a digital label for the sampling points, is used to accurately identify the sensor physical port (such as the ECG signal channel, left ventricular pressure signal channel, or aortic pressure signal channel) to which each sampling point belongs in the mixed data stream. The physiological signals of heart failure animals are transient and multi-channel in parallel. Without a rigorous data stream initialization mechanism, subsequent algorithms will be unable to extract specific waveforms from the interwoven data stream for processing.

[0036] Based on this, the embodiments of the present invention can... The collected electrical signals are converted into a structured data set to ensure that each data sampling point has traceable physical attributes.

[0037] Specifically, embodiments of the present invention can use double buffer storage technology to convert analog electrical signals from different channels into discrete sampling point sequences, and maintain the linear arrangement of multiple signals on a unified time axis through equal-interval sampling.

[0038] For example, in an embodiment of the present invention, obtaining a sequence of original sampling points for multi-source signals, including potential values, timestamps, and source channel indexes, includes: simultaneously polling and collecting the potential values ​​of electrocardiogram (ECG) signals and pressure signals of the experimental target using an ADC acquisition device, wherein the pressure signals include left ventricular pressure signals and aortic pressure signals; marking the ECG signals and pressure signals according to their source channel indexes to finally obtain the original sampling point sequence, wherein the source channels include ECG signal channels and non-ECG signal channels; each sampling point in the original sampling point sequence has a corresponding potential value, timestamp, and source channel index.

[0039] The data acquisition frequency can be set to The corresponding sampling period for The non-ECG signal channels are the left ventricular pressure signal channels or aortic pressure signal channels. ECG signals and pressure signals together constitute multi-source signals.

[0040] Because the hardware sampling frequency is fixed, when the sampling period As the voltage increases, the loss of voltage evolution information between adjacent sampling points becomes more severe, leading to a decrease in fidelity and a negative correlation.

[0041] Thus, by synchronously acquiring multi-source data and initializing the original sampling point sequence, the embodiments of the present invention can effectively establish a data space with physical consistency, thereby providing an accurate data foundation for subsequent feature extraction.

[0042] S2: Lock the feature interval in the original sampling point sequence. The composite feature value of multiple consecutive sampling points within the feature interval is greater than the preset threshold. The composite feature value is positively correlated with the voltage change rate of the corresponding sampling point and the local variance of the sampling points within the window, and negatively correlated with the local mean.

[0043] It should be noted that the signals from animals with heart failure are often mixed with noise caused by breathing or equipment vibration, and a simple slope threshold can easily cause false triggering at baseline fluctuations. This step uses a multi-factor feature enhancement model to lock in the feature interval, specifically the QRS complex in the electrocardiogram signal. The QRS complex is a composite waveform composed of the Q wave, R wave, and S wave. It is a set of waveforms representing ventricular depolarization in the electrocardiogram and is a core feature for determining the onset of cardiac contraction. Its duration is usually only [missing information]. When locking the characteristic range, it is necessary not only to analyze the rate of change of voltage, but also to introduce local variance to evaluate the energy stability of the signal, thereby automatically distinguishing high-frequency noise from the true depolarization waveform start point.

[0044] Specifically, in this embodiment, sampling points with source channel index of ECG signal channel are obtained from the original sampling point sequence. A window of sampling points is constructed with the current sampling point as the center. The mean and variance of potential values ​​of all sampling points in the window are obtained as the local mean and local variance of the sampling points in the window, so as to calculate the composite feature value of the current sampling point.

[0045] The sampling point window can be preset to 5, but the specific setting can be adjusted according to actual needs.

[0046] For example, in an embodiment of the present invention, the composite feature value of the sampling point is obtained as follows:

[0047] ;

[0048] For the first The composite feature value of each sampling point For the first The potential value at each sampling point For the first The potential value at each sampling point The sampling period is The preset physical correction factor can be set to a range of values. to , For the first Local variance of sampling points within a sampling window For the first Local mean of sampling points within a sampling window To prevent constants with a denominator of zero, It is an absolute value function.

