Noise filtering for electrophysiological signals

Abstract

Systems and methods for estimating and filtering electrophysiological signals are described herein. In some examples, a noise filtering system can receive at least one electrophysiological signal. A signal segment extractor of the system can extract a signal segment of interest from the electrophysiological signal. The system employs a signal segment noise calculator to evaluate the extracted signal segment of interest to estimate noise in the signal segment of interest. The estimated noise can be provided to a signal segment filter of the system to determine a substitute noise estimate for at least one remaining signal segment of the electrophysiological signal for use in noise filtering the at least one remaining signal segment. The signal segment noise calculator can be configured to filter the signal segment of interest based on the estimated noise, and can combine the filtered signal segments to provide a filtered electrophysiological signal.

Description

Technical Field

[0001] This disclosure relates to noise filtering of electrophysiological signals. Background Technology

[0002] Electrophysiological signals are sensed in a variety of applications, including electroencephalography (EEG), electromyography (EMG), electrocardiography (ECG), and electrooculography (EOG). These signals are frequently contaminated by noise, such as power line noise. This contamination can degrade signal quality, which in some applications can affect subsequent signal processing and diagnosis. Summary of the Invention

[0003] This disclosure relates to noise filtering of electrophysiological signals.

[0004] In one example, one or more non-transitory computer-readable media may include data and machine-readable instructions executable by a processor. The data may include electroanatomical data that can characterize electrophysiological signals measured from a patient. The machine-readable instructions may include: a signal segment extractor programmed to extract a signal segment of interest from the electrophysiological signal; a signal segment noise calculator programmed to evaluate the extracted signal segment of interest to estimate noise in the signal segment of interest; and a signal segment filter programmed to determine an alternative noise estimate for at least one remaining signal segment of the electrophysiological signal and to filter the at least one remaining signal segment based on the alternative noise estimate to remove noise therefrom.

[0005] In another example, a system may include: at least one sensor configured to measure at least one electrophysiological signal from a location on tissue associated with a patient; a memory configured to store machine-readable instructions and data representing the measured at least one electrophysiological signal; and at least one processor configured to access the memory and to execute the machine-readable instructions. The machine-readable instructions may include: a signal segment extractor programmed to evaluate the signal morphology of the at least one electrophysiological signal to identify a signal segment of interest; and a signal segment noise calculator programmed to convert the signal segment of interest into corresponding frequency domain data having discrete frequency bins of the signal in the signal segment of interest, and to evaluate the frequency domain data to estimate noise in the signal segment of interest. The machine-readable instructions may also include a signal segment filter programmed to calculate an alternative noise estimate of at least one remaining segment of the at least one electrophysiological signal based on the estimated noise in the signal segment of interest, and to remove noise in the at least one remaining segment based on the alternative noise estimate and to remove the noise in the signal segment of interest based on the estimated noise, to provide a noise-filtered version of the at least one electrophysiological signal.

[0006] In another example, one approach includes: extracting a segment of interest from an electrophysiological signal measured from a patient; and converting the segment of interest into corresponding frequency domain data using a discrete Fourier transform (DFT), the corresponding frequency domain data having discrete frequency ranges of the signal in the segment of interest. The transformation may include calculating a set of DFT coefficients for each of these signals used for the segment of interest. The approach may further include: evaluating the frequency domain data to identify frequencies of noise signals in the signal within the segment of interest; selecting DFT coefficients of the noise signals from the set of DFT coefficients based on the identified frequencies to estimate noise in the segment of interest; determining an alternative noise estimate for at least one remaining segment of the electrophysiological signal; and filtering the at least one remaining segment based on the alternative noise estimate to remove noise therefrom. Attached Figure Description

[0007] Figure 1 An example of a noise filtering system for electrophysiological signals is depicted.

[0008] Figure 2 An example of a signal processing system for electrophysiological signals is described.

[0009] Figure 3 Examples of electrophysiological signals and spectral mapping are depicted.

[0010] Figures 4A-4E An example of electrophysiological signal and noise filtering mapping is shown.

[0011] Figure 5 An example of electrophysiological signal mapping is shown.

[0012] Figure 6 An example of an electrophysiological monitoring system that can implement electrophysiological noise filtering is described.

[0013] Figure 7 An example of electrophysiological mapping is shown.

[0014] Figure 8 Another example of electrophysiological mapping is shown.

[0015] Figure 9 An example of a method for filtering noise from electrophysiological signals is shown. Detailed Implementation

[0016] This disclosure relates to signal filtering that can be applied to reduce noise from sensed electrophysiological signals. For example, an electrophysiological noise filter can be configured to remove white noise (e.g., Gaussian noise), fixed-frequency noise (e.g., electric field noise), harmonic noise, and / or transient noise (e.g., spike noise). As used herein, the term "noise" can refer to unwanted signals or artifacts in an electrophysiological signal.

[0017] In some examples, systems and methods are provided for removing noise from electrophysiological signals, such as those that can be stored in memory as electrical data. The electrical data may correspond to a digital representation of one or more electrophysiological signals that have been measured from an individual's body via one or more electrodes or other sensors (e.g., contact or non-contact sensors), such as monopolar or bipolar electrographs or other electrophysiological signals. For example, a signal segment extractor (e.g., executable program code) is employed to extract a segment of interest from the electrophysiological signal and the extracted segment is provided to a signal segment noise calculator. In some examples, the identified segment of interest may correspond to a portion of the electrophysiological signal that does not contain noise distortion features (such as transient features (e.g., spikes) and / or QRS features).

[0018] A signal segment noise calculator (e.g., executable code) can be configured to evaluate extracted signal segments of interest and estimate noise in those segments. A signal segment filter (e.g., executable code) can be configured to remove noise from at least one remaining segment of the electrophysiological signal based on the estimated noise in the signal segment of interest. For example, the signal segment filter is configured to compute alternative estimates for the remaining signal segments. Thus, noise identified in the signal segment of interest can be used to provide alternative noise estimates for each remaining signal segment of interest, and the signal segment filter can be configured to remove noise from each remaining signal segment based on the alternative noise estimates. The signal segment filter can be configured to remove noise from the remaining signal segments based on the alternative noise estimates and to remove noise from the signal segment of interest based on the estimated noise, thereby removing noise (e.g., electric field noise) from the electrophysiological signal. By employing a portion of the electrophysiological signal that does not include noise distortion features for noise estimation, the accuracy of noise estimation in the electrophysiological signal can be improved, and thus diagnostic accuracy can be improved.

