Multi-lead low-sampling-rate pacing pulse detection method

Abstract

The invention discloses a multi-lead low-sampling-rate pacing pulse detection method. The method comprises the following steps: step 1, carrying out filtering processing on electrocardiosignals; 2, signal absolute values are superposed; step 3, performing low-pass filtering; step 4, potential pace-making pulse detection; and step 5, pacing pulse screening. According to the invention, low-memory and low-power-consumption pace-making pulse detection can be realized on the premise of ensuring the identification precision.

Description

Technical Field

[0001] The present invention relates to the technical field of electrocardiogram signal detection, and particularly to a multi-lead low-sampling-rate pacing pulse detection method. Background Art

[0002] Cardiac pacemakers can effectively treat diseases including bradycardia, heart failure, atrioventricular block, and atrial fibrillation, and are widely used clinically. With the improvement of the economic level and the enhancement of people's health awareness, the number of patients implanted with pacemakers is increasing. After a pacemaker is implanted in the human body for a period of time, follow-up is required to determine whether its working state meets the patient's needs. Currently, the detection and evaluation of the pacemaker state in clinical practice are mainly analyzed through electrocardiogram (ECG).

[0003] According to the "American National Standards Institute / Association for the Advancement of Medical Instrumentation Diagnostic Electrocardiogram Equipment Standard" (ANSI / AAMI EC11), the detected pacemaker pulse duration must meet 0.1 ms to 2 ms. The "Medical Electrical Equipment - Part 2-27: Particular Requirements for the Basic Safety and Essential Performance of Electrocardiogram Monitoring Equipment" (IEC 60601-2-27) requires that the duration of the pacing pulse is 0.5 ms to 2.0 ms. Since the typical pacing pulse duration is between microseconds and a few milliseconds, a Nyquist frequency of at least 4 kHz is required to reliably capture the high-frequency content of the pacing pulse. Traditional pacing detection methods include both hardware and software. The hardware detection scheme is to add a dedicated pacing detection circuit, which increases the additional hardware cost, and due to the limitations of the detection principle, the accuracy of pacing pulse detection is relatively low. The software detection scheme is to perform detection based on the morphology of the pacing pulse at a high sampling rate. This high-sampling-rate detection method has a relatively high requirement for memory, and at the same time, it will also increase power consumption and affect the battery life of the device. Therefore, a high-precision and low-power detection method is needed to be applicable to long-term operation of portable and wearable electrocardiogram devices to monitor the function of the pacemaker. Summary of the Invention

[0004] To solve the above-mentioned deficiencies of the prior art, the present invention provides a multi-lead low-sampling-rate pacing pulse detection method, which can identify pacing pulses with extremely few resources while ensuring accuracy.

[0005] The technical solution adopted by the present invention to solve its technical problems is:

[0006] A multi-lead low-sampling-rate pacing pulse detection method includes the following steps:

[0007] 1) Signal filtering process: Perform high-pass filtering on the original electrocardiogram signals of each lead to filter out power frequency interference, QRS wave, P wave, T wave electrocardiogram components, and low-frequency noise, and retain the pacing signal;

[0008] 2) Signal absolute value and superposition: Take the absolute value of the signals filtered by each channel above, and then superpose each lead to further amplify the pacing pulse;

[0009] 3) Low-pass filtering: Perform multi-stage FIR low-pass filtering on the superposed signal to filter out high-frequency noise;

[0010] 4) Potential pacing pulse detection: According to the electrocardiogram signal after preprocessing in 3), find the positive peak points; according to the detected positive peak points, determine the threshold; according to the threshold, find the peak points exceeding the threshold as the potential pacing pulse position sequence, and record paceCandi[k] as the k-th pacing pulse position, k = 1, 2,..., P;

[0011] 5) Pacing pulse screening: Screen the pacing pulses detected in 4), and judge whether there are misdetected pacing pulses and missed detected pacing pulses according to the amplitude. If so, search for the missed pacing pulses and remove the misdetected pacing pulses; confirm whether there are missed low-amplitude pacing pulses. If so, search for the missed pacing pulses according to the backward search algorithm.

[0012] Furthermore, in 1), perform high-order FIR high-pass filtering on the original electrocardiogram signals of each lead to filter out power frequency interference, QRS waves, P waves, T wave electrocardiogram components and low-frequency noise, and retain the pacing signal.