[0049] The local variance of the sampling points within the sampling window is used to measure the fluctuation energy of the signal. The constant can be set to... The specific settings can be customized according to actual needs. Since the intensity of myocardial electrical signals varies physically among different experimental targets (such as rats, rabbits, or dogs), the sensitivity of the detection logic is adjusted using a physical correction factor.

[0050] In this relationship, the rate of change of voltage Used to measure the intensity of instantaneous changes in electrical potential. Since the QRS complex represents ventricular depolarization, its rate of voltage change is significantly higher than the baseline signal. Therefore, the larger this value is, the greater the likelihood that the current sampling point is within the characteristic region of the ECG signal.

[0051] The term constitutes a dimensionless energy perturbation term. Since the voltage change rate increases dramatically and the signal energy distribution changes significantly when the signal enters the true depolarization region, the ratio of the potential variance to the square of the mean can amplify the energy mutation characteristics during myocardial depolarization while eliminating the influence of dimensions. By combining the energy perturbation term and the voltage change rate, isolated noise on the stable baseline can be effectively suppressed, resulting in a more focused feature interval on the true depolarization region, thus distinguishing the true myocardial depolarization waveform from isolated high-frequency noise on the stable baseline.

[0052] Based on the above steps, the composite feature value of each sampling point can be obtained. If the ECG signal enters the QRS complex, the voltage change rate increases sharply, which will lead to... The threshold is quickly exceeded, thus triggering the feature interval locking logic.

[0053] For example, in an embodiment of the present invention, locking the feature interval in the original sampling point sequence includes: if continuous The composite feature values ​​of all sampling points exceed the threshold. The sequence of sampling points is used as the feature interval, where the threshold is... , The preset number.

[0054] in, The frequency and duration of the depolarization process can be set according to the sampling frequency and the duration of the depolarization process. In this embodiment of the invention, it can be set to 6, but it can be set according to actual needs.

[0055] It should be further noted that in order to reduce the impact of QRS group waveform distortion, it is necessary to reconstruct the waveform curve of the locked feature interval. During the reconstruction process, a set of discrete skeletons is needed to support its shape, i.e., control points. Therefore, this embodiment of the invention needs to generate multiple control points within the feature interval.

[0056] For example, in an embodiment of the present invention, generating multiple control points within a feature interval includes: using the starting index of the feature interval... Starting from the sampling step size, interpolation is performed between adjacent sampling points to generate... The coordinates of control points that provide physical guidance, wherein the number of control points... satisfy , This is the terminating index of the feature interval. This is the starting index of the feature interval.

[0057] Thus, this embodiment of the invention locks the depolarization interval through multi-factor feature enhancement and generates control points to guide reconstruction, which can effectively filter high-frequency interference and provide a reliable geometric skeleton for sub-millisecond waveform restoration.

[0058] S3: Generate multiple control points within the feature interval, and construct a cubic B-spline reconstruction curve within the feature interval based on the control points; adjust the potential values ​​of the control points by minimizing the energy value of the evaluation function to obtain the simulated potential of each sampling point in the final reconstruction curve.

[0059] The energy value of the evaluation function is positively correlated with the second derivative of the reconstructed curve and with the difference between the simulated potential and the potential value at the sampling point in the reconstructed curve.

[0060] It should be noted that the feature intervals identified based on the above steps only provide the coordinates of the electrocardiogram (ECG) signal. However, since the number of sampling points is typically only 2-3, which is extremely small, it is necessary to supplement the missing information by simulating the physical evolution of energy bursts and decays during myocardial depolarization. In the processing of ECG signals in animal models of heart failure, feature sites refer to key instantaneous points with clear bioelectrophysiological significance, mainly referring to key inflection points in the QRS complex, including the Q point, R peak, and S point. However, the control points initially constructed based on the above steps are usually limited by the amplitude of the low-frequency raw sampling points and cannot directly reflect the true depolarization intensity. Furthermore, ordinary linear connections cannot simulate the smooth evolution of bioelectrical signals.