[0019] For example, the systems and methods disclosed herein are used as part of a diagnostic and / or therapeutic workflow to identify and treat arrhythmias (e.g., cardiac resynchronization therapy (CRT), premature ventricular contractions (PVC), ventricular tachycardia (VT), and premature atrial contractions (PAC)) based on electrical activity acquired from a patient. In some examples, patient electrical activity includes noninvasive body surface measurements of body surface electrical activity. Alternatively or concurrently, patient electrical activity may include invasive measurements of cardiac electrical activity, including epicardial measurements and / or endocardial measurements. While many examples herein are described in the context of cardiac electrical signals, it should be understood that the methods disclosed herein are equally applicable to other electrophysiological signals, such as electroencephalography (EEG), electromyography (EMG), electrooculography (EOG), etc.

[0020] Figure 1An example of a noise filtering system 100 that can be configured to filter input electrical signal data 102 to remove noise therein is depicted. The noise filtering system 100 can be implemented as hardware (e.g., circuitry and / or devices), software (e.g., a non-transitory medium having machine-readable instructions), or a combination of hardware and software. The input electrical data 102 can be stored in memory and corresponds to one or more electrophysiological signals, such as physiological signals acquired by one or more sensors. For example, sensors can be applied to non-invasively measure electrical activity, such as by being positioned on the surface of a patient's body, such as the patient's head (e.g., for electroencephalography), the patient's chest cavity (e.g., for electrocardiography), or other non-invasive locations. Examples of sensors that can be used to acquire electrical activity on the body surface are disclosed in U.S. Patent No. 9,549,683 and International Application PCT / US20091063803. In other examples, the input electrical signal data 102 is acquired invasively, such as by one or more electrodes positioned within the patient's body (e.g., on a lead or basket catheter, etc., during an EP study). In other examples, the input electrical signal data 102 includes both non-invasively acquired electrical signals and invasively acquired electrical signals. In some examples, such as acquiring one or more electrophysiological signals in real time during the procedure.

[0021] The noise filtering system 100 can be used to remove noise from one or more electrophysiological signals before low-pass filtering and transient feature identification and / or removal (such as spikes, e.g., pacing spikes). Therefore, in some examples, the noise filtering system 100 is used to provide initial line filtering before applying other types of filtering to one or more electrophysiological signals. In some examples, the noise filtering system 100 can be used to remove unwanted signals from other signals generated by signal sources such as therapeutic devices or navigation systems.

[0022] The noise filtering system 100 can be configured to evaluate one or more signal segments before and / or after a signal segment of interest in one or more electrophysiological signals to remove noise from these signal segments based on the estimated noise of the signal segment of interest. As used herein, the term "segment of interest" can refer to a portion of an electrophysiological signal in the frequency domain that has discrete frequency components that are not within a corresponding frequency interval (e.g., frequency range). The signal segment of interest can be selected by the user (e.g., in response to user input) or automatically.

[0023] The corresponding frequency range can be defined by the user based on the noise filtering application for which the noise filtering system 100 will be used. In some examples, if the noise filtering system 100 is used as part of an electrophysiological monitoring system, a given frequency range corresponds to a frequency range excluding the QRS signal component of the electrophysiological signal. Therefore, in some examples, the corresponding frequency range may correspond to a frequency range from about 0 Hz to about 20 Hz. Thus, in some examples, the noise filtering system 100 is configured to filter line noise, such as power line noise with a frequency of about 50 Hz (e.g., as in Europe, China, India, etc.) or about 60 Hz (e.g., as in the United States).

[0024] In some examples, the noise filtering system 100 may include a machine learning algorithm configured to improve the filtering of the noise filtering system 100 based on historical noise filtering data (e.g., previous noise filtering settings). Any of a variety of techniques can be used for the machine learning algorithm, including support vector machines, regression models, self-organizing maps, fuzzy logic systems, data fusion processes, rule-based systems, or artificial neural networks. In other examples, different machine learning algorithms may be used. The machine learning algorithm can be trained based on historical noise filtering data to adjust the noise filtering settings of the noise filtering system 100. The noise filtering settings may include signal morphological characteristics, such as those used herein for signal segment extraction. In other examples, other settings of the noise filtering system 100 may be adjusted using machine learning algorithms (e.g., fill functions, transform functions, window sizes, etc.).

[0025] For example, noise filtering system 100 includes a signal segment extractor 104 for extracting a segment of interest from an electrophysiological signal. For instance, signal segment extractor 104 includes program code configured to evaluate portions of the electrophysiological signal based on a moving window function 106. Moving window function 106 may have a defined window size representing the number of samples (e.g., sampling frequency) and duration. Moving window function 112 may include a Hann window, Hamming window, Blackman window, Nuttall window, Blackman-Nuttall window, Blackman-Harris window, or another window type function for signal sampling. Moving window function 106 may be configured to slide relative to the electrophysiological signal to sample at least one portion of the signal. Signal segment extractor 104 may be configured to evaluate each sampled portion to determine whether the corresponding portion of the electrophysiological signal will be identified or labeled as a segment of interest.

[0026] For example, signal segment extractor 104 is configured to evaluate the signal characteristics (e.g., morphology) of each signal segment to identify the signal segment of interest. In some examples, signal segment extractor 104 is configured to compare the amplitude and / or slope of each signal segment relative to a signal threshold (e.g., amplitude and / or slope threshold) to identify the signal segment of interest. The corresponding signal segment can be identified or labeled as the signal segment of interest based on this comparison. For example, if the amplitude of a corresponding signal segment is less than or equal to an amplitude threshold, that corresponding signal segment is identified or labeled as the signal segment of interest. In another or alternative example, if the slope of a corresponding signal segment is greater than or equal to a slope threshold, that corresponding signal segment can be identified or labeled as the signal segment of interest. Therefore, in some examples, signal segment extractor 104 is configured to evaluate electrophysiological signals to identify signal segments that do not include noise distortion features (e.g., transient features such as spikes) and / or QRS signal components.

[0027] In some examples, the signal segment extractor 104 is configured to extract signal segments of interest for noise estimation based on user input data 108. The user input data 108 may identify the signal segment processing interval (e.g., a time interval relative to the electrophysiological signal) used to extract the signal segments of interest. For example, the intervals of the electrophysiological signal relative to time are presented on a display via a graphical user interface (GUI). The GUI may generate an electrophysiological signal with sliding scales that the user can interact with (e.g., via a user input device providing user input data 108) to select the signal segments of interest and thus identify the segment processing intervals used for extraction by the signal segment extractor 104.

