CN115881276B - Time-frequency double-bar code characteristic image generation method of electrocardiosignal and storage medium - Google Patents
Time-frequency double-bar code characteristic image generation method of electrocardiosignal and storage medium Download PDFInfo
- Publication number
- CN115881276B CN115881276B CN202310151498.0A CN202310151498A CN115881276B CN 115881276 B CN115881276 B CN 115881276B CN 202310151498 A CN202310151498 A CN 202310151498A CN 115881276 B CN115881276 B CN 115881276B
- Authority
- CN
- China
- Prior art keywords
- electrocardiosignal
- imf
- signal
- matrix
- noise
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
The invention relates to a time-frequency double-bar code characteristic image generation method of an electrocardiosignal and a storage medium, wherein the method comprises the steps of intercepting an electrocardiosignal x (t) obtained based on measurement of electrocardiograph equipment, designing a window function to obtain a complete electrocardiosignal period sample, normalizing the intercepted signal to realize pretreatment of data, performing EMD (empirical mode decomposition) on the processed signal and providing a correlation coefficient to determine a boundary point of noise-dominant IMF and signal-dominant IMF in the signal; and then, obtaining a time domain signal by filtering the electrocardiosignal, obtaining a frequency domain signal by fast Fourier transformation, and constructing an electrocardiosignal feature matrix by the time domain signal and the frequency domain signal together to generate a time-frequency double-bar code feature image of the electrocardiosignal. The invention can reliably remove noise components in the signals, screen useful IMF through the correlation coefficient, reduce interference and construct double bar codes by using the time domain signals of the electrocardiosignals after noise filtering and the frequency domain signals thereof.
Description
Technical Field
The invention relates to the technical field of image generation, in particular to a time-frequency double-bar code characteristic image generation method of electrocardiosignals.
Background
With the development of society, the pace of life of people is accelerated, and heart and cardiovascular diseases are one of the main diseases endangering human bodies due to various factors such as high working pressure, numerous bad living habits and the like. The Electrocardiosignal (ECG) of human body contains the law, physiology and pathology information of heart activity, so that the analysis of the electrocardiosignal characteristics in clinical medical practice is an important basis for diagnosing heart and cardiovascular diseases. At present, the analysis of electrocardiosignals mainly completes the diagnosis of diseases by analyzing the characteristics of electrocardiograms by doctors. Electrocardiogram manifestation an electrocardiogram is a one-dimensional waveform.
The electrocardiosignals are easy to be interfered by environment electromagnetic interference in the process of collecting the electrocardiosignals, so that the collected electrocardiosignals contain noise. Because the electrocardiosignals contain noise due to the field environment, the limb actions and the muscle contraction of the human body, in order to filter noise interference in the electrocardiosignals, a hardware filter is designed to realize the noise filtering process of the electrocardiosignals. The design of the filter requires the knowledge of the frequency range of the noise in advance, the electrocardiosignal needs to be known deeply, and if the frequencies of the noise and the signal are heavy, the noise with overlapped frequencies cannot be filtered by the design of the filter so as to keep the useful components of the electrocardiosignal.
Disclosure of Invention
The time-frequency double-bar code characteristic image generation method for the electrocardiosignals can at least solve the technical problems, and realize characteristic image generation of the electrocardiosignals, namely, the time-frequency double-bar code characteristic image generation.
In order to achieve the above purpose, the present invention adopts the following technical scheme:
a time-frequency double-bar code characteristic image generation method of electrocardiosignals, which comprises the following steps,
step A: for an electrocardiograph signal x (t) obtained based on electrocardiograph device measurement, the unit is millivolt; t is time, the unit is seconds, a window function is designed to intercept to obtain a complete electrocardio cycle sample, and the intercepted signal is normalized;
and (B) step (B): performing Empirical Mode Decomposition (EMD) on the normalized electrocardiosignal to obtain IMFs of natural mode components from high frequency to low frequency, calculating correlation coefficients of each IMF and an original signal, and determining boundary points of the noise-dominant IMF and the signal-dominant IMF by using the correlation coefficients;
step C: filtering out noise-dominant IMF, reserving signal-dominant IMF to perform electrocardiosignal noise filtering, and performing fast Fourier transform on the noise-filtered electrocardiosignal to obtain a frequency domain signal of the noise-filtered electrocardiosignal;
step D: construction of time-domain bar code matrix H for filtered electrocardiosignals 1 For the electrocardiosignals after noise filteringConstructing a frequency domain bar code matrix H2 by using the frequency domain signals, and combining the time domain bar code matrix H1 and the frequency domain bar code matrix H 2 Alignment is carried out, an electrocardiosignal characteristic matrix H is constructed, and the numerical value in H is linearly mapped to gray image pixel value [0, 255 ]]The output of the two-dimensional electrocardiosignals is an image, and the one-dimensional electrocardiosignals are converted into two-dimensional time-frequency double-bar code characteristic images.
Further, the step a: for an electrocardiograph signal x (t) obtained based on electrocardiograph device measurement, the unit is millivolt; t is time, the unit is seconds, a window function is designed to intercept to obtain a complete electrocardio cycle sample, and the intercepted signal is normalized, and the method comprises the following specific steps:
s101: the electrocardiosignal x (t) obtained based on the measurement of the electrocardiograph equipment is expressed in the form of a sequence, namely an electrocardiosignal sequence x a (n) signal interception of its design window function w (n);
where n=0, 1,2,3 … is the sample size, n=t×f; w (N) is a window function, N 1 、N 2 The starting point and the end point of the window function are respectively;
then the intercepted electrocardiosignal sequence x b (n)＝x a (n) ×w (n) in units of: millivolts, x a (n) is an electrocardiosignal sequence, and the unit is: millivolts, x b (n) is a one-dimensional array sequence, the position value range of the elements in the array is [0, N]Sequence length n+1, n=n 2 -N 1 ；
S102: for the intercepted electrocardiosignal sequence x b (n) performing a normalization operation;
wherein x is b (n) is the intercepted electrocardiosignal sequence, x c (n) is normalized electrocardiosignal sequence, x min For the intercepted electrocardio signal sequenceMinimum amplitude, x max The maximum amplitude of the intercepted electrocardiosignal sequence is obtained.
