A multi-channel electrophysiological mapping system and method for in-vitro cardiac tissue based on microelectrode array

Abstract

The present application relates to the technical field of medical devices, in particular to a multi-channel electrophysiological mapping system and method for isolated heart tissue based on a microelectrode array, wherein the system comprises: a microelectrode array for contacting and recording original electrical activity signals of excitable tissue or cells; a pressure detection module for collecting systolic and diastolic pressure of the heart tissue; an electrical signal acquisition module for filtering the original electrical signals transmitted by the microelectrode array in combination with the systolic and diastolic pressure to obtain net electrical activity signals, the electrical signal acquisition module comprising a first adaptive filter and a second adaptive filter, the first adaptive filter being used for motion artifact filtering of the original electrical signals according to the systolic and diastolic pressure of the heart tissue, and the second adaptive filter being used for series mode interference filtering of the original electrical signals. The above-mentioned system innovatively filters in combination with the systolic and diastolic pressure signals, effectively improving the signal-to-noise ratio and quality of the net electrical activity signals.

Description

Technical Field

[0001] This invention relates to the field of medical device technology, specifically to a multi-channel electrophysiological mapping system and method for ex vivo heart tissue based on a microelectrode array. Background Technology

[0002] In cardiac electrophysiology research, drug safety evaluation for the heart, and exploration of arrhythmia mechanisms, high-precision electrical excitation mapping of cardiac tissue is crucial. Traditional electrophysiological research tools mainly rely on patch-clamp techniques. Although this technique is the "gold standard" for studying ion channels, it requires a high level of operator skill, making it difficult to maintain stable recordings over long periods, and data reproducibility is subject to stringent experimental conditions. Furthermore, patch-clamp techniques typically record the electrical activity of only a single cell at a time, failing to simultaneously acquire spatial electrical conduction information at the cell population or tissue level. The resulting data is relatively limited in scope and cannot accurately reflect the overall electrical activity characteristics of the heart as a functional syncytiosome.

[0003] To overcome the above-mentioned shortcomings, some existing technologies have introduced microelectrode array technology into the field of electrophysiological mapping of cardiac tissue. This technology synchronously records the field potential signals of multiple sites on the tissue surface through multiple electrode contacts arranged in an array. It can intuitively depict spatial information such as the origin of excitation, conduction path, velocity and dispersion, and has significant advantages such as relatively simple operation, high data throughput and long-term stable recording.

[0004] However, when microelectrode array technology is applied to the mapping of ex vivo heart tissue, the mechanical pulsation of the heart tissue causes changes in the contact state between the microelectrode array and the heart tissue, resulting in motion artifacts in the electrical signal. On the other hand, in the electrophysiological mapping of small-sized heart tissues (such as mouse, rat, and ex vivo cultured cardiomyocyte tissues), the small size of the microelectrode array leads to cross-mode interference between the electrodes due to parasitic capacitance and electromagnetic induction. This results in poor quality of the electrical signal acquired by the microelectrode array, making it impossible to achieve high-resolution electrical signal mapping on a small area. Furthermore, while removing motion artifacts and crosstalk, it is impossible to completely preserve the true field potential signal, which is prone to signal distortion or conduction mapping interruption problems. Summary of the Invention

[0005] To address the technical problem of poor signal quality in existing microelectrode array technology for mapping isolated cardiac tissue, this application provides a multi-channel electrophysiological mapping system for isolated cardiac tissue based on a microelectrode array, comprising:

[0006] Microelectrode arrays are used to contact and record raw electrical activity signals of excitable tissues or cells;

[0007] The pressure detection module includes a miniature balloon placed in the ventricle and a high-precision pressure sensor for collecting systolic and diastolic pressures of the heart tissue.

[0008] An electrical signal acquisition module is used to filter the original electrical signal transmitted by the microelectrode array in combination with systolic and diastolic pressure to obtain a net electrical activity signal. The electrical signal acquisition module includes a first adaptive filter and a second adaptive filter. The first adaptive filter is used to perform motion artifact filtering on the original electrical signal based on the systolic and diastolic pressure of the heart tissue, and the second adaptive filter is used to perform crosstalk interference filtering on the original electrical signal.

[0009] The above system innovatively combines systolic and diastolic pressure signals, uses a first adaptive filter to dynamically remove motion artifacts based on the systolic and diastolic pressures of the heart tissue, and then uses a second adaptive filter to specifically filter out crosstalk interference in the original electrical signal. While suppressing artifacts and crosstalk, it completely preserves the real physiological field potential signal, effectively improving the signal-to-noise ratio and quality of the net electrical activity signal.