[0013] In 2), the absolute value superposition of the electrocardiogram signal after high-pass filtering includes:

[0014] y(n) = |y 1 (n)| + |y 2 (n - 1)| +... + |y N (n)|

[0015] where y(n) is the superposed signal and N is the number of leads.

[0016] Still further, in 4), according to the positive peak points, determine the threshold Threshold, and the process is as follows: Taking the detected positive peak points as the center and W as the window length, calculate the area of the signal in this window area;

[0017]

[0018] where K represents the set coefficient.

[0019] Further, in the above 4), peak points exceeding the threshold are found according to the threshold as a potential pacing pulse position sequence, and the process is as follows: Starting from the i-th peak point, with the current detection point p(i) as the center, calculate the signal area areaVal within the region [i - 0.25W, i + 0.25W]. Compare this area areaVal with the threshold Threshold. If it exceeds half of the threshold, mark this peak point as the potential pacing pulse position paceCandi[k], where k is the k-th potential pacing pulse detected. Continue to detect the next peak point until all peak points are detected.

[0020] Further, in the above 5), it is determined whether there are missed or mis-identified pacing pulses between two adjacent pacing pulses according to the amplitude, and the process is as follows: Search for the maximum peak point between two adjacent pacing pulses p[k] and p[k + 1]. If its amplitude exceeds K1 times the average amplitude of the two pacing pulses p[k] and p[k + 1], and the distance from both of these two pacing pulses is at least T1, then mark this peak point as a valid pacing pulse; if the amplitude of the pacing pulse p[k + 1] is less than K1 times the average of the adjacent two pacing pulses p[k] and p[k + 2], then this pacing pulse is a false positive pacing pulse and this pacing pulse is removed. Repeat this screening process T times.

[0021] Further, in the above 5), it is confirmed whether there are missed low-amplitude pacing pulses. If so, search for the missed pacing pulses according to the backward search algorithm, and the process is as follows: Taking the identified pacing pulses as the center, calculate the set average eaECG of the data with a length of T3. Calculate the intervals between all adjacent pacing pulses, find the pacing pulse intervals less than the period T2 and calculate the average inter-pulse interval avgPP; for each pacing pulse interval [p[k], p[k + 1]] greater than the period T2, search for the first K2 maximum peak points between these two pacing pulses as candidate pacing pulses, where K2 = (p[k + 1] - p[k]) / avgPP. Taking the candidate pacing pulses as the center, take the data cdECG with a length of T3 and perform cross-correlation analysis with the set average eaECG. If the correlation coefficient Corcoef is greater than 0.5, regard the candidate pacing pulse as a valid pacing pulse, as:

[0022]

[0023] where P is the number of identified pacing pulses, and p(k) is the position of the k-th pacing pulse;

[0024]

[0025] where m(cdECG) is the average value of cdECG, and m(eaECG) is the average value of eaECG.

[0026] The present invention filters out the electrocardiogram components and low-frequency noise through high-pass filtering, retains the pacing pulses, superimposes the absolute values of each lead, and detects the pacing pulses through dynamic threshold, and finally the position of the pacing pulses can be detected, improving the accuracy of pacing pulse detection.

[0027] The beneficial effect of the present invention is that this method enables the detection of pacing pulses under low sampling rates. Description of the Drawings

[0028] Figure 1 is a flow chart of the pacing pulse detection method of the present invention;

[0029] Figure 2 is a waveform diagram after high-pass filtering;

[0030] Figure 3 is a waveform diagram after superimposing the absolute values of each lead;

[0031] Figure 4 is a waveform diagram after low-pass filtering;

[0032] Figure 5 is a schematic diagram of the pacing pulse detection result. Detailed Embodiments

[0033] The present invention will be further described below with reference to the drawings.

[0034] Referring to Figures 1 to 5 , a multi-lead low-sampling-rate pacing pulse detection method, through signal filtering processing, signal absolute value superimposition, low-pass filtering, potential pacing pulse detection, and pacing pulse screening, the position of the pacing pulses can finally be obtained, and the detection of pacing pulses under low sampling rates is realized.

