阅读说明：本技术 心脏快速击发的检测和定位 (Detection and location of rapid heart firing ) 是由 贾平 曾清国 T·G·雷斯科 楼青 于 2019-03-14 设计创作，主要内容包括：用于心脏快速击发(例如,心房快速击发)检测的系统和方法在收集到的心脏波形数据的通道上执行频率分析,并测试该数据的具有频率高于与心脏纤颤(例如,心房纤颤)或其它心律失常活动相关联的基线频率波群成分的离群频率波群成分。心脏快速击发起源的解剖区域能够经由图形显示实时地显示在心外膜表面映射图上,以辅助治疗。在进行此类检测之前,能够执行QRST波群移除,以确保心室活动不影响心房快速击发分析。还公开了一种用于QRST波群移除的基于频率的方法。(Systems and methods for cardiac rapid fire (e.g., atrial rapid fire) detection perform frequency analysis on channels of collected cardiac waveform data and test the data for outlier frequency complexes having frequencies above a baseline frequency complex associated with cardiac fibrillation (e.g., atrial fibrillation) or other arrhythmic activity. The anatomical region from which the heart is rapidly fired can be displayed on the epicardial surface map in real time via a graphical display to assist in treatment. Prior to making such detections, QRST complex removal can be performed to ensure that ventricular activity does not affect the atrial rapid fire analysis. A frequency-based method for QRST complex removal is also disclosed.)

1. A method of detecting a heart rapid firing activity of a heart, the method comprising:

collecting cardiac waveform data from a plurality of channels;

performing a frequency analysis on the collected cardiac waveform data for each of the plurality of channels over a moving window;

identifying a channel exhibiting a rapid firing frequency peak during a given time window; and

the channels identified as fast fire in a given time window are mapped to one or more spatial regions of the heart.

2. The method of claim 1, wherein frequency analysis further comprises removing QRST components from each channel of the collected cardiac waveform data.

3. The method of claim 1 or 2, wherein the plurality of channels correspond to an arrangement of electrodes positioned across a body surface of a patient.

4. The method of any of the preceding claims, further comprising generating a graphical output indicative of at least one of a time and/or an anatomical location of a quick fire activity.

5. The method of any of the preceding claims, further comprising controlling delivery of therapy based on the identified anatomical location of the quick fire activity.

6. The method of any of the preceding claims, wherein the channel exhibiting a rapid firing frequency comprises a channel having a frequency that is at least one standard deviation greater than an average baseline frequency.

7. The method of any of the preceding claims, further comprising determining that a quick fire event has occurred based on:

providing a frequency value as a frequency threshold, either manually as a user input or via an automatic threshold generator;

comparing the dominant frequency of each channel during a given time window to a frequency threshold;

determining a channel exhibiting a dominant frequency greater than a frequency threshold as a rapid-fire channel; and

anatomically positioning a rapid firing activity based on the determined rapid firing channel.

8. The method of claim 7, wherein the frequency threshold is a value in a range of about 8Hz to about 10 Hz.

9. The method of claim 7, further comprising:

providing, either manually as user input or via automatic determination, an integer value as a channel number threshold indicative of a minimum channel number;

the method further includes determining that a quick fire event has occurred based on the number of channels determined to be quick fire channels during the given time window exceeding a channel number threshold.

10. The method of any of claims 1-6, further comprising determining that a quick fire event has occurred based on:

analyzing the frequency profile of each of a plurality of given channels to test for changes in the dominant frequency over a plurality of windowed time frames; and

identifying one or more of the plurality of given channels as fast firing channels during one or more time frames in which the dominant frequency of the identified channel is in the higher frequency range based on the identified channel showing significant movement of the dominant frequency from the lower frequency range to the higher frequency range or from the higher frequency range to the lower frequency range.

11. The method of claim 10, wherein the lower frequency range is about 3Hz to about 8Hz, and wherein the higher frequency range is about 8Hz to about 12 Hz.

12. The method of any of claims 1-6, further comprising: a rapid fire event is determined to have occurred based on a comparison of the frequency profile of a given channel to the frequency profiles of one or more spatially adjacent channels during a given time window.

13. A system, comprising:

a non-transitory memory for storing electrical data and machine readable instructions representing a plurality of ECG signals; and

a processor to access non-transitory memory and execute machine-readable instructions, the instructions comprising:

heart rapid fire detection code programmed to perform heart rapid fire detection to determine an outlier dominant frequency of a plurality of ECG signals;

code programmed to store in memory rapid-fire data characterizing the detected rapid-fire heart.

14. The system of claim 13, wherein the cardiac flash data is provided to specify a time, a channel, and/or an epicardial surface region that exhibits cardiac flash.

15. The system of claim 13 or 14, wherein the instructions further comprise:

code to convert the plurality of ECG signals to frequency domain ECG signals; and

code programmed to remove ventricular components from the frequency domain ECG signals, the heart quick-fire detection code operating on the plurality of ECG signals after removal of ventricular components.

16. The system of claim 13, 14 or 15, further comprising a display that visualizes the graphical representation based on the cardiac rapid fire data.

17. The system of claim 13, 14 or 15, further comprising a therapy system that delivers therapy to the patient based on the cardiac rapid fire data.

18. The system of any of claims 13, 14, 15, 16, or 17, wherein the heart rapid fire detection code determines a frequency at which an outlier dominant frequency of a set of channels exhibits at least one standard deviation greater than a mean baseline frequency.

19. A frequency domain-based method of removing QRST from a cardiac signal, the method comprising:

performing a frequency analysis on an electrocardiogram signal originating from one of a plurality of electrocardiogram channels to generate a signal frequency curve;

performing a frequency analysis on the identified QRST frequency template to generate a template frequency curve;

subtracting the template frequency curve from the signal frequency curve to generate a frequency curve of the QRST-removed electrocardiogram signal corresponding to the electrocardiogram signal; and

repeating the signal frequency curve generation and subtraction for other ones of the plurality of electrocardiogram channels.

20. The method of claim 19, wherein the amount of power removed from the electrocardiogram signal in the electrocardiogram signal with the QRST removed is based on the number of QRS complexes provided in the electrocardiogram signal, and wherein the template frequency curve is not normalized.

Technical Field

The present disclosure relates to detection and analysis of cardiac waveforms.

Background

Electrocardiography (ECG) systems monitor the electrical activity of a patient's heart via invasive or external electrodes. Electrophysiological (EP) protocols use a single or multiple catheters within the heart to assess the electrical activity and conduction pathways of the heart.

Disclosure of Invention

The present disclosure relates to detection and analysis of cardiac waveforms.

As one example, a method of detecting cardiac fast firing (cardiac fast firing) activity in real time includes collecting cardiac waveform data from a plurality of channels. For an example of detecting atrial rapid firing activity, QRST components may be removed from each channel of the collected cardiac waveform. A frequency analysis is performed on each channel over a moving window. Channels exhibiting a rapid firing frequency peak during a particular window are identified. The channels identified as rapid fire within a particular time frame are mapped to one or more epicardial surface regions and a graphical output is provided indicating the time and epicardial location of the rapid fire activity.