Further, in the step S101, N 1 、N 2 The selection rule of (1) is that the electrocardiosignal sequence x is used a Taking one R peak of (N) as a reference, taking 0.4 second data forward, setting electrocardiosignal sampling frequency f=360 Hz, thus taking 144=0.4×360 points forward, namely N 1 Data points at 144 before the R peak; at the same time, the data of the last 0.5 seconds of the 2 nd R peak of the backward number of the R peak is taken, so 180=0.5×360 points are taken backward, namely N 2 Is the 180 th data point after the 2 nd R peak, counted backward.
Further, the step B: the EMD is carried out on the normalized electrocardiosignal to obtain IMFs of natural mode components from high frequency to low frequency, the correlation coefficient of each IMF and the original signal is calculated, and the boundary points of the noise-dominant IMF and the signal-dominant IMF are determined by the correlation coefficient, which comprises the following steps:
s103: for normalized electrocardiosignal sequence x c (N) performing EMD decomposition operation, setting maximum number of decomposed IMFs N IMF =7, and calculate each IMF and x c A correlation coefficient of (n);
wherein x is c (n) is normalized electrocardiosignal sequence, c i (n) is the ith IMF, ρ obtained by EMD decomposition i Represents x c (n) correlation coefficient with the ith IMF, μ x Sum sigma x Is x c Mean and standard deviation of (n), μ c Sum sigma c C is i Mean and scale of (n)Difference in accuracy, N a +1 is normalized electrocardiosignal sequence x c The data length of (N), where N a =n; whereas EMD decomposition includes decomposing the number of IMFs N IMF N can be obtained by EMD decomposition IMF IMFs and 1 residual component R (n);
s104: from the first maximum point of the curve of the set of correlation coefficients ρ, which corresponds to the element index K in the set f Boundary points for noise-dominant IMF and signal-dominant IMF;
wherein the set of correlation coefficients ρ= (ρ) 1 ，ρ 2 ，…，ρ m ) The number of set elements m=n IMF The method comprises the steps of carrying out a first treatment on the surface of the The definition of noise-dominant IMF and signal-dominant IMF is, formerly, K f The IMF of order is used as noise dominant IMF, N is used as f The IMF after the order is used as a signal dominant IMF; if the curve of the correlation coefficient set rho has no maximum value, selecting the first 3-order IMF as a noise-dominant IMF, and selecting the IMFs after 3-order as signal-dominant IMFs.
Further, the step C: filtering noise-dominant IMF, reserving signal-dominant IMF to perform electrocardiosignal noise filtering, and performing fast Fourier transform on the noise-filtered electrocardiosignal to obtain a frequency domain signal of the noise-filtered electrocardiosignal, which specifically comprises the following steps:
s105: filtering out noise dominant IMF, retaining signal dominant IMF, and performing signal reconstruction together with residual component R (n) obtained by EMD decomposition;
wherein x is d (n) is the filtered electrocardiosignal sequence, c i (N) is the ith IMF obtained by EMD decomposition, M is the number N of IMFs obtained by decomposition IMF ，K f For the boundary points of the noise-dominant IMF and the signal-dominant IMF, R (n) is the remaining component of the EMD decomposition;
s106: the electrocardiosignal sequence x after noise filtering d (n) performing a fast Fourier transform to obtain x d Frequency domain signal x of (n) e (n)。
Further, the step D: opposite filteringConstructing a time domain bar code matrix H1 of the noisy electrocardiosignal, and constructing a frequency domain bar code matrix H of the frequency domain signal of the noisy electrocardiosignal 2 Matrix H of time domain bar code 1 And frequency domain bar code matrix H 2 Alignment is carried out, an electrocardiosignal characteristic matrix H is constructed, and the numerical value in H is linearly mapped to gray image pixel value [0, 255 ]]The medium output is an image, so that one-dimensional electrocardiosignals are converted into two-dimensional time-frequency double-bar code characteristic images; in particular comprising the following steps of the method,
s107: the electrocardiosignal sequence x after noise filtering d (N) replication N c Parts and consist of N c The electrocardiosignal sequence x after noise filtering d (N) constructing a two-dimensional matrix to obtain an N c ×(N b +1) size matrix H 1 ；
Wherein N is b ＝N a ，N b +1 is the filtered electrocardiosignal sequence x d Data length of (n), H 1 The filtered electrocardiosignal bar code matrix is a time domain bar code matrix; n (N) c For a time-domain bar code matrix H 1 At the same time, the number of copies is also the number of the lines of the electrocardiosignal sequence x after noise filtering d Expanding the data of (n) to expand a one-dimensional matrix to a two-dimensional matrix;
s108: taking the frequency domain signal x e Front N of (N) e Data points, obtain frequency domain signal x f1 (n) and at the same time let x f2 (n)＝x f1 (n) x is f1 (n) and x f2 (n) cross-fusing to form a processed frequency domain signal x f (n)；
Wherein x is e (n) is represented by x d (N) frequency domain transforming to obtain N d +1 is the frequency domain signal x e (N) data length, N d Numerical size and N of (2) b The values of (2) are substantially the same, i.e. N d ＝N b ，x f1 (n)＝(x f1 (0)，x f1 (1)，…，x f1 (N e ) X) f2 (n)＝(x f2 (0)，x f2 (1)，…，x f2 (N e ))，x f (n)＝(x f1 (0)，x f2 (0)，x f1 (1)，x f2 (1)…，x f1 (N e )，x f2 (N e ) Obtaining a processed frequency domain signal according to a cross fusion rule;
s109: the processed frequency domain signal x f (N) replication N c Parts, and is composed of N c The processed frequency domain signal x f (N) constructing a two-dimensional matrix to obtain an N c ×(N f +2) size matrix H 2 ；
Wherein N is f ＝2×N e ，N f +2 is the processed frequency domain signal x f Data length of (n), H 2 The frequency domain bar code matrix is the processed frequency domain signal bar code matrix;
s110: matrix H of time-domain bar codes 1 And frequency domain bar code matrix H 2 Alignment processing is carried out to obtain an electrocardiosignal characteristic matrix H, and the numerical value in H is linearly mapped to gray image pixel value [0, 255 ]]The output image is the time-frequency double-bar code characteristic image of the electrocardiosignal;
wherein H is an electrocardiosignal characteristic matrix, and the size of the matrix is 2N c ×(N b +1), in the process of H 1 And H 2 If H at the time of alignment 2 The number of columns of the matrix is less than H 1 Column number of matrix, then to H 2 Element supplementation is carried out on the matrix; the supplement rule is that the average value of the first two elements of the supplement element position is taken as the value of the supplement element, so that H 1 Matrix sum H 2 The columns of the matrix are equal, i.e. H after supplementation 2 The matrix size becomes N c ×(N b +1), and constructing an electrocardiosignal characteristic matrix H according to a formula (9);
in another aspect, the invention discloses a computer readable storage medium storing a computer program which, when executed by a processor, causes the processor to perform the steps of the method as described above.