[0061] Therefore, embodiments of the present invention can establish an evolutionary reconstruction evaluation function, utilizing... Function coercion The reconstructed curve satisfies the acceleration constraint of myocardial potential evolution at the sub-millisecond level, which allows the reconstructed curve to physically compensate for the control point potential, restore the R-wave peak morphology unique to heart failure animals, and reduce the possibility of millisecond-level shifts in characteristic sites within the characteristic interval.

[0062] Specifically, since the cubic B-spline reconstruction curve is essentially a weighted sum of basis functions and corresponding control point potential coefficients, each control point potential coefficient acts as an anchor point with physical traction. Therefore, by changing the value of the control point potential coefficient, the fluctuations, slope, and curvature of the reconstruction curve within a corresponding time period can be precisely controlled. In this embodiment of the invention, the control point potential coefficient can be changed through an evaluation function, thereby constraining the curve shape. The logical basis is that the integral of myocardial potential acceleration represents the numerical energy consumed in the reconstruction process; the lower this energy, the more the waveform conforms to natural electrophysiological laws.

[0063] It should be further explained that, since the influence of sampling points at different locations within the characteristic interval on the R-wave varies, corresponding weights can be set for each sampling point.

[0064] For example, the corresponding point confidence weight can be preset according to the timestamp of the sampling point. The closer the timestamp of the sampling point is to the center timestamp within the feature interval, the greater the corresponding point confidence weight. The specific settings can be made according to actual needs.

[0065] For example, the influence of a sampling point on the morphology of the reconstructed curve can be dynamically allocated based on the dynamic activity of the sampling point within the feature interval. Specifically, the point confidence weight is used to control the degree of influence of the sampling point on the morphology of the reconstructed curve. The more drastic the signal change and the more concentrated the energy of the sampling point, the more critical the electrophysiological information it contains, and therefore the greater the corresponding point confidence weight.

[0066] For example, in an embodiment of the present invention, the point confidence weight of the sampling points within the feature interval is determined by referring to the following relation:

[0067] ;

[0068] In the above relation, For the first The point confidence weight of each sampling point For the first The composite feature value of each sampling point and These represent the maximum and minimum values ​​of the composite eigenvalues ​​of all sampling points within the feature interval. and These are the preset upper and lower weight limits.

[0069] Optionally, in embodiments of the present invention, It can be set to 1.5. It can be set to 0.5, and the specific value can be dynamically adjusted according to the noise level of the experimental environment.

[0070] In this relation, due to composite eigenvalues It is positively correlated with the rate of change of potential and local variance. Therefore, when the sampling point is located near the steep rising edge of the QRS complex or near the peak of the R wave, it leads to its corresponding The weight of the point confidence level is significantly increased. As it increases, a positive correlation is observed. The point confidence weight is used to balance the physical inertia of the curve with the fidelity of discrete sampling points in the evaluation function. When When the curve increases, the system will force a reconstruction. At this point, it is necessary to get closer to the original sampling point to prevent feature site reduction due to excessive smoothing.

[0071] Thus, by introducing dynamic point confidence weights, this embodiment of the invention effectively resolves the contradiction between physical smoothness and local fidelity during the reconstruction process, thereby providing accurate weight support for the precise restoration of sub-millisecond feature points.

[0072] For example, in an embodiment of the present invention, the method for obtaining the energy value of the evaluation function can be referred to the following relation:

[0073] ;

[0074] The energy value of the evaluation function for reconstructing the curve. To control the point potential coefficient, The potential value of the characteristic interval changes with time The acceleration of change, i.e., the second derivative of the reconstructed curve, To preset the smoothing weight, optionally, in this embodiment, it can be set to... , To control the number of points, To reconstruct the simulated potential at the sampling points in the curve, To reconstruct the potential values ​​at the sampling points in the curve, As the credibility weight of the location, This is the terminating index of the feature interval. This is the starting index of the feature interval. This is a minimization optimization algorithm.