As another example, a system includes a processor and a non-transitory memory to store electrical data and machine-readable instructions representing a plurality of ECG signals. The processor accesses the non-transitory memory and executes the machine-readable instructions. The instructions include heart rapid fire detection code programmed to perform heart rapid fire detection to determine an outlier dominant frequency of the plurality of ECG signals. The code is further programmed to store heart rapid fire data in the memory to specify a time, a channel, and/or an epicardial surface region exhibiting rapid firing of the heart. The display visualizes the graphical representation based on the heart quick fire data.

As yet another example, a method may include frequency-domain removal of a QRST complex (complex) from a cardiac waveform signal.

Drawings

Fig. 1 depicts an example system for detecting and analyzing cardiac waveforms and performing cardiac rapid fire detection.

Fig. 2 illustrates an example of a graphical representation of frequency analysis curves (plot) and chest (thoracic) channel positions for use in cardiac rapid fire detection.

Fig. 3A-3C, 4A-4C, and 5A-5C depict larger versions of the curves and representations of fig. 2.

Fig. 6 depicts a frequency plot of channels determined to exhibit rapid firing of the heart.

Fig. 7 depicts a thoracic graphical representation illustrating a channel determined to exhibit rapid firing of the heart.

Fig. 8 depicts a thoracic graphical representation illustrating a channel determined to exhibit rapid firing of the heart.

Fig. 9 depicts a graphical map (map) of a portion of the epicardial surface illustrating regions determined to exhibit rapid firing of the heart.

Fig. 10 is a flow diagram depicting an example method of removing a QRST complex from a cardiac waveform.

Fig. 11 depicts an example of multiple waveforms demonstrating identification of P-waves based on QRST template regions.

Fig. 12 depicts plots of cardiac waveforms before and after removal of the QRST portion.

Fig. 13 depicts an example of a system that may be used to perform diagnostics (including cardiac rapid fire detection) and/or therapy.

FIG. 14 is a flow chart of an example method of rapid heart fire detection.

Detailed Description

Rapid heart firing refers to abnormal electrical activity of the heart at a higher frequency than that associated with fibrillation activities such as atrial fibrillation and ventricular fibrillation. An example of a fast firing of the heart is a fast firing of the atria, which is an electrophysiological signal originating in the atria and which is faster than the remaining atrial chambers. Rapid firing of the heart may occur in brief episodes of only a few seconds in duration, with each episode being located in one continuous region of the heart (or multiple separate regions of the heart) rather than significantly across the entire heart surface. Thus, detection of rapid firing of the heart involves, in part, finding outlier frequency activity that is typically in a predetermined frequency range (e.g., about 8-15Hz) and, in the case of detection using body surface measurements of cardiac electrical activity (BSM), only in a subset of BSM channels (e.g., ECG channels). Some patients may exhibit very fast baseline cardiac activity, e.g., very fast baseline atrial activity, meaning that outlier activity indicative of a fast firing of the heart (e.g., a fast firing of the atria) may be found in a higher frequency range (e.g., about 10-15Hz) for such patients. The precise boundary between rapid cardiac firing and baseline fibrillation activity may vary from patient to patient.

The detection and localization of the anatomical origin of the heart's rapid firing activity is of clinical significance, such as presenting one or more potential targets for ablation or other therapy to correct or mitigate cardiac dysfunction. For example, by allowing targeted therapy while a patient is undergoing an EP protocol, detection and localization of a heart's rapid-fire onset can therefore be important to the patient's treatment. Such detection and localization may also be used as a screener to suggest further diagnostic studies. Detection and localization may be accomplished in an off-line analysis or in real-time. By "real-time" is meant that the heart is detected and located within seconds of the occurrence of the event, as opposed to, for example, being detected and located substantially after collection of such data (e.g., after minutes or hours) during subsequent offline analysis of the collected data.

The present disclosure relates to detection and analysis of cardiac waveforms, including detecting rapid firing of the heart, and in some examples doing so in real-time. Detection and analysis may also include detection and removal of QRST complexes from the cardiac waveform to improve atrial rapid fire detection. The detected waveforms and associated analysis may also be used to drive output to a display corresponding to an interactive graphical map (e.g., a Graphical User Interface (GUI)). The GUI may, for example, alert the physician of one or more detected rapid fire events and/or may display one or more graphical maps, e.g., corresponding to the chest and/or epicardial surface, indicating one or both of the detected rapid fire body surface locations or cardiac locations, which may then be used, for example, as a guide for an ablation procedure in order to correct or mitigate cardiac dysfunction.

As used herein, an "electrocardiogram signal" ("ECG signal") refers to a plot (graph) of voltage over time recorded for one or more channels, each channel based on cardiac electrical signals sensed by electrodes. The ECG signal can be generated from Body Surface Measurements (BSM). The systems and methods described herein can display and highlight a region of interest (i.e., a portion of the surface of the heart) corresponding to a likely source of rapid percussive electrical activity and thus to a potential treatment target without displaying ECG signals, reconstructing electrograms (electograms) on the surface of the heart from ECG signals, or generating cycle length maps or dominant frequency maps. Thus, the present system and method eliminates the need for the operator to manually select any particular beat (beat) for analysis, as the system and method can automatically and simultaneously process the continuously collected ECG signals to produce a graphical display output illustrating the rapidly fired heart surface area, all of which can be done in real time. In contrast, systems and methods that rely on solving an inverse problem to reconstruct an electrogram on the surface of the heart may require the operator to consciously and deliberately select beats to reconstruct. However, in some examples, the ECG signal can be used to reconstruct electrograms over the cardiac envelope that are calculated by solving an inverse problem based on electrical signals acquired from a set of non-invasive body surface measurements and geometry data related to body surface measurement locations relative to the cardiac envelope.

In some examples, a Computed Tomography (CT) scan of a patient may be performed to establish a heart and torso geometry in the form of a transfer matrix a. From the transfer matrix A, an inverse transfer matrix A can be calculated^-1It may also be referred to as an influence coefficient matrix. Computing the inverse transfer matrix A is described in U.S. Pat. No.9,256,166 to Rudy et al^-1The patent is incorporated herein by reference. This inverse transfer matrix A^-1The effect of each electrode position on each heart surface position is defined. Inverse transfer matrix A^-1Is itself a matching table, providing a correspondence between Body Surface Measurement (BSM) channels (e.g. from electrodes on the vest) and heart surface locations, in other words a torso-heart relationship related to the contribution of electrical activity from each location on the heart to the measured potential at each location on the torso and thus to each individual BSM channel. A subset of BSM channels that exhibit outlier components in the high frequency spectrum may then be determined. This determined subset of BSM channels may be referred to as a quick fire channel. The inverse transfer matrix A may then be examined^-1To determine the absolute maximum coefficient or a few such coefficients in each of said columns corresponding to the determined fast excitation channel. The absolute maximum coefficient corresponds to the cardiac surface location having the greatest contribution to the corresponding BSM channel. Thus, the inverse transfer matrix A can be used^-1The particular column checked in (1) is limited to the column corresponding to the channel for which rapid firing activity was detected. A graphical representation of the heart surface or a region of interest of the heart surface may then be generated, highlighting a particular location, such as by examining the inverse transfer matrix A^-1As determined by the fast firing channel column in order to find the largest coefficient in each column. This generated graphical representation indicates a quick firing activityThe position of (a).