According to the technical scheme, the time-frequency dual-bar code characteristic image generation method of the electrocardiosignal comprises the steps of firstly, intercepting an electrocardiosignal x (t) (with the unit of millivolts; t is time and the unit of seconds) obtained based on measurement of an electrocardiograph device, designing a window function to obtain a complete electrocardiosignal period sample, normalizing the intercepted signal to realize data preprocessing, carrying out EMD decomposition on the processed signal, and providing a correlation coefficient to determine boundary points of noise-dominant IMF and signal-dominant IMF in the signal. And then, obtaining a time domain signal by filtering the electrocardiosignal, obtaining a frequency domain signal by fast Fourier transformation, and constructing an electrocardiosignal feature matrix by the time domain signal and the frequency domain signal together to generate a time-frequency double-bar code feature image of the electrocardiosignal. The result shows that the EMD decomposition method can effectively reduce the interference of noise components on the actual value of the electrocardiosignal. The time-frequency double-bar code characteristic image generation of the electrocardiosignal generated by the time-domain signal and the frequency-domain signal of the reconstructed electrocardiosignal can highlight the information contained in the electrocardiosignal, and a novel method is provided for generating the electrocardiosignal characteristic image.
In general, the invention adopts empirical mode decomposition to realize the noise reduction process of the electrocardiosignal, extracts useful components of the electrocardiosignal, obtains a time domain signal of the electrocardiosignal, and simultaneously carries out frequency domain analysis on the time domain signal to obtain a frequency domain signal of the electrocardiosignal. And generating a time-frequency double-bar code characteristic image of the electrocardiosignal by using the time-domain signal and the frequency-domain signal, and converting the one-dimensional waveform image into a two-dimensional time-frequency double-bar code characteristic image. The dual-bar code image simultaneously comprises the time domain features and the frequency domain features of the electrocardiosignal, and better highlights the features of the electrocardiosignal. The invention can reliably remove the noise component in the signal and effectively reduce the interference of the noise component on the actual value of the electrocardiosignal. The useful IMF is screened through the correlation coefficient, the interference is reduced, and the double bar codes are constructed by using the time domain signals and the frequency domain signals of the electrocardiosignals after noise filtering, so that the time-frequency double bar code characteristic image generation method of the electrocardiosignals is realized.
Drawings
FIG. 1 is a flow chart of a method for generating a dual bar code of an electrocardiosignal feature image in accordance with an embodiment of the invention;
FIG. 2 is a diagram of an electrocardiographic signal sequence data according to an embodiment of the present invention;
FIG. 3a is a diagram of truncated and normalized normal electrocardiographic signal data according to an embodiment of the present invention;
FIG. 3b is a graph of truncated and normalized ventricular premature beat electrocardiosignal data according to an embodiment of the invention;
FIG. 3c is a graph of truncated and normalized atrial premature beat electrocardiosignal data of an embodiment of the invention;
FIG. 4a is an IMF diagram of normal electrocardiographic signal EMD decomposition according to an embodiment of the present invention;
FIG. 4b is an IMF diagram of an EMD decomposition of ventricular premature beat cardiac signals according to an embodiment of the present invention;
FIG. 4c is an IMF diagram of an EMD decomposition of an atrial premature beat signal according to an embodiment of the present invention;
FIG. 5a is a graph of correlation coefficients of normal electrocardiographic signals according to an embodiment of the present invention;
FIG. 5b is a graph of correlation coefficients of ventricular premature beat electrocardiosignals according to an embodiment of the invention;
FIG. 5c is a graph of correlation coefficients of atrial premature beat electrocardiosignals according to an embodiment of the invention;
FIG. 6a is a graph of filtered and noisy electrocardiographic data of a normal electrocardiographic signal according to an embodiment of the present invention;
FIG. 6b is a graph of filtered and noisy electrocardiographic data of ventricular premature beat electrocardiographic signals according to an embodiment of the present invention;
FIG. 6c is a graph of filtered and noisy electrocardiographic data of an atrial premature beat electrocardiograph according to an embodiment of the present invention;
FIG. 7a is a frequency domain signal data plot of a normal electrocardiographic signal according to an embodiment of the present invention;
FIG. 7b is a frequency domain signal data plot of a ventricular premature beat cardiac electrical signal according to an embodiment of the invention;
FIG. 7c is a frequency domain signal data plot of an atrial premature beat cardiac electrical signal according to an embodiment of the present invention;
FIG. 8 is a schematic diagram of a feature image matrix of an electrocardiograph signal according to an embodiment of the present invention;
FIG. 9a is a dual bar code plot of a normal electrocardiographic signal according to an embodiment of the present invention;
FIG. 9b is a dual bar code plot of a ventricular premature beat cardiac electrical signal according to an embodiment of the invention;
fig. 9c is a dual bar code plot of an atrial premature beat cardiac electrical signal according to an embodiment of the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments of the present invention.
As shown in fig. 1, the method for generating a time-frequency dual-bar code feature image of an electrocardiograph signal according to the embodiment specifically includes the following steps:
step A: an electrocardiograph signal x (t) (unit is millivolt; t is time and unit is second) obtained based on electrocardiograph equipment measurement is intercepted by a window function to obtain a complete electrocardiograph period sample, and the intercepted signal is normalized;
as shown in FIG. 2, which is an electrocardiosignal sequence data graph, the data length is 650000 sampling points, and the data shown in FIG. 2 is that the data acquisition frequency is 360Hz
The minute electrocardiosignals comprise normal electrocardiosignals, ventricular premature beat electrocardiosignals and atrial premature beat electrocardiosignals. The small plot in the figure is an enlarged plot showing sample points 1000 to 2000, with a data length of 1000 sample points showing data for 3R peaks.