[0010] Based on the above system, this application also provides a multi-channel electrophysiological mapping method for ex vivo cardiac tissue based on a microelectrode array, comprising the following steps:

[0011] Extract the AC component or derivative features of systolic and diastolic pressures collected at multiple past sampling times, construct a pressure reference vector, and use it as the reference signal for the filter;

[0012] The raw electrical signal acquired by the target electrode is used as the main input of the filter;

[0013] The artifact interference component in the original electrical signal is estimated using the weight vector of the filter, and the artifact interference component is subtracted from the original electrical signal to obtain the artifact filtering error. The weight vector is adjusted with the minimization of the artifact filtering error as the termination condition, and the final filtered electrical signal is output.

[0014] The primary filtered electrical signal of the target electrode is used as the main input of the second adaptive filter;

[0015] A reference signal vector is constructed from the original electrical signals collected by at least one electrode adjacent to the target electrode at multiple sampling times in the past.

[0016] The crosstalk component in the first-filtered electrical signal is estimated based on the weight vector of the second adaptive filter, and the crosstalk component is subtracted from the first-filtered electrical signal to obtain the crosstalk filtering error. The iteration terminates when the crosstalk filtering error is minimized. The weight vector of the second adaptive filter is iteratively adjusted to output the final second-filtered electrical signal, i.e., the net electrical activity signal.

[0017] The aforementioned mapping method, through two-stage adaptive filtering, effectively separates and removes artifact interference and crosstalk components from adjacent electrodes in the original electrical signal, significantly improving the signal-to-noise ratio of the net electrical activity signal. The first-stage filtering accurately estimates and eliminates artifact interference, laying a clean signal foundation for subsequent processing. The second-stage filtering innovatively utilizes historical sampling data from adjacent electrodes to construct a reference signal vector and dynamically adjusts the weight vector using a least mean square algorithm, achieving adaptive suppression of crosstalk components. This hierarchical filtering strategy not only ensures the targeted and efficient nature of the filtering process but also makes the adjustment of the weight vector more targeted by using artifact filtering error and crosstalk filtering error as iteration termination conditions. The final output net electrical activity signal more realistically and accurately reflects the electrophysiological activity characteristics of isolated cardiac tissue, providing high-quality data support for subsequent mapping analysis. Attached Figure Description

[0018] Figure 1 This is an overall structural diagram of the calibration system provided by the present invention.

[0019] Figure 2 Electrochemical signal diagram of iPSC-CMs cells obtained by the system provided in this application.

[0020] Figure 3 This is an overall flowchart of the calibration method provided by the present invention. Detailed Implementation

[0021] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] In cardiac electrophysiology research, drug safety evaluation for the heart, and exploration of arrhythmia mechanisms, high-precision electrophysiological mapping of cardiac tissue is crucial. Traditional extracellular recording techniques often suffer from low spatial resolution and a limited number of channels, making it difficult to comprehensively capture the complex spatiotemporal characteristics of electrical activity in cardiac tissue. However, multi-channel electrophysiological mapping systems for ex vivo cardiac tissue based on microelectrode arrays (MEAs) can achieve simultaneous, high-resolution, multi-site recording of surface electrical activity in ex vivo cardiac tissue slices or the entire heart (such as the Langendorff perfused heart) by integrating a high-density, microscale electrode array on a chip.

[0023] Based on microelectrode array technology, this application provides a multi-channel electrophysiological mapping system for ex vivo cardiac tissue based on a microelectrode array, referencing... Figure 1 ,include:

[0024] The activity maintenance module is used to maintain the biological activity of heart tissue. It includes a temperature control unit and a perfusion unit. The temperature control unit is used to control and provide feedback on the temperature in the cell tank or bath to provide an environmental temperature that can maintain the biological activity of heart tissue (mainly in vitro cultured cardiomyocytes, etc.). The perfusion unit is used to precisely perfuse culture solution or drugs into heart tissue or the container for culturing heart tissue, and works with the temperature control unit to maintain the biological activity of heart tissue.

[0025] Microelectrode arrays are used to contact and record raw electrical activity signals of excitable tissues or cells;

[0026] The pressure detection module is used to collect the systolic and diastolic pressure of the heart tissue and the perfusion pressure of the perfusion subsystem, and feed the perfusion pressure back to the perfusion unit so that the perfusion unit can control the perfusion pressure in a closed loop.