Thus, the systems and methods described herein can be implemented, for example, by determining the inverse transfer matrix A^-1Of the heart, several selected ECG channels (e.g., corresponding to certain ECG electrodes in the dorsal ventricle with many such electrodes) are "mapped" to corresponding heart surface locations. The selected (e.g. vest) channel can be identified by the high frequency component in its ECG signal. Thus, selected (e.g., vest) channels with high frequency components may be "mapped" to locations on the surface of the heart to generate, for example, a graphical depiction of possible locations indicating the origin of detected rapid firings. More discussion of the relationship between BSM channels and cardiac surface locations may be found, for example, in U.S. Pat. No.9,549,683 to Jia et al and U.S. Pat. No.9,186,515 to Ramanathan et al, which are incorporated herein by reference.

Fig. 1 depicts an example of a system 10 for detecting and analyzing cardiac flash and generating a graphical map that may be visualized on a display 12 indicating the location of the detected cardiac flash. The system 10 includes a memory 14, which may include one or more non-transitory machine-readable media. The system 10 also includes a processor 16, which processor 16 may include one or more processing cores to access memory and execute corresponding instructions exposed within the processor block 16.

In the example of fig. 1, the memory 14 stores electrophysiological (e.g., ECG) data 18. In some examples, the ECG data 18 corresponds to a raw (e.g., unfiltered and pre-processed) ECG signal that is non-invasively measured via sensors placed on an exterior surface of the patient's body (e.g., an arrangement of body surface sensors distributed in a non-invasive manner across an exterior surface of the patient's body, such as the patient's chest or a portion thereof, e.g., two hundred and fifty-two sensors distributed substantially uniformly around the chest). Various measurement systems (not shown in fig. 1, but see measurement system 566) may be used to acquire body surface electrical measurements, which may be used to provide ECG data 18, which ECG data 18 may correspond to real-time data obtained when this method is implemented, or which ECG data 18 may correspond to data already obtained a priori, such as part of a previous Electrophysiology (EP) protocol, or acquired during another intervention.

The processor 16 executes machine readable instructions including a heart rapid fire detector 20 to detect a heart rapid fire event in the ECG data 18. As an example, the cardiac rapid fire detector 20 processes raw (e.g., non-linearly filtered) ECG data 18 for one or more selected time intervals for each of a plurality of input channels. The heart rapid fire detector 20 employs a frequency-based approach to identify and locate rapid fire episodes to corresponding locations on certain recording channels and/or epicardial surfaces. The determined heart rapid-fire time and location may be stored in memory as heart rapid-fire event data 22, the heart rapid-fire event data 22 specifying a timestamp (index), channel, epicardial surface region, or other label for the heart rapid-fire event determined via the heart rapid-fire detector 20.

As another example, the heart rapid fire detector 20 includes a heart rapid fire channel determination code 24 that may utilize one or more tests to determine whether a channel exhibits heart rapid fire activity over a period of time, as described herein. The heart rapid fire determination code 24 may perform frequency analysis on the data from the signal collection channels. As an example, the heart rapid-fire channel determination code 24 may employ statistical analysis of the composite signal, may perform single-channel temporal detection, or may perform multi-channel spatial detection to determine that a heart rapid-fire episode has occurred and isolate each channel exhibiting heart rapid-fire activity.

The heart rapid fire detector 20 further comprises a mapping function 28 to map the channel determined to exhibit heart rapid fire activity to an epicardial surface region to graphically indicate the anatomical origin of the heart rapid fire event. For example, mapping function 28 may use inverse transfer matrix A^-1To detect the anatomical region as a source of rapid firing, in some real-time examplesDetection is done in real time once a rapid fire is detected in certain body surface (e.g., vest) channels.

As another example, when system 10 is configured to detect atrial rapid firing, processor 16 may also execute instructions corresponding to QRST detection and removal function 32. The QRST detection and removal function 32 processes the cross-channel ECG data signal to remove ventricular signal components, facilitating analysis of the atrial signal, including the heart's rapid firing detector. The QRST detection and removal function 32 may, for example, generate a QRST template that combines the QRS complex and the T wave into a single template region of interest.

For example, the QRST detection and removal function 32 may perform Principal Component Analysis (PCA) on a region of interest of the cardiac waveform, such as may be selected automatically or manually in response to user input identifying intervals of signals corresponding to QRST complexes. Thus, PCA can be used to generate QRST template definitions, which can be applied across time frames, such as by time stepping the template relative to the ECG data to be searched to determine correlation coefficients. The peak correlation coefficients are used to identify potential locations where the template matches the data. The correlation coefficient may be compared to a threshold to identify a corresponding region of interest for each of the plurality of channels.

The QRST detection and removal function 32 may remove each region of interest (i.e., each corresponding to a QRST complex) and perform spline interpolation to automatically connect adjacent P-waves. As an example, the interpolation may be implemented as a shape-preserving piecewise cubic interpolation (e.g., Piecewise Cubic Hermite Interpolation Polynomial (PCHIP) or another spline interpolation function). Such an interpolation function keeps the interpolated values monotonic (e.g., increasing or decreasing) based on the endpoint value used for such interpolation. Thus, the QRST complex is replaced in the analyzed cardiac waveform by a substitute signal portion that does not have high frequency components that would interfere with the analysis of the waveform for other purposes. The processor 16 may also effect baseline removal and/or removal of bad input channels prior to performing the QRST detection and removal function 32, such as disclosed herein.

As another example, the QRST detection and removal function 32 may use a frequency-based approach to remove frequency components associated with QRST complexes from the frequency analysis curve for each channel. For example, a fast fourier transform may be applied to the ECG signal for each channel, and the frequencies corresponding to the QRST complexes may be removed from each frequency domain ECG signal. The processor 16 may be configured to remove the QRST complex prior to detecting rapid firing of the heart by the detector 20.

The output generator 34 may be utilized to generate one or more graphical outputs 36 that can be presented on the display 12. For example, the output generator 34 can display a plurality of ECG signals, such as may be acquired for a plurality of measurement locations distributed across a body surface (invasive or non-invasive), or derived from measurements of electrical activity on a surface (e.g., an outer surface and/or an inner surface) of a patient's body, such as disclosed herein. The output may also include a graphical, textual, or audible notification or alert indicating the detection of a rapid firing of the heart and/or the time(s) of such rapid firing.

The output generator 34 may also include a user interface 38, which user interface 38 may be used to set parameters to control which ECG signals are included in the output 36 in response to user input, and to otherwise interact with and select portions of the electrophysiological (e.g., ECG) data 18, such as disclosed herein. For example, a baseline frequency parameter for cardiac rapid fire detection may be manually specified using the user interface 38. As another example, the user interface 38 may be used to manually specify a portion of a cardiac waveform to use as a QRST template for a QRST detection and removal process. To specify such a portion, the output generator 34 may generate a set of calipers that are placed at the start time and stop time of the selected interval.

The output generator 34 is also capable of generating one or more electrophysiological maps in the graphical output 36 that can be presented on the display 12. For example, the output generator 34 may generate an activation map or other map representing arrhythmic activity, such as based on the channel after QRST removal. This may be for a selected set of signals distributed across the surface or for the entire surface and for one or more time intervals of interest, which may be selected in response to user input. Examples of the types of output visualizations and maps that may be generated may be found in U.S. patent No.9,427,169 and/or U.S. patent application publication No. 2014/0200822. The output may also include a graphical map illustrating the channel from which the quick fire event was detected. The output may also include a graphical representation of a region of the heart surface showing an estimated or determined location of the origin of the cardiac rapid fire event.

