WO2021096378A1

WO2021096378A1 - Conditioning, quality assessment, and change detection of ecg signals

Info

Publication number: WO2021096378A1
Application number: PCT/RS2019/000027
Authority: WO
Inventors: Srdjan VLASKALIC; Darko BOLJEVIC; Petar Belicev; Dejan VUKAJLOVIC; Lana POPOVIC MANESKI; Bojan RAJKOVIC
Original assignee: Vlaskalic Srdjan; Boljevic Darko
Priority date: 2019-11-14
Filing date: 2019-11-14
Publication date: 2021-05-20

Abstract

A method for the detection of acute myocardial infarction by comparing potentially ischemic ECG signal with a reference ECG signal consists of a procedure to control the quality of the recorded ECG signals and to select the best interval to be sent to the remote diagnostic center and a procedure for correction of the effects of the heart rhythm change of recorded ECG signals relative to reference ECG signals. The reference ECG signals of the patient are modified so that signal shapes correspond to approximately the same heart rate as the recorded ECG signals. Recorded ECG signals and modified reference signals are graphically presented so that the on- call physician can quickly and accurately detect the existence of an urgent cardiac condition, such as an acute myocardial infarction, and contact the patient, based on the difference between the current and the reference ECG, and take apropriate action.

Description

CONDITIONING, QUALITY ASSESSMENT, AND CHANGE DETECTION OF ECG SIGNALS

CROSS REFERENCE TO RELATED APPLICATIONS [0001] This patent application may be related to EU patent EP1659936 A1 (Bojovic et al. 2003, titled “APPARATUS AND METHOD FOR CORDLESS RECORDING AND TELECOMMUNICATION TRANSMISSION OF THREE SPECIAL ECG LEADS AND THEIR PROCESSING”), herein incorporated by reference in its entirety.

TECHNICAL FIELD

[0002] Described herein are methods and devices for diagnosis of cardiac conditions. In particular, described herein are handheld devices for diagnosis of cardiac conditions. For example, described herein are three-lead cardiac signal acquisition and processing devices configured to assess a patient’s risk of a serious condition such as AMI (Acute Myocardial Infarction, or heart attack) or cardiac ischemia (the underlying physiological process in AMI).

BACKGROUND ART

TECHNICAL PROBLEM

[0003] Acute Myocardial Infarction (AMI, heart attack) remains a leading cause of mortality in the developed world. Finding accurate and cost effective solutions for AMI diagnosis is vital. Survival of patients having AMI dramatically depends on treatment delay (symptom onset to medical treatment time) (De Luca, G. et al. Time delay to treatment and mortality in primary angioplasty for acute myocardial infarction. Circulation 2004;109:1223-5). It is very important that this most critical period be minimized (Lenfant C. et al.: Considerations for a national heart attack alert program, Clin. Cardiol. 1990 August; 13 (8 Suppl 8): VIII 9-11). In the AMI setting, the conventional 12-lead ECG is not only the most important piece of information, but it is also nearly as important as all other information combined (Goldman, L. et al. Triage of patients with acute chest pain and possible cardiac ischemia: the elusive search for diagnostic perfection. Ann Intern Med. 2003 Dec 16;139(12):987-95)

[0004] One of the main problems with using personal ECG devices with integrated electrodes for recording three or more ECG leads used for emergency cardiac diagnostics that the user can apply at any location to self-record his/her ECG and send it to a remote diagnostic center via a commercial telecommunications network is the quality of recorded signals. Specifically, when the patient is recording his or her own ECG by applying the device with the chest electrodes to his or her chest while touching the finger electrodes while holding the device, it is difficult to ensure the stability of the device during the recording, so that, as a rule, it produces a signal that is contaminated with signal interference that are a result of moving the device or changing the electrode pressure on the chest. This type of interference causes changes in the basic level of the signal, referred as BLW (Base Line Wander). When baseline changes are slow and mild, they do not introduce significant errors when reconstructing 12 standard ECG leads and do not interfere with physician’s diagnostic interpretation of ECG recordings.

However, when the baseline changes are fast and significant, then they introduce significant errors in the reconstruction of the 12 standard ECG leads that may affect the diagnostic interpretation of the ECG recordings.