[0027] An electrical signal acquisition module is used to filter the original electrical activity signal transmitted by the microelectrode array in combination with systolic and diastolic pressure to obtain a net electrical activity signal. The electrical signal acquisition module includes a first adaptive filter and a second adaptive filter. The first adaptive filter is used to perform motion artifact filtering on the original electrical activity signal based on the systolic and diastolic pressure of the heart tissue, and the second adaptive filter is used to perform crosstalk interference filtering on the original electrical signal.

[0028] The high-density microelectrode array (e.g., 64 to 256 channels, electrode spacing 100-200 μm) of the aforementioned system can directly and simultaneously record electrical signals at multiple sites on cell populations, tissue sections, or the whole heart, laying the foundation for comprehensive and accurate capture of the electrophysiological activity of cardiac tissue. Simultaneously, the pressure detection module not only acquires the systolic and diastolic pressures of the cardiac tissue in real time, reflecting its mechanical activity state, but also monitors and performs closed-loop control of the perfusion pressure of the perfusion subsystem, ensuring a stable perfusion environment and providing a suitable physiological environment for isolated cardiac tissue, thus guaranteeing the reliability of experimental results. The electrical signal acquisition module (front-end low-noise amplifier + 24-bit ADC, sampling rate 10-50 kHz) innovatively combines systolic and diastolic pressure signals, employing a first adaptive filter to dynamically remove motion artifacts based on the systolic and diastolic pressures of the cardiac tissue, and then using a second adaptive filter to specifically filter out crosstalk interference in the original electrical signal, effectively improving the signal-to-noise ratio and quality of the net electrical activity signal.

[0029] Specifically, the pressure detection module typically measures perfusion pressure using a separate pressure sensor (sampling rate 1kHz~10kHz, accuracy ≥12 bits, response time ≤1ms) monitored at the aortic cannulation site or in the perfusion tubing (sampling rate 1kHz, used for closed-loop control of perfusion flow rate). This reflects the driving pressure of the perfusion fluid pumped into the coronary artery. The measurement of cardiac systolic and diastolic pressure is generally achieved using a miniature balloon (outer diameter 0.3~0.8mm, length 1~3mm, made of elastic silicone; insertion method: inserted into the left ventricle via the apex, inflated slightly to conform to the ventricular wall after filling with 10~30μL of saline) placed within the ventricle. A small amount of fluid is injected into the balloon, causing it to slightly inflate. During cardiac systole, the myocardium compresses the balloon; during diastole, the myocardium relaxes, and the balloon rebounds. The change in fluid pressure within the balloon directly reflects the change in left ventricular pressure. This pressure is transmitted non-destructively to an external pressure sensor via a catheter, thus providing the cardiac systolic and diastolic pressure.

[0030] Furthermore, in order to ensure that the heart tissue remains active during the isolated heart experiment, this system also includes a temperature control module and a perfusion module. The temperature control module is used to provide an environmental temperature that can maintain the activity of isolated heart tissue, especially cultured cardiomyocytes, while the perfusion module is used to provide perfusion solution and drugs to the heart tissue for pharmacological studies.

[0031] Furthermore, some isolated heart experiments require the application of electrical stimulation to cardiac tissue or cells to induce specific electrical activities, such as action potentials or arrhythmias. Therefore, this system is also equipped with a stimulation intervention module. This module provides adjustable electrical stimulation signals, including stimulation amplitude, frequency, and pulse width, allowing for precise setting of stimulation parameters according to experimental needs, thereby meeting the requirements for studying the electrophysiological characteristics of cardiac tissue under different experimental conditions.

[0032] Furthermore, the system provided in this application also integrates a data processing module for analyzing the electrical activity characteristic parameters of cardiac tissue based on the filtered net electrical activity signal, and obtaining more information about the electrical activity of cardiac tissue, such as the location of the excitation pacemaker, conduction direction, conduction velocity, electrical conduction dispersion, repolarization dispersion, etc., which are all important parameters for cardiac research.