S101: electrocardiogram signal obtained based on electrocardiograph device measurementx (t), expressed in the form of a sequence, i.e. an electrocardiosignal sequence x a (n) signal interception of its design window function w (n)
Where n=0, 1,2,3, … is the sampling amount, n=t×f, f is the sampling frequency of the electrocardiograph signal x (t), and the unit is hertz. w (N) is a window function, N 1 、N 2 The start and end of the window function, respectively. According to the characteristics of the electrocardiosignals, N is used for intercepting a complete electrocardiosignal 1 、N 2 The selection rule of (1) is that the electrocardiosignal sequence x is used a One R peak of (N) is used as a reference (the complete electrocardiosignal contains a QRS wave, and the R peak is the peak point of the QRS wave), and 0.4 seconds of data is taken forward, and the electrocardiosignal sampling frequency f=360 Hz is set in the invention, so 144=0.4×360 points, namely N, are taken forward 1 Is the 144 th data point before the R peak. At the same time, the data of the last 0.5 seconds of the 2 nd R peak of the backward number of the R peak is taken, so 180=0.5×360 points are taken backward, namely N 2 Is the 180 th data point after the 2 nd R peak, counted backward. Then the intercepted electrocardiosignal sequence x b (n)＝x a (n). Times.w (n) (in millivolts), x a (n) is an electrocardiosignal sequence (unit is millivolt) and x is b (n) is a one-dimensional array sequence, the position value range of the elements in the array is [0, N]Sequence length n+1, n=n 2 -N 1 。。x b The 3R peaks are included in (n) and more information is included than one R peak. Meanwhile, the data of 0.4 seconds in the forward direction of the 0 th R peak and the data of 0.5 seconds in the backward direction of the 2 nd R peak are determined according to the characteristics of electrocardiosignals, so that more information can be contained, and the situation that the R peak data before the 0 th R peak and the R peak data after the 2 nd R peak are contained in the current interception is avoided, so that the intercepted signals are not interfered by signals in other time periods is avoided. The R peak on the electrocardiosignal needs to be identified, and the identification method is not considered here, and the R peak on the electrocardiosignal is assumed to be marked, so that the method is carried out under the assumed condition.
S102: for the intercepted electrocardiosignal sequence x b (n) performing a normalization operation;
wherein x is b (n) is the intercepted electrocardiosignal sequence, x c (n) is the normalized electrocardiosignal sequence (no unit, no unit for the subsequent signal sequence due to normalization here), x min To the minimum amplitude value of the intercepted electrocardiosignal sequence, x max The maximum amplitude of the intercepted electrocardiosignal sequence is obtained.
As shown in fig. 3a, 3b and 3c, the data diagram of the 3 classes of electrocardiosignal after interception and normalization is shown. Namely, according to the method of step S101, a section of electrocardiosignal containing 3R peaks is intercepted from the data shown in FIG. 2, and normalized according to the method of step S101 to obtain a normalized electrocardiosignal sequence x c (n) and FIGS. 3a, 3b and 3c are x respectively c (n) corresponding data map. For comparison, 3 classes of electrocardiographic signal data are intercepted and normalized herein, wherein fig. 3a shows a normal electrocardiographic signal, fig. 3b shows a ventricular premature electrocardiographic signal, and fig. 3c shows an atrial premature electrocardiographic signal.
And (B) step (B): empirical mode decomposition (empirical mode decomposition, EMD) is carried out on the normalized electrocardiosignals to obtain natural mode components (intrinsic mode function, IMF) from high frequency to low frequency, correlation coefficients of the IMFs and the original signals are calculated, and boundary points of the noise-dominant IMFs and the signal-dominant IMFs are determined by the correlation coefficients;
s103: for normalized electrocardiosignal sequence x c (N) performing EMD decomposition operation, setting maximum number of decomposed IMFs N IMF =7, and calculate each IMF and x c A correlation coefficient of (n);
wherein x is c (n) is normalized electrocardiosignal sequence, c i (n) is the ith IMF, ρ obtained by EMD decomposition i Represents x c (n) correlation coefficient with the ith IMF, μ x Sum sigma x Is x c Mean and standard deviation of (n), μ c Sum sigma c C is i Mean and standard deviation of (N), N a +1 is normalized electrocardiosignal sequence x c The data length of (N), where N a =n. Whereas EMD decomposition is a decomposition method, the parameters include decomposing the number N of IMFs IMF N can be obtained by EMD decomposition IMF IMF and 1 residual component R (n). Here, the maximum decomposition IMF number N is set IMF =7 is according to x c (n) signal characteristics. If N IMF Too small a setting will result in x c (N) inadequate decomposition of N IMF Too large a setting will result in x c (N) the decomposition time is too long, the operation speed of the method is reduced, so N is selected on the premise of keeping the decomposition sufficient and the operation speed of the method IMF ＝7。
As shown in fig. 4a, 4b and 4c, the IMF diagram is an EMD decomposition of 3 electrocardiographic signals. From the figure, it can be seen that the normalized electrocardiosignal sequence x c And (n) decomposing into 7 IMF components, so that the composition components of the electrocardiosignal can be clearly decomposed. Wherein, the IMFs corresponding to the normal electrocardiosignal, the ventricular premature heart signal and the atrial premature heart signal are respectively shown in figures 4a, 4b and 4 c. Fig. 4c shows that there are only 6 IMF components because they meet the partial EMD decomposition termination condition, which is a default parameter, and therefore do not wait until the 7 th IMF is completed before stopping.
S104: from the first maximum point of the curve of the set of correlation coefficients ρ, which corresponds to the element index K in the set f Boundary points for noise-dominant IMF and signal-dominant IMF;
wherein the set of correlation coefficients ρ= (ρ) 1 ，ρ 2 ，…，ρ m ) The number of set elements m=n IMF . The definition of noise-dominant IMF and signal-dominant IMF is, formerly, K f The IMF of order is used as noise dominant IMF, N is used as f The IMF after the order is used as the signal dominant IMF. If the curve of the correlation coefficient set rho has no maximum value, selecting the first 3-order IMF as a noise-dominant IMF, and selecting the IMFs after 3-order as signal-dominant IMFs.
As shown in fig. 5a, 5b and 5c, the correlation coefficient graphs of the class 3 electrocardiosignals are shown, wherein fig. 5a, 5b and 5c are respectively corresponding correlation coefficient graphs of the normal electrocardiosignals, the ventricular premature beat electrocardiosignals and the atrial premature beat electrocardiosignals. The noise-dominant IMF and signal-dominant IMF sub-points are determined from the first maxima point on the graph, i.e. the functions implemented in step S104. As can be seen from fig. 3a, 3b, 3c, 4a, 4b, 4c, 5a, 5b, and 5c, the noise levels in the 3 types of electrocardiographic signals are different, so that the adaptive adjustment of the demarcation point parameters can be realized through step S104 without manual adjustment.