[0035] As Figure 1 shown, a multi-lead low-sampling-rate pacing pulse detection method provided by the present invention includes the following steps: 1) Signal filtering processing; 2) Signal absolute value and superimposition; 3) Low-pass filtering; 4) Potential pacing pulse detection; 5) Pacing pulse screening.

[0036] In the signal filtering processing step of 1), as Figure 2 shown, the original electrocardiogram signals of each lead are subjected to high-pass filtering to filter out electrocardiogram components such as power frequency interference, QRS waves, P waves, T waves, and low-frequency noise, and retain the pacing signals. In this embodiment, a 55 Hz second-order FIR high-pass filter is used to filter out noise signals such as baseline drift, power frequency interference, normal electrocardiogram signal components, and low-frequency interference;

[0037] In the signal absolute value and superimposition step of 2), as Figure 3As shown, take the absolute value of the signals filtered by the above-mentioned leads, and then superimpose the leads to further amplify the pacing pulses, where the number of leads numLead satisfies 3 ≤ numLead ≤ 12;

[0038] In the low-pass filtering step of 3) above, as Figure 4 shown, perform multi-stage (second-order and above) FIR low-pass filtering on the superimposed signal to filter out high-frequency noise. In this embodiment, use a 5Hz second-order FIR low-pass filter;

[0039] In the pacing pulse detection step of 4), according to the electrocardiogram signal after preprocessing in 3), find the positive peak point. Taking the detected positive peak point as the center and W as the window length, calculate the area of the signal within this window region to determine the threshold Threshold:

[0040]

[0041] where K represents a set coefficient. Starting from the i-th peak point, taking the current detection point p(i) as the center, calculate the signal area areaVal within the region [i - 0.25W, i + 0.25W]. Compare this area areaVal with the threshold Threshold. If it exceeds half of the threshold, mark this peak point as the potential pacing pulse position paceCandi[k], where k is the k-th potential pacing pulse detected. Continue to detect the next peak point until all peak points are detected;

[0042] In the pacing pulse screening step of 5) above, search for the maximum peak point located between two adjacent pacing pulses p[k] and p[k+1]. If its amplitude exceeds K1 times the average amplitude of the two pacing pulses p[k] and p[k+1], and the distance from both of these two pacing pulses is at least T1, then mark this peak point as a valid pacing pulse; if the amplitude of the pacing pulse p[k+1] is less than K1 times the average of the two adjacent pacing pulses p[k] and p[k+2], then this pacing pulse is a false positive pacing pulse and this pacing pulse is removed. Repeat this screening process T times to obtain a pacing pulse position sequence. Centered on the identified pacing pulses, calculate the set average value eaECG of the data with a length of T2; calculate the intervals between all adjacent pacing pulses, find the pacing pulse intervals less than the period T3 and calculate the average interval avgPP. For each pacing pulse interval [p[k], p[k+1]] greater than the period T3, find the first K2 maximum peak points between these two pacing pulses as candidate pacing pulses, where K2 = (p[k+1] - p[k]) / avgPP. Centered on the candidate pacing pulses, take the data cdECG with a length of T2 and perform cross-correlation analysis with the set average value eaECG. If the correlation coefficient Corcoef is greater than 0.5, regard the candidate pacing pulse as a valid pacing pulse, which is:

[0043]

[0044] where P is the number of identified pacing pulses, and p(k) is the position of the k-th pacing pulse;

[0045]

[0046] where m(cdECG) is the average value of cdECG, and m(eaECG) is the average value of eaECG. As Figure 5 shown, the effect of the present invention on detecting the position of a segment of pacing pulses is demonstrated.

[0047] In the above-described embodiments, the technical solutions of the present invention have been described in detail. The present invention can effectively detect the position of pacing pulses at a low sampling rate.

[0048] The content described in the embodiments of this specification is only an enumeration of the implementation forms of the inventive concept of the present invention. The protection scope of the present invention should not be regarded as limited to the specific forms stated in this embodiment. The protection scope of the present invention also extends to equivalent technical means that can be thought of by those of ordinary skill in the art according to the inventive concept of the present invention.