[0028] In some examples, electrical signals of different frequencies may be applied to the patient's body. For example, the electrical signals may be generated by a pacing device (e.g., a pacemaker). Input electrical signal data 102 may include data characterizing the applied electrical signal. Signal segment extractor 104 may be configured to evaluate the applied electrical signal to determine common noise in the applied electrical signal. Signal segment extractor 104 may be configured to extract the signal segment of interest based on the determined common noise in the applied electrical signal.

[0029] The signal segment extractor 104 can be configured to provide the signal segment of interest to the signal segment noise calculator 110 for noise estimation. In some examples, the signal segment extractor 104 is configured to provide signal timing information that identifies the start and stop times of the signal segment processing interval. The signal segment noise calculator 110 can be configured to use the signal timing information to retrieve the signal segment of interest stored in memory. The signal segment noise calculator 110 can be configured to use a fixed window function 112 to sample the signal segment of interest.

[0030] The signal segment noise calculator 110 can be configured to apply a transform function 112 to transform the signal segment of interest into a frequency domain representation. For example, the transform function 112 can be configured to apply a Discrete Fourier Transform (DFT) to transform the signal segment of interest into a frequency domain representation. The signal segment noise calculator 110 can be configured to identify signals present in the signal segment of interest and represent the signals in the frequency domain as a discrete frequency domain representation. The signal segment noise calculator 110 can be configured to calculate the set of DFT coefficients for each signal present in the signal segment of interest. The signal segment noise calculator 110 can be configured to calculate the corresponding DFT coefficients for each signal in the signal segment of interest to convert the signal segment of interest into corresponding frequency domain data. The corresponding frequency domain data can have discrete frequency ranges of the signals in the signal segment of interest.

[0031] Each DFT coefficient provided by the signal segment noise calculator 110 can specify the amplitude and phase in a discrete frequency domain representation, which together define the signal segment of interest in the time domain. The signal segment noise calculator 110 can be configured to represent the signal segment of interest in a complex exponential manner according to the DFT series synthesis equation:

[0032]

[0033] Where S[k] are the DFT coefficients, N is the period of the discrete-time signal s[n], and k is the exponent.

[0034] The signal segment noise calculator 110 can be configured to determine the set of DFT coefficients based on the DFT analysis equation:

[0035]

[0036] in <n>Indicates the sum over any N consecutive integers.

[0037] The signal segment noise calculator 110 can be configured to identify, from the set of DFT coefficients, DFT coefficients representing noise that can be applied to the signal segment of interest. For example, the signal segment noise calculator 110 is configured to use the set of DFT coefficients to convert the signal segment of interest into a corresponding frequency domain representation. The frequency domain representation may correspond to a frequency domain spectrum representing the power of the frequency content present in the signal segment of interest. The signal segment noise calculator 110 can be configured to evaluate the frequency domain spectrum to determine the frequency of the noise signal.

[0038] For example, the signal segment noise calculator 110 is configured to compare each frequency interval in the spectrum with a frequency interval threshold. In some examples, the signal segment noise calculator 110 is configured to set the frequency interval threshold based on a frequency interval criterion. The frequency interval criterion may specify the frequency of the noise signal (e.g., the frequency corresponding to power line noise). In some examples, the frequency interval criterion is provided as user input data 108 or as part thereof. The signal segment noise calculator 110 may be configured to identify the frequency of interest in response to determining that the frequency of interest corresponding to the frequency of the noise signal is equal to the frequency interval threshold.

[0039] The signal segment noise calculator 110 can be configured to select DFT coefficients of a noise signal from the set of DFT coefficients based on an identified frequency of interest to estimate the noise in the signal segment of interest. Therefore, the signal segment noise calculator 110 can be configured to estimate the DFT coefficients of the noise signal, and thereby estimate the phase and amplitude of the noise signal present in the signal segment of interest based on the identified frequency of interest. Thus, the signal segment noise calculator 110 can be configured to estimate the noise in the signal segment of interest, which can be stored in memory as signal segment noise data.

[0040] In some examples, the signal segment noise calculator 110 can be configured to apply a fill function 114 to the estimated noise, which corresponds to filling the estimated noise to change the frequency resolution of the estimated noise. Because the extended window function 116 used to sample at least one remaining segment of the electrophysiological signal can have a different resolution than the estimated noise, the signal segment noise calculator 110 can be configured to fill the estimated noise to provide a filled version of the estimated noise. For example, the fill function 114 is configured to scale the DFT coefficients by a given window scaling value based on the window characteristics of the extended window function 116 to provide scaled DFT coefficients, which corresponds to changing the frequency resolution of the estimated noise.

[0041] A segment noise calculator 110 can be configured to provide a scaled estimated noise (corresponding to scaled DFT coefficients) to a segment filter 118. The segment filter 118 can be configured to apply an extended window function 116 to select at least one remaining segment of the electrophysiological signal for noise filtering. The segment filter 118 can employ a noise estimation function 120 configured to provide an alternative noise estimate for at least one remaining segment based on the scaled estimated noise. For example, the noise estimation function 120 is configured to interpolate the scaled estimated noise in the segment of interest into at least one remaining segment of the electrophysiological signal, extrapolating the noise in the at least one remaining segment. In some examples, the noise estimation function 120 is configured to extend the phase and magnitude of the scaled estimated noise and thus interpolate that phase and magnitude into at least one remaining segment of the electrophysiological signal, extrapolating the noise in the at least one remaining segment of the electrophysiological signal. The alternative noise estimate can represent the interpolated phase and magnitude of the scaled estimated noise in the remaining at least one segment of interest.

[0042] In some examples, the noise estimation function 120 is configured to estimate noise forward and backward relative to the signal segment of interest based on a scaled estimated noise to provide alternative noise estimates for the remaining signal segments. For example, if the signal segment of interest is located in time between a first remaining signal segment and a second remaining signal segment, the noise estimation function 120 is configured to predict backward based on the scaled estimated noise relative to the signal segment of interest to provide an alternative noise estimate for the first remaining signal segment, and predict forward to provide an alternative noise estimate for the second remaining signal segment. Thus, the noise estimation function 120 can provide an alternative noise estimate for each remaining signal segment based on the estimated noise of the signal segment of interest.