[0005] The technical problem to be solved by the present invention is how to ensure that the signals recorded by the personal device for recording, processing and transmission of at least 3 ECG leads have good signal quality with a low BLW (Base Line Wander) interference level and are suitable for diagnostic interpretation, primarily in the detection of acute myocardial infarction (AMI). Eliminating the BLW interference usually comprises filtering, which may often include additional distortion of the signal. The other part of the problem is evaluating BLW content of the ECG signal for deciding if the signal is suitable for diagnostic interpretation. Yet another part of the problem is the detection of ECG signal changes between the ECG signal with suspected AMI and a prior ECG signal in the case when the two signals have shapes affected by different heart rate.

STATE OF THE ART

[0006] The technical problems described are common for the ECG technologies that enable self-recording of the ECG signal and remotely provide 12 standard ECG leads to physicians for diagnostic interpretation. This comprises technologies that use recording of special 3 ECG leads and reconstruction of the 12 standard ECG leads, like one that is described in the patent document EP1659936 A1 (Bojovic 2003), and use a recording device without cables, or Cardiosecur device (Personal Medsystems, Frankfurt, Germany) that uses 4 cables to record 3 leads. This also comprises technologies that use devices directly recording 12 standard ECG leads, such as Heartviewpl2 (Aerotel medical systems, Holon, Israel) or Smartheart (SHL Telemedicine, Tel Aviv, Israel).

[0007] One approach to reduce the problem of BLW interference in a personal ECG device is the ergonomic design of such device that is tailored to the morphology of the patient's body and provides mechanical stability and stable electrical contact with the user's skin.

[0008] Personal device with integrated electrodes for recording three ECG leads sent to a remote diagnostic center where the cardiologist is provided with 12 reconstructed ECG leads described in patent document EP1659936 A1 (Bojovic 2003) made with three integrated chest electrodes, two measuring and a common ground, which ensures the stability of the device by leaning the device against the chest at three points. In this embodiment, the chest electrodes must be high enough (about 8-10 mm) to avoid leaning the device body against the patient chest due to the morphology of the patient (curvature of the chest area), thereby compromising the stability of the device and the contact of the chest electrodes with the skin.

[0009] Patent Document US20100174204, Danteny describes a personal device that has a concave surface to adapt to patient morphology and provide good electrical contact with the chest electrodes.

[00010] Another approach is to use some BLW elimination procedure such as filtering signals using high-pass filters. Existing procedures are effective when the BLW is slow and mild, but with faster and more significant BLW they introduce artificial interference and modify the ST segment of the signal as shown in the publication (Buendia-Fuentes F, et al. High-Bandpass Filters in Electrocardiography: A Source of Error in the Interpretation of the ST Segment. ISRN Cardiol. 2012; 2012: 706217, 1-10), which is critical for the diagnosis of acute myocardial infarction. Artificially inserted ST segment changes in recorded leads are transmitted to 12 reconstructed standard ECG leads and may lead to errors in diagnostic interpretation of ECG recordings.

[00011] Therefore, there is a need for a method to perform a quantitative assessment of the quality of the signals recorded using a personal device and, on this basis, to send to the remote diagnostic center only signals of a quality enabling reliable processing without artificial interference and accurate reconstruction 12 standard ECG leads, while preventing the transmission of signals that have significant interference that, in the process of reconstruction of 12 standard ECG leads, can cause misdiagnosis.