[0075] In this relationship, due to the local support properties of cubic B-splines, when When the weight increases, it leads to The system automatically reverts to the original sampling point, allowing the reconstructed curve to restore the peak while maintaining continuous acceleration, thus exhibiting a positive correlation.

[0076] For example, in this embodiment of the invention, the simulated potential of each sampling point in the final reconstructed curve is obtained by adjusting the potential value of the control point by minimizing the energy value of the evaluation function. This includes: dynamically iteratively adjusting the potential coefficients of each control point in the feature interval by executing a minimization optimization algorithm in the energy value of the evaluation function to obtain a set of optimal coefficient vectors that minimize the total energy value of the evaluation function; based on the solved optimal coefficient vectors, generating the final reconstructed curve with continuous second derivative on the time axis by linear weighted combination of cubic B-spline basis functions to obtain the simulated potential corresponding to each sampling point in the feature interval.

[0077] The generated final reconstruction curve is as follows: ,in The characteristic interval over time The final reconstruction curve of the change, The basis functions are cubic B-spline functions, which are existing technologies and will not be elaborated upon here.

[0078] See also Figure 2 As shown, Figure 2 This is a schematic diagram of feature interval locking and sub-millisecond waveform physical reconstruction provided in an embodiment of the present invention.

[0079] Figure 2 This invention demonstrates the specific process of feature interval locking and physical reconstruction. Due to the limited sampling frequency of experimental-grade acquisition equipment, ideal bioelectrical signals suffer severe peak reduction under low-frequency discrete sampling. This embodiment of the invention generates multiple control points with physical guidance through the above steps, providing a mathematical framework for waveform evolution. Based on this, this embodiment utilizes an evaluation function model to generate a high-fidelity reconstructed curve that fits the acceleration continuity constraint of myocardial depolarization. Compared to the original discrete sampling points, the reconstructed curve recovers the lost sub-millisecond peaks, freeing feature sites from the constraints of discrete sampling values, thereby achieving accurate restoration of waveform extrema.

[0080] In this way, the embodiments of the present invention balance the physical inertia of the curve with the fidelity of discrete sampling points through the evolutionary reconstruction model, which can effectively recover the peak features lost under low-frequency sampling, thereby providing a high-fidelity signal for subsequent accurate heart rate calculation.

[0081] S4: Calculate the real-time heart rate in the reconstructed sequence and determine the phase correction offset for time alignment.

[0082] For example, calculating the real-time heart rate in the reconstructed sequence and determining the phase correction offset for time alignment includes: connecting the final reconstructed curve of the feature interval with the original sampling point sequence outside the feature interval to obtain the real-time heart rate in the reconstructed sequence; calculating the phase correction offset based on the difference between the real-time heart rate and the reference heart rate; correcting the timestamps of the sampling points of non-ECG signal channels in the reconstructed sequence; and aligning the timestamps of multi-source signals to achieve abnormal detection of cardiac function indicators.

[0083] It should be noted that the R wave is the highest-amplitude and sharpest upward peak in the QRS complex of an electrocardiogram (ECG). Physically, it represents the peak of large-scale depolarization of the ventricular myocardium and is the trigger point for the heart to contract and pump blood. Because the heart's beats are highly rhythmic, two adjacent R waves... The time difference between peak values ​​(i.e., the RR interval) physically encompasses a complete physical cycle of the heart from contraction to relaxation and then to the next contraction. Therefore, it is the most stable and reliable benchmark for determining the instantaneous cardiac cycle.

[0084] Heart rate fluctuations in animals with heart failure are dramatic, and the amount of phase drift is physically coupled with the length of the cardiac cycle. If the real-time heart rate is calculated directly from the original sampling points, errors in the instantaneous heart rate calculation will occur due to the absence or offset of the R wave peak, resulting in a directional error in the correction amount and causing a logical misalignment between the left ventricular pressure curve and the electrocardiogram curve.