As disclosed herein, in some examples, the ECG data 18 is spatially and temporally consistent across the surface over which the ECG signal is measured or derived. Thus, an ECG signal may be generated for the entire heart surface over one or more time intervals. The output generator 34 can employ a user interface to set parameters for the graphical map and otherwise interact with and select portions of the electrophysiological data 18 in response to user input, such as disclosed herein.

Data collection for rapid cardiac firing detection

Although the cardiac rapid fire detection of the present application does not require the performance of an EP protocol, the use of the described cardiac rapid fire detection may be performed in the context of an EP protocol or similar diagnostic or therapeutic protocol. EP protocols or similar protocols generally involve: first, a period of patient and physician preparation during which the patient prepares for the procedure and the physician involved performs catheterization and/or other preparation; second, appropriate protocols during which measurements may be taken and therapy (e.g., cardiac ablation or drug delivery) may be applied; and third, a rest period after the protocol during which the patient remains at rest and observed, and during which data may continue to be collected. Prior to the procedure, the patient may be equipped with an array of cardiac sensors, such as electrophysiological sensors, which may for example be applied as a vest, such that a plurality of such sensors are distributed over the surface of the chest. As an example, more than one hundred sensors (e.g., two hundred and fifty-two sensors) may be applied. In other examples, different numbers and arrangements of sensors may be used, such as an arrangement of electrodes configured to sense cardiac electrical activity. The signals collected by the applied sensors can be monitored and analyzed for rapid cardiac firing at any of the above-described stages of the procedure.

Heart rapid percussion detection

Fig. 2 depicts an example overview of a first portion of a heart rapid onset detection method, which corresponds to the functionality of the heart rapid onset detector 20 of fig. 1. The middle row of fig. 2 shows three power graphs 54, 50, 66, each of which is illustrated in larger versions in fig. 4B, 5B and 6B, respectively. The power graphs 54, 50, 66 illustrate examples of rapid firing activity in windowed segments from collected electrophysiological data before, during, and after a detected rapid firing event from the Right Atrial Appendage (RAA), respectively. Each graph 54, 50, 66 contains a number of frequency curves (e.g., about 252 curves), one for each electrophysiological channel, each curve corresponding to a frequency spectrum of a cardiac waveform measured from a patient. Each frequency curve may be obtained, for example, by performing a Fast Fourier Transform (FFT) on a window of the collected time-domain electrophysiological data; other frequency conversions may suffice. The length of the window may be selected to be, for example, two seconds, five seconds, ten seconds, or twenty seconds. The frequency curve may have the QRST component (or more specifically, the ventricular QRST component) removed, e.g., using one or more of the methods described herein, in order to better present the frequency components attributable only to atrial electrical activity.

The upper row of fig. 2 shows three graphs 52, 58, 64, each graph being a composite power spectrum derived from substantially all of the curves in the corresponding graphs 54, 60, 66 of the middle row of fig. 2. For example, plot 52 may be obtained by summing all of the frequency plots of plot 54, or by averaging, or by using any other suitable method for obtaining a composite result; graphs 58 and 64 may be similarly obtained from graphs 60 and 66, respectively. Larger versions of graphs 52, 58, 64 are shown in fig. 3A, 4A, and 5A, respectively.

The lower row of fig. 2 shows three graphical representations 66, 62, 68 of the patient's thorax illustrating the electrode distribution thereon, each triangle in each illustrated grid corresponding to the position of an electrode of the set of electrodes from which the electrophysiological data of the middle and upper rows of the graph was obtained. Thus, each triangular surface face represents a channel for data collection. Each triangle is shaded according to the dominant frequency of the corresponding channel. The lighter shaded channels in the graphical representation 62 indicate a detected quick fire event. Larger versions of the graphical representations 66, 62, 68 are shown in fig. 3C, 4C and 5C, respectively.

As an example, the cardiac rapid fire detection method of the present application involves performing a channel-by-channel frequency analysis on collected cardiac waveforms. Based on the frequency analysis, channels exhibiting a rapid firing frequency peak during a particular window may be identified. The channel may then be mapped to one or more epicardial surface regions in order to locate the anatomical origin of the rapid firing of the heart. The quick fire determination (i.e., the binary determination that a quick fire event has occurred) and/or the time, channel, and/or anatomical region associated with the quick fire event may be stored as cardiac quick fire event data 22 (in fig. 1). Inputs to the heart rapid fire detection method include segments of ECG data for time intervals, a list of bad channels (e.g., channels known to be broken or channels exhibiting too high an impedance to deliver useful data), and various input parameters that may be manually or automatically defined.

When analyzing the frequency curve for any given time window to indicate rapid firing activity, all frequency components below about 2Hz, which effectively correspond to, for example, healthy heart activity and baseline drift, may be ignored. For patients with atrial fibrillation, the remaining frequency components will generally show the dominant frequencies of the analyzed channels clustered in one or two frequency complexes. As shown in the frequency analysis graph 54 of FIG. 2 or FIG. 3B, most of the channels each have a corresponding dominant frequency peak in the 4-8Hz range, constituting a baseline complex 70, corresponding to the cardiac activity of fibrillation (e.g., atrial fibrillation). In the example shown, the average dominant frequency of all analyzed channels is 5.12 Hz. There are no significant outlier frequency complexes in graph 54, indicating that no rapid firing activity was detected within the analyzed time frame.

In contrast, the frequency analysis graph 60 in fig. 2, shown in larger form in fig. 4B, from a later time frame than the earlier graph 54, exhibits two dominant complexes 72, 74 of the frequency spectrum. In this example, a second outlier complex 74 (corresponding to atrial rapid fire) is centered at about 10Hz, except for the 4-8Hz baseline complex 72 during atrial fibrillation. Detection of an outlier complex 74 within a frequency range that is higher than the frequency range of the previously established baseline complex 70 may trigger an alert and/or additional analysis that a quick fire event has been detected to locate the anatomical origin of the quick fire event. One or more time stamps indicative of the time of the detected rapid fire activity may be stored in memory, for example, as event data 22 linked with corresponding time and/or frequency domain ECG data. The frequency analysis graph 66 of fig. 2, shown in larger form in fig. 5B, again exhibits only one dominant frequency burst 76, indicating that by the time of the analyzed time frame, the detected quick fire event has ceased.

Rapid firing of the heart is characterized by outlier frequencies showing such frequencies on a subset of channels (e.g., relatively few channels). Because rapid firing of the heart is a local event, it tends to be limited to a relatively small number of channels, for example, about ten of the two hundred and fifty-two channels in the array. In some examples, the outlier frequency cluster has an average dominant frequency that is more than one standard deviation higher in frequency than an average dominant frequency of the baseline cluster corresponding to fibrillating cardiac activity, for example, which may ignore low frequency components (e.g., frequency components below 2 Hz). In other examples, other multiples of the standard deviation of the mean baseline frequency complex may be utilized, such as more than two standard deviations higher, more than three standard deviations higher, or more than four standard deviations higher, and so forth. In some examples, an outlier cluster has an average dominant frequency that is more than one standard deviation higher in frequency, e.g., more than two standard deviations higher, e.g., more than three standard deviations higher, e.g., more than four standard deviations higher, than the average dominant frequency of all channels (ignoring frequency components below 2 Hz).