Step C: filtering out noise-dominant IMF, reserving signal-dominant IMF to perform electrocardiosignal noise filtering, and performing fast Fourier transform on the noise-filtered electrocardiosignal to obtain a frequency domain signal of the noise-filtered electrocardiosignal;
s105: filtering out noise dominant IMF, retaining signal dominant IMF, and performing signal reconstruction together with residual component R (n) obtained by EMD decomposition;
wherein x is d (n) is the filtered electrocardiosignal sequence, c i (N) is the ith IMF obtained by EMD decomposition, M is the number N of IMFs obtained by decomposition IMF ，K f For the boundary points of the noise-dominated IMF and the signal-dominated IMF, R (n) is the remaining component of the EMD decomposition. Noise-dominated IMF based on the noise of the electrocardiosignalAnd filtering the electrocardiosignals without entering the reconstruction process of the electrocardiosignals, thereby realizing the noise reduction process. Although the signal-dominant IMF is noisy, the influence of the signal-dominant IMF is small and can be ignored, so the signal-dominant IMF is selected as a main part of reconstruction of the electrocardiosignal.
As shown in fig. 6a, 6b and 6c, the filtered and noisy electrocardiosignal data graph is a 3-class electrocardiosignal, namely a filtered and noisy electrocardiosignal sequence x d (n) a data map. Compared with fig. 3a, 3b and 3c, the noise-filtering electrocardiosignal 'sawtooth effect' is weakened, the signal is smoother, and the characteristics of the signal are maintained. Wherein, the data graphs of the noise-filtered electrocardiosignals corresponding to the normal electrocardiosignals, the ventricular premature beat electrocardiosignals and the atrial premature beat electrocardiosignals are respectively shown in fig. 6a, 6b and 6 c.
S106: the electrocardiosignal sequence x after noise filtering d (n) performing a fast Fourier transform to obtain x d Frequency domain signal x of (n) e (n)；
Among them, the fast fourier transform is a method, i.e., a high-efficiency fast mathematical transform method of fourier transform. The signal in the time domain can be converted into a frequency domain signal which is easy to analyze through fast Fourier transform, namely, the electrocardiosignal sequence x after noise filtering d (n) change from time domain to frequency domain signal x e (n)。
As shown in fig. 7a, 7b and 7c, the frequency domain signal data diagram of the class 3 noise-filtered electrocardiosignal is shown. The frequency domain signals are also different among different electrocardiosignals, wherein the frequency domain signal data graphs of the noise filtering electrocardiosignals corresponding to the normal electrocardiosignals, the ventricular premature beat electrocardiosignals and the atrial premature beat electrocardiosignals are respectively shown in fig. 7a, 7b and 7 c. Fig. 7a and 7c are similar in that the two electrocardiographs are similar in nature, with the difference that the time interval between R peaks of the normal electrocardiograph can be considered equal, and the time interval between R peaks of the atrial premature electrocardiograph is shorter than the time interval between R peaks of the normal electrocardiograph, so that the difference in the frequency domain is not obvious. But figures 7a and 7b together with figure 7c have a significant difference in that their signal characteristics can be highlighted by the frequency domain signal.
Step D: construction of time-domain bar code matrix H for filtered electrocardiosignals 1 Constructing a frequency domain bar code matrix H for the frequency domain signals of the electrocardiosignals after noise filtering 2 Matrix H of time domain bar code 1 And frequency domain bar code matrix H 2 Alignment is carried out, an electrocardiosignal characteristic matrix H is constructed, and the numerical value in H is linearly mapped to gray image pixel value [0, 255 ]]The output of the two-dimensional electrocardiosignals is an image, and the one-dimensional electrocardiosignals are converted into two-dimensional time-frequency double-bar code characteristic images.
S107: the electrocardiosignal sequence x after noise filtering d (N) replication N c Parts and consist of N c The electrocardiosignal sequence x after noise filtering d (N) constructing a two-dimensional matrix to obtain an N c ×(N b +1) size matrix H 1 ；
Wherein N is b ＝N a ，N b +1 is the filtered electrocardiosignal sequence x d Data length of (n), H 1 The method is an electrocardiosignal bar code matrix after noise filtering, namely a domain bar code matrix. N (N) c For a time-domain bar code matrix H 1 At the same time, the number of copies is also the number of the lines of the electrocardiosignal sequence x after noise filtering d The data of (n) is expanded, and one-dimensional matrix is expanded to two-dimensional matrix. Here, the filtered electrocardiosignal sequence x d (n) copying and reconstructing the two-dimensional matrix to better emphasize the filtered electrocardiosignal sequence x d (n) if only x is used d (n) constructing a one-dimensional matrix, it is only one line when the matrix is converted into an image output, and it is not obvious enough that it is difficult to recognize the characteristic information contained therein by both a human and a machine.
Taking the frequency domain signal x e Front N of (N) e Data points, obtain frequency domain signal x f1 (n) and at the same time let x f2 (n)＝x f1 (n) x is f1 (n) and x f2 (n) cross-fusing to form a processed frequency domain signal x f (n)；
Wherein x is e (n) is represented by x d (N) frequency domain transforming to obtain N d +1 is the frequency domain signal x e (N) data length, N d Numerical size and N of (2) b The values of (2) are substantially the same, i.e. N d ＝N b ，x f1 (n)＝(x f1 (0)，x f1 (1)，…，x f1 (N e ) X) f2 (n)＝(x f2 (0)，x f2 (1)，…，x f2 (N e ))，x f (n)＝(x f1 (0)，x f2 (0)，x f1 (1)，x f2 (1)…，x f1 (N e )，x f2 (N e ) And (3) obtaining the processed frequency domain signal according to the cross fusion rule. Due to the nature of the fast fourier transform, the resulting frequency domain signal x e (n) is a symmetrical signal, x e The latter half and the former half of (N) are so that the former N is intercepted e Data points. And the cross fusion is carried out, so that the information contained in the intercepted frequency domain signal can be amplified to be twice of the original information, and the frequency domain characteristics of the electrocardiosignal are more highlighted.
S109: the processed frequency domain signal x f (N) replication N c Parts, and is composed of N c The processed frequency domain signal x f (N) constructing a two-dimensional matrix to obtain an N c ×(N f +2) size matrix H 2 ；
Wherein N is f ＝2×N e ，N f +2 is the processed frequency domain signal x f Data length of (n), H 2 The frequency domain bar code matrix is the processed frequency domain signal bar code matrix.