[0085] Based on this, embodiments of the present invention can use the true peak values ​​of sampling points in the reconstructed sequence obtained by the above steps to calculate the real-time heart rate, ensuring the accuracy of multi-index fusion analysis.

[0086] Furthermore, this embodiment can achieve real-time heart rate calculation by extracting the extreme points of the reconstructed sequence from the feature interval.

[0087] For example, in an embodiment of the present invention, obtaining the real-time heart rate in the reconstructed sequence includes: determining the peak times of two adjacent R waves in the reconstructed sequence by using the first derivative zero-finding method, taking the difference between the peak times as the instantaneous cardiac cycle, and obtaining the real-time heart rate in the reconstructed sequence based on the instantaneous cardiac cycle.

[0088] For example, in an embodiment of the present invention, the phase correction offset is calculated based on the difference between the real-time heart rate and the reference heart rate, as shown in the following formula:

[0089] ;

[0090] This refers to the phase correction offset for non-ECG signal channels. This is the fixed delay constant for the non-ECG signal channel. To preset the heart rate regulation coefficient, For real-time heart rate, The average heart rate over the 24 hours prior to modeling the experimental target is the baseline heart rate.

[0091] In this relationship, the fixed delay constant of the non-ECG signal channel can optionally be set to to ; A preset heart rate adjustment coefficient is used to linearly convert the relative change in heart rate into a fine-tuning amount in the time dimension, thereby achieving dynamic tracking of phase shift. Its unit is milliseconds. In this embodiment of the invention, the value may optionally be [value]. .

[0092] Since an increase in heart rate alters the propagation characteristics of fluid in the duct, and real-time heart rate reflects the rate of myocardial contraction, when real-time heart rate increases, the electromechanical coupling process of the heart accelerates, leading to an increase in the phase drift of the pressure waveform. Therefore, when real-time heart rate increases, the correction offset also increases, showing a positive correlation.

[0093] Thus, the embodiments of the present invention can effectively reduce time delay through the dynamic phase correction model, thereby providing aligned physical coordinates for the subsequent extraction of cardiac function fusion indicators.

[0094] S5: Timestamp alignment of multi-source signals to enable abnormal detection of cardiac function indicators.

[0095] It should be noted that after completing sub-millisecond waveform reconstruction and phase correction in the preceding steps, the obtained reconstructed sequence curve has extremely high physical confidence. It eliminates amplitude compression under low-frequency sampling and removes physical delay artifacts. This high-quality curve provides an accurate data base for subsequent logical identification. ECG signals are electrical signals, and the propagation speed of ECG signals is close to the speed of light, with almost no delay. Pressure signals are mechanical signals transmitted through saline fluid in the catheter. In non-ECG signal channels, the fluid is damped, the catheter is elastic, and the pressure sensor experiences signal conversion delay.

[0096] Therefore, by aligning the timestamps of multi-source signals, data preparation can be made for subsequent detection of abnormal cardiac function indicators.

[0097] For example, in an embodiment of the present invention, timestamp alignment of multi-source signals includes: subtracting the phase correction offset from the sampling point timestamps of non-ECG signal channels to obtain an aligned non-ECG signal sequence.

[0098] The signal acquired by the non-ECG signal channel is the pressure signal. Because pressure waves travel relatively slowly physically, their peak value in the original sequence is lagging on the time axis. By subtracting the phase correction offset, the entire pressure curve can be shifted forward on the time axis, thereby achieving alignment of the multi-source signal sequence.

[0099] For example, in an embodiment of the present invention, cardiac function feature parameters are extracted based on the aligned multi-source signal sequence to assist in the abnormal detection of cardiac function indicators.

[0100] It is understood that, based on the above steps, the embodiments of the present invention can realize signal reconstruction and alignment of multi-source data, and the obtained data is a high-fidelity curve that has been restored and accurately aligned at the sub-millisecond level. Based on this, abnormal detection of cardiac function indicators can be accurately achieved.