[0043] In some examples, the signal segment filter 118 includes a segment filtering function 122 configured to remove noise from each remaining signal segment based on an alternative noise estimate provided by the noise estimation function 120. The segment filtering function 122 can be configured to subtract the alternative noise estimate of each remaining signal segment of interest from the corresponding signal segment to filter (e.g., remove) the noise therein and provide a filtered remaining signal segment. For example, the segment filtering function 122 is configured to subtract noise from the signal segment of interest based on the estimated noise to filter the signal segment of interest against noise, thereby providing a filtered signal segment of interest. The segment filtering function 122 can be configured to combine (e.g., suture) the filtered remaining signal segments and the filtered signal segments to provide a noise-filtered representation of the electrophysiological signal. The noise-filtered electrophysiological signal can be provided (and stored in memory) as noise-filtered signal data 124.

[0044] The noise-filtered signal data 124 can be used for further signal processing (e.g., signal conditioning), such as low-pass filtering or other filtering, to provide filtered electrophysiological signal data. Additional signal processing techniques can also be used to provide such filtered electrophysiological signal data. As an example, signal processing techniques may include electrogram reconstruction on the epicardium or other capsules, such as by solving an inverse solution based on geometric and electrical data measured on the body surface.

[0045] In some examples, the noise filtering system 100 is configured to filter noise in each channel used for measuring (e.g., capturing) electrical activity from a person based on estimated noise in the signal segment of interest. For example, the measurement system may be used to capture electrical activity from a human body via a sensor (e.g., an electrode). Each sensor may thus define a corresponding channel. The measurement system may be configured to provide the electrical activity captured from the human body for each channel as part of the input electrical signal data 102. A given electrophysiological signal provided by the corresponding channel may be selected and analyzed by the noise filtering system 100 as described herein to estimate the noise in the signal segment of interest for the given electrophysiological signal. In some examples, the electrophysiological signals of any number of channels may be presented on a display relative to time by a GUI with graphical elements, which the user can interact with to select a given electrophysiological signal for noise estimation. In other examples, a signal segment extractor 104 is configured to select a given electrophysiological signal.

[0046] The noise filtering system 100 can be configured to estimate the noise in each channel based on the estimated noise in the signal segment of interest for the respective channel. For example, the noise estimation function 120 is configured to interpolate the estimated noise in the signal segment of interest for a given electrophysiological signal into each electrophysiological signal from the corresponding remaining channel, as described herein, and extrapolate the noise in each electrophysiological signal. The segment filtering function 122 can then remove the noise from each electrophysiological signal from the corresponding remaining channel based on the noise estimated by the noise estimation function 120 to provide a noise-filtered electrophysiological signal. In some examples, the noise-filtered electrophysiological signal can be provided as noise-filtered signal data 124.

[0047] In some examples, sensors may be arranged on the surface of a person's body, and for each of multiple spatial regions, the sensors may be grouped into two or more suitable subsets. Each spatial region may include a subset of sensors. Electrophysiological signals from a given sensor in each spatial region may be selected and evaluated by a corresponding noise filtering system (such as noise filtering system 100) to estimate noise in the signal segment of interest of the electrophysiological signal. Each corresponding noise filtering system may use the estimated noise to filter the electrophysiological signal from the given sensor and the electrophysiological signals from the remaining sensors in each corresponding spatial region to provide a noise-filtered electrophysiological signal. Thus, in some examples, for each spatial region, a corresponding noise filtering system (such as noise filtering system 100) may be employed to provide noise estimation and filtering in the same or similar manner as described herein.

[0048] In some examples, the analysis system described herein can be configured to generate a first graphic mapping of electroanatomical activity based on electrophysiological signals provided by sensors in one or more first spatial regions among a plurality of spatial regions. The analysis system can be configured to generate a second graphic mapping of electroanatomical activity based on electrophysiological signals provided by sensors in one or more second spatial regions among a plurality of spatial regions. The analysis system can be configured to evaluate the first and second graphic mappings of electroanatomical activity to determine whether noise has been adequately filtered for the respective one or more spatial regions among the plurality of spatial regions. For example, if the difference between the first and second graphic mappings is greater than a difference threshold, the analysis system can be configured to output an indication on a GUI that one of the first or one or more second spatial regions is contaminated by noise. In some examples, the analysis system can be configured to cause the noise filtering system 100 to extract a signal segment different from the signal segment of interest used for noise estimation.

[0049] Therefore, the noise filtering system 100 can be used to improve the signal quality of electrophysiological signals in the initial preprocessing stage, which can reduce signal morphology that may lead to misdiagnosis of arrhythmias. For example, electrograms are often contaminated by power line noise. Power line noise can overlap with the frequency of the QRS component of the electrogram, which can distort the power line noise. Using the entire electrogram for noise estimation will result in incorrect line filtering results (e.g., incorrect nose estimation) because the noise may be distorted by the QRS portion of the electrogram. Furthermore, such as with electrograms Figure 1 Transient events such as pacing spikes during capture can also distort power line noise estimation. By employing a noise filtering system 100 to provide power line noise estimation of the electrogram based on a portion of the electrogram excluding QRS signal components and / or transient events, accurate estimation of power line noise is allowed. Therefore, the noise filtering system 100 can accurately estimate power line noise within the electrogram for improved power line noise filtering.

[0050] Figure 2 An example of a signal processing system 200 for electrophysiological signals is depicted. In some examples, the signal processing system 200 may be implemented on a processing device (e.g., a digital signal processor, a field-programmable gate array, a computer, or other processing equipment). The signal processing system 200 can be used to process electrophysiological signals provided by a measurement system via a corresponding channel, which is used to capture electrical activity from a human body via one or more sensors. Thus, in examples employing multiple sensors to capture electrical activity from a human body, a corresponding signal processing system 200 may be employed to provide the signal processing as described herein.

[0051] exist Figure 2 In one example, the signal processing system 200 includes a noise filtering system 202 that receives input electrical signal data 204. In some examples, the noise filtering system 202 may correspond to, for example... Figure 1 The noise filtering system 100 is shown. Therefore, in Figure 2 The following description of the example can be referenced. Figure 1 Examples. Noise filtering system 202 can be configured to filter input electrical signal data 204 to provide noise-filtered signal data 206, as described herein with respect to noise filtering system 100. Therefore, in some examples, input electrical signal data 204 and noise-filtered signal data 206 can correspond to input electrical signal data 102 and noise-filtered signal data 124, as... Figure 1 As shown herein. For example, noise filtering system 202 is configured to receive electrophysiological signals and filter the electrophysiological signals based on estimated noise from a given segment of interest of the electrophysiological signal, as described herein.