[00012] The main diagnostic sign of AMI (Acute Myocardial Infarction) or ischemia (a major physiological phenomenon in AMI) is the elevation of the ST segment. However, there are other conditions having elevation of the ST segment that may mimic acute myocard ial infarction (Wang, K. et al. ST-segment elevation in conditions other than acute myocardial infarction. The New England journal of medicine 2003, 34922: 2128-35). This is strongly influencing the accuracy of physician’s interpretation of the ECG with suspected acute myocardial infarction (Bimbaum, Y. The burden of nonischemic ST-segment elevation. Journal of electrocardiology 2007, 40 1 :6-9). This problem may be improved by comparing the ECG with suspected AMI to a prior ECG recording (Fesmire, FM, et al. Diagnostic and prognostic importance of comparing the initial to the previous electrocardiogram in patients admitted for suspected acute myocardial infarction. South Med J. 1991 Jul;84(7):841-6; Lee, T, et al. Impact of the availability of a prior ECG on the triage of the patient with acute chest pain. J Gen Intern Med. 1990;5:381-388). The personal ECG system described in patent document EP1659936 Al (Bojovic 2003) may provide physicians with a previously recorded ECG signals (prior ECG) which may contribute to a more accurate interpretation of the ECG recording with suspected AMI. Namely, the calibration recording comprises a 12-lead ECG that can be used as prior ECG in the later use of the diagnostic system. However, the authors of the present invention have observed that presenting ECG readers with potentially ischemic and prior (non-ischemic) ECG recordings consecutively results in significantly improved specificity, but non-significant change in sensitivity. This is explained by the fact that when evaluating two consecutive ECG recordings, one potentially ischemic and one prior (non-ischemic), the readers are not able to notice small but significant changes in the ST segment. This suggests that simultaneously displaying potentially ischemic and a prior (non-ischemic) ECG recording by displaying signals one over the other would significantly increase sensitivity.

[00013] However, presenting signals one over the other has a problem because the cardiac cycles do not have the same duration due to the different heart rhythm of the potentially ischemic ECG recording and the prior non-ischemic ECG recording, which may lead to the inability to recognize small ST changes. Namely, the length of the heart beat signal changes when heart rate is changed. A clinical study (Kligfield, PD. et al. QTc behavior during treadmill exercise as a function of the underlying QT-heart rate relationship. Journal of Electrocardiology 1995, 28 Suppl, 206-10) showed that the QT interval (the interval from the onset of the QRS complex to the end of the T wave) varies from 377 milliseconds at rest, over 318 milliseconds at medium intensity exercise, to 266 milliseconds at peak exercise intensity. It is clear that lengths of QT intervals in the recorded ECG signal and in the prior ECG recording can be very different. Thus, the correction of ECG interval length differences due the heart rate differences in presenting signals one over the other would significantly contribute to diagnostic accuracy. A problem in such correction is that the change in length of different parts (complexes and waves) of the heart beat due to heart rate change is different. Specifically, the change in the JT interval (the interval from the end of the QRS complex to the end of the T wave) is the major contributor to the change in the QT interval while there are no changes in the duration of the QRS complex or are very small and difficult to detect (Spodick, D.H. Reduction of QT-interval imprecision and variance by measuring the JT interval. American Journal of Cardiology 1992, 70 1, 103). A solution to this problem would be a selective correction of the cardiac cycle intervals.

[00014] Therefore, there is a need for a graphical representation in which the measured ECG signals are displayed simultaneously with the prior ECG of the patient, which would contribute to a more efficient interpretation of the ECG recording as the physician may visually compare the ECG recording with the patient's prior ECG recording and more easily observe changes in the ECG recording relative to the prior ECG recording. In the graphical representation it is possible to display recorded ECG signals and reference (prior) ECG signals one over the other in two different colors or one above the other. In both cases, there is a need to synchronize the signal so that the QRS complexes of both signals are in the same place on the timeline. In the general case, such synchronization cannot be achieved because the recorded ECG signals and reference ECG signals do not have the same periodicity, that is, they have different heart rhythms and thus different duration of heart cycles. Therefore, there is a need to synchronize the reference ECG signal with the recorded ECG signal.

DISCLOSURE OF INVENTION [00015] The present invention is a method of quantitatively assessing the quality of recorded signals using a personal device for recording three or more ECG leads and, based on such quality assessment, to send to the remote diagnostic center only signals that allow reliable processing without artificial interference and accurate reconstruction 12 standard ECG leads and their graphical representation in comparison with the patient's prior ECG signals, which are synchronized in time with the recorded ECG signals, thus providing reliable diagnostic interpretation.

SIGNAL CONDITIONING

[00016] The process of ensuring good quality of the recorded ECG signals has two phases. In the first phase, three or more ECG leads are recorded using a personal device that is significantly longer than the signal that will later be sent to a remote diagnostic center. The length of the recorded signals TS can be 30 to 60 seconds as this does not significantly impair the convenience of using a personal ECG. Only a shorter interval T with the best quality of recorded signals is sent to the remote diagnostic center, which is usually about 5 seconds, which is sufficient to allow physicians a reliable ECG interpretation. The BLW content in the 5 seconds interval is approximated with a 5^th degree polynomial function and that polynomial function is subtracted from the 5 seconds interval to suppress the baseline wander content from this interval.

