Method for extracting electrocardiogram waveform based on electrocardiogram

Abstract

The application discloses a method for extracting electrocardiogram waveform based on electrocardiogram, S1, acquiring electrocardiogram, correcting the electrocardiogram, and obtaining a front view image of the electrocardiogram; S2, acquiring a target selected region image selected by a user based on the front view image for extracting electrocardiogram waveform; S3, performing binaryzation processing on the target selected region image to obtain a target selected region binaryzation image; and S4, extracting electrocardiogram waveform feature points in the target selected region binaryzation image to construct electrocardiogram waveform.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of medical simulation, in particular to a method for extracting electrocardiogram waveform based on electrocardiogram. BACKGROUND

[0002] Electrocardiogram is a technique for recording the electrical activity of the heart during each cardiac cycle on the body surface, and is one of the most commonly used clinical examinations, widely used, and identifying electrocardiogram waveform to diagnose illness is a skill that medical personnel need to master. In the existing teaching system, teaching is usually based on clinical electrocardiogram cases, but there are limited clinical electrocardiogram cases and teaching cases, so the present application aims to provide a method for autonomously extracting electrocardiogram waveform based on clinical electrocardiogram to simulate different electrocardiogram cases. SUMMARY

[0003] In order to achieve the above-mentioned purpose, the present application provides a method for extracting electrocardiogram waveform based on electrocardiogram, comprising the steps of:

[0004] S1, acquiring an electrocardiogram, correcting the electrocardiogram to obtain a front view image of the electrocardiogram;

[0005] S2, acquiring a target selection image selected by a user based on the front view image for extracting electrocardiogram waveform;

[0006] S3, performing binaryzation processing on the target selection image to obtain a target selection binaryzation image, comprising:

[0007] S31, performing gray scale processing on the target selection image to obtain a target selection gray scale image;

[0008] S32, setting a binaryzation threshold, performing binaryzation processing on the target selection gray scale image to obtain a target selection binaryzation image;

[0009] S4, extracting electrocardiogram feature points in the target selection binaryzation image to construct an electrocardiogram waveform, comprising:

[0010] S41, taking a pixel point in the target selection binaryzation image as an electrocardiogram sampling point, acquiring a coordinate of the electrocardiogram sampling point;

[0011] S42, based on the coordinate of the electrocardiogram sampling point, screening the electrocardiogram sampling point, and marking electrocardiogram feature points, comprising:

[0012] If there is no electrocardiogram sampling point on the current coordinate axis, marking the electrocardiogram feature point of the current coordinate axis as an abnormal electrocardiogram feature point;

[0013] If there is one ECG waveform sample point on the current coordinate axis, mark the current ECG waveform sample point as the valid ECG waveform feature point of the current coordinate axis;

[0014] If there are two or more ECG waveform sample points on the current coordinate axis, perform collection processing on the continuous ECG waveform sample points, take the center coordinate point of each collection as the center ECG waveform sample point of each collection, calculate the weight of the center ECG waveform sample point of each collection respectively, and select the center ECG waveform sample point with the highest weight as the valid ECG waveform feature point of the current coordinate axis;

[0015] The weight calculation expression is:

[0016] ;

[0017] Wherein, W is the weight, Num is the number of ECG waveform sample points contained in the current collection, and diff is the vertical coordinate difference between the center ECG waveform sample point of the current collection and the valid ECG waveform feature point of the previous coordinate axis.

[0018] S43, extract the valid ECG waveform feature point, and complete the ECG waveform construction.

[0019] Further, in step S42, if there are two or more ECG waveform sample points on the current coordinate axis, before taking the center coordinate point of each collection as the center ECG waveform sample point of each collection after performing collection processing on the continuous ECG waveform sample points, it further includes the following steps: setting a height threshold, judging whether the height of each collection has a value greater than the height threshold, if yes, reducing the binary threshold, and returning to step S32.

[0020] Further, in step S42, if there are two or more ECG waveform sample points on the current coordinate axis, before taking the center coordinate point of each collection as the center ECG waveform sample point of each collection after performing collection processing on the continuous ECG waveform sample points, it further includes the following steps: setting a collection number threshold, judging whether the number of collections contained on the current coordinate axis is greater than the collection number threshold, if yes, reducing the binary threshold, and returning to step S32.