[0033] Specifically, in the actual application of the system, the raw electrical activity signals acquired by the microelectrode array are affected by many interference factors and contain a large amount of noise. These include cross-mode interference between microelectrodes, electromagnetic radiation from the external environment, and motion artifacts caused by the mechanical activities of the heart tissue itself (such as contraction and relaxation) (which are different from common-mode interference caused by external environmental radiation). Among these, cross-mode interference between microelectrodes is particularly prominent, especially in small, high-resolution microelectrode arrays (such as those used in small-scale scenarios like mice, rats, and cultured cardiomyocytes). Since the electrode spacing of the microelectrode array is usually small (e.g., tens to hundreds of micrometers), when one electrode records a strong electrical signal, its signal may be conducted to adjacent electrodes through the surrounding electrolyte solution or tissue, causing the signal components of that electrode to be mixed into the signals recorded by adjacent electrodes. This affects the accurate judgment of the true electrical activity of each electrode site, especially when analyzing excitation conduction pathways and locating pacemakers. This cross-mode interference will seriously reduce the spatial resolution and reliability of the mapping results.

[0034] To address this, the system provided by this invention eliminates motion artifacts in the original electrical activity signal using a first adaptive filter, specifically including the following steps:

[0035] S1. Extract the AC component or derivative features of systolic and diastolic pressures collected at multiple past sampling times, construct a pressure reference vector, and use it as the reference signal for the filter.

[0036] For example, considering that mechanical artifacts are usually caused by pressure “fluctuations” rather than the underlying hydrostatic pressure, the AC component of the systolic and diastolic pressures can be extracted as a reference signal for discretely sampled systolic and diastolic pressures. Its exchange component is:

[0037]

[0038] On the other hand, friction or displacement at the electrode-tissue interface is often most closely related to the velocity or acceleration of myocardial contraction, thus allowing for the extraction of systolic and diastolic blood pressure. The first or second derivative is used as a reference signal:

[0039] The first derivative represents the velocity characteristic of pressure change, reflecting the intensity of cardiac tissue contraction:

[0040]

[0041] The second derivative represents the acceleration characteristic of pressure changes, reflecting the instantaneous abrupt changes at the onset of cardiac tissue contraction and relaxation, which are often the moments that produce sharp artifacts.

[0042]

[0043] In practical applications, the AC component is typically linearly combined with the first / second derivative, or used as multiple independent reference inputs of the first adaptive filter to achieve optimal artifact fitting. Assuming we choose the first derivative as the core feature, the reference signal for the first adaptive filter is:

[0044]

[0045] Furthermore, in the process of converting mechanical pressure changes into electrical artifacts, there is a time delay (mechanical contraction usually lags slightly behind electrical excitation, and the sensor response also has a phase difference). Therefore, the input and reference signals of the filter cannot simply be single values ​​at the current moment. It should be a historical state vector (i.e., a pressure reference vector) containing the past L sampling times:

[0046]

[0047] Where L is the order of the filter. The higher the order, the stronger the ability to fit complex artifacts, but the greater the computational cost.

[0048] S2. Use the raw electrical signal acquired by the target electrode as the main input of the first adaptive filter. ;

[0049] S3. Estimate the artifact interference component in the original electrical signal using the weight vector of the first adaptive filter. :

[0050]

[0051] in, The weight vector of the filter The transpose of ;

[0052] The artifact filtering error is obtained by subtracting the artifact interference component from the original electrical signal.

[0053]

[0054] The weight vector is adjusted based on minimizing the artifact filtering error as the iteration termination condition:

[0055]

[0056] In the formula, This is the step size factor, which controls the convergence speed.

[0057] Once the iteration is complete, a primary filtered electrical signal with artifact interference removed can be obtained.

[0058] Furthermore, in order to address the problem of abrupt changes during severe contraction, it is possible to... Designed as a dynamic adaptive variable related to the derivative of systolic and diastolic blood pressure, for example, using a step size factor in the following form:

[0059]

[0060] In the formula, The initial step size is determined empirically. It is a relatively small constant to ensure that the step size is always a non-zero value. The adjustment coefficient is determined by the second derivative of the pressure waveform, and for example, it is inversely proportional to or has a complex exponential relationship with the second derivative of the pressure waveform, as expressed by the following normalization function:

[0061]

[0062] By designing the step size factor as a dynamic adaptive variable related to the pressure derivative, the step size factor can be adaptively reduced according to the severity of the pressure change (i.e., the magnitude of the second derivative of pressure) when the pressure signal changes abruptly due to the violent contraction of cardiac tissue. This avoids over-adjustment and convergence oscillation caused by excessively large step size at the moment of abrupt change, improves the stability of the filtering process, and ensures a relatively fast convergence speed in the steady state. Thus, while solving the problem of abrupt changes during violent contraction, it also takes into account the convergence speed and filtering accuracy of the filter.