The establishment of the baseline cluster frequency range and the subsequent detection of the dominant frequency within the outlier cluster frequency range may be performed in a number of ways, each of which may be used alone or in combination: setting a threshold value analysis, a bimodal distribution statistical analysis of the composite signal, a single-channel time analysis and/or a multi-channel spatial analysis.

As shown in the example of fig. 2, outlier channels do not continuously produce higher frequencies. In contrast, even with such a channel, the quick firing activity is brief. Thus, the analysis (e.g., by the detector 20) may be configured to monitor each channel independently and capture data for a subset of channels that exhibit an abrupt switch from their baseline frequency state to a higher dominant frequency. Over time, most or all of the channels may share a common baseline spectrum. However, during a local quick fire event, a channel of a small cluster, which may or may not be a continuous channel, will show a dominant frequency away from the baseline spectrum. In this way, a sudden shift of the average dominant frequency in a small subset of channels (but not their neighbors) from baseline to a higher frequency spectrum (e.g., about 2Hz higher) over a short duration will identify these channels as fast-fire channels.

Set threshold analysis test

As one example of cardiac rapid fire detection, a frequency value may be provided as a frequency threshold, either manually as user input or via an automatic threshold generator (e.g., to processor 16 in fig. 1), and the dominant frequency of the respective channel compared to the threshold within each windowed time frame. As an example, the threshold may be set to, for example, 8Hz, 9Hz, or 10 Hz. Channels exhibiting a dominant frequency greater than a threshold are determined to be fast firing channels and therefore may be used to anatomically localize the fast firing activity. Setting the frequency threshold too low may result in the determination of false positives, while setting the frequency threshold too high may result in missed detection of a quick fire event.

In addition to the frequency threshold, a second threshold indicating a minimum number of channels may be set again, either as user input or via automatic determination. Then, a quick fire event is determined to have occurred only if a threshold number of channels meet a dominant frequency threshold criterion within a time frame.

In other examples, other criteria may be utilized to set and/or change the threshold or threshold range for the quick fire channel. For example, the type of fibrillation (e.g., atrial and/or ventricular), the patient's demographics (e.g., gender, weight, height), and/or the number and distribution of sensors may be used to set or modify the threshold or threshold range.

Testing using bimodal distribution statistical analysis of composite signals

As another example of cardiac rapid fire detection, a composite frequency analysis signal may be generated from frequency analysis of multiple channels during a time frame being analyzed, e.g., by summing or averaging the frequency analysis of all good channels, resulting in a composite analysis, as shown in graphs 52, 58, or 64. Statistical analysis using any of several known tests (e.g., tests performed by Haldane, Larkin, Benett, Tokeshi, or Holzmann and Vollmer) may then be performed to determine whether the composite frequency curve exhibits a bimodal distribution (see fig. 4A) after showing only a unimodal distribution earlier (see fig. 3A, 5A) (and/or before showing only a unimodal distribution later), again ignoring frequency components below about 2 Hz. Detection of a statistically significant bimodal distribution in the composite curve (as in curve 58) triggers a determination that a rapid firing event has occurred. Channels having a dominant frequency greater than a determined anti-mode (anti mode) frequency (e.g., a minimum frequency value or range between modes) during the fast firing event time frame(s) may be labeled as fast firing channels and thus may be used to anatomically position the fast firing activity.

Single channel time analysis test

As yet another example of rapid heart attack detection, the frequency profile of each channel (e.g., excluding the bad channels) may be analyzed to test the change in dominant frequency over time (i.e., over multiple windowed time frames). Any single channel showing significant movement of the dominant frequency from a lower frequency range (e.g., 3-8Hz for an example of atrial fibrillation) to a higher frequency range (e.g., 8-12Hz), or vice versa, may be labeled as a fast-fire channel during the timeframe(s) in which the dominant frequency is in the higher frequency range, and thus may be used to anatomically locate the fast-fire activity.

In some examples, rather than an accurate frequency value being used as a threshold as in the above-described set threshold analysis, relative motion over time (i.e., differences in dominant frequency values) is indicative of a rapid firing onset. However, as with setting the threshold analysis, it may be set that a binary determination that a quick fire episode has occurred is made only when a threshold number of channels have been determined (e.g., within a certain time period) to have met the dominant frequency difference criteria. It may also be desirable for at least a certain number of channels (e.g., 2 channels, 3 channels, or 5 channels) to be directly adjacent to each other to ensure that detected rapid fire channels are present in the cluster, as would be expected for a rapid fire event. The difference criteria and/or the channel number threshold(s) may be either provided manually as user input or may be automatically generated, e.g., adaptively or as default settings.

Multi-channel spatial analysis testing

As yet another example of cardiac rapid fire detection, the frequency curve of each channel may be compared to the frequency curves of one or more spatially adjacent channels (i.e., channels in which corresponding ECG electrodes or other sensors are located in close spatial proximity to one another with respect to placement on the patient's body) during the same time frame. Any single channel that shows a significantly higher dominant frequency during this time frame than the adjacent channels may be labeled as a fast firing channel within this time frame and thus may be used to anatomically localize the fast firing activity.

In another example, rather than the frequency values being exact as in the above-described set threshold analysis being used as thresholds, or the temporal frequencies for a single channel as in the above-described single channel temporal analysis changing, the thresholds may correspond to the difference in dominant frequency values between adjacent channels over a single time period, which difference is indicative of a rapid firing onset. As an example, the difference threshold may be set to 4Hz, 5Hz, 6Hz, or 7 Hz.

As with the other analyses, it may be provided that the binary determination that a rapid-fire episode has occurred is made only when a threshold number of channels have been determined to have satisfied the spatially dominant frequency difference criterion. As an example, the number of channels threshold may be set to 2 channels, 5 channels, or 10 channels. It may also be desirable for at least a certain number of channels (e.g., 2 channels, 3 channels, or 5 channels) to be directly adjacent to each other to ensure that detected rapid fire channels are present in the cluster, as would be expected for a rapid fire event. The difference criteria and/or the channel number threshold(s) may be either provided manually as user input or may be automatically generated, e.g., adaptively or as default settings.

Combined quick fire determination test

The various tests described above can be combined in various ways. For example, a channel may be determined to exhibit rapid firing activity only if it meets the criteria of a given two or more of the above tests, etc.

Anatomical positioning of rapid firing events

As described above, once the quick fire detection is performed, the quick fire may be positioned to only those channels exhibiting quick firing activity. FIG. 6 illustrates a frequency plot 78 with only the quick fire channel plotted. In the example shown, it can be seen that the peak dominant frequency among the rapid fire channels is at about 9.5 Hz. Fig. 7 illustrates a graphical representation 80 of the torso, similar to the graphical representations 56, 62, 68 in the third row of fig. 2 (also shown in fig. 3C, 4C, 5C), indicating the quick fire channel as a lighter shaded triangle. The representation 80 may be displayed as part of the graphical output 36 as shown in fig. 1 to indicate to the user where from among all analyzed channels (e.g., among 252 channels) to fire quickly. In the illustrated example 80 of fig. 7, a region in the middle of the chest exhibits rapid firing.