[00017] The selection of the best interval T for sending is made on the basis of a quantitative measure of the quality of the signal MK, which is calculated as the difference between the ECG signal and the polynomial approximation of the signal in a time window T that moves on the recorded signals of the length TS from their beginning to the end with a step dT. This procedure allows a shorter quality interval T to be selected from the interval TS of the recorded signal. This phase of the process is shown graphically in Figure 1.

QUALITY ASSESSMENT

[00018] The second phase of the procedure is a quantitative assessment of the quality of MK1 of the measured ECG signals at the selected interval T. MKR is a threshold of the quality measure MK1 of the measured ECG signals in the selected interval T. The decision whether the signals are good enough for diagnostic interpretation is based on MKR. The measure of the quality MKl is calculated as the mean value of the signal content in the frequency range 0.05 Hz to 1 Hz in a window of 3 seconds. When the MKl quality measure is above the MKR threshold, the measured ECG signals quality may allow reliable signal processing without artificial interference and accurate reconstruction of 12 standard ECG leads. These signals are sent to a remote diagnostic center where the reconstructed 12 standard ECG leads are provided to physicians for diagnostic interpretation. If the MKl quality measure is below the MKR threshold, this indicates that the measured ECG signals at the selected interval T have significant interferences and that reconstructed 12 standard ECG leads may not allow a reliable ECG interpretation. Therefore, the measurement is rejected as unusable for further processing and analysis, signals are not sent to the remote diagnostic center and the user is notified to repeat the measurement. This phase of the process is shown graphically in Figure 2.

[00019] The procedure is implemented on a smartphone with a user application that enables wireless communication with a personal ECG device via a Bluetooth connection and can activate, monitor and control the recording process.

[00020] In the present method, the implementation of this procedure achieves two objectives. The first objective is to select the interval with the best signal quality from the measured ECG signals. The second objective is to check the quality of the selected interval and to prevent further processing and analysis of ECG signals that cannot be interpreted diagnostically or may lead to misdiagnosis. This method improves the quality of the signals recorded by personal ECG devices and increases the reliability of interpretation of the recorded ECG signals.

DETECTION OF ECG SIGNAL CHANGES WITH CORRECTING THE EFFECTS OF THE HEART RHYTHM CHANGE

[00021] In addition to the quality of ECG signals, a very significant influence on the performance of physicians in the interpretation of the ECG recordings is the way of graphically presenting the recorded ECG signal and the prior ECG signal of the patient in order to detect ECG signal changes that are suggestive of AMI.

[00022] The method of solving the problem of correcting effects of the heart rhythm change of the recorded ECG signals relative to reference ECG signals is performed by stretching or shrinking (i.e. contracting or expanding) the time axis of the prior ECG signals in order to synchronize the heart cycle intervals with the recorded ECG signals without distorting the prior ECG signals morphology, especially in the ST segment (characteristic points and intervals on the ECG heart cycle are shown in Fig. 3) which is crucial for the detection of acute myocardial infarction. Therefore, the time intervals in which the QRS complexes are located (the intervals between Q and J points in the ECG recording) are first denoted. The denoted QJ intervals are not subject to the stretching or shrinking procedure of the time axis because the change in these intervals with the heart rhythm is negligible. The signal stretching/shrinking procedure applies to the remaining intervals of the reference ECG signals (the time intervals between the J point of the observed cardiac cycle and the Q point of the next cardiac cycle). The aforementioned intervals are shrinked (contracted) by time axis if the heart rate of the reference ECG recordings is less than the heart rate of the recorded ECG signals. When the heart rate of the reference ECG recordings is greater than the heart rate of the recorded ECG signals then the intervals mentioned are stretched (expanded). The percentage of shrinking or stretching of the time axis of the time intervals in the reference ECG signals is determined by comparing the lengths of the cardiac cycle intervals dependent on the heart rate in the recorded and reference ECG signals. In this way, the duration of the ST segment of the reference ECG signals is approximately the same as that of the recorded ECG signals. Modified reference ECG signals are graphically presented one over the other with the recorded ECG signals by synchronizing them to adjust the QRS complexes of the reference ECG signals at the same location on the time axis where the QRS complexes of the measured ECG signals are located. This is achieved by overlapping some characteristic points (for example, R points) on the reference and measured ECG signals, or by the autocorrelation method of the reference and measured ECG signals

[00023] The whole correction procedure can be applied to complete ECG signals containing multiple heart cycles or can be applied to representative heart cycles of reference and measured ECG signals obtained as medians of multiple individual heart cycles.