[0021] Further, in step S42, if there is no ECG waveform sample point on the current coordinate axis, after marking the current coordinate axis ECG waveform feature point as an abnormal ECG waveform feature point, it further includes the following steps: setting an abnormal ECG waveform feature point number threshold, counting the total number of abnormal ECG waveform feature points, judging whether the total number of abnormal ECG waveform feature points is greater than the abnormal ECG waveform feature point number threshold, if yes, increasing the binary threshold, and returning to step S32.

[0022] Further, the binarization threshold is set to include an initial threshold and an offset threshold, and the offset threshold is set to adjust the binarization threshold, and the specific steps are:

[0023] The initial value of the offset threshold is set to 0,

[0024] When the binarization threshold is increased, the offset threshold is increased by a set unit amount;

[0025] When the binarization threshold is decreased, the offset threshold is decreased by a set unit amount.

[0026] Further, the setting step of the initial threshold is:

[0027] The maximum gray value and the minimum gray value of the target selected region gray image are obtained, and the average value of the maximum gray value and the minimum gray value is taken as the initial threshold.

[0028] Further, it further includes a step S44 of correcting the abnormal electrocardiogram feature point in combination with the effective electrocardiogram feature points before and after the abnormal electrocardiogram feature point on the coordinate axis.

[0029] Further, the step of correcting the abnormal electrocardiogram feature point adopts a gradient method.

[0030] Further, the step of correcting the abnormal electrocardiogram feature point adopts a cubic spline interpolation method.

[0031] The beneficial effects of the present application are:

[0032] The present application obtains a flat electrocardiogram by correcting a clinical electrocardiogram, and a user can independently select a corresponding waveform on the electrocardiogram as a target selected region image according to the waveform features on the electrocardiogram, and further performs binarization processing to sample the pixel points in the target selected region image, and according to the conditions of the electrocardiogram sampling points on different coordinate axes, corresponding marking and extraction methods are respectively adopted, so as to obtain effective electrocardiogram feature points, so that the electrocardiogram is clear and accurate, and the target selected region binarization image obtained after the binarization processing can filter the cases of too many or too few electrocardiogram sampling points, and the extraction of the electrocardiogram feature points is more convenient. BRIEF DESCRIPTION OF DRAWINGS

[0033] Figure 1 is a method flow diagram for extracting an electrocardiogram waveform based on an electrocardiogram according to an embodiment of the present application.

[0034] Figure 2 is a flow diagram of step S3 according to an embodiment of the present application.

[0035] Figure 3 is a flow diagram of step S4 according to an embodiment of the present application.

[0036] Figure 4 is a flowchart diagram of step S42 of an embodiment of the present application.

[0037] Figure 5 is a flowchart diagram of step S42 of another embodiment of the present application.

[0038] Figure 6 is a schematic diagram of a clinical electrocardiogram sample example.

[0039] Figure 7 is a schematic diagram of different examples of a target selection binarization image of an embodiment of the present application.

[0040] Figure 8 is a schematic diagram of calculating the weight of the central electrocardiogram waveform sampling point of a collection of the present application. DETAILED DESCRIPTION

[0041] In order to make the purpose, technical solutions and advantages of the present application clearer, the technical solutions in the present application will be described clearly and completely below in combination with the drawings of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments of the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the protection scope of the present application.

[0042] In the actual application of existing electrocardiogram teaching, the electrocardiogram used is mainly from actual clinical cases of various hospitals, and is usually stored as a picture (jpg, tif, png, etc.) or PDF format by scanning or photographing and applied to teaching. Due to the limitation of scanning or photographing, the stored electrocardiogram may be folded, deformed or distorted, so that the electrocardiogram waveform is not flat. At the same time, under the mode of photographing or scanning, other noise points may be generated. In addition, the grid lines on the electrocardiogram for marking the time domain and amplitude of the electrocardiogram waveform will interfere with the extraction of the waveform. Therefore, in order to improve the defects mentioned above, an electrocardiogram waveform extraction method based on electrocardiogram is provided in an embodiment of the present application.