S110: matrix H of time-domain bar codes 1 And frequency domain bar code matrix H 2 The alignment processing is carried out to obtain an electrocardiosignal characteristic matrix H, and the numerical value in H is linearly mapped to a gray level image pixel value [0 ],255]the output image is the time-frequency double-bar code characteristic image of the electrocardiosignal;
wherein H is an electrocardiosignal characteristic matrix, and the size of the matrix is 2N c ×(N b +1), in the process of H 1 And H 2 If H at the time of alignment 2 The number of columns of the matrix is less than H 1 Column number of matrix, then to H 2 The matrix is subjected to element supplementation, wherein the supplementation rule is that the average value of the first two elements of the supplementation element position is taken as the value of the supplementation element, so that H 1 Matrix sum H 2 The columns of the matrix are equal, i.e. H after supplementation 2 The matrix size becomes N c ×(N b +1), and constructing an electrocardiosignal characteristic matrix H according to a formula (9).
As shown in fig. 8, a schematic diagram of a feature matrix of the electrocardiograph signal is shown. Taking the data of the normal electrocardiographic signals as an example, the process from step S107 to step S110 is as shown in fig. 8. Transforming one-dimensional time domain signal and one-dimensional frequency domain signal into one 2N c ×(N b +1) provides data for generating a time-frequency dual bar code signature and also amplifies the signature of the electrocardiographic signal.
As shown in fig. 9a, 9b and 9c, the images are dual barcode images of 3 types of electrocardiosignals, wherein fig. 9a, 9b and 9c are respectively time-frequency dual barcode characteristic images corresponding to normal electrocardiosignals, ventricular premature beat electrocardiosignals and atrial premature beat electrocardiosignals. Comparing fig. 9a, 9b, and 9c, it can be seen that the time-frequency dual-bar code feature maps generated by class 3 electrocardiographic signals each have distinct features and are different from each other. In fig. 9a, the upper half is uniformly distributed with 3 distinct white bars, and the lower half is changed from left to right from gradient to black by the white bars, and finally becomes black. In fig. 9b, the upper half is more obvious by a black vertical bar, the lower half is changed from a white vertical bar to black from left to right, but gradient change is not presented, the black is finally formed, the middle is obviously formed by the black vertical bar, and the left side of the black vertical bar is a dense white vertical bar. In fig. 9c, the upper half has 3 distinct white bars, but the upper half is not uniform, the lower half is changed from left to right from gradient to black by the white bars, and finally, the upper half is black, but the middle is distinct from the black bars.
The invention discloses a time-frequency double-bar code characteristic image generation method of an electrocardiosignal, which comprises the following steps: (1) An electrocardiograph signal x (t) (unit is millivolt; t is time and unit is second) obtained based on electrocardiograph equipment measurement is intercepted by a window function to obtain a complete electrocardiograph period sample, and the intercepted signal is normalized; (2) Empirical mode decomposition (empirical mode decomposition, EMD) is carried out on the normalized electrocardiosignals to obtain natural mode components (intrinsic mode function, IMF) from high frequency to low frequency, correlation coefficients of the IMFs and the original signals are calculated, and boundary points of the noise-dominant IMFs and the signal-dominant IMFs are determined by the correlation coefficients; (3) Filtering out noise-dominant IMF, reserving signal-dominant IMF to perform electrocardiosignal noise filtering, and performing fast Fourier transform on the noise-filtered electrocardiosignal to obtain a frequency domain signal of the noise-filtered electrocardiosignal; (4) Construction of time-domain bar code matrix H for filtered electrocardiosignals 1 Constructing a frequency domain bar code matrix H for the frequency domain signals of the electrocardiosignals after noise filtering 2 Matrix H of time domain bar code 1 And frequency domain bar code matrix H 2 Alignment is carried out, an electrocardiosignal characteristic matrix H is constructed, and the numerical value in H is linearly mapped to gray image pixel value [0, 255 ]]The output of the two-dimensional electrocardiosignals is an image, and the one-dimensional electrocardiosignals are converted into two-dimensional time-frequency double-bar code characteristic images. The invention can reliably remove the noise component in the signal and effectively reduce the interference of the noise component on the actual value of the electrocardiosignal. The useful IMF is screened through the correlation coefficient, the interference is reduced, and the double bar codes are constructed by using the time domain signals and the frequency domain signals of the electrocardiosignals after noise filtering, so that the time-frequency double bar code characteristic image generation method of the electrocardiosignals is realized.
In yet another aspect, the invention also discloses a computer readable storage medium storing a computer program which, when executed by a processor, causes the processor to perform the steps of any of the methods described above.
In yet another aspect, the invention also discloses a computer device comprising a memory and a processor, the memory storing a computer program which, when executed by the processor, causes the processor to perform the steps of any of the methods described above.
In yet another embodiment provided herein, there is also provided a computer program product containing instructions which, when run on a computer, cause the computer to perform the steps of any of the methods of the above embodiments.
It may be understood that the system provided by the embodiment of the present invention corresponds to the method provided by the embodiment of the present invention, and explanation, examples and beneficial effects of the related content may refer to corresponding parts in the above method.
Those skilled in the art will appreciate that all or part of the processes in the methods of the above embodiments may be implemented by a computer program for instructing relevant hardware, where the program may be stored in a non-volatile computer readable storage medium, and where the program, when executed, may include processes in the embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in the various embodiments provided herein may include non-volatile and/or volatile memory. The nonvolatile memory can include Read Only Memory (ROM), programmable ROM (PROM), electrically Programmable ROM (EPROM), electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double Data Rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous Link DRAM (SLDRAM), memory bus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), among others.
The technical features of the above embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.