[00024] The solutions of all three parts of the technical problem described above, signal conditioning with BLW suppression, signal quality assessment and correcting effects of the heart rhythm change may be applied to multiple personal ECG technologies that remotely provide 12 standard ECG leads to physicians: both technologies that use reconstruction of the 12 standard ECG leads, like one described in the patent document EP1659936 A1 (Bojovic 2003) or Cardiosecur (Personal Medsystems, Frankfurt, Germany), and technologies that directly record 12 standard ECG leads, such as Heartview pl2 (Aerotel medical systems, Holon, Israel) or Smartheart (SHL Telemedicine, Tel Aviv, Israel).

Brief description of the drawings

[00025] The invention is described in detail in the embodiment shown in the draft in which:

[00026] FIG. 1 - Block diagram of the procedure for searching for a 5 second interval over recorded 30 second ECG signals in which the recorded signals are of the best quality.

[00027] FIG. 2 - Block diagram of a procedure that quantitatively determines the quality of recorded ECG signals over a selected 5-second interval and decides whether the signals are acceptable for further processing and analysis.

[00028] FIG. 3 - Characteristic points and intervals of the ECG signal

[00029] FIG. 4 - Block diagram of a procedure for correcting effects of the heart rhythm change

[00030] FIG. 5a - Application example of the method for correcting effects of the heart rhythm change: superimposed reference and recorded ECG signals before the application of the method

[00031] FIG. - 5b Application example of the method for correcting effects of the heart rhythm change: superimposed reference and recorded ECG signals after application of the method

BEST MODE FOR CARRYING OUT OF THE INVENTION A. Signal conditioning

[00032] In the case of performing a procedure to ensure good quality of recorded ECG signals measured using a personal device with integrated electrodes for recording three ECG leads, which are sent to a remote diagnostic center where 12 ECG leads are reconstructed, described in patent document EP 1659936 A1 (Bojovic 2003), the procedure is performed in two stages. The first phase is the procedure by which a 5-second interval is selected from the recorded 30-second ECG signals in having the best quality of the ECG signals. The process is carried out in the following steps, which are illustrated in Figure 1.

[00033] The input data are signals of 3 ECG leads, recorded by a three-channel ECG device with integrated electrodes in digital form, with a sampling frequency of 500 Hz. In a window of T = 5 seconds, the signals at each lead are approximated by the fifth degree polynomial to approximate the BLW content, and that polynomial function is subtracted from the 5 seconds interval to suppress the BLW content from this interval. Then, the quality measure MK is calculated as the sum of the absolute differences of the measured signal (£/) of the lead i and at time t_j and the approximate signals by the fifth degree polynomial VPi(t_j). When the sampling frequency of the signal is 500 Hz then the step on the time axis is 2 milliseconds, and in a window of 5 seconds there are 2500 points and MK is calculated by the formula:

2500.

[00034] In the next step, the window of T = 5 seconds is moved by 0.2 seconds and the value of the MK quality measure is again calculated. The procedure is repeated until the end of the signal is reached, and the interval with the lowest MK value is taken as the best quality interval. As a result of this procedure, a minimum value of the quality measure MK0 is obtained and the beginning of the interval TpO of 5 seconds is calculated. In this way, the criterion for selecting the best quality interval is the part of the BLW remaining after “flattening” by subtracting the polynomial. The presented procedure is illustrated in the diagram in Figure 1.