[0052] Signal processing system 200 includes a transient feature detector 208, which can be configured to identify the location of one or more transient features (e.g., spike noise) that may be present in each electrophysiological signal. For example, a transient feature is a spike that may correspond to a naturally occurring biological event (e.g., a cardiac arrhythmia such as fibrillation), or the spike may be a pacing spike induced by a device. In some examples, transient feature detector 208 is programmed to transmit transient location information of transient features in a given electrophysiological signal to noise filtering system 202 for noise estimation. For example, noise filtering system 202 is configured to use the transient location information of transient features to extract the signal segment of interest from a given electrophysiological signal. The transient location information may include timing or spatial location information of the transient feature.

[0053] In some examples, the transient feature detector 208 includes a transient feature removal function 210. The transient feature removal function 210 can be configured to remove transient features from the noise-filtered signal data 206 based on transient location information. The transient feature removal function 210 can be programmed to pass the noise-filtered signal data 206, which contains no transient features, to a low-pass filter 212. The low-pass filter 212 can be configured to attenuate or block frequencies above a predetermined cutoff frequency in the noise-filtered signal data 206 to remove high-frequency signals (e.g., muscle artifacts and / or external interference). The low-pass filter 212 can be configured to provide filtered input electrical signal data 214. Additional signal processing techniques can be applied to the filtered input electrical signal data 214. For example, electrogram reconstruction can be performed on the epicardium or other capsules, such as by solving an inverse kinematics based on geometric data and the filtered input electrical signal data 214.

[0054] In some examples, the low-pass filter 212 can be implemented as an elliptic low-pass filter. By combining the noise filtering system 202 with the elliptic low-pass filter, high-frequency components in the graphical mapping of the heart surface that can be generated by a mapping generator such as those described herein can be reduced, resulting in a more accurate representation of electrical activity following the spikes, which improves patient diagnosis.

[0055] Figure 3 An example of electrophysiological signals and spectral mapping 300 is depicted. First mapping 302 shows a composite map of electrophysiological signals measured from a human body (e.g., via sensors distributed on the surface of a patient's body). Second mapping 304 shows the spectral response of the electrophysiological signals in first mapping 302, highlighting the QRS component of each electrophysiological signal in first mapping 302. Third mapping 306 shows the spectral response of the electrophysiological signals in first mapping 302, highlighting the spike components within each electrophysiological signal. As described herein, noise filtering systems (e.g., such as...) Figure 1 The noise filtering system 100 shown or such Figure 2 The noise filtering system 202 shown can be used in the first plot 302 to apply a moving window function (e.g., short window DFT (SWDFT)) from left to right to identify the signal segment 308 of interest for a given electrophysiological signal.

[0056] like Figure 3 As shown, the signal segment of interest 308 can be located between a first residual signal segment of interest 310, which includes QRS components, and a second residual signal segment 312, which includes spike components. A noise filtering system (e.g., systems 100, 202) can use the signal segment of interest 308 to estimate the noise in the signal segment of interest 308, and use the estimated noise to filter noise in the residual signal segments 312 and 314. The noise filtering system can also remove noise from the signal segment of interest 308 based on the estimated noise. The noise filtering system can combine the noise-filtered segments 308, 310, and 312 to provide a noise-filtered electrophysiological signal for further signal processing (e.g., spike component removal, low-pass filtering, etc.). In some examples, the noise filtering system can be configured to use the estimated noise from the signal segment of interest 308 to filter noise in each residual electrophysiological signal in the first mapping 302 for further signal processing, as described herein.

[0057] Figures 4A-4E An example of electrophysiological signal and noise filtering mapping 400 is depicted. Figures 4A-4E In the examples, the first plot 402 shows an unfiltered electrophysiological signal measured from the human body (e.g., via a sensor), and the second plot 404 shows an example of estimated system (e.g., global) noise in the electrophysiological signal 402. The third plot 406 shows an example of a system noise-filtered version of the electrophysiological signal in the first plot 402 based on the estimated system noise removed.

[0058] Figure 408 depicts an example of the estimated noise for the signal segment of interest, and Figure 410 shows an example of a noise-filtered version of signal 402 in response to the removal of the estimated noise 408 according to the method described herein. For example, a noise filtering system (e.g., such as...) Figure 1 The noise filtering system 100 shown or such Figure 2 The noise filtering system 202 shown can be configured to receive the electrophysiological signal from the first figure 402 and calculate the estimated noise as shown in Figure 408 based on the segment of interest of the electrophysiological signal. The noise filtering system can use the estimated noise to remove noise from the segment of interest and the remaining segment of the electrophysiological signal to provide a noise-filtered electrophysiological signal, as shown in Figure 410.

[0059] Figure 5 Another example of electrophysiological signal mapping 500 is depicted. In some examples, multiple electrophysiological signals measured from the human body can be provided to a GUI for display on a monitor. First mapping 502 shows a composite map of multiple electrophysiological signals without spike components displayed on a monitor by the GUI. Second mapping 504 shows a composite map of multiple electrophysiological signals with spike components displayed on a monitor by the GUI. Because spike components can distort multiple electrophysiological signals, they can be removed from multiple electrophysiological signals.

[0060] For example, the GUI can generate GUI elements that a user can use to interact with multiple electrophysiological signals. The user can further interact with the GUI elements to select or flag to define time intervals 506 corresponding to spike components, in order to exclude spike components from further processing (such as inverse reconstruction). For example, a transient feature removal function of the signal processing system (e.g., transient feature removal function 210, such as...) Figure 2 The signal (shown) can be configured to exclude spike components from the electrophysiological signal in response to a time interval 506 that defines the spike components. A third figure 508 shows an electrophysiological signal without spike components and filtered for noise by a noise filtering system (e.g., system 100, 202), as described herein.

[0061] Figure 6 An example of an electrophysiological monitoring system 600 capable of implementing electrophysiological noise filtering as disclosed herein is depicted. System 600 may include an analysis system 602 employing a filter system 604 as disclosed herein (e.g., corresponding to...). Figure 1 (Noise filtering system 100). In some examples, filter system 604 may correspond to, for example, Figure 2 The signal processing system 200 is shown. The filter system 604 can be applied in real time, such as a noise filter 606, during an electrophysiological study of a patient, or the filter system can be implemented with respect to stored electrical measurement data previously acquired for a given patient. In some examples, the sensed electrical activity can be used to generate one or more graphical representations (e.g., graphical mapping of electroanatomical activity) based on the sensed electrical activity, which can be provided to a display 608.