[0101] Specifically, when extracting cardiac function characteristic parameters to obtain abnormal cardiac function index detection results, the peak voltage of the R-wave can be obtained based on the reconstructed sequence to reflect the electrophysiological intensity of ventricular depolarization; the electromechanical delay index can be obtained based on the time difference between the start timestamp of the ECG signal after time axis alignment and the rise timestamp of the pressure signal. If the peak voltage of the R-wave and / or the electromechanical delay index exceed the healthy baseline range of the experimental target, it is determined that the cardiac function index of the experimental target is abnormal. The healthy baseline range of the experimental target can be set according to actual needs, and this embodiment of the invention does not impose too many restrictions.

[0102] Please see Figure 3 As shown, Figure 3 This is a schematic diagram comparing the anomaly detection performance of the present invention and the prior art, provided as an embodiment of the present invention.

[0103] Figure 3 This invention visually demonstrates the performance difference between the present invention and existing technologies in heart failure diagnosis. Existing technologies suffer from aliasing due to insufficient sampling rate, resulting in significant amplitude compression. Because the peak height fails to reach the heart failure diagnostic threshold, misdiagnosis is easily made when encountering minor waveform anomalies. The present invention restores physical features through the aforementioned steps, restoring the compressed amplitude to its true level. Based on the high-quality restored curve, the extracted cardiac function feature parameters (such as peak voltage) have extremely high physical confidence. This high-quality data foundation lays a solid foundation for improving the identification of abnormal central functional indicators in heart failure models, assisting researchers in model construction.

[0104] It is understood that the data results produced by the above-mentioned steps in the embodiments of the present invention can provide staff with high-fidelity physiological signal reconstruction data and multi-source signal alignment reference, thereby providing a high-quality data foundation for the construction of animal models of heart failure.

[0105] Thus, by fusing and analyzing the characteristics of the restoration curve, the embodiments of the present invention can effectively improve the sensitivity of judging minute pathological waveform abnormalities, thereby achieving accurate monitoring of cardiac function under the heart failure model.

[0106] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for detecting abnormality of a cardiac function index of a heart failure animal model, characterized by, include: Obtain the raw sample point sequence of multi-source signals, which includes potential values, timestamps, and source channel indices; The feature interval in the original sampling point sequence is locked. The composite feature value of multiple consecutive sampling points within the feature interval is greater than a preset threshold. The composite feature value is positively correlated with the voltage change rate of the corresponding sampling point and the local variance of the sampling points within the window, and negatively correlated with the local mean. Multiple control points are generated within the feature interval, and a cubic B-spline reconstruction curve within the feature interval is constructed based on the control points. The potential values ​​of the control points are adjusted by minimizing the energy value of the evaluation function to obtain the simulated potential of each sampling point in the final reconstruction curve. The method for obtaining the energy value of the evaluation function includes: ； The energy value of the evaluation function for reconstructing the curve. To control the point potential coefficient, The potential value of the characteristic interval changes with time The acceleration of change, i.e., the second derivative of the reconstructed curve, To preset smoothing weights, To control the number of points, , These are the simulated potential and potential value at the sampling points in the reconstructed curve, respectively. As the credibility weight of the location, This is the terminating index of the feature interval. This is the starting index of the feature interval. This is a minimization optimization algorithm; The formula for calculating the confidence weight of a location is: ； In the above relation, For the first The point confidence weight of each sampling point For the first The composite feature value of each sampling point and These represent the maximum and minimum values ​​of the composite feature value of all sampling points within the feature interval. and These are the preset upper and lower weight limits; The final reconstructed curve of the feature interval is connected with the original sampling point sequence outside the feature interval to obtain the real-time heart rate in the reconstructed sequence; the phase correction offset is calculated based on the difference between the real-time heart rate and the reference heart rate, and the timestamps of the sampling points of non-ECG signal channels in the reconstructed sequence are corrected. The timestamps of the multi-source signals are aligned to realize the detection of abnormal cardiac function indicators. The final reconstruction curve is where is a final reconstruction curve that varies over time for a feature interval, is a cubic B-spline basis function.