B. Quality assessment

[00035] The next phase is to evaluate the quality of the selected 5 seconds, based on which it is decided whether the signals are of good enough quality to allow reliable signal processing without artificial interference, and still allow accurate reconstruction of 12 standard ECG leads. In this phase, we introduce the measure of the quality MK1, the mean value of the signal content in the frequency range 0.05 Hz to 1 Hz in a window of 3 seconds. [00036] Specifically, high-pass filtering with a cutoff frequency of 1 Hz is known to result in the ECG signal with the almost completely eliminated BLW effect. However, such a filtered signal cannot be used to detect ischemia since it introduces significant ST segment distortion.

[00037] When the sampling frequency of the signal is 500 Hz then the step on the time axis is 2 milliseconds and in the window of 3 seconds there are 1500 points, so MK1 is calculated by the formula: i = 1,3; j = 1, 1500

[00038] - is a signal of 3 seconds in length of the lead i and time instant t_j, which represents the ECG signal filtered in a conventional manner before displaying or printing.

[00039] - is a signal of 3 seconds in length of the lead i and time instant t_j, filtered by a high-pass filter with 1 Hz cut-off, representing an ECG signal approximately without BLW interference.

[00040] In this way, the difference between these two signals, on which the calculation of the MK1 quality measure is based, is an approximation of the BLW signal size.

[00041] Alternatively, a filtered signal with a band-pass filter using 0.05 Hz AND 1 Hz cutoff frequencies may be introduced as a measure of quality.

[00042] The 3-second window moves along the 5-second signal in 0.2-second increments, using the minimum MK1 value as a measure of quality. The calculated value of MK1 is compared with the defined MKR threshold. If the MK1 is less than the threshold, the MKR signals go into further processing and are used to reconstruct the standard 12 ECG leads. When MK1 is greater than MKR, the user is informed that the signals are not acceptable for use for diagnostic purposes due to poor quality. The process presented is illustrated by the diagram in Figure 2. [00043] The 5-second signal length displayed is chosen since it is one of the usual signal intervals displayed with conventional ECGs. The essence of the invention will not change by choosing another display interval. Also, the 3-second signal length used to evaluate signal quality was chosen given that physicians intuitively use a series of 3 consecutive heart beats to evaluate the ST segment, which most often coincides with a 3-second interval. The essence of the invention will not change by choosing another interval to display an estimate of signal quality.

[00044] In the example shown, the threshold MKR = 0.1 mV is used and the sampling frequency of the signal is 500 Hz and the essence of the invention will not change if some other values are used.

C. Detection of ECG signal changes with correcting effects of the heart rhythm change

[00045] In the case of performing a procedure that solves the problem of correcting effects of the heart rhythm change of recorded ECG signals relative to reference ECG signals on a personal device with integrated electrodes for recording three ECG leads sent to a remote diagnostic center where 12 reconstructed ECG leads described in patent document EP1659936 A1 (Bojovic 2003) are disclosed to the cardiologist, the procedure is performed in five steps.

[00046] In the first step, corresponding representative signals over a single cardiac cycle are generated from the reference and recorded ECG signals for 5 seconds. A representative cardiac cycle is obtained by dividing the signals into segments of the median length of the RR interval of the selected signal for 5 seconds, where the beginning of each segment at points is Q - 20 milliseconds. Thus, the duration of representative cardiac cycles of the reference signals TR0 and of the measured signals TR1 are obtained, respectively. Hence TR0 is the mean value of the RR interval of the reference ECG signal and TR1 is the mean value of the RR interval of the recorded ECG signal. The voltage value is obtained as the median voltage value at each point on the time axis of the individual segments, when the signals on the segments are aligned to match the R points. The R and Q points in the ECG signals can be marked by the physician or can be automatically determined using one of the existing methods such as Pan-Tompkins (Pan, J. et al. A Real-Time QRS Detection Algorithm. IEEE Transactions on Biomedical Engineering 1985, BME-32: 230-236) or similar (Arzeno, NM. et al. Analysis of First-Derivative Based QRS Detection Algorithms. IEEE Transactions on Biomedical Engineering 2008, 5:478-484; Lyon, A. et al. Computational techniques for ECG analysis and interpretation in light of their contribution to medical advances. Journal of the Royal Society 2018, Interface, 15 138), and can then be manually corrected by the physician. Segment alignment can be accomplished by using other characteristic signal points such as a Q point, or alternatively by performing a segment autocorrelation method that ensures that the QRS segment complexes are set to be in the same place on the time axis.