[0062] Analysis system 602 can be implemented as a computer, such as a laptop computer, desktop computer, server, tablet computer, workstation, etc. Analysis system 602 may include memory 610 for storing data and machine-readable instructions. Memory 610 can be implemented as, for example, a non-transitory computer storage medium, such as volatile memory (e.g., random access memory), non-volatile memory (e.g., hard disk drive, solid-state drive, flash memory, etc.), or a combination thereof. The instructions can be programmed to perform one or more methods, such as those described herein. Figure 1 The examples disclosed.

[0063] The analysis system 602 may also include a processing unit 612 to access the memory 610 and execute machine-readable instructions stored in the memory. The processing unit 612 may be implemented as, for example, one or more processor cores. In this example, although the components of the analysis system 602 are shown as implemented on the same system, in other examples, different components may be distributed across different systems and communicate, for example, via a network.

[0064] System 600 may include a measurement system 614 to acquire electrophysiological information of patient 616. Figure 6 In one example, sensor array 618 includes one or more electrodes that can be used to record a patient's electrical activity. As an example, sensor array 618 may correspond to an arrangement of body surface electrodes distributed above and around the patient's chest cavity for measuring electrical activity associated with the patient's heart (e.g., as part of an ECM procedure). In some examples, sensor array 618 may contain about 200 or more sensors (e.g., about 252 sensors), each corresponding to a node defining a corresponding channel. Examples of non-invasive sensor arrays that can be used are shown and described in U.S. Patent No. 9,655,561, filed December 22, 2011, or International Patent Application PCT / US2009 / 063803, filed November 10, 2009. This non-invasive sensor array corresponds to one example of a complete complement to sensors that may include one or more sensing areas. As another example, sensor array 618 may include a dedicated arrangement of electrodes corresponding to a single sensing area or multiple discrete sensing areas. Alternatively or additionally, the sensor array 618 may include invasive sensors that can be inserted into the patient's body, such as via a catheter or other probe device.

[0065] The measurement system 614 receives sensed electrical signals from a corresponding sensor array 618. The measurement system 614 may include appropriate control and signal processing circuitry (e.g., filters and safety circuitry) 620 to provide corresponding electrical measurement data 622, which describes the electrical activity of each of the plurality of input channels detected by the sensors in the sensor array 618. In some examples, the electrical measurement data 622 corresponds to, for example,... Figure 1 The input electrical signal data shown is 102.

[0066] Measurement data 622 can be stored in memory 610 as analog or digital information. Appropriate timestamps and channel identifiers can be used to index the corresponding measurement data 622 to facilitate its evaluation and analysis. As an example, each sensor in the sensor array 618 can simultaneously sense electrical activity on the body surface and provide corresponding measurement data 622 for one or more user-selected time intervals. Therefore, measurement data 622 can represent spatially and temporally consistent electrical information based on the location of the sensors in array 618 on and / or within the body of the patient 616. Analysis system 602 is programmed to process the electrical measurement data 622 and generate one or more outputs. These outputs can be stored in memory 610 and provided to display 608 or other types of output devices. As disclosed herein, the type of output and information presented can vary depending on, for example, the user's application requirements.

[0067] As mentioned, the analysis system 602 is programmed to employ a noise filter 606 to remove noise and / or transients from the measured electrical activity, which can improve the accuracy of the processing and analysis performed by the analysis system 602. The noise filter 606 can, for example, be implemented to perform any of the filter functions or combinations disclosed herein (see, for example...). Figures 1-2 (and corresponding descriptions). Therefore, noise filter 606 can be applied to remove noise, transients, or other signal characteristics from the signal stored in memory as measurement data 622. Filter system 604 can be programmed to provide, for example, filtered signal data 624 stored in memory 610, which includes the results from noise filter 606, such as combining measurement data 622 with other parameter data. In some examples, filtered signal data 624 corresponds to, for example... Figure 1 The noise-filtered signal data 124 shown is as follows: Figure 2 The noise-filtered signal data 206 is shown. In other examples, the filtered signal data 624 may correspond to, for example... Figure 2 The filtered input electrical signal data 214 is shown.

[0068] In some examples, the filter system 604 can be programmed to interface with a graphical user interface (GUI) 626 stored as executable instructions in memory 610. The GUI 626 can thus provide an interactive user interface, such as being used to selectively define time intervals for processing electrophysiological signals in response to user input 628. The GUI 626 can be programmed to provide data that can be presented as interactive graphics on display 608. For example, the GUI 626 can be programmed to generate GUI elements (e.g., checkboxes, radio buttons, sliders, etc.) that a user can use to select or mark to define or select time intervals corresponding to the filter window to be applied to the input signal provided in measurement data 622 for noise estimation as described herein.

[0069] The analysis system 602 can also generate output to be graphically displayed on the display 608, representing filtered or unfiltered waveforms of one or more signals. As disclosed herein, the waveforms can represent a graphical representation of the filtered or unfiltered input channels, similar to... Figures 3-5 The waveforms shown herein. Alternatively, the output waveforms may represent filtered or unfiltered graphical representations of the reconstructed waveforms, as disclosed herein. As another example, analysis system 602 may include mapping system 630, which may be programmed to generate electroanatomical mappings based on filtered signal data. Mapping system 630 may include mapping generator 632, which may be programmed to generate mapping data representing graphics (e.g., electrical or electroanatomical mappings) based on measurement data 622. Mapping generator 632 may be programmed to generate mapping data to visualize mappings spatially superimposed on a graphical representation of an anatomical structure (e.g., the heart) via display 608.

[0070] In some examples, the mapping system 630 includes a reconstruction component 634 programmed to reconstruct cardiac electrical activity by combining measurement data 622 with geometric data 636 through inverse computation. The inverse computation employs a transformation matrix and reconstructs the electrical activity sensed on the patient's body by the sensor array 618 onto an anatomical capsule, such as the epicardial surface, endocardial surface, or other capsule. Examples of inverse algorithms that can be implemented by the reconstruction component 634 are disclosed in U.S. Patents 7,983,743 and 6,772,004. The reconstruction component 634, for example, calculates coefficients of a transfer matrix to determine cardiac electrical activity on the cardiac capsule based on the body surface electrical activity represented by the electrical measurement data 622. Since the reconstruction on the capsule can be sensitive to noise on the corresponding input channels, a filter system 604 helps remove noise in the channels that may distort the signal morphology, which can improve patient diagnosis (e.g., for patients with arrhythmias such as CRT, PVC, VT, and PAC).