FIG. 8 shows another graphical representation 82 in which channels are based on inverse transfer matrix A according to the channels^-1Mapped to anatomical regions of the heart, and shaded, the inverse transfer matrix A^-1The calculations may be performed in real time as described in U.S. patent No.9,259,166 to Rudy et al. For example, the upper left shadow channel is known to correspond to the upper right atrium; the shadow channel in the upper right is known to correspond to the right auricle; the remaining shaded channel near the base corresponds to the lower right atrium. Squares around certain channels indicate that these channels are determined to show rapid firing activity, in this example, with a dominant frequency between 8.75Hz and 10.25 Hz. Applying the solution to the inverse problem including a transfer matrix linking certain regions on the body surface to certain regions on the heart may result in a graphical representation 84 of a portion of the heart anatomy, such as shown in fig. 9, where the quick-fire regions are depicted using the shading scheme of fig. 8. Since the matching table giving the mapping between the electrodes and the heart surface area is already known to the patient before the procedure involving the detection of the rapid fire, the matching heart position corresponding to the body surface position is known immediately upon the detection of the rapid fire on a certain area of the body surface. The anatomical graphical map 84 may be displayed as part of the graphical output 36 as shown in fig. 1 to indicate to the user the anatomical location from which the quick fire originated. Because the graphical map 82 may immediately shadow regions on the cardiac anatomy (here, on the atrial anatomy) based on a predetermined matching table, the quick fire map 82 may be displayed immediately upon detection of a quick fire event, thereby facilitating real-time use and assisted therapeutic intervention during the procedure.

The number of shaded points in the example map of fig. 9 is used in the illustration as a way to "draw" the anatomical region. Various shading or visualization techniques may be used to render the graphical map based on the quick fire data.

Time domain QRST detection and removal

As disclosed herein, the systems and methods disclosed herein can detect and remove QRST complexes. For example, instead of treating QRS and T as separate entities, the methods herein treat them as a single entity.

Fig. 10 is a flow diagram depicting an example time-domain method 100 of removing a QRST complex from a cardiac waveform, such as may be implemented by the QRST detection and removal function 32 of fig. 1. At 102, the QRST detection and removal function performs Principal Component Analysis (PCA) on the selected region of interest of the cardiac waveform. The region of interest may be selected automatically or manually in response to user input identifying the interval of the signal corresponding to the QRST complex. Thus, PCA can be used to generate QRST template definitions. At 104, the template is applied across time frames to identify matching regions of interest by correlation, such as by time stepping the template relative to the determined correlation coefficients. The peak correlation coefficients are used to identify potential locations where the template matches the data. The correlation coefficient may be compared to a threshold to identify a corresponding region of interest for each of the plurality of channels.

At 106, each matching region of interest (i.e., each region corresponding to a QRST complex) is removed from the cardiac waveform and interpolation (e.g., spline interpolation) is performed to automatically connect adjacent P-waves. As an example, the interpolation may be implemented as a shape-preserving PCHIP function or another spline interpolation function. Such an interpolation function keeps the interpolated values monotonic (e.g., either increasing or decreasing) based on the endpoint value used for such interpolation. Thus, in the analyzed cardiac waveform, the QRST complex is partially replaced by a substitute signal that does not have high frequency components that would interfere with the waveform analysis for purposes of rapid fire detection. Prior to performing the QRST detection and removal method 100, baseline removal and/or removal of bad input channels may be performed.

As an example, a QRST complex may be defined to generate a template ROI, such as shown in the curve 110 of the ECG data in fig. 11. For example, a QRST template 112 is computed and a matching ROI is detected 114 via correlation, as described above. The P-wave may be automatically calculated 116 between two adjacent ROIs, as shown in fig. 11. Since a single template is used for the entire QRST complex, a single set of calipers can be used to identify the region of interest.

As a simplified example, QRST detection and removal may manually define a QRST complex once per interval in response to user input. In another example, the QRST detection and removal functionality may implement a semi-automatic or fully automatic method, such as by automatic template matching with some standard QRST complexes or pre-selected or even pre-detected QRSTs.

The QRST detection and removal function (e.g., function 32, 578) operates to remove the QRS and T regions of the cardiac waveform such that the residual signal amplitude in the QRS and T regions is no better than the residual signal amplitude of the P-wave or results in a residual signal within the P-wave. To reduce artifacts caused by the QRS and T signals on the P signal during subsequent filtering to obtain a frequency bandwidth of interest (e.g., about 4-15Hz), QRST detection and removal may remove the QRST region by interpolation with the low frequency signal. This can be achieved by using a method such as monotonic cubic spline interpolation between the beginning and end of each QRST ROI.

Fig. 12 illustrates an example case where the time domain approach described herein successfully detects and removes QRST. As can be seen from the coverage curve 124, one shows the original signal, one shows the signal after QRST removal, which has been removed by spline interpolation of QRST.

As another approach, instead of defining a QRST complex and performing template matching, the QRST detection and removal function may define the P-wave for one beat and then fill in (pad) any signals outside the P-wave. For example, this method may work in a beat-by-beat manual frame.

Atrial signals in the QRST complex

For a normal sinus rhythm heart, there is no potential atrial signal in the QRST complex. However, for arrhythmias like atrial fibrillation and for rapid firing of the atria, atrial signals may be present during the QRST complex. To use atrial signals during the QRST complex in atrial rapid fire detection, a reliable QRST subtraction method such as that described below may be used to remove ventricular portions of the signals, enabling analysis of atrial signals, including atrial signals that may reside in the QRST complex.

One example of mitigating impairment of atrial signals in a region of interest (QRST) is the identification of good QRST complexes, with or without signal averaging (e.g., "clean" QRST complexes), during normal sinus rhythm. The QRST detection and removal function may perform template matching between a clean QRST complex and a QRST complex in atrial fibrillation. By not performing any ROI averaging in defining the template, the QRST detection and removal function may subtract the contribution of a clean QRST complex from the QRST complex of each arrhythmia, so that the remaining signal within the QRST interval will include the atrial signal.

To reduce user interaction during map creation, the user may pick one template per procedure. For example, to define the QRST complex, the beginning and end of the interval definition are placed where the signal is flat or where there is less heart activity. Baseline drift due to respiratory motion, etc., may alter the template profile, so the baseline removal step may be performed prior to QRST removal. Baseline removal may also precede automatic bad channel identification to reduce baseline wander effects on that portion of the overall process.

As another example, the QRST detection and removal process may be implemented as a method. The method includes performing a principal component analysis on the selected region of interest with respect to a plurality of ECG signals to define a QRST template. The method also includes correlating the interval of the QRST template relative to each of the plurality of ECG signals to identify a matching region of interest. The method also includes removing the identified matching region of interest from each of the plurality of ECG signals using interpolation. For example, the region of interest is manually selected in response to a user input, or automatically selected. As another example, a QRST template defines a single template that is applied to each ECG signal in a given time interval. As another example, the interpolation implemented by the method includes monotonic cubic spline interpolation to connect P-waves together for adjacent beats. As another example, prior to removing the identified matching regions of interest, the method further includes averaging the templates across the regions of interest. As another example, prior to removing the identified matching region of interest, the method further includes adjusting the template to account for baseline drift in the ECG signal. As another example, the ECG signal includes time intervals exhibiting atrial fibrillation. In this example, the method further comprises: identifying a clean QRST complex during a sinus rhythm without atrial fibrillation; performing template matching between a clean QRST complex and a QRST complex during atrial fibrillation; and removing the clean QRST template from the ECG signal. In some examples, the method further comprises automatically determining each P-wave as a region between two adjacent QRST regions of interest. As another example, the method further includes detecting an R peak for each ECG signal and using the detected R peaks to locate intervals containing QRST complexes.