2. The method for detecting abnormal cardiac function indicators in an animal model of heart failure according to claim 1, characterized in that, The acquisition of the original sampling point sequence of the multi-source signal, including potential values, timestamps, and source channel indexes, includes: The ADC acquisition device polls and acquires the electrocardiogram (ECG) potential value and pressure potential value of the experimental target at the same time. The pressure signal includes the left ventricular pressure signal and the aortic pressure signal. The ECG and pressure signals are marked by their source channel index to obtain the original sampling point sequence. The source channel includes ECG signal channel and non-ECG signal channel. Each sampling point in the original sampling point sequence has a corresponding potential value, timestamp and source channel index.

3. The method for detecting abnormal cardiac function indicators in an animal model of heart failure according to claim 1, characterized in that, The methods for obtaining the composite feature values ​​include: Obtain the sampling points in the original sampling point sequence whose source channel index is the ECG signal channel. Construct a window centered on the sampling point and calculate the composite feature value of the sampling point: ； , The first Composite feature value and potential value of each sampling point , The first Local variance and local mean of the sampling points within a sampling window For the first The potential value at each sampling point The sampling period is As a preset physical correction factor, To prevent constants with a denominator of zero, It is an absolute value function.

4. The method for detecting abnormal cardiac function indicators in an animal model of heart failure according to claim 1, characterized in that, The process of locking the feature regions in the original sampling point sequence includes: If continuous The composite feature values ​​of all sampling points exceed the threshold. The sequence of sampling points is used as the feature interval, where the threshold is... , The preset number.

5. The method according to claim 1, wherein the method is characterized by, The generation of multiple control points within the feature interval includes: with the start index of the feature interval generated by interpolation between adjacent sampling points according to a sampling step size a number of control points with a physical guiding effect, wherein the number of control points satisfies , is the end index of the feature interval, is the start index of the feature interval.

6. The method for detecting abnormal cardiac function indicators in an animal model of heart failure according to claim 1, characterized in that, The process of adjusting the potential value of the control point by minimizing the energy value of the evaluation function to obtain the simulated potential of each sampling point in the final reconstructed curve includes: By executing the minimization optimization algorithm in the energy value of the evaluation function, the potential coefficients of each control point in the feature interval are dynamically iteratively adjusted to obtain a set of optimal coefficient vectors that minimize the total energy value of the evaluation function. Based on the solved optimal coefficient vectors, the final reconstruction curve with the second derivative continuity on the time axis is generated by the linear weighted combination of cubic B-spline basis functions, and the simulated potentials corresponding to each sampling point in the feature interval are obtained.

7. The method for detecting abnormal cardiac function indicators in an animal model of heart failure according to claim 1, characterized in that, The process of obtaining the real-time heart rate from the reconstructed sequence includes: The peak times of two adjacent R waves in the reconstructed sequence are determined by the first derivative zero-finding method. The difference between the peak times is taken as the instantaneous cardiac cycle, and the real-time heart rate in the reconstructed sequence is obtained based on the instantaneous cardiac cycle.

8. The method for detecting abnormal cardiac function indicators in an animal model of heart failure according to claim 1, characterized in that, The calculation of the phase correction offset based on the difference between the real-time heart rate and the reference heart rate includes: ； This refers to the phase correction offset for non-ECG signal channels. This is the fixed delay constant for the non-ECG signal channel. To preset the heart rate regulation coefficient, For real-time heart rate, The average heart rate over the 24 hours prior to modeling the experimental target is the baseline heart rate.

9. The method for detecting abnormal cardiac function indicators in an animal model of heart failure according to claim 1, characterized in that, The step of aligning multi-source signals with time stamps to detect abnormal cardiac function indicators includes: Based on the aligned multi-source signal sequence, cardiac function feature parameters are extracted to assist in the detection of abnormal cardiac function indicators.

Images