[00047] In this way, ECG signals of two representative cardiac cycles are obtained. One set is median signals [VM0 (i), i = 1,3] for a representative cardiac cycle of reference ECG signals with duration TR0 and the other set is median signals [VM1 (i), i = 1,3] for measured ECG signals with duration TR1.

[00048] The second step is to determine the QJ0 duration interval of the QRS complex in the reference set of representative signals VM0 (i), and the Q J 1 interval of the duration of the QRS complex in the measured set of representative signalsVMl (i). Points Q and J are indicated by the physician or can be automatically determined using one of the existing methods, and can then be manually corrected by the physician.

[00049] The third step is the linear interpolation or extrapolation of the representative ECG reference signals VM0 (i) over the duration interval TR0-QJ0, which starts at point J, by stretching or shrinking this interval using the scaling factor K = (TR1-QJ1) / ( TR0-QJ0). In this way, representative ECG reference signals VM0 (i) in the invariant interval QJ0 is merged with the scaled signals of the interval TR0-QJ0 on the interval TR1-QJ1 and a modified representative set of reference ECG signals VMM0 (i) having approximately the same ST segment length as the ST segment in the measured set of representative signalsVMO (i) is obtained. The interval at which the aforementioned procedure applies may be any interval longer than the interval between points J and Tb (the beginning of the T wave, Figure 3), such as the intervals between points J and Tmax or between points J and Tend. The interval scaling (shrinking / stretching) factor can also be calculated using any intervals longer than the intervals between points J and Tb. For example, the coefficient K can be calculated as K = (QTmaxl-QJl) / (QTmaxO-QJO) or as K = (QTendl-QJl) / (QTendO-QJO). Also interpolation or extrapolation can be nonlinear such that the parts of the interval have different scaling factors K.

[00050] The fourth step is the reconstruction of standard 12 ECG leads for reference ECG signals based on the VMMO (i) signal, and for the measured ECG signals based on the VM1 (i) signal, according to the procedure described in patent document EP 1659936 A1 (Bojovic 2003). In this way, the reconstruction produces the standard 12 ECG leads of the modified reference ECG signals [VS0 (i), i = 1,12] and the measured ECG signals [VS1 (i), i = 1,12]. The present embodiment is applicable to the measurement and 12 ECG leads reconstruction technology described in patent document EP1659936 A1 (Bojovic 2003), and also to other similar technologies recording at least 3 ECG leads and reconstructing 12 ECG leads, like Cardiosecur (Personal Medsystems, Frankfurt, Germany).

[00051] In another embodiment, this reconstruction step may be performed prior to said third step, and then step three is subsequently applied to the reconstructed standard 12 ECG leads.

[00052] In another embodiment, all steps of the presented method may be applied directly to the 12 lead ECG. This embodiment is applicable to the technologies with 12 ECG leads reconstruction, but also to technologies directly recording 12 standard ECG leads, such as Heartview pl2 (Aerotel medical systems, Holon, Israel) or Smartheart (SHL Telemedicine, Tel Aviv, Israel).

[00053] The fifth step is a comparative graphical representation of the received signals VS0 (i) and VS1 (i). In the described embodiment, the reference signals VS0 (i) and the measured signals VS1 (i) are displayed one over the other in two different colors by matching the position of the QRS complex and vertically matching the level of the PQ segment. In addition, there is an option that the physician can interactively, if desired, move the measured signals VS1 (i) relative to the reference signals VS0 (i). In this way, the overlapping of the QRS complex and the vertical displacement of the measured VS1 (i) signals relative to the reference signals VS0 (i) can be fine-tuned. This graphical display and interactive signal shift capability allows the physician to easily detect any changes that occurred on the ST segment of the measured ECG signals relative to the reference ECG signals, which speeds up the interpretation of the recordings and increases the accuracy of detection of acute myocardial infarction. [00054] Fig. 5 shows an example of the application of the method for correcting effects of the heart rhythm change: superposed reference and recorded ECG signal before (Fig. 5 a) and after (Fig. 5 b) application of the method. It can be seen that before the correction is applied, the QRS complexes of the transmitted and reference signals are correctly positioned, while the ST segment and T wave are significantly shifted along the time axis. After the correction is applied, all segments of the heartbeat are correctly positioned.