As disclosed, one or more non-transitory computer-readable media store instructions for performing any variation of the QRST detection and removal method.

Frequency domain QRST detection and removal

The length of the QRST complex can vary greatly over time at intervals (i.e., from beat to beat). Thus, frequency domain QRST removal may involve performing a frequency analysis as described above on the identified QRST frequency template and subtracting the resulting template frequency curve from the frequency curve of the corresponding channel. This is done channel by channel. For each time window, the number of QRS complexes should be provided to ensure that the correct amount of power contributed by the QRS complexes is removed; thus, the template frequency curve should not be normalized.

Acquisition, output display and processing

Fig. 13 depicts an example of a system 550 that may be used to generate outputs to process body surface signals to characterize a patient's arrhythmic activity and perform rapid fire detection. In some examples, the system 550 may generate a graphical map (e.g., a body surface map or a map of a heart model) 594 and/or display the processed electrical signals. The system may also provide information in other formats to provide guidance to the user indicating one or more of the calculated signal characteristics and information derived from such calculated signal characteristics.

As disclosed herein, the system 550 has application in various stages of patient care. By way of example, the system may be used as part of a patient screening process (e.g., as part of a diagnostic and/or treatment planning procedure) or to perform post-treatment evaluations. Further, the system 550 may be used as part of a therapy protocol, such as determining parameters for delivering therapy to a patient (e.g., delivery location, amount and type of therapy). For example, a catheter having one or more therapy delivery devices 556 secured thereto may be inserted into the body 554 so as to contact the patient's heart 552 endocardially or epicardially. Those skilled in the art will understand and appreciate the various types and configurations of therapy delivery devices 556 that may be used, which may vary depending on the type of procedure and therapy. For example, the treatment device 556 may be configured to deliver electrical therapy, chemotherapy, sonic therapy, thermal therapy, or any combination thereof.

By way of further example, the therapy delivery device 556 may include one or more electrodes at the tip of the ablation catheter configured to generate heat for ablating tissue in response to electrical signals (e.g., radio frequency energy) supplied by the therapy system 558. In other examples, the therapy delivery device 556 may be configured to deliver cooling to perform ablation (e.g., cryoablation), deliver a chemical (e.g., a drug), ultrasound ablation, high frequency radio frequency ablation, or a combination thereof. In still other examples, the therapy delivery device 556 may include one or more electrodes at the tip of a pacing catheter to deliver electrical stimulation, such as for pacing the heart, in response to electrical signals (e.g., pacing current pulses) supplied by the therapy system 558. Other types of therapy can also be delivered via the therapy system 558 and the invasive therapy delivery device 556 positioned within the body.

As another example, the therapy system 558 may be located external to the patient's body 554 and configured to control therapy delivered by the device 556. For example, the therapy system 558 includes a control system (e.g., hardware and/or software) 560, and the control system 560 may communicate (e.g., supply) electrical signals via an electrically conductive link electrically connected between the delivery device (e.g., one or more electrodes) 556 and the therapy system 558. Control system 560 may control parameters (e.g., current, voltage, repetition rate, trigger delay, sense trigger amplitude) of signals supplied to device 556 for delivering therapy (e.g., ablation or stimulation) to one or more locations of heart 552 via electrode(s) 554. The control system 560 may set the treatment parameters and apply stimulation based on automatic, manual (e.g., user input), or a combination of automatic and manual (e.g., semi-automatic control). One or more sensors (not shown) may also transmit sensor information back to the therapy system 558. The position of the device 556 relative to the heart 552 may be determined and tracked intra-operatively via imaging modalities (e.g., fluoroscopy, X-ray), mapping system 562, direct vision, and the like. The location of the device 556 and the therapy parameters can thus be combined to determine corresponding therapy delivery parameters.

Before, during, and/or after providing therapy via the therapy system 558, electrophysiological information for the patient may be obtained with another system or subsystem. In the example of fig. 1313, sensor array 564 includes one or more body surface electrodes that may be used to measure patient electrical activity. As one example, the sensor array 564 may correspond to a high density arrangement of body surface sensors (e.g., greater than approximately one hundred electrodes, e.g., greater than approximately two hundred electrodes, e.g., two hundred fifty-two electrodes) distributed over a portion of a patient's torso (e.g., the chest cavity) to measure electrical activity associated with the patient's heart (e.g., as part of an electrocardiography mapping procedure). Examples of high density body surface non-invasive devices that may be used as sensor array 564 are shown and described in U.S. patent No.9,655,561 and international publication No. wo 2010/054352. Other arrangements and numbers of sensing electrodes may be used as sensor array 564. For example, the array may be a reduced set of electrodes that do not cover the entire torso of the patient and are designed to measure electrical activity for a specific purpose (e.g., an array of electrodes specifically designed for analyzing atrial fibrillation and/or ventricular fibrillation) and/or for monitoring a predetermined spatial region of the heart. In other examples, an array with a conventional or modified 12-lead ECG or a single electrode may be implemented as the sensor array 564 to provide body surface electrical signals.

In some examples, one or more sensors may also be located on a device 556 that is inserted into the patient's body. Such sensors may be used alone or in combination with the non-invasive sensor array 564 to map electrical activity against endocardial surfaces (such as the walls of a heart chamber) as well as against epicardial surfaces. In addition, such electrodes may also be used to help locate device 556 within heart 552, which may be registered to an image or map generated by system 550. Alternatively, such positioning may be accomplished without emitting signals from electrodes within the heart 552 or on the heart 552.

In each such example method for acquiring patient electrical information, including invasive, non-invasive, or a combination of invasive and non-invasive sensing, the sensor array(s) 564 provide the sensed electrical information to the corresponding measurement system 566. Measurement system 566 may include a corresponding control 568, control 568 configured to provide electrical measurement data 570 (e.g., ECG signals) descriptive of electrical activity detected by the sensors in sensor array 564. For example, the signal processing circuitry of the measurement system 566 may convert the measured analog signal(s) into corresponding digital information. The measurement system 566 may also process digital information corresponding to one or more electrophysiological signals from the sensor array 564 and remove non-arrhythmic features from each such signal and provide pre-processed data stored in memory as electrical measurement data 570.

Control 568 may also be configured to control a data acquisition process (e.g., at a predefined sampling rate) for measuring electrical activity and providing measurement data 570. In some examples, control 568 may control acquisition of measurement data 570 separately from operation of therapy system 558 (if implemented), such as in response to user input. In other examples, the measurement data 570 may be acquired concurrently and synchronously with the delivery of therapy by the therapy system, such as to detect electrical activity of the heart 552 occurring in response to the application of a given therapy (e.g., as a function of therapy parameters). For example, an appropriate timestamp may be used to index the temporal relationship between the corresponding measurement data 570 and the use of the treatment parameters to facilitate their evaluation and analysis.

Mapping system 562 is programmed to combine measurement data 570 corresponding to sensed body surface electrical activity of heart 552 to provide corresponding output data 574. The output data 574 may represent or characterize the detected ECG signals on the body surface and/or within the heart. The output data may also represent information derived from the measured signal, such as disclosed herein.