[0043] As shown in Figures 1-4 , an electrocardiogram waveform extraction method based on electrocardiogram is provided in an embodiment of the present application, which comprises the following steps:

[0044] S1, acquiring an electrocardiogram, correcting the electrocardiogram to obtain a front view image of the electrocardiogram;

[0045] In this step, the electrocardiogram is first corrected to obtain a relatively flat and normal front view image of the electrocardiogram, which provides a basis for the subsequent electrocardiogram waveform extraction step, as shown in Figure 6As shown, it is a clinical electrocardiogram sample example, which can be usually corrected by image perspective transformation (also known as projection mapping) method.

[0046] S2, obtaining a target selection image selected by a user based on the front view image for extracting an electrocardiogram waveform;

[0047] In this step, the user is a medical student or other related personnel who needs to master electrocardiogram technology; the target selection image is a selection area selected by the user for extracting the electrocardiogram waveform, and the user extracts the electrocardiogram waveform in the target selection image range by determining the target selection image. In this step, the user can select a waveform region on the electrocardiogram as the target selection image for extracting the electrocardiogram waveform according to the waveform characteristics and teaching needs on the front view image of the corrected electrocardiogram, which can be manually selected or the electrocardiogram can be divided into regions to select a corresponding region for extracting the electrocardiogram waveform, such as dividing the electrocardiogram into wave bands according to the cardiac cycle.

[0048] S3, performing binaryzation processing on the target selection image to obtain a target selection binaryzation image;

[0049] In this step, the target selection image is binaryzation processed by setting a binaryzation threshold value to set the pixel value of the pixel point on the target selection image to 0 or 255, so that the entire image presents a clear black and white effect, which is convenient for sampling the electrocardiogram waveform in the target selection image region. This step specifically includes:

[0050] S31, performing grayscale processing on the target selection image to obtain a target selection grayscale image;

[0051] S32, setting a binaryzation threshold value, performing binaryzation processing on the target selection grayscale image to obtain a target selection binaryzation image.

[0052] As shown in FIG. 1, Figure 7 FIG. 2 shows target selection binaryzation image examples obtained after setting different binaryzation threshold values and performing binaryzation processing, wherein Figure 7 the electrocardiogram waveform sampling point boundary in (a) is clear, indicating that the binaryzation threshold value is set reasonably; Figure 7 the electrocardiogram waveform sampling point in (b) is too much, the interface is complex, indicating that the binaryzation threshold value is set too large and needs to be reduced; Figure 7 the electrocardiogram waveform sampling point in (c) is too few, which is not conducive to extracting the complete electrocardiogram waveform, indicating that the binaryzation threshold value is set too small and needs to be increased.

[0053] S4, extracting electrocardiogram waveform feature points in the target selection binaryzation image to construct an electrocardiogram waveform, including:

[0054] S41, taking the pixel point in the target selected area binary image as an electrocardiogram waveform sampling point, obtaining the coordinate of the electrocardiogram waveform sampling point;

[0055] S42, based on the coordinate of the electrocardiogram waveform sampling point, screening the electrocardiogram waveform sampling point, marking the electrocardiogram feature point, including:

[0056] If there is no electrocardiogram waveform sampling point on the current coordinate axis, marking the current coordinate axis electrocardiogram feature point as an abnormal electrocardiogram feature point;

[0057] If there is one electrocardiogram waveform sampling point on the current coordinate axis, marking the current electrocardiogram waveform sampling point as the effective electrocardiogram feature point of the current coordinate axis;

[0058] If there are two or more electrocardiogram waveform sampling points on the current coordinate axis, the continuous electrocardiogram waveform sampling points are collected, the center coordinate point of each collection is taken as the center electrocardiogram waveform sampling point of each collection, the weight of each center electrocardiogram waveform sampling point is calculated, and the center electrocardiogram waveform sampling point with the highest weight is selected as the effective electrocardiogram feature point of the current coordinate axis;

[0059] Wherein, the weight calculation expression is:

[0060] ;

[0061] Wherein: W is the weight, Num is the number of electrocardiogram waveform sampling points contained in the current collection, and diff is the vertical coordinate difference between the center electrocardiogram waveform sampling point of the current collection and the effective electrocardiogram feature point of the previous coordinate axis;

[0062] Here, the setting principle of the weight is further explained. The weight (W) includes distance weight (W1) and quantity weight (W2). The distance weight can be understood as the distance weight of the center electrocardiogram waveform sampling point of the current collection from the effective electrocardiogram feature point of the previous coordinate axis, and the quantity weight can be understood as the quantity weight of the electrocardiogram waveform sampling points contained in the current collection. Wherein:

[0063] The distance weight (W1) expression is:

[0064] ;

[0065] The quantity weight (W2) expression is:

[0066] ;

[0067] As shown in Figure 8 , in combination with the electrocardiogram waveform form shown in Figure 7 , when there are two or more electrocardiogram waveform sampling points on the coordinate axis, the continuous electrocardiogram waveform sampling points are collected, as shown in Figure 8As shown in the figure, the current coordinate axis is used to collect the continuous ECG waveform sampling points, which are divided into four collections, namely Collection 1~Collection 4, and the number of ECG waveform sampling points in each collection is Num1~Num4, respectively. The weight of each collection is Num1x0.01, Num2x0.01, Num3x0.01, and Num4x0.01, respectively. The distance weight is calculated by the distance between the center ECG waveform sampling point of each collection and the effective ECG waveform feature point of the previous coordinate axis, Figure 8 The point (M) is the effective ECG waveform feature point of the previous coordinate axis, and the center ECG waveform sampling point of each collection is the center coordinate point of each collection. The difference between the center ECG waveform sampling point of each collection and the point (M) in the vertical coordinate is calculated, and is represented by diff1~diff4, respectively. The distance weight of each collection is 1 / 2 diff1 +1, 1 / 2 diff2 +1, 1 / 2 diff3 +1, 1 / 2 diff4 +1.

[0068] S43, extract the effective ECG waveform feature point, and complete the ECG waveform construction.

[0069] In this embodiment, the corrected ECG obtained by clinical acquisition is used to obtain a relatively flat ECG front view image. The user selects the corresponding waveform region as the target selection image, and then uses binary processing to obtain the target selection binary image. The ECG waveform feature points are extracted by obtaining the ECG waveform sampling points in the target selection binary image, so as to construct the ECG waveform. In the whole method step, by setting the binary threshold, the pixel points presented in the target selection image region can be adjusted. When the binary threshold is too large, the pixel points in the target selection image region are too many, and when the binary threshold is too small, the pixel points in the target selection image region are too few. Therefore, the setting of the binary threshold determines the presentation of the ECG waveform sampling points, and the acquisition and accuracy of the ECG waveform sampling points play a key role in the construction of the ECG waveform. Too many ECG waveform sampling points not only complicate the screening process, but also may cause the deviation of the screened ECG waveform feature points, affecting the waveform curve, and too few ECG waveform sampling points will also cause too few ECG waveform feature points, which is not conducive to the construction of complete ECG waveform.

[0070] To avoid the case of too many ECG waveform sampling points, in some embodiments, in the case of two or more ECG waveform sampling points existing on the current coordinate axis, after the step of performing the collection processing on the continuous ECG waveform sampling points, before the step of taking the center coordinate point of each collection as the center ECG waveform sampling point of each collection, the collection is further processed, and the specific steps are as follows: a height threshold is set, it is judged whether the height of each collection has a value greater than the height threshold, yes, the binarization threshold is reduced, and the step S32 is returned. In this step, the height threshold can be set to a proportion of the height of the target selected image, for example, half of the height of the target selected image. When there are ECG waveform sampling points with a height greater than the set height threshold in the collection, it is considered that there are too many ECG waveform sampling points at present, it is determined that the current target selected region binary image is invalid, the binarization threshold needs to be reduced, and the step S32 is returned to perform the binarization processing on the target selected image again.

[0071] In other embodiments, in the case of two or more ECG waveform sampling points existing on the current coordinate axis, after the step of performing the collection processing on the continuous ECG waveform sampling points, before the step of taking the center coordinate point of each collection as the center ECG waveform sampling point of each collection, the collection can also be processed as follows:

[0072] A collection number threshold is set, it is judged whether the number of collections contained on the current coordinate axis is greater than the collection number threshold, yes, the binarization threshold is reduced, and the step S32 is returned. In this step, by pre-setting the collection number threshold, the number of collections on the current coordinate axis is counted, and when the sum of the number of collections is greater than the collection number threshold, it is considered that there are too many ECG waveform sampling points at present, it is determined that the current target selected region binary image is invalid, the binarization threshold needs to be reduced, and the step S32 is returned to perform the binarization processing on the target selected image again.

[0073] To avoid the case of too few ECG waveform sampling points and difficult to extract complete cardiac waveform, in some embodiments, in the case of no ECG waveform sampling point existing on the current coordinate axis, after the step of marking the ECG waveform feature point on the current coordinate axis as an abnormal ECG waveform feature point, the step of setting an abnormal ECG waveform feature point number threshold, counting the total number of abnormal ECG waveform feature points, judging whether the total number of abnormal ECG waveform feature points is greater than the abnormal ECG waveform feature point number threshold, yes, increasing the binarization threshold, and returning to the step S32 is further included. In this step, by setting the abnormal ECG waveform feature point number threshold, the case of too many abnormal ECG waveform feature points can be screened out, and the abnormal ECG waveform feature point number threshold can be set to half of the number of horizontal coordinate points of the target selected image.

[0074] Figure 5 The flowchart of step S42 in other embodiments is shown, from Figure 5As can be seen, in the case that there are two or more ECG waveform sampling points on the current coordinate axis, the height threshold and the number threshold of the set can be set at the same time, and the height of each set and the number of the current set are judged respectively, so as to filter the case that the ECG waveform sampling points are too much, adjust the binary threshold, return to step S32, and obtain the target selected binary image again. As can be seen from the two judgment methods, the order of the two judgment methods is not limited. The height threshold, the number threshold of the set, and the number threshold of the abnormal ECG waveform feature points mentioned in the above steps can be set by the user. Through the setting of each threshold, the corresponding ECG waveform can be quickly screened and constructed.

[0075] In the above steps, the binary threshold is adjusted when the ECG waveform sampling points are too much or too few, and the target selected image is binarized again in step S32. In order to facilitate the adjustment of the binary threshold, the binary threshold is set to include an initial threshold and an offset threshold in this embodiment. The offset threshold is set to adjust the binary threshold. The specific setting steps include:

[0076] The initial threshold is set as follows: the maximum gray value and the minimum gray value of the target selected gray image are obtained, and the average of the maximum gray value and the minimum gray value is taken as the initial threshold.

[0077] The offset threshold is initially set to 0. When certain conditions are met, the offset threshold is increased or decreased by a set unit amount, so as to achieve the purpose of adjusting the binary threshold. Specifically:

[0078] When the binary threshold is increased, the offset threshold is increased by a set unit amount;

[0079] When the binary threshold is decreased, the offset threshold is decreased by a set unit amount.

[0080] The unit amount is set by the user, such as 10 as a unit amount. If the pixel points in the target selected binary image region are too much, i.e. the ECG waveform sampling points are too much, it means that the binary threshold is too large, and the offset threshold is decreased by 10 as a unit amount. If the pixel points in the target selected binary image region are too few, i.e. the ECG waveform sampling points are too few, it means that the binary threshold is too small, and the offset threshold is increased by 10 as a unit amount.

[0081] In some embodiments, in order to make the ECG waveform line smooth, step S44 is further included, which combines the effective ECG waveform feature points before and after the abnormal ECG waveform feature points to correct the abnormal ECG waveform feature points. The correction method can use the gradient method or the cubic spline interpolation method.

Claims

1. A method for extracting electrocardiogram waveforms based on electrocardiogram, characterized in that, The method comprises the steps of: S1, acquiring an electrocardiogram, correcting the electrocardiogram, and obtaining a front view image of the electrocardiogram; S2, acquiring a target selection image selected by a user based on the front view image for extracting an electrocardiogram waveform; S3, performing binaryzation processing on the target selection image to obtain a target selection binaryzation image, comprising: S31, performing grayscale processing on the target selection image to obtain a target selection grayscale image; S32, setting a binaryzation threshold, performing binaryzation processing on the target selection grayscale image to obtain a target selection binaryzation image; S4, extracting electrocardiogram waveform feature points in the target selection binaryzation image to construct an electrocardiogram waveform, comprising: S41, taking a pixel point in the target selection binaryzation image as an electrocardiogram waveform sampling point, and acquiring a coordinate of the electrocardiogram waveform sampling point; S42, based on the coordinate of the electrocardiogram waveform sampling point, screening the electrocardiogram waveform sampling point, and marking an electrocardiogram waveform feature point, comprising: if there is no electrocardiogram waveform sampling point on a current coordinate axis, marking the current coordinate axis electrocardiogram waveform feature point as an abnormal electrocardiogram waveform feature point; if there is one electrocardiogram waveform sampling point on the current coordinate axis, marking the current electrocardiogram waveform sampling point as an effective electrocardiogram waveform feature point of the current coordinate axis; if there are two or more electrocardiogram waveform sampling points on the current coordinate axis, performing collection processing on the continuous electrocardiogram waveform sampling points, taking a central coordinate point of each collection as a central electrocardiogram waveform sampling point of each collection, respectively calculating a weight of the central electrocardiogram waveform sampling point of each collection, and selecting a central electrocardiogram waveform sampling point with the highest weight as the effective electrocardiogram waveform feature point of the current coordinate axis; wherein the weight calculation expression is: ; wherein W is the weight, Num is the number of electrocardiogram waveform sampling points contained in the current collection, and diff is a vertical coordinate difference value between the central electrocardiogram waveform sampling point of the current collection and the effective electrocardiogram waveform feature point of the previous coordinate axis; S43, extracting the effective electrocardiogram waveform feature point to complete the construction of the electrocardiogram waveform.

2. The method for extracting electrocardiogram waveform based on electrocardiogram according to claim 1, characterized in that, In step S42, before taking the central coordinate point of each collection as the central electrocardiogram waveform sampling point of each collection after performing collection processing on the continuous electrocardiogram waveform sampling points when there are two or more electrocardiogram waveform sampling points on the current coordinate axis, the method further comprises the step of: setting a height threshold, judging whether the height of each collection has a value greater than the height threshold, if yes, reducing the binaryzation threshold, and returning to step S32.

3. The method for extracting electrocardiogram waveform based on electrocardiogram according to claim 1, characterized in that, In step S42, before taking the central coordinate point of each collection as the central electrocardiogram waveform sampling point of each collection after performing collection processing on the continuous electrocardiogram waveform sampling points when there are two or more electrocardiogram waveform sampling points on the current coordinate axis, the method further comprises the step of: setting a collection number threshold, judging whether the number of the collections contained on the current coordinate axis is greater than the collection number threshold, if yes, reducing the binaryzation threshold, and returning to step S32.

4. The method for extracting electrocardiogram waveform based on electrocardiogram according to claim 1, characterized in that, In step S42, if there is no ECG sample point on the current coordinate axis, after marking the ECG feature point on the current coordinate axis as an abnormal ECG feature point, the method further comprises the steps of: setting an abnormal ECG feature point quantity threshold, counting the total number of abnormal ECG feature points, and determining whether the total number of abnormal ECG feature points is greater than the abnormal ECG feature point quantity threshold. If yes, the binarization threshold is increased, and the method returns to step S32.

5. The method of extracting electrocardiogram waveforms based on electrocardiogram according to any one of claims 1 to 4, characterized in that, The binarization threshold is set to include an initial threshold and an offset threshold, and the offset threshold is set to adjust the binarization threshold. The specific steps are as follows: The initial value of the offset threshold is set to 0, When the binarization threshold is increased, the offset threshold is increased by a set unit amount; When the binarization threshold is decreased, the offset threshold is decreased by a set unit amount.

6. The method of extracting electrocardiogram waveforms based on electrocardiogram according to claim 5, characterized in that, The setting steps of the initial threshold are as follows: The maximum and minimum gray values of the target selected region gray image are obtained, and the average of the maximum and minimum gray values is taken as the initial threshold.

7. The method for extracting electrocardiogram waveform based on electrocardiogram according to claim 1, characterized in that, The method further comprises step S44, which combines the effective ECG feature points before and after the abnormal ECG feature point to correct the abnormal ECG feature point.

8. The method of claim 7, wherein, The step of correcting the abnormal ECG feature point uses the gradient method.

9. The method of claim 7, wherein, The step of correcting the abnormal ECG feature point uses the cubic spline interpolation method.

Images