Claims

1. A multi-lead low sampling rate pacing pulse detection method, characterized in that: The method comprises the following steps: 1) Signal filtering: Perform high-pass filtering on the original ECG signals of each lead to remove power frequency interference, QRS wave, P wave, T wave ECG components and low-frequency noise, and retain the pacing signal; 2) Signal absolute value and superposition: Take the absolute value of the filtered signals of each channel, and then superimpose the leads to further amplify the pacing pulse; 3) Low-pass filtering: The superimposed signal is subjected to multi-order FIR low-pass filtering to filter out high-frequency noise; 4) Potential pacing pulse detection: according to the ECG signal after preprocessing in 3), find the positive peak point; according to the detected positive peak point, determine the threshold; according to the threshold, find the peak point exceeding the threshold as the potential pacing pulse position sequence, and record paceCandi[k] as the kth pacing pulse position, k=1,2,…,P; 5) Pacing pulse screening: Screen the pacing pulses detected in 4) and determine whether there are falsely detected pacing pulses and missed pacing pulses based on the amplitude. If so, search for missed pacing pulses and remove falsely detected pacing pulses; confirm whether there are missed low-amplitude pacing pulses. If so, search for missed pacing pulses based on the backward search algorithm.

2. The multi-lead low sampling rate pacing pulse detection method according to claim 1, characterized in that: In the above 1), the original ECG signal of each lead is subjected to high-order FIR high-pass filtering to filter out power frequency interference, QRS wave, P wave, T wave ECG components and low-frequency noise, and retain the pacing signal.

3. The multi-lead low sampling rate pacing pulse detection method as claimed in claim 1, characterized in that: In the above 2), performing absolute value superposition on the ECG signal after high-pass filtering includes: y(n)=|y1(n)|+|y2(n-1)|+...+|y N (n)| Among them, y(n) is the superimposed signal, and N is the number of leads.

4. The multi-lead low sampling rate pacing pulse detection method according to claim 1 or 2, characterized in that: In the above 4), according to the positive peak point, the threshold Threshold is determined as follows: taking the detected positive peak point as the center and W as the window length, the area of ​​the signal in the window area is calculated; Wherein, K represents the set coefficient.

5. The multi-lead low sampling rate pacing pulse detection method according to claim 4, characterized in that: In the above 4), the peak points exceeding the threshold are searched as potential pacing pulse position sequences according to the threshold, and the process is as follows: starting from the i-th peak point, with the current detection point p(i) as the center, the signal area areaVal in the region [i-0.25W, i+0.25W] is calculated, and the area areaVal is compared with the threshold Threshold. If it exceeds half of the threshold, the peak point is recorded as the potential pacing pulse position paceCandi[k], where k is the k-th potential pacing pulse detected; continue to detect the next peak point until all peak points are detected.

6. The multi-lead low sampling rate pacing pulse detection method according to claim 4, characterized in that: In the above 5), it is determined whether there is missed recognition or misrecognition between two adjacent pacing pulses based on the amplitude: the maximum peak point between two adjacent pacing pulses p[k] and p[k+1] is searched. If its amplitude exceeds K1 times the average amplitude of the two pacing pulses p[k] and p[k+1], and is at least T1 away from the two pacing pulses, then this peak point is marked as a valid pacing pulse. If the amplitude of the pacing pulse p[k+1] is less than K1 times the average amplitude of the two adjacent pacing pulses p[k] and p[k+2], then the pacing pulse is a false positive pacing pulse and is removed; repeat this screening process T times.

7. The multi-lead low sampling rate pacing pulse detection method according to claim 6, characterized in that: In the above 5), the missed pacing pulses are searched according to the backward search algorithm, and the process is as follows: with the identified pacing pulse as the center, the collective average eaECG of the data with a length of T3 is calculated, the intervals between all adjacent pacing pulses are calculated, the pacing pulse intervals less than the period T2 are found and the average interval avgPP is calculated; for each pacing pulse interval [p[k], p[k+1]] greater than the period T2, the first K2 maximum peak points between the two pacing pulses are found as candidate pacing pulses, where K2 = (p[k+1]-p[k]) / avgPP; with the candidate pacing pulse as the center, the data cdECG with a length of T3 is taken, and a cross-correlation analysis is performed between it and the collective average eaECG. If the correlation coefficient Corcoef is greater than 0.5, the candidate pacing pulse is regarded as a valid pacing pulse. Where P is the number of identified pacing pulses, p(k) is the kth pacing pulse position; Among them, m(cdECG) is the average value of cdECG, and m(eaECG) is the average value of eaECG.

Images