As one example, the mapping system 562 includes a cardiac rapid fire detection function 577, such as corresponding to the cardiac rapid fire detector 20 (e.g., as disclosed herein with respect to fig. 2-9). Mapping system 562 may also include a QRST detection and removal function 578, such as corresponding to QRST detection and removal function 32 (e.g., as disclosed with respect to fig. 10-15). Each of the functions 577 and 578 may be applied to ECG data presented as electrical measurement data 570. As mentioned, in some examples, the cardiac rapid fire detection function 577 and the QRST detection and removal function 578 operate on raw ECG data (e.g., acquired via non-invasive electrodes to measure electrical signals across a body surface) to detect cardiac (e.g., atrial) rapid fire and remove QRST signals from the raw signals, respectively.

The mapping system 562 includes an output generator to provide output data 574 to visualize one or more intervals of ECG signals on a display 592 based on electrical measurement data acquired for a patient at one or more time intervals (e.g., before, after, or during an EP protocol or treatment protocol). In examples where the sensor array 564 includes multiple electrodes, the output data 574 may include a selected set of channels of ECG signals measured via the sensors 564 on the surface of the patient's body. The parameters may be set to identify a subset of signals that satisfy one or more user-configurable parameters (e.g., via GUI 590). Some examples of output displays that may be provided by output generator 586 are disclosed with respect to fig. 3C, 4C, 5C, 7, 8, and 9. Thus, the output generator generates output data to display a graphical representation of a time domain curve, a frequency domain curve, a channel disposed on the torso, or a region mapped to the epicardial surface.

In some examples, the computed data may be mapped to a geometric surface of a heart model. As disclosed herein, the map may be calculated based on electrical data acquired non-invasively via one or more electrodes in a sensor array 564 distributed over the surface of the patient's body 554.

Since the measurement system 566 may concurrently measure electrical activity of a predetermined region of the torso or the entire torso (e.g., where the sensor array 564 includes a plurality of electrodes covering the entire thorax of the patient's body 554), the resulting output data (e.g., the ECG signal and/or the electrocardiogram) may thus also be capable of representing concurrent cardiac electrical data in a temporally and spatially consistent manner. The time interval for calculating the output data/map may be selected based on user input. Additionally or alternatively, the selected interval may be synchronized with the application of therapy by the therapy system 558. As disclosed herein, an indication of the presence or absence of stable arrhythmic activity may be calculated from body surface electrical signal(s) without performing electrogram reconstruction based on patient geometry.

In other examples where additional information may be obtained and the geometry data 572 may be obtained, the system may include an electrogram reconstruction 580 programmed to calculate an inverse solution and provide a corresponding reconstructed electrogram based on the processed signal and the geometry data 572. For example, the geometry data 572 may correspond to a mathematical model, such as may be a generic model or a model that is constructed based on image data obtained for the patient (e.g., via an imaging modality such as CT, MRI, biplane X-ray, etc.) and provides spatial coordinates for the patient's heart 552 and electrodes on the sensor array. Thus, the reconstructed electrogram may correspond to electrocardiographic activity across the pericardium (cardiac envelope) and may include static (three-dimensional at a given time) and/or dynamic (e.g., four-dimensional maps that vary over time). Examples of inverse algorithms that may be used in the system 550 include the inverse algorithms disclosed in U.S. patent nos. 7,983,743 and 6,772,004. Accordingly, EGM reconstruction 580 may reconstruct the body surface electrical activity measured via sensor array 564 onto multiple locations (e.g., greater than 1000 locations, such as about 2000 locations or more) on the pericardium. In other examples, the mapping system 562 can calculate electrical activity on a sub-region of the heart based on invasively measured electrical activity, such as via a basket catheter or other form of measurement probe (e.g., on or attached to the device 556).

Parameters associated with the graphical representation corresponding to the output visualization of the computed map (such as including selection time intervals, types of information to be presented in the visualization, etc.) may be selected in response to user input via the corresponding visualization GUI 590.

Additionally, if included in the system 550, the therapy system 558 may use the output data 574. The control achieved may be fully automatic, semi-automatic (partially automatic and responsive to user input), or manual based on the output data 574. In some examples, the control system 560 for the therapy system 558 may utilize the output data to control one or more therapy parameters. As an example, control 560 can control delivery of ablation therapy to a cardiac site (e.g., epicardial or endocardial wall) based on the herein disclosed rapid fire data that has been determined by function 577. For example, delivery of therapy may be automatically terminated in response to detecting no rapid firing of the heart (e.g., atrium) after a period of time or no stable driving activity after a period of time. In other examples, an individual user may view a map generated in a display to manually control a therapy system based on visualized information. Other types of treatments and devices may also be controlled based on the output data.

Fig. 14 is a flow diagram depicting an example method 1400 of detecting rapid firing activity of a heart (e.g., an atrium). The method 1400 includes collecting cardiac waveform data from a plurality of channels, for example, as may be obtained from an electrode array on a body surface (e.g., a chest) of a patient, e.g., more than one hundred channels, e.g., more than two hundred channels, e.g., two hundred and fifty-two channels. The QRST component may be removed 1404 from each channel of the collected cardiac waveform using, for example, one or a combination of time-based or frequency-based methods described herein. In some examples, only the ventricular QRST component is removed, such that signals originating from the atrium remain in the filtered signal. A frequency analysis is performed 1406 for each channel over the moving window. As an example, the length of the window may be, for example, two seconds, or five seconds, or ten seconds, or twenty seconds. For example, using one or a combination of the tests described herein for rapid firing activity detection, channels exhibiting a rapid firing frequency peak during a particular window are identified 1408. When a heart rapid firing activity is detected, a notification or warning may be issued. The channel identified as quick fire in a particular time frame is mapped 1410 to one or more epicardial surface regions and a graphical output indicating the time and epicardial location of the quick fire activity may be provided 1412, e.g., via a visual display. As described herein, the displayed graphical map may be used to guide therapy, such as ablation or drug delivery, and/or may be used to automatically control therapy delivery.

In view of the foregoing structural and functional description, those skilled in the art will recognize that portions of the present invention may be embodied as a method, data processing system, or computer program product. Accordingly, these portions of the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, portions of the present invention may be a computer program product on a computer usable storage medium having computer readable program code on the medium. Any suitable computer readable medium may be utilized including, but not limited to, static and dynamic memory devices, hard disks, optical memory devices, and magnetic memory devices.

Certain embodiments of the present invention have also been described herein with reference to block illustrations of methods, systems, and computer program products. It should be understood that the illustrated blocks, and combinations of blocks in the illustrations, can be implemented by computer-executable instructions. These computer-executable instructions may be provided to one or more processors of a general purpose computer, special purpose computer, or other programmable data processing apparatus (or a combination of devices and circuits) to produce a machine, such that the instructions, which execute via the processors, implement the functions specified in the block or blocks.

These computer-executable instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instructions which implement the function specified in the flowchart block or blocks. The 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 which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks.

What has been described above is an example. It is, of course, not possible to describe every conceivable combination of components or methodologies, but one of ordinary skill in the art will recognize that many further combinations and permutations are possible. Accordingly, the present invention is intended to embrace all such alterations, modifications and variations that fall within the scope of the present application, including the appended claims. Where the disclosure or claims recite "a," "an," "a first," or "another" element or the equivalent thereof, then it should be interpreted to include one or more such elements, neither requiring nor excluding two or more such elements. As used herein, the term "including" means "including but not limited to," and the term "comprising" means "including but not limited to. The term "based on" means "based at least in part on".