[0063] Specifically, the process of crosstalk interference filtering performed by the second adaptive filter provided by the present invention is as follows:

[0064] S4. Use the first filtered electrical signal of the target electrode after passing through the first adaptive filter as the main input of this filter. This filtered electrical signal contains both the actual physiological signal at that location and crosstalk from other channels.

[0065] S5. Construct a reference signal vector from the original electrical signals collected by at least one electrode adjacent to the target electrode at multiple sampling times in the past;

[0066] There are two main factors contributing to crosstalk between electrodes in a microelectrode array: one is capacitive coupling (the dominant factor). The extremely close proximity of the traces between microelectrodes creates distributed capacitance, allowing high-frequency components (such as spike signals from neuronal firing) to easily crosstalk to adjacent channels through these tiny capacitances. The other is common impedance coupling. All electrodes typically share a common reference electrode (ground electrode). Due to the impedance of the electrolyte and traces, a current change in one channel can cause fluctuations in the common terminal voltage, thus affecting all channels. Based on the above analysis, crosstalk between electrodes can be characterized by an impedance network model composed of the following factors:

[0067] Electrode interface impedance : Constant phase angle element or impedance at the electrode-tissue / solution interface;

[0068] Coupling capacitor Mutual capacitance between adjacent traces or electrode points;

[0069] Solution resistance : The resistive network in the electrolyte;

[0070] Substrate capacitor If silicon-based or thin-film processes are used, the signal will also be induced through the substrate.

[0071] Each term in this impedance network model introduces crosstalk to the original electrical signal of the microelectrode array. The magnitude of the crosstalk depends not only on the specific physical parameters of the microelectrode array but also on the frequency and amplitude of the original electrical signal. Since many of the specific physical parameters of the microelectrode array are difficult to measure, the essence of iterative filtering is to mathematically inversely approximate and dynamically simulate the physical transfer function determined by this complex impedance network model. Specifically, the coupling equations between channels can be derived based on Kirchhoff's Current Law (KCL). For example, the interference contribution of channel j to channel i can be simplified as follows:

[0072]

[0073] in, The original electrical activity signal of electrode j. It is a frequency-dependent coupling coefficient, determined by the various impedances in the impedance network model described above. The weight vector, which the filter iteratively adjusts, is essentially a mathematical inverse approximation and dynamic simulation of the physical transfer function determined by the complex impedance network model. Once the weight vector approaches The filter can accurately calculate and subtract crosstalk components from the mixed signal.

[0074] Therefore, in the second adaptive filter, the original electrical signals of the N electrodes adjacent to the target electrode over the past L sampling times are used to construct a reference signal vector. It is used as a reference signal for filtering:

[0075]

[0076] in, This represents the original electrical signal of electrode 1 at time t; the meanings of the other components can be deduced similarly.

[0077] S6. Based on the weight vector of the second adaptive filter Estimate the crosstalk component in a first-order filtered electrical signal :

[0078]

[0079] The crosstalk component is then subtracted from the primary filtered electrical signal to obtain the crosstalk filtering error.

[0080]

[0081] In the weight vector Each component represents a coupling coefficient in the impedance network model.

[0082] The weight vector is iteratively adjusted using the least mean square algorithm to minimize the crosstalk filtering error.

[0083]

[0084] in, This is the iteration step size.

[0085] The final output is the secondary filtered electrical signal, i.e., the net electrical activity signal.

[0086] In this way, the filter actually "learns" the complex impedance coupling relationship between the electrodes. Whenever a signal transition is detected in an adjacent channel, the algorithm will automatically cancel out the spurious component in the target channel according to the learned coupling ratio.

[0087] Furthermore, in practical applications, the crosstalk elimination methods provided in steps S4 to S6 may occasionally filter out genuine physiological signals. The electrical excitation of cardiac tissue is conducted in a wave-like manner, meaning that the signals recorded by adjacent electrodes not only include "instantaneous crosstalk" caused by capacitive / impedance coupling, but also the genuine physiological action potentials arriving a few milliseconds later. Since the system directly feeds the raw signals from adjacent electrodes as reference signals to the second adaptive filter, if the LMS algorithm is used directly to minimize error, the filter will also treat the genuine physiological signals conducted from adjacent electrodes as noise and cancel them out, leading to severe signal distortion or a broken conduction map.