[0071] Mapping generator 632 can generate corresponding mappings of electrical activity using reconstructed electrical data calculated via an inverse method. This mapping can represent the electrical activity of the patient's heart on display 608, such as a mapping corresponding to a reconstructed electrogram (e.g., a potential mapping). Alternatively or additionally, analysis system 602 can calculate other electrical characteristics from the reconstructed electrogram, such as activation maps, repolarization maps, propagation maps, or other electrical characteristics that can be calculated from measurement data. The type of mapping can be set in response to user input 628 via GUI 626.

[0072] Figures 7-8 Graphical mapping diagrams 700 and 800 of electrical activity are shown. Each graphical mapping diagram can be generated by a mapping diagram generator (such as mapping diagram generator 632, etc.). Figure 6 (As shown) generated. Figure 7 A graphic mapping map 700 is shown, generated based on electrical signals measured from the patient's body surface, which have been filtered according to different signal processing schemes as described herein. The electrical signals used to generate the graphic mapping map 700, measured from the patient's body surface, have been line-filtered using an infinite impulse response (IR) filter. As shown at 702, the graphic mapping map 700 includes high-frequency components. Figure 8 This paper presents a graphical mapping map 800 generated based on electrical signals measured from the patient's body surface, which have been based on, as described herein, etc. Figure 2 The processing scheme described in the example is filtered. For example, a graphic mapping map 800 can be generated based on filtered input electrical signal data 214, which can be provided by the signal processing system 200. Because the signal processing system 200 uses a noise filtering system 202 and an elliptic low-pass filter as a low-pass filter 212 to filter the input electrical signal 204, this combination reduces high-frequency components in the graphic mapping map 800, as shown at 802, thereby producing a more accurate representation of electrical activity after the spikes, which improves patient diagnosis.

[0073] In view of the structural and functional features described above, reference will be made to Figure 9 To better understand the exemplary method. Although for the purpose of simplifying the explanation, Figure 9 The exemplary methods shown and described are executed sequentially; however, it should be understood and recognized that the exemplary methods are not limited to the order shown, as some actions may occur in different orders, multiple times, and / or simultaneously with the actions shown and described herein in other examples.

[0074] Figure 9 An example of a method 900 for filtering noise from electrophysiological signals is shown. At least some portions of method 900 may be implemented by hardware, software, or a combination of hardware and / or software, as described herein. In some examples, the method may be implemented by a noise filtering system (e.g., such as...). Figure 1 The noise filtering system 100 shown or such Figure 2 The noise filtering system 202 shown is implemented. Method 900 can be implemented at 902 by extracting from the electrophysiological signal (e.g., via signal segment extractor 104, such as...). Figure 1 (As shown) Start with the signal segment of interest. At 904, it can be (e.g., via a signal segment noise calculator 110, such as...) Figure 1 (As shown) The extracted signal segment of interest is evaluated to estimate the noise in the signal segment of interest. At 906, it can be (e.g., via signal segment filter 118, as shown) Figure 1 (As shown) The estimated noise in the signal segment of interest is used to provide an alternative noise estimate for at least one remaining segment of the electrophysiological signal. At 908, it can be (e.g., via signal segment filter 118, as shown) Figure 1 (As shown) At least one residual signal segment of the electrophysiological signal is filtered based on the alternative noise estimation to remove noise from at least one residual signal segment, thereby providing at least one filtered residual signal segment. At 910, it can be (e.g., via signal segment filter 118, as shown) Figure 1 (As shown) Based on the estimated noise, the signal segment of interest in the electrophysiological signal is filtered to remove noise from the signal segment of interest, thereby providing a filtered signal segment of interest. At 912, it is possible (e.g., via signal segment filter 118, such as...) Figure 1 (As shown) At least one filtered residual signal segment and a filtered signal segment of interest are combined to provide a noise-filtered electrophysiological signal (e.g., as part of noise-filtered signal data 124, such as...). Figure 1 (As shown).

[0075] In view of the foregoing structural and functional description, those skilled in the art will understand that portions of the systems and methods disclosed herein may be embodied as methods, data processing systems, or computer program products, such as non-transitory computer-readable media. Therefore, these portions of the methods disclosed herein may take the form of a completely hardware implementation, a completely software implementation (e.g., in a non-transitory machine-readable medium), or an implementation combining software and hardware. Furthermore, portions of the systems and methods disclosed herein may be computer program products on a computer-usable storage medium having computer-readable program code on the medium. Any suitable computer-readable medium may be used, including but not limited to static and dynamic storage devices, hard disks, optical storage devices, and magnetic storage devices.

[0076] This document also describes certain implementations with reference to block diagrams of methods, systems, and computer program products. It should be understood that the illustrated blocks and combinations thereof can be implemented by computer-executable instructions. These computer-executable instructions can be provided to one or more processors of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus (or combination of apparatus and circuitry) to produce a machine such that the instructions, executed via the processor, implement the functions specified in one or more blocks.

[0077] These computer-executable instructions may also be stored in a computer-readable storage medium capable of instructing a computer or other programmable data processing apparatus to operate in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of art including instructions that implement the functions specified in one or more flowchart blocks. These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, such that the instructions executing on the computer or other programmable apparatus provide steps for implementing the functions described herein.

[0078] The examples described above are merely examples. It is impossible to describe every conceivable combination of structures, components, or methods, but those skilled in the art will understand that many other combinations and arrangements are possible. Therefore, this invention is intended to cover all such changes, modifications, and variations that fall within the scope of this application (including the appended claims). If the disclosure or claims refer to an element or its equivalent, such as "a," "an," "first," or "another," it should be interpreted as including one or more such elements, neither requiring nor excluding two or more such elements. As used herein, the term "comprising" means including but not limited to, and the term "including" means including but not limited to. The term "based on" means at least partially based on.< / n>

Claims

1. One or more non-transitory computer-readable media having data and machine-readable instructions executable by a processor, the data including electroanatomical data characterizing electrophysiological signals measured from a patient, the machine-readable instructions including: A signal segment extractor, the signal segment extractor being programmed to extract the signal segment of interest from the electrophysiological signal; A signal segment noise calculator, which is programmed to evaluate an extracted signal segment of interest to estimate the noise in the signal segment of interest; and A signal segment filter is programmed to determine an alternative noise estimate of at least one remaining segment of the electrophysiological signal based on the estimated noise in the signal segment of interest, and to filter the at least one remaining segment based on the alternative noise estimate to remove noise therein, the at least one remaining segment being different from the extracted signal segment.