[00055] In the example described, the overall procedure is implemented in the form of software on a smartphone, but alternatively it can be implemented on a remote server with which the patient smartphone communicates.

INDUSTRIAL APPLICABILITY

[00056] The method for the detection of acute myocardial infarction by comparison with a reference ECG signal, intended for use with mobile personal ECG devices with integrated electrodes for recording, processing and transmission of three ECG leads, can be used for emergency cardiac diagnosis by the user, at any location, to self-record his/her three-lead ECG and send it to a remote diagnostic center via a commercial telecommunications network, where the on-call physician is given an insight into the reconstruction of standard 12 ECG leads. The measured ECG is then displayed one over the other with the patient's reference ECG, so that the on-call physician can quickly and accurately detect the existence of an urgent cardiac condition, such as an acute myocardial infarction, contact the patient and take appropriate action. The above procedure can also be applied to similar diagnostic systems based on personal ECG devices that can provide physicians with a 12-channel ECG signal, such as Heartview pl2 (Aerotel medical systems, Holon, Israel), Smartheart (SHL Telemedicine, Tel Aviv, Israel) and Cardiosecur (Personal Medsystems, Frankfurt, Germany)

Claims

CLAIMS:

1. A method for ECG signal conditioning, the method comprising: acquiring a first interval of the ECG signal; selecting a second interval of the ECG signal shorter than the first interval; approximating the baseline wander signal of the second interval by interpolating a polynomial function; suppress the baseline wander signal from the second interval by subtracting the polynomial function; selecting the best second interval within the first interval having a minimal difference between the ECG signal and the polynomial function.

2. The method according to claim 1, where the first interval of the ECG signal is 30 seconds.

3. The method according to claim 1, where the second interval of the ECG signal is 5 seconds.

4. The method according to claim 1, where the polynomial function is a 5^th order polynomial.

5. A method for assessing the quality of the ECG signal, the method comprising: selecting a first interval of the ECG signal; selecting a second interval of the ECG signal shorter than the first interval; obtaining a first filtered signal by applying a high-pass filter to the second interval of the ECG signal using a first cut-off frequency; obtaining a second filtered signal by applying a high-pass filter to the second interval of the ECG signal using a second cut-off frequency; obtaining the approximate baseline wander signal of second interval of the ECG signal by subtracting second filtered signal from the first filtered signal; determining the baseline wander content of the second interval by calculating the average value of the baseline wander signal in the second interval; determining the quality of the best second interval using the minimal baseline wander content. assessing the quality of the first interval using the quality of the best second interval within the first interval.

6. The method according to claim 5, where the first interval of the ECG signal is 5 seconds.

7. The method according to claim 5, where the second interval of the ECG signal is 3 seconds.

8. The method according to claim 5, where the first cut-off frequency is 0.05 Hz.

9. The method according to claim 5, where the second cut-off frequency is 1 Hz.

10. A method for detection of ECG signal changes, the method comprising: acquiring a first set of at least three ECG leads from a patient's chest and hands at a first time; acquiring a second set of at least three orthogonal ECG leads from the patient's chest and hands at a second time; performing a beat alignment in a processor of the first and second sets of at least three orthogonal leads to synchronize representative beats from the first and second sets of at least three orthogonal leads; compensating for heart rate difference between the first and second set by expanding or contracting the variable portion dependent on the heart rate of the representative heartbeat of the first set using a scaling factor; graphically presenting the compensated representative heartbeat of the first set simultaneously to the representative heartbeat of the second set.

11. The method according to claim 10, where the variable portion of the heart beat is the J-Tend interval.

12. The method according to claim 10, where the variable portion of the heart beat is the J-Q interval.

13. The method according to claim 10, where the scaling factor is determined by comparing the lengths of the J-Tend intervals of the representative beats of first and second sets.

14. The method according to claim 10, where the scaling factor is determined by comparing the lengths according to the J-Q intervals of the representative heart beats of first and second sets.

15. The method according to claim 10, where the scaling factor is determined by comparing the lengths of the J-Tmax intervals of the representative beats of first and second sets.

16. The method according to claim 10, where the scaling factor is determined by maximizing the correlation of the shapes of the variable portions of the representative heart beats of first and second sets.