[0088] In response, this application further improves the above crosstalk filtering process. Specifically, in step S5, the sampling time of the original electrical signal participating in the construction of the reference signal vector is limited to a very short time before the current time (the crosstalk response window length). For example, setting the system sampling frequency to... (For example, 10kHz), the minimum conduction delay time of the action potential of a cardiomyocyte between adjacent electrodes is The crosstalk response window length is... Must meet:

[0089]

[0090] Reference signal vector The form becomes:

[0091]

[0092] Because crosstalk caused by capacitive and impedance coupling is almost "zero-delay" in the sampling system, while the conduction velocity of real electrophysiological excitation in tissue (typically about 0.3 ~ 1 m / s) causes a physical delay of milliseconds between adjacent electrodes (the minimum conduction delay time can be calibrated experimentally). Therefore, the reference time window of the crosstalk filter is strictly limited to include only the extremely short time when crosstalk may occur, thereby avoiding the inclusion of subsequent real physiological signals in the error cancellation process.

[0093] On the other hand, the crosstalk filtering process of the second adaptive filter can also introduce a gating mechanism to avoid erroneous elimination of real physiological electrical signals. The specific implementation method is as follows:

[0094] S7. During the process of constructing the reference signal vector in step S5, the original electrical signals of any adjacent electrode j are... Action potential wavefront detection was performed, and the absolute value of the first-order difference was used as the wavefront characteristic quantity. :

[0095]

[0096] Set wavefront recognition threshold This threshold is based on the typical action potential depolarization rate of isolated cardiac tissue. And the dynamic determination of the environmental noise baseline.

[0097] Based on wavefront identification threshold and wavefront characteristics Define gate function :

[0098] when At that time, it is determined that a physiological signal wavefront appears at the adjacent electrode j, and then let ;

[0099] when When this is determined to be background noise or crosstalk, then let ;

[0100] S8. In step S6, the gating function is combined with the iterative formula of the weight vector of the least mean square algorithm. When the wavefront feature detected by any electrode adjacent to the target electrode is greater than the wavefront recognition threshold, the gating function is zero. Only when the wavefront feature of all adjacent electrodes is less than or equal to the wavefront recognition threshold is the gating function equal to one. The iterative formula of the weight vector can be in the following form:

[0101]

[0102] In the formula, Step size factor For global gating state functions:

[0103]

[0104] Where N is the number of electrodes contained in the microelectrode array, and the value of the global gated state function is... Depending on the wavefront characteristics of all electrodes, the global gating state function takes the value of zero when the wavefront characteristics of any electrode are greater than the wavefront recognition threshold. The global gating state function takes the value of one only when the wavefront characteristics of all electrodes are less than or equal to the wavefront recognition threshold.

[0105] The amplitude of the electrical signal generated by crosstalk will be significantly smaller than that of the electrical signal generated by the electrical activity of the heart tissue. In the time domain, this is manifested as a gradient feature that is less than a certain threshold. When the fluctuation of the electrical signal exceeds the certain threshold, it indicates that the electrical signal originates from the electrophysiological activity of the heart tissue, and the physiological action potential is conducted to the reference electrode. The filter stops updating the weight vector to prevent over-filtering of the target channel's physiological signal.

[0106] Furthermore, the value of the global gating state function can be limited not only to the original electrical signal of the electrode adjacent to the target electrode, but also extended to the original electrical signal of all electrodes in the entire microelectrode array. When the wavefront feature of any electrode is greater than the wavefront recognition threshold, it indicates that the heart tissue is undergoing electrophysiological activity, and therefore the filter stops updating the weight vector.

[0107] Furthermore, to prevent misjudgment during the wide band of the action potential, when triggered... At this time, a "refractory period window" (e.g., 10~50ms) can be set to force the action potential to remain within this window. .

[0108] Through the above-mentioned gating mechanism, during the resting period without significant physiological activity, the gating function takes the value of one, the filter iterates normally, and accurately learns and tracks the capacitance and impedance coupling coefficients between the electrodes. When a strong physiological action potential is conducted to the reference electrode, the gating function takes the value of zero, the weight update term is instantly cleared, the filter stops approximating this huge "error" that represents the real physiological activity, and effectively prevents over-filtering of the physiological signal of the target channel.