The above embodiments are only for illustrating the technical solution of the present invention, and are not limiting; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims (5)
1. A time-frequency double-bar code characteristic image generation method of electrocardiosignals is characterized by comprising the following steps,
step A: for an electrocardiograph signal x (t) obtained based on electrocardiograph device measurement, the unit is millivolt; t is time, the unit is seconds, a window function is designed to intercept to obtain a complete electrocardio cycle sample, and the intercepted signal is normalized;
and (B) step (B): performing Empirical Mode Decomposition (EMD) on the normalized electrocardiosignal to obtain IMFs of natural mode components from high frequency to low frequency, calculating correlation coefficients of each IMF and an original signal, and determining boundary points of the noise-dominant IMF and the signal-dominant IMF by using the correlation coefficients;
step C: filtering out noise-dominant IMF, reserving signal-dominant IMF to perform electrocardiosignal noise filtering, and performing fast Fourier transform on the noise-filtered electrocardiosignal to obtain a frequency domain signal of the noise-filtered electrocardiosignal;
step D: construction of time-domain bar code matrix H for filtered electrocardiosignals 1 Constructing a frequency domain bar code matrix H for the frequency domain signals of the electrocardiosignals after noise filtering 2 Matrix H of time domain bar code 1 And frequency domain bar code matrix H 2 Alignment is carried out, an electrocardiosignal characteristic matrix H is constructed, and the numerical value in H is linearly mapped to gray image pixel value [0, 255 ]]The medium output is an image, so that one-dimensional electrocardiosignals are converted into two-dimensional time-frequency double-bar code characteristic images;
the specific steps of the step A are as follows:
s101: the electrocardiosignal x (t) obtained based on the measurement of the electrocardiograph equipment is expressed in the form of a sequence, namely an electrocardiosignal sequence x a (n) signal interception of its design window function w (n);
where n=0, 1,2,3 … is the sample size, n=t×f; w (N) is a window function, N 1 、N 2 The starting point and the end point of the window function are respectively;
then the intercepted electrocardiosignal sequence x b (n)＝x a (n) ×w (n) in units of: millivolts, x a (n) is an electrocardiosignal sequence, and the unit is: millivolts, x b (n) is a one-dimensional array sequence, the position value range of the elements in the array is [0, N]Sequence length n+1, n=n 2 -N 1 ；
S102: for the intercepted electrocardiosignal sequence x b (n) performing a normalization operation;
wherein x is b (n) is the intercepted electrocardiosignal sequence, x c (n) is normalized electrocardiosignal sequence, x min To the minimum amplitude value of the intercepted electrocardiosignal sequence, x max The maximum amplitude value of the intercepted electrocardiosignal sequence;
the step D specifically includes the following steps,
s107: the electrocardiosignal sequence x after noise filtering d (N) replication N c Parts and consist of N c The electrocardiosignal sequence x after noise filtering d (N) constructing a two-dimensional matrix to obtain an N c ×(N b +1) size matrix H 1 ；
Wherein N is b ＝N a ，N b +1 is the filtered electrocardiosignal sequence x d Data length of (n), H 1 The filtered electrocardiosignal bar code matrix is a time domain bar code matrix; n (N) c For a time-domain bar code matrix H 1 At the same time, the number of copies is also the number of the lines of the electrocardiosignal sequence x after noise filtering d Expanding the data of (n) to expand a one-dimensional matrix to a two-dimensional matrix;
s108: taking the frequency domain signal x e Front N of (N) e Data points, obtain frequency domain signal x f1 (n) and at the same time let x f2 (n)＝x f1 (n) x is f1 (n) and x f2 (n) cross-fusing to form a processed frequency domain signal x f (n)；
Wherein N is d +1 is the frequency domain signal x e (N) data length, N d Numerical size and N of (2) b The values of (a) and (b) are the same, x f1 (n)＝(x f1 (0)，x f1 (1)，…，x f1 (N e ) X) f2 (n)＝(x f2 (0)，x f2 (1)，…，x f2 (N e ))，
x f (n)＝(x f1 (0)，x f2 (0)，x f1 (1)，x f2 (1)…，x f1 (N e )，x f2 (N e ) Obtaining a processed frequency domain signal according to a cross fusion rule;
s109: the processed frequency domain signal x f (N) replication N c Parts, and is composed of N c The processed frequency domain signal x f (N) constructing a two-dimensional matrix to obtain an N c ×(N f +2) size matrix H 2 ；
Wherein N is f ＝2×N e ，N f +2 is the processed frequency domain signalx f Data length of (n), H 2 The processed frequency domain signal bar code matrix is a frequency domain bar code matrix;
s110: matrix H of time-domain bar codes 1 And frequency domain bar code matrix H 2 Alignment processing is carried out to obtain an electrocardiosignal characteristic matrix H, and the numerical value in H is linearly mapped to gray image pixel value [0, 255 ]]The output image is the time-frequency double-bar code characteristic image of the electrocardiosignal;
wherein H is an electrocardiosignal characteristic matrix, and the size of the matrix is 2N c ×(N b +1), in the process of H 1 And H 2 If H at the time of alignment 2 The number of columns of the matrix is less than H 1 Column number of matrix, then to H 2 Element supplementation is carried out on the matrix; the supplement rule is that the average value of the first two elements of the supplement element position is taken as the value of the supplement element, so that H 1 Matrix sum H 2 The columns of the matrix are equal, i.e. H after supplementation 2 The matrix size becomes N c ×(N b +1), and constructing an electrocardiosignal characteristic matrix H according to a formula (9);
2. the time-frequency dual-bar code feature image generating method of electrocardiographic signals according to claim 1, characterized in that: n in the step S101 1 、N 2 The selection rule of (1) is that the electrocardiosignal sequence x is used a Taking one R peak of (N) as a reference, taking 0.4 second data forward, setting electrocardiosignal sampling frequency f=360 Hz, thus taking 144=0.4×360 points forward, namely N 1 Data points at 144 before the R peak; at the same time, the data of the last 0.5 seconds of the 2 nd R peak of the backward number of the R peak is taken, so 180=0.5×360 points are taken backward, namely N 2 Is the 180 th data point after the 2 nd R peak, counted backward.
3. The time-frequency dual-bar code feature image generation method of an electrocardiograph signal according to claim 2, characterized in that: the step B is as follows: the EMD is carried out on the normalized electrocardiosignal to obtain IMFs of natural mode components from high frequency to low frequency, the correlation coefficient of each IMF and the original signal is calculated, and the boundary points of the noise-dominant IMF and the signal-dominant IMF are determined by the correlation coefficient, which comprises the following steps:
s103: for normalized electrocardiosignal sequence x c (N) performing EMD decomposition operation, setting maximum number of decomposed IMFs N IMF =7, and calculate each IMF and x c A correlation coefficient of (n);
wherein x is c (n) is normalized electrocardiosignal sequence, c i (n) is the ith IMF, ρ obtained by EMD decomposition i Represents x c (n) correlation coefficient with the ith IMF, μ x Sum sigma x Is x c Mean and standard deviation of (n), μ c Sum sigma c C is i Mean and standard deviation of (N), N a +1 is normalized electrocardiosignal sequence x c The data length of (N), where N a =n; whereas EMD decomposition includes decomposing the number of IMFs N IMF N can be obtained by EMD decomposition IMF IMFs and 1 residual component R (n);
s104: from the first maximum point of the curve of the set of correlation coefficients ρ, which corresponds to the element index K in the set f Boundary points for noise-dominant IMF and signal-dominant IMF;
wherein the set of correlation coefficients ρ= (ρ) 1 ，ρ 2 ，…，ρ m ) The number of set elements m=n IMF The method comprises the steps of carrying out a first treatment on the surface of the Noise dominated IMFThe definition of signal dominant IMF is, previously, K f The IMF of order is used as noise dominant IMF, K f The IMF after the order is used as a signal dominant IMF; if the curve of the correlation coefficient set rho has no maximum value, selecting the first 3-order IMF as a noise-dominant IMF, and selecting the IMFs after 3-order as signal-dominant IMFs.