2. The one or more non-transitory computer-readable media of claim 1, wherein the signal segment extractor is programmed to apply a moving window function to sample a portion of the electrophysiological signal and evaluate the signal morphology of the sampled portion of the electrophysiological signal to determine whether the portion of the electrophysiological signal will be identified as the signal segment of interest.

3. The one or more non-transitory computer-readable media according to claim 2, wherein the moving window function includes a Hamming window.

4. The one or more non-transitory computer-readable media of claim 3, wherein the signal segment noise calculator includes a transformation function programmed to convert a sampled signal segment of interest into corresponding frequency domain data having discrete frequency ranges of the signal in the signal segment of interest.

5. One or more non-transitory computer-readable media according to claim 4, wherein the transform function is programmed to apply a discrete Fourier transform (DFT) to the sampled signal segment of interest to convert the sampled signal segment of interest into the corresponding frequency domain data.

6. The one or more non-transitory computer-readable media of claim 5, wherein the signal segment noise calculator is programmed to calculate a set of DFT coefficients, including phase and amplitude, for each signal in the signal segment of interest to convert the sampled signal segment of interest into the corresponding frequency domain data.

7. The one or more non-transitory computer-readable media according to claim 6, wherein the signal segment noise calculator is programmed to: Evaluate the corresponding frequency domain data to identify the frequency of noise signals in the corresponding signals of the signal segment of interest; The DFT coefficients of the noise signal are selected from the set of DFT coefficients based on the identified frequencies; and The noise in the signal segment of interest is estimated based on the selected DFT coefficients.

8. One or more non-transitory computer-readable media according to claim 7, wherein the signal segment filter comprises: An extended window function, which is programmed to sample at least one remaining segment of the electrophysiological signal; and A noise estimation function, programmed to extrapolate or interpolate the estimated noise in the segment of interest to the at least one remaining segment, to provide the alternative noise estimate of the at least one remaining segment based on selected DFT coefficients of the noise signal.

9. One or more non-transitory computer-readable media according to claim 8, wherein the signal segment noise calculator is programmed to scale selected DFT coefficients to scale the estimated noise in the signal segment of interest in the noise estimate of at least one remaining signal segment.

10. One or more non-transitory computer-readable media according to claim 8 or claim 9, wherein the signal segment filter comprises a segment filtering function programmed to subtract the noise in the at least one remaining signal segment from the alternative noise estimate of the at least one remaining signal segment to filter the electrophysiological signal against the noise.

11. The one or more non-transitory computer-readable media of claim 10, wherein the noise in the electrophysiological signal is line noise having a frequency of one of 50 Hz and 60 Hz, and the signal segment of interest does not include either spikes or QRS complexes.

12. One or more non-transitory computer-readable media according to claim 11, The electroanatomical data includes multiple electrophysiological signals measured from the patient via a set of sensors, and the electrophysiological signals correspond to a given electrophysiological signal. The noise estimation function is programmed to provide an alternative noise estimate of the remaining electrophysiological signal among the multiple electrophysiological signals based on the estimated noise in the signal segment of interest of the given electrophysiological signal. The segment filtering function is programmed to subtract the noise in the remaining electrophysiological signal from the alternative noise estimate of the remaining electrophysiological signal to filter the remaining electrophysiological signal against the noise.

13. The one or more non-transitory computer-readable media of claim 12, wherein the machine-readable instructions comprise a plurality of noise filtering systems, the plurality of noise filtering systems comprising respectively the signal segment extractor, the signal segment noise calculator, and the signal segment filter, and the set of sensors are arranged in a plurality of spatial regions, and The respective noise filtering systems in the plurality of noise filtering systems are adapted for each spatial region and are configured as follows: Noise in the signal segment of interest of the corresponding electrophysiological signal measured by the sensor in the corresponding spatial region of the plurality of spatial regions, the corresponding electrophysiological signal corresponding to the given electrophysiological signal; An alternative noise estimate of the electrophysiological signal measured by the remaining sensors in the corresponding spatial region is calculated based on the estimated noise in the corresponding electrophysiological signal segment of interest. as well as The electrophysiological signal is filtered against the noise by subtracting the noise in the electrophysiological signal measured by the remaining sensors in the corresponding spatial region from the alternative noise estimate of the electrophysiological signal.

14. The one or more non-transitory computer-readable media of claim 13, wherein the signal segment of interest is a first signal segment of interest, the alternative noise is a first alternative noise, and the data comprises electrical signal data characterizing a signal generated by a treatment device or navigation system, wherein the signal segment extractor is programmed to extract a second signal segment of interest from the signal, the signal segment noise calculator is programmed to evaluate the second signal segment of interest to estimate the noise in the second signal segment of interest, and the signal segment filter is programmed to determine a second alternative noise estimate of at least one remaining signal segment of the signal, and to filter the at least one remaining signal segment of the signal based on the second alternative noise estimate to remove noise therefrom.

15. One or more non-transitory computer-readable media according to claim 13 or 14, wherein the data includes electrical signal data characterizing electrical signals measured and applied to the patient's body, wherein the signal segment extractor is programmed to: The measured electrical signal is evaluated to determine the common noise in the measured electrical signal; and The signal segment of interest is extracted based on the identified common noise in the measured electrical signal.

16. The machine-readable instructions of any one or more non-transitory computer-readable media according to any one of claims 13 to 15, further comprising a calibration system programmed to: A first graphic mapping of electroanatomical activity is generated based on electrophysiological signals provided by corresponding sensors in one or more first spatial regions of the plurality of spatial regions. A second graphic mapping of electroanatomical activity is generated based on electrophysiological signals provided by corresponding sensors in one or more second spatial regions of the plurality of spatial regions. as well as The first and second graphic mapping maps of the electroanatomical activity are evaluated to determine the quality of noise filtering for each of the first and second spatial regions among the plurality of spatial regions.

17. One or more non-transitory computer-readable media according to any one of claims 13 to 16, wherein the corresponding noise filtering system further comprises a machine learning algorithm programmed to improve noise filtering of the corresponding noise filtering system based on historical noise filtering data.

Images