[0109] Figure 2 The system provided in this application uses the above-mentioned filtering method to obtain an electrical signal image of iPSC-CMs cells. The image clearly records multiple electrical activity cycle signals, showing that the electrical signals collected by each electrode have successfully eliminated crosstalk between electrodes.

[0110] Furthermore, based on the above system, this application also provides a multi-channel electrophysiological mapping method for ex vivo cardiac tissue based on a microelectrode array, referencing... Figure 3 This includes the following steps:

[0111] A1. Extract the AC component or derivative features of systolic and diastolic pressures collected at multiple past sampling times, construct a pressure reference vector, and use it as the reference signal for the filter;

[0112] A2. Use the raw electrical signal collected by the target electrode as the main input of the filter;

[0113] A3. Use the weight vector of the filter to estimate the artifact interference component in the original electrical signal, and subtract the artifact interference component from the original electrical signal to obtain the artifact filtering error. Use the minimization of the artifact filtering error as the termination condition for the iteration, adjust the weight vector, and output the final first-stage filtered electrical signal.

[0114] A4. Use the primary filtered electrical signal of the target electrode as the main input of the second adaptive filter;

[0115] A5. Construct a reference signal vector from the original electrical signals collected by at least one electrode adjacent to the target electrode at multiple sampling times in the past;

[0116] A6. Estimate the crosstalk component in the first-filtered electrical signal based on the weight vector of the second adaptive filter, and subtract the crosstalk component from the first-filtered electrical signal to obtain the crosstalk filtering error. Use the minimum crosstalk filtering error as the termination condition for iteration, and iteratively adjust the weight vector of the second adaptive filter (such as the least mean square algorithm) to output the final second-filtered electrical signal, i.e., the net electrical activity signal.

[0117] Furthermore, in step A5 above, to prevent the filter from filtering out the real physiological signals of adjacent electrodes as noise, the sampling time of the original electrical signals of the adjacent electrodes participating in the construction of the reference signal vector is less than the crosstalk response window length, which is determined by the system sampling frequency and the minimum conduction delay time. The specific mathematical form can be found in the descriptions of steps S4 to S8.

[0118] Furthermore, gating mechanisms can be introduced to prevent the filter from treating the real physiological signals from adjacent electrodes as noise. Specifically, this includes:

[0119] A8. For any electrode adjacent to the target electrode, obtain the first-order absolute value of the original electrical signal of that electrode as a wavefront characteristic quantity.

[0120] A9. When a gating function is introduced into the weight vector update process, the gating function is set to zero when the wavefront feature of any adjacent electrode is greater than the wavefront recognition threshold, and the weight vector stops updating. When the wavefront feature of all adjacent electrodes is less than or equal to the wavefront recognition threshold, the gating function is set to one, and the weight vector is iteratively adjusted.

[0121] The specific definition and mathematical form of the gating function can be found in the descriptions in steps S4 to S9 above.

[0122] In conclusion, 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 spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A multi-channel electrophysiological mapping system for ex vivo cardiac tissue based on a microelectrode array, characterized in that, include: Microelectrode arrays are used to contact and record raw electrical activity signals of excitable tissues or cells; The pressure detection module includes a miniature balloon and a pressure sensor placed in the ventricle of the heart to collect systolic and diastolic pressures of the heart tissue. An electrical signal acquisition module is used to filter the original electrical signal transmitted by the microelectrode array in combination with systolic and diastolic pressure to obtain a net electrical activity signal. The electrical signal acquisition module includes a first adaptive filter and a second adaptive filter. The first adaptive filter is used to perform motion artifact filtering on the original electrical signal based on the systolic and diastolic pressure of the heart tissue, and the second adaptive filter is used to perform crosstalk interference filtering on the original electrical signal.

2. The system according to claim 1, characterized in that, The process of motion artifact filtering by the first adaptive filter includes the following steps: Extract the AC component or derivative features of systolic and diastolic pressures collected at multiple past sampling times, construct a pressure reference vector, and use it as the reference signal for the first adaptive filter; The original electrical signal acquired by the target electrode is used as the main input of the first adaptive filter; The artifact interference component in the original electrical signal is estimated using the weight vector of the first adaptive filter, and the artifact interference component is subtracted from the original electrical signal to obtain the artifact filtering error. The weight vector is adaptively adjusted with the goal of minimizing the artifact filtering error, and the final filtered electrical signal is output.