4. The time-frequency dual-bar code feature image generating method of electrocardiographic signals according to claim 3, wherein: the step C: filtering noise-dominant IMF, reserving signal-dominant IMF to perform electrocardiosignal noise filtering, and performing fast Fourier transform on the noise-filtered electrocardiosignal to obtain a frequency domain signal of the noise-filtered electrocardiosignal, which specifically comprises the following steps:
s105: filtering out noise dominant IMF, retaining signal dominant IMF, and performing signal reconstruction together with residual component R (n) obtained by EMD decomposition;
wherein x is d (n) is the filtered electrocardiosignal sequence, c i (N) is the ith IMF obtained by EMD decomposition, M is the number N of IMFs obtained by decomposition IMF ，K f For the boundary points of the noise-dominant IMF and the signal-dominant IMF, R (n) is the remaining component of the EMD decomposition;
s106: the electrocardiosignal sequence x after noise filtering d (n) performing a fast Fourier transform to obtain x d Frequency domain signal x of (n) e (n)。
5. A computer readable storage medium storing a computer program which, when executed by a processor, causes the processor to perform the steps of the method of any one of claims 1 to 4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310151498.0A CN115881276B (en) | 2023-02-22 | 2023-02-22 | Time-frequency double-bar code characteristic image generation method of electrocardiosignal and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310151498.0A CN115881276B (en) | 2023-02-22 | 2023-02-22 | Time-frequency double-bar code characteristic image generation method of electrocardiosignal and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115881276A CN115881276A (en) | 2023-03-31 |
| CN115881276B true CN115881276B (en) | 2023-06-09 |
Family
ID=85761552
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310151498.0A Active CN115881276B (en) | 2023-02-22 | 2023-02-22 | Time-frequency double-bar code characteristic image generation method of electrocardiosignal and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115881276B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116712056B (en) * | 2023-08-07 | 2023-11-03 | 合肥工业大学 | Characteristic image generation and identification method, equipment and storage medium for electrocardiogram data |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9275013B2 (en) * | 2012-03-16 | 2016-03-01 | Qualcomm Incorporated | System and method for analysis and reconstruction of variable pulse-width signals having low sampling rates |
| US9918651B2 (en) * | 2014-09-02 | 2018-03-20 | The Regents Of The University Of Michigan | Electrocardiogram data analysis method for rapid diagnosis |
| CN110742599B (en) * | 2019-11-01 | 2022-05-10 | 广东工业大学 | Electrocardiosignal feature extraction and classification method and system |
| CN110680308B (en) * | 2019-11-04 | 2021-04-20 | 北京理工大学 | Electrocardiosignal denoising method based on fusion of improved EMD and threshold method |
| CN112906578B (en) * | 2021-02-23 | 2023-09-05 | 北京建筑大学 | Method for denoising bridge time sequence displacement signal |
| CN114176596A (en) * | 2021-12-28 | 2022-03-15 | 合肥工业大学 | Magnetocardiogram signal denoising method for improving empirical mode decomposition permutation entropy |
-
2023
- 2023-02-22 CN CN202310151498.0A patent/CN115881276B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN115881276A (en) | 2023-03-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Sameni et al. | A nonlinear Bayesian filtering framework for ECG denoising | |
| Sayadi et al. | A model-based Bayesian framework for ECG beat segmentation | |
| Kaur et al. | ECG signal analysis and arrhythmia detection using wavelet transform | |
| DE60029776T2 (en) | MONITORING DEVICE USING WAVELET TRANSFORMATIONS FOR CARDIAC RATIO ANALYSIS | |
| US11147516B2 (en) | Methods and systems to quantify and remove asynchronous noise in biophysical signals | |
| Patro et al. | Effective feature extraction of ECG for biometric application | |
| Lu et al. | Feature selection for ECG signal processing using improved genetic algorithm and empirical mode decomposition | |
| CN108113665B (en) | Automatic noise reduction method for electrocardiosignal | |
| Jeyarani et al. | Analysis of noise reduction techniques on QRS ECG waveform-by applying different filters | |
| CN115881276B (en) | Time-frequency double-bar code characteristic image generation method of electrocardiosignal and storage medium | |
| CN111920407B (en) | Electrocardiogram feature extraction method, system, device and medium based on wavelet transformation | |
| Kaur et al. | On the detection of cardiac arrhythmia with principal component analysis | |
| CN110693488A (en) | Electrocardiosignal processing system, electrocardiosignal processing method, electronic equipment and readable storage medium | |
| Khoma et al. | Outlier Correction in ECG-Based Human Identification | |
| JP2023120282A (en) | Biological signal processing device and data generation method for automatic analysis | |
| Lee et al. | Adaptive ecg signal compression method based on look-ahead linear approximation for ultra long-term operating of healthcare iot devices (sci) | |
| EP1129661A2 (en) | Device for processing body signals | |
| Hegde et al. | A review on ECG signal processing and HRV analysis | |
| CN116172572A (en) | SHAP value weighted sum segmentation HDBSCAN-based heart beat clustering method and system | |
| Nallikuzhy et al. | Spatially enhanced ECG using patient-specific dictionary learning | |
| Agrawal et al. | Comparative study of ECG feature extraction methods | |
| Khandait et al. | ECG signal processing using classifier to analyses cardiovascular disease | |
| CN106682630A (en) | Feature sampling method based on electrocardiographic signal | |
| Rajendran et al. | Classification of heart disease from ECG signals using Machine Learning | |
| Banta et al. | A novel convolutional neural network for reconstructing surface electrocardiograms from intracardiac electrograms and vice versa |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |