Method, device and computer program product for determining an isoelectric point

Abstract

Method for determining an isoelectric point in an electrocardiogram (1) comprising the steps: a) recording (6) an electrocardiogram (1); b) performing a wavelet transformation (7) of the electrocardiogram (1); c) determining a first wavelet coefficient and a second wavelet coefficient from the wavelet transform (7) which partially overlap each other, or using an already existing first wavelet coefficient and an already existing second wavelet coefficient which partially overlap each other; d) determining an R-wave (R) in the wavelet transform of the electrocardiogram (1) and selecting (11) a data range lying before the R-wave (R); e) checking (13) the selected data range of the wavelet transform for a value below a predetermined first threshold value (SWA) of a first characteristic variable which depends on the first wavelet coefficient; f) checking (13) the selected data range of the wavelet transform for a value below a predetermined second threshold value (SWB) of a second characteristic variable which depends on the second wavelet coefficient; g) wherein if the two threshold values ​​(SWA, SWB) are undershot in a data point of the selected data range, the determined data point is defined as an isoelectric point (17), wherein the method with steps a) to g) is repeatedly carried out and the first threshold value (SWA) and the second threshold value (SWB) are determined as a function of previous amplitude values ​​or previous values ​​of the wavelet coefficients.

Description

[0001] The present invention relates to a method for determining an isoelectric point, a device for carrying out the method and a computer program product for carrying out the method and / or for controlling the device.

[0002] An electrocardiogram (ECG) reflects the condition and function of the heart and, as an established technique, represents an important tool for medical diagnostics. The electrocardiogram provides a direct link to the condition and function of the heart, is non-invasive, and provides a quantitative assessment. Long-term ECG monitoring can also be used to assess performance during sports and in extreme situations, to detect stress, or to monitor alertness while driving. For this purpose, an ECG module is applied close to the body, for example, integrated into clothing. However, corresponding devices must be adapted in terms of size and weight, continuously improved in terms of quality and functionality, and optimized for low cost and low energy consumption. Since the goal is to continually increase the efficiency, safety, and autonomy of device function, new, high-performance real-time ECG detection algorithms are necessary.

[0003] Due to its ability to capture high-frequency signal components with precise timing and low-frequency signal components with precise frequency, wavelet-based time-frequency analysis is particularly well-suited for biosignals such as electrocardiograms. For example, QRS complexes in the ECG can be identified efficiently and with high reliability using so-called "modulus maximum pairs" in the frequency-typical scales and serve as the basis for further signal analysis steps. For example, the publication US 2012 / 123232 A1 also discloses a method for reliable R-wave detection.

[0004] Biosignal analysis in the image domain of time-frequency transformation supports subsequent signal processing steps such as filtering or correlating. However, this approach is disadvantageous in that it involves working in an unvisual image domain. Detection of additional ECG segments—such as P waves and T waves—cannot be achieved with the same degree of certainty as R-wave detection. Furthermore, measuring the amplitude of the observed biosignal (which corresponds to a voltage value) is not possible in the image domain. Thus, the course of the isoelectric line and a direct comparison to the level of the ST segment cannot be recorded. However, these parameters are important for a physiological assessment of the biosignal and must be determined indirectly when applying time-frequency analysis.

[0005] The present invention is therefore based on the object of proposing a method and a device for determining an isoelectric point in an electrocardiogram, which avoids the disadvantages mentioned, and with which the isoelectric point in an electrocardiogram can be detected quickly and with high reliability.

[0006] This object is achieved according to the invention by a method according to claim 1, a device according to claim 7 and a computer program product according to claim 9. Advantageous embodiments and further developments are described in the dependent claims.

[0007] A method for determining an isoelectric point or an isoelectric line (i.e., several consecutive isoelectric points) in an electrocardiogram comprises several steps. First, the electrocardiogram is recorded, and a wavelet transform of the electrocardiogram is performed. A first wavelet coefficient and a second wavelet coefficient are determined from the wavelet transform, which partially overlap one another, or an already existing first wavelet coefficient and an already existing second wavelet coefficient, which partially overlap one another, are used. In the resulting wavelet transform of the electrocardiogram, the R wave is determined, and a data range prior to the R wave is selected to be checked for the presence of the isoelectric point.The selected data range is then examined to determine whether a first characteristic value falls below a first predetermined threshold, the first characteristic value being dependent on the first wavelet coefficient. Furthermore, the selected data range of the wavelet transform is checked to determine whether a second predetermined threshold of a second characteristic value falls below a second predetermined threshold. The second characteristic value is dependent on the second wavelet coefficient. If both the first threshold and the second threshold are undershot in a data point of the selected data range, this data point is defined as the isoelectric point. The described method is repeated using the described steps. The first threshold and / or the second threshold are determined as a function of previous amplitude values ​​or previous values ​​of the first wavelet coefficient and the second wavelet coefficient.

[0008] By examining the wavelet transform, which is a linear transformation with conservation of signal energy, in which frequency components of a signal can be analyzed and localized in the spatial and temporal domains, the isoelectric point or isoelectric line can be reliably detected without shape recognition. The use of parameters that depend on the wavelet coefficients obtained as part of the wavelet transform allows the procedure to be carried out quickly and efficiently. Finally, R-wave detection, which can be carried out with a high degree of reliability using modern methods, also makes it easy to define the data range to be examined. An atrial-ventricular conduction can thus be found in the image area of ​​the electrocardiogram and measured for each heartbeat directly in a transmitted digital signal, i.e. in the electrocardiogram.This also allows the presence of a sinus rhythm to be determined if the R wave and the isoelectric point are reliably detected. Typically, this repeats continuously. This allows the isoelectric point to be continuously determined, making it very easy to monitor the presence of sinus rhythm.

[0009] It can be provided that the first parameter and / or the second parameter is a sum of two wavelet coefficients. This allows for an increase in detection reliability.

[0010] Typically, the test for undershooting the first and second thresholds is performed simultaneously. Parallel testing allows the process to be performed quickly, especially since the determined data point is defined as the isoelectric point even if both thresholds are undershot simultaneously. Preferably, the two wavelet coefficients are determined during the wavelet transformation to enable the process to be performed very quickly, typically in real time.

[0011] The first characteristic value can be equal to the absolute value of the first determined wavelet coefficient, and the second characteristic value can be equal to the sum of the absolute values ​​of the first wavelet coefficient and the second wavelet coefficient. Forming absolute values ​​facilitates further processing, while summing absolute values ​​results in a value that is easy to verify.

[0012] Preferably, the first wavelet coefficient is the wavelet coefficient of scale 3, and the second wavelet coefficient is the wavelet coefficient of scale 4, since these coefficients cover a frequency range typically of interest in the context of electrocardiogram analysis. Of course, other wavelet coefficients can also be used, with their scales typically differing by 1.

[0013] Alternatively or additionally, the first threshold may be no more than one-tenth of the total energy of the electrocardiogram, and / or the second threshold may be no more than one-fifth of the total energy of the electrocardiogram. These maximum values ​​are favorable for a simple and reliable determination of the isoelectric point. The two thresholds can be set either before or after the electrocardiogram is recorded.

[0014] Once the isoelectric point has been detected, a corresponding electrocardiogram amplitude for the determined isoelectric point can be determined from the electrocardiogram. Preferably, both this amplitude and the determined isoelectric point are displayed on an output unit to make the information recognizable to a user of the method.

[0015] A device for performing the above-described method according to one of the preceding claims comprises a recording unit, a computing unit, and an output unit. The recording unit is configured to record the electrocardiogram, while the computing unit serves to perform the wavelet transformation and determine the isoelectric point.

[0016] It can also be provided that the first wavelet coefficient and the second wavelet coefficient are stored in a memory of the device, preferably in a ring memory, and the computing unit is designed to load the coefficients from there or to overwrite the stored values ​​of the coefficients when the two coefficients are determined again.

[0017] The device is typically a device for performing a long-term ECG or a cardiac pacemaker.

[0018] A computer program product contains a sequence of instructions stored on a machine-readable medium for carrying out the described method and / or for controlling the described device when the computer program product runs on a computing unit.

[0019] Embodiments of the invention are illustrated in the drawings and are described below with reference to Fig. 1 to 4 explained.

[0020] They show: Fig. 1 an idealized electrocardiogram; Fig. 2 a real voltage curve of the electrocardiogram and wavelet coefficients determined from it; Fig. 3 a flowchart of a method for determining an isoelectric point and Fig. 4 is a schematic representation of a device attached to a body for performing the Fig. 3 described procedure.

[0021] In Fig. Figure 1 shows an example of an electrocardiogram 1, i.e., a time series of electrical signals from cardiac muscle fibers. Different voltages are recorded along a time axis, with peaks usually designated by the letters P, Q, R, S, and T for easier differentiation. Such a complex of peaks repeats periodically in electrocardiogram 1.

[0022] QRS complexes, often referred to as R waves or ventricular oscillations, follow one another, with intervals between two consecutive ventricular oscillations being referred to as RR intervals 2. If the heart beats with a sinus rhythm, the duration of each RR interval 2 deviates by no more than 15 percent from the duration of an immediately preceding RR interval 2.

[0023] Between a P wave and the QRS complex, an isoelectric curve 4 is shown in the electrocardiogram 1 due to the conduction of an excitation, i.e., the atria of the heart are fully excited, while the ventricles of the heart are still completely unexcited. The isoelectric curve 4 is composed of several isoelectric points and is often referred to as the baseline or zero line of a biosignal such as the electrocardiogram 1.

[0024] The length of this isoelectric waveform 4 or this isoelectric segment and its distance to the following QRS complex as part of a PQ interval 3 vary only slightly, depending, among other things, on the heart rate. Ultimately, both are determined by a physiological process of cardiac activity. The signal segments 4 represent a characteristic of physiological cardiac excitation, which physicians call sinus beat or sinus rhythm. This characteristic is to be automatically detected using the method described in more detail below.

[0025] This occurs directly in the image domain of the wavelet transform without shape recognition. Two selected wavelet coefficients are sampled within a fixed time range before each detected R wave to determine whether they simultaneously fall below a threshold. As soon as the two selected wavelet coefficients fall below this threshold for the first time, a time of atrial-ventricular conduction is present.

[0026] This allows the position of the baseline of electrocardiogram 1 to be read directly from the recorded voltage-time curve and used as a reference point for later analysis steps such as measuring an ST line.

[0027] Such a voltage curve of the electrocardiogram 1 is exemplary for a single heartbeat in Fig. 2. Recurring features are provided with identical reference numerals in this and the following figures. The actual voltage curve differs from that shown in Fig. 1 idealized curve slightly. Two wavelet coefficients are shown below electrocardiogram 1. In the middle is the wavelet coefficient of scale 3 and in the lower section of the Fig. 2 shows the wavelet coefficient of scale 4.

[0028] These wavelet coefficients can be implemented very efficiently within an automatic analysis of electrocardiogram 1. R-wave detection is necessary in any case and can be performed very reliably, although a detection algorithm using a wavelet transformation of the biosignal, i.e., electrocardiogram 1, is often performed for this purpose anyway.

[0029] Fig. Figure 3 shows a flowchart of a method for the computational analysis of electrocardiogram 1, which is performed under time-critical conditions, i.e., during real-time or signal-compressed signal processing. The goal of the method in real-time operation is to locate the isoelectric point in electrocardiogram 1, which can be used to determine whether the measured current heartbeat can be attributed to normal cardiac activity, i.e., a sinus rhythm.

[0030] After starting the method, a first detection threshold SWA and a second detection threshold SWB are set in step 5. These thresholds can be set variably, for example, depending on previous amplitude or coefficient values, but can also be statically specified. Setting the thresholds can also be done during a learning phase, and this setting should guarantee a sufficiently smooth signal profile in the time windows under consideration. In other embodiments, setting the detection thresholds can also be done at a later time.

[0031] In step 6, a signal is recorded, i.e., the electrocardiogram 1 is measured at a sampling rate of 250 sps (samples per second) in the illustrated embodiment. The electrocardiogram 1 is discretized, and the loaded digital values ​​are stored for wavelet transformation. Subsequently, in a further step 7, the measured electrocardiogram 1 is subjected to wavelet transformation. Relevant wavelet coefficients, which were defined in advance, are stored in ring buffers, which are continuously updated in correlation with the recorded voltage-time curve of the biosignal.

[0032] The R-wave is detected in step 8, for example, using "modulus maximum pairs" on scales 2 and 3 of the wavelet transform. The individual major fluctuations detected in electrocardiogram 1, referred to as R-triggers, are stored and consecutively fed for further processing. The detection trigger for the R-wave lies within the QRS complex, and step 9 checks whether this detection trigger is present. If so, the method continues with step 10 and sets a counter n to one beat of the oldest R-trigger detected. This means that the time of occurrence, i.e., a beat number in the discrete time signal, is set to counter n.

[0033] If no detection trigger is found, the procedure checks in step 16 whether further data is available and, if so, repeats the procedure starting with step 6. If no further data is available, the procedure is terminated.

[0034] In step 11, a data range or window boundaries are selected to be examined for the presence of the isoelectric point. In the illustrated embodiment, this data range is located at a starting data point n-10, i.e., 10 beats before the corresponding R-trigger, and an ending data point n-25, i.e., 25 beats before the corresponding R-trigger. Depending on the instantaneous heart rate, the search interval can also be varied in other embodiments. However, the search range may extend to a maximum of one end of the T wave of the previous beat.

[0035] In this defined range, the wavelet coefficients of scales 3 and 4, either stored in a ring buffer or determined during the wavelet transformation, are to be tested. These wavelet coefficients capture energy components of electrocardiogram 1 between 9 Hz and 27 Hz, or between 4 Hz and 13 Hz. It is advantageous that the corresponding filter lengths of 13 sampling points and 29 sampling points (corresponding to 52 ms and 116 ms, respectively) have a smoothing effect and overlap each other (centered). Electrocardiogram 1 is thus examined at two different resolutions.

[0036] The search begins at the starting point 40 ms before the R-trigger and continues up to a maximum of 100 ms before the R-trigger. In step 12, a sum is first calculated from the absolute values ​​of the two wavelet coefficients, and in step 13 it is examined whether, at the same time, for one of the data points, the absolute value of the wavelet coefficient of scale 3 is less than the first threshold SWA and the calculated sum is less than the second threshold SWB. If the wavelet coefficient of scale 3 falls below the power threshold SWA and, at the same time, the sum of the two coefficients of scales 3 and 4 is below the threshold SWB, the isoelectric line 4 or the isoelectric point is deemed to have been detected in step 17. The search is then terminated, and an amplitude belonging to the determined data point, i.e. the associated voltage value, is taken directly from the recorded biosignal, i.e., electrocardiogram 1, in step 18.Finally, in step 19, the determined rate and amplitude of the isoelectric point are output. Subsequently, in step 16, the presence of further data is checked to verify the presence of the isoelectric point. A tenth of the total energy of electrocardiogram 1 has proven to be favorable for the SWA threshold, and a fifth of the total energy of electrocardiogram 1 has proven favorable for the SWB threshold.

[0037] If the specified conditions are not met in step 13, the counter n is decremented in step 14, i.e., decreased by 1. If the counter now reaches the specified end point of the search in the interval before the R-trigger, the process continues with step 16. Otherwise, the processing loop is run through again, starting with step 12, with the counter n-1 decremented and the wavelet coefficients reloaded accordingly.

[0038] Once all sections of electrocardiogram 1 have been processed, the procedure ends.

[0039] The method can be stored as a computer program on a machine-readable medium and run on the computer or machine. The computer program can also control a device for carrying out the method, as described in more detail below.

[0040] The method can be performed on a computer or microcontroller and serves to step-by-step process the externally or intracardially recorded, digitized electrocardiogram 1 in the image area with the aim of detecting certain invariant shape segments associated with the conduction of excitation from the atria to the ventricles and to indicate the time of their occurrence. For this purpose, a continuous pattern search is performed in a physiologically defined interval before the detection of ventricular excitation. Two superimposed wavelet coefficients are tested for a threshold violation, thus determining the temporal position and amplitude of an isoelectric segment 4, which is typical of a heartbeat with a physiological excitation pattern, i.e., a sinus beat.

[0041] This determines the baseline of electrocardiogram 1 beat by beat and makes it available for measuring other ECG segments, such as the ST segment, in real time. Furthermore, the determined baseline can be used in a subsequent procedural step for baseline correction or, as part of rhythm analysis, to separate sections of electrocardiogram 1 with sinus rhythm from other sections with other rhythm types.

[0042] Alternatively or additionally, if sinus rhythm is detected, isolated ventricular extrasystoles can also be further evaluated diagnostically in subsequent procedural steps, for example to calculate parameters of a so-called “heart rate turbulence” or sections with undisturbed sinus rhythm can be used to determine a so-called “heart rate variability” without distorting signal manipulation such as filtering, beat replacement or interpolation.

[0043] Unnecessary storage or telemetric transmission or evaluation of an undisturbed sinus rhythm can thus be suppressed and the charging-free operating time of an implanted device with Holter function or other wearable devices can be extended.

[0044] In further embodiments, an implanted cardiac pacemaker can also be controlled based on the described method, in particular to avoid unnecessary stimulation stimuli. The effectiveness of an antiarrhythmic therapy, in particular a cardioversion, can be monitored by the real-time detection of ECG segments with undisturbed sinus rhythm. This allows a cardiac situation to be correctly determined beat by beat, thus controlling therapeutic interventions by implanted devices with therapeutic functions, in particular with regard to cardiac function or unnecessary defibrillation shocks.

[0045] In further embodiments, the isoelectric point can also be searched for after the R wave using the described method, for example at another physiologically defined location of a cardiac revolution such as the ST segment or after a detected end of the T wave.

[0046] In Fig.Figure 4 shows a schematic view of a Holter ECG device 20, which is worn on the body by a user 21 and supplied with electrical energy by a battery, but in further embodiments can also be implanted. The Holter ECG device 20 enables ECG analysis under time-critical conditions, which is carried out using simple hardware in a space-, mass-, and energy-saving manner. The illustrated Holter ECG device 20 is integrated into the clothing of the user 21 and can operate autonomously for extended periods without an external power supply. The reliability of the biosignal analysis determines whether the signal is permanently stored, as well as the frequency and length of necessary radio transmissions to a base station. A recording unit 22, a computing unit 23, and an output unit 24 are arranged in the Holter ECG device 20 and are electrically connected to one another.

[0047] The recording unit 22 has a plurality of electrodes attached to the body of the user 21 and is used to record the electrocardiogram 1. The recorded data is evaluated by the computing unit 23, in the present case a microcontroller, using the described method, i.e. the isoelectric point is determined and output graphically or in text form on the output unit 24, in the illustrated embodiment a display, together with the amplitude at the isoelectric point determined from the electrocardiogram 1. The display can also be configured to show the entire recorded electrocardiogram 1. In further embodiments, a data transmission device for wireless or wired transmission of the data can also be provided on the long-term ECG device 20. This data transmission device can also be a connection for a memory card.The determined wavelet coefficients are stored and kept in a ring buffer 25 and can be loaded from there when required.

[0048] In further embodiments, the output unit 24 may also have only individual diodes instead of the display or in addition to the display.

[0049] Only features of the various embodiments disclosed in the exemplary embodiments can be combined with one another and claimed individually.

Claims

[1] Method for determining an isoelectric point in an electrocardiogram (1) comprising the steps: a) recording (6) an electrocardiogram (1); b) performing a wavelet transformation (7) of the electrocardiogram (1); c) determining a first wavelet coefficient and a second wavelet coefficient from the wavelet transform (7) which partially overlap each other, or using an already existing first wavelet coefficient and an already existing second wavelet coefficient which partially overlap each other; d) determining an R-wave (R) in the wavelet transform of the electrocardiogram (1) and selecting (11) a data range lying before the R-wave (R); e) checking (13) the selected data range of the wavelet transform for a value below a predetermined first threshold value (SWA) of a first characteristic variable which depends on the first wavelet coefficient; f) checking (13) the selected data range of the wavelet transform for a value below a predetermined second threshold value (SWB) of a second characteristic variable which depends on the second wavelet coefficient; g) wherein if the two threshold values ​​(SWA, SWB) are undershot in a data point of the selected data range, the determined data point is defined as an isoelectric point (17), wherein the method with steps a) to g) is repeatedly carried out and the first threshold value (SWA) and the second threshold value (SWB) are determined as a function of previous amplitude values ​​or previous values ​​of the wavelet coefficients. [2] Method according to claim 1, characterized bythat the first characteristic and / or the second characteristic is a sum of two wavelet coefficients. [3] Method according to claim 1 or claim 2, characterized by that the first characteristic is equal to an absolute value of the first determined wavelet coefficient and the second characteristic is equal to the sum of absolute values ​​of the first wavelet coefficient and the second wavelet coefficient. [4] Method according to one of the preceding claims, characterized by that the first wavelet coefficient is the wavelet coefficient of scale 3 and the second wavelet coefficient is the wavelet coefficient of scale 4. [5] Method according to one of the preceding claims, characterized by that the first threshold value (SWA) is at most one tenth of a total energy of the electrocardiogram and / or the second threshold value (SWB) is at most one fifth of the total energy of the electrocardiogram (1). [6] Method according to one of the preceding claims, characterized by that a corresponding amplitude for the determined isoelectric point is determined from the electrocardiogram (1) and preferably the amplitude and the isoelectric point are displayed on an output unit (24). [7] Device (20) which is arranged to carry out the method according to one of claims 1-7, with a recording unit (22) which is designed to record the electrocardiogram (1), a computing unit (23) which is designed to carry out the wavelet transformation and to determine the isoelectric point and an output unit (24) which is designed to output the data point of the isoelectric point. [8] Device (20) according to claim 7, characterized bythat the computing unit (23) is designed to store the determined first wavelet coefficient and the second wavelet coefficient in a ring memory (25) or to load them from the ring memory (25). [9] Computer program product containing a sequence of instructions stored on a machine-readable carrier, which is configured to carry out the method according to one of claims 1-6 and / or to control the device (20) according to claim 7 or claim 8 when the computer program product runs on a computing unit.

Images