3. The system according to claim 2, characterized in that, The process of crosstalk filtering by the second adaptive filter includes the following steps: The primary filtered electrical signal of the target electrode is used as the main input of the second adaptive filter; A reference signal vector is constructed from the original electrical signals collected by at least one electrode adjacent to the target electrode at multiple sampling times in the past. The crosstalk component in the first-filtered electrical signal is estimated based on the weight vector of the second adaptive filter, and the crosstalk component is subtracted from the first-filtered electrical signal to obtain the crosstalk filtering error. The weight vector is adaptively adjusted with the goal of minimizing the artifact filtering error, and the final second-filtered electrical signal, i.e., the net electrical activity signal, is output.

4. The system according to claim 3, characterized in that, The sampling time of the original electrical signals of the adjacent electrodes that participate in constructing the reference signal vector is less than the crosstalk response window length, which is determined by the system sampling frequency and the minimum conduction delay time.

5. The system according to claim 3, characterized in that, The second adaptive filter's steps for crosstalk interference filtering also include: For any electrode adjacent to the target electrode, the absolute value of the first-order difference of the original electrical signal of that electrode is obtained as a wavefront characteristic quantity. A gating function is introduced into the weight vector update. When the wavefront feature of any adjacent electrode is greater than the wavefront recognition threshold, the gating function is set to zero and the weight vector stops updating. When the wavefront feature of all adjacent electrodes is less than or equal to the wavefront recognition threshold, the gating function is set to one, and the least mean square algorithm iteratively adjusts the weight vector.

6. The system according to claim 3, characterized in that, The second adaptive filter's steps for crosstalk interference filtering also include: Obtain the absolute value of the first-order difference of the original electrical signal of any electrode as a wavefront characteristic quantity; When a gating function is introduced into the weight vector update process, the gating function is set to zero when the wavefront feature of any electrode is greater than the wavefront identification threshold, and the weight vector stops updating. When the wavefront feature of all electrodes is less than or equal to the wavefront identification threshold, the gating function is set to one, and the least mean square algorithm iteratively adjusts the weight vector.

7. The system according to claim 5 or 6, characterized in that, When the gating function is zero, a time window is started to keep the gating function at zero throughout the action potential refractory period time window.

8. The system according to claim 1, characterized in that, The microballoon has an outer diameter of 0.3~0.8mm, a length of 1~3mm, and is made of silicone. After being filled with 0.5~5μL of physiological saline, it can conform to the ventricular wall.

9. A multi-channel electrophysiological mapping method for ex vivo cardiac tissue based on a microelectrode array, used to control the system according to any one of claims 1-8, characterized in that, Includes the following steps: Extract the AC component or derivative features of systolic and diastolic pressures collected at multiple past sampling times, construct a pressure reference vector, and use it as the reference signal for the filter; The raw electrical signal acquired by the target electrode is used as the main input of the filter; The artifact interference component in the original electrical signal is estimated by using the weight vector of the filter, and the artifact interference component is subtracted from the original electrical signal to obtain the artifact filtering error. The weight vector is adaptively adjusted with the goal of minimizing the artifact filtering error, and the final filtered electrical signal is output. The primary filtered electrical signal of the target electrode is used as the main input of the second adaptive filter; A reference signal vector is constructed from the original electrical signals collected by at least one electrode adjacent to the target electrode at multiple sampling times in the past. The crosstalk component in the first-filtered electrical signal is estimated based on the weight vector of the second adaptive filter, and the crosstalk component is subtracted from the first-filtered electrical signal to obtain the crosstalk filtering error. The weight vector is adaptively adjusted with the goal of minimizing the artifact filtering error, and the final second-filtered electrical signal, i.e., the net electrical activity signal, is output.

10. The method according to claim 9, characterized in that, The sampling time of the original electrical signals of the adjacent electrodes that participate in constructing the reference signal vector is less than the crosstalk response window length, which is determined by the system sampling frequency and the minimum conduction delay time.

11. The method according to claim 9, characterized in that, The second adaptive filter's steps for crosstalk interference filtering also include: For any electrode adjacent to the target electrode, the absolute value of the first-order difference of the original electrical signal of that electrode is obtained as a wavefront characteristic quantity. When a gating function is introduced into the weight vector update process, the gating function is set to zero when the wavefront feature of any adjacent electrode is greater than the wavefront recognition threshold, and the weight vector stops updating. When the wavefront feature of all adjacent electrodes is less than or equal to the wavefront recognition threshold, the gating function is set to one, and the weight vector is iteratively adjusted.

Images