CN107595276A - A kind of atrial fibrillation detection method based on single lead electrocardiosignal time-frequency characteristics - Google Patents
A kind of atrial fibrillation detection method based on single lead electrocardiosignal time-frequency characteristics Download PDFInfo
- Publication number
- CN107595276A CN107595276A CN201710725679.4A CN201710725679A CN107595276A CN 107595276 A CN107595276 A CN 107595276A CN 201710725679 A CN201710725679 A CN 201710725679A CN 107595276 A CN107595276 A CN 107595276A
- Authority
- CN
- China
- Prior art keywords
- waveform
- window
- heartbeat
- electrocardiosignal
- mrow
- 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.)
- Granted
Links
- 206010003658 Atrial Fibrillation Diseases 0.000 title claims abstract description 29
- 238000001514 detection method Methods 0.000 title claims abstract description 16
- 238000000034 method Methods 0.000 claims abstract description 35
- 239000011159 matrix material Substances 0.000 claims description 12
- 238000012549 training Methods 0.000 claims description 11
- 230000008859 change Effects 0.000 claims description 5
- 230000008569 process Effects 0.000 claims description 5
- 238000012360 testing method Methods 0.000 claims description 5
- 108010076504 Protein Sorting Signals Proteins 0.000 claims description 3
- 230000002159 abnormal effect Effects 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000000354 decomposition reaction Methods 0.000 claims description 3
- 239000000203 mixture Substances 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 2
- 230000001746 atrial effect Effects 0.000 abstract description 6
- 230000004044 response Effects 0.000 abstract description 5
- 230000002861 ventricular Effects 0.000 abstract description 5
- 238000005516 engineering process Methods 0.000 abstract description 4
- 230000036541 health Effects 0.000 abstract description 3
- 239000000284 extract Substances 0.000 abstract description 2
- 230000010365 information processing Effects 0.000 abstract description 2
- 206010003119 arrhythmia Diseases 0.000 description 4
- 230000006793 arrhythmia Effects 0.000 description 4
- 201000010099 disease Diseases 0.000 description 3
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 3
- 238000000605 extraction Methods 0.000 description 3
- 238000007689 inspection Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 238000000513 principal component analysis Methods 0.000 description 2
- 230000033764 rhythmic process Effects 0.000 description 2
- 208000024891 symptom Diseases 0.000 description 2
- 206010008190 Cerebrovascular accident Diseases 0.000 description 1
- 206010019280 Heart failures Diseases 0.000 description 1
- 208000006011 Stroke Diseases 0.000 description 1
- 206010042434 Sudden death Diseases 0.000 description 1
- HCHKCACWOHOZIP-UHFFFAOYSA-N Zinc Chemical compound [Zn] HCHKCACWOHOZIP-UHFFFAOYSA-N 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 208000029078 coronary artery disease Diseases 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 238000012217 deletion Methods 0.000 description 1
- 230000037430 deletion Effects 0.000 description 1
- 230000009191 jumping Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000001314 paroxysmal effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Landscapes
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
A kind of atrial fibrillation detection method based on single lead electrocardiosignal time-frequency characteristics, belong to the crossing domain of the information processing technology and medical treatment ＆ health.The distinguishing feature of the present invention is aiming at single lead electrocardiosignal with higher noise, each heartbeat waveform inside same signal is simply clustered using Pearson came relative coefficient, and then the noise jamming of heartbeat waveform is eliminated using the method for average, and then find out the representative beat waveform of signal, it is then based on the time-frequency characteristics that Matching Pursuits algorithms extract representative waveform, the classification for whole piece electrocardiosignal.In addition, the present invention also incorporates the feature (time-frequency characteristics for combining HRV feature and heartbeat waveform) of phase between RR, the characteristics of thus having concurrently based on atrial activity and based on two methods of ventricular response, there is higher accuracy rate, good robustness.
Description
Technical field
The invention belongs to the information processing technology and the crossing domain of medical treatment ＆ health, is related to a kind of dividing for single lead electrocardiosignal
Class method, more particularly to a kind of atrial fibrillation detection method based on single lead electrocardiosignal time-frequency characteristics.
Background technology
Auricular fibrillation (abbreviation atrial fibrillation) is most common continuation arrhythmia cordis, the incidence of disease in population be 1~
2%.With the increase at age, the incidence of disease of atrial fibrillation is also continuously increased, and crowd is up to 10% within more than 75 years old.Atrial fibrillation can increase including
The onset risk of a variety of diseases such as sudden death, apoplexy, heart failure and coronary heart disease, serious threat is formed to the life and health of people.
Because part atrial fibrillation is paroxysmal, and may not any manifest symptom, therefore easily in early stage in early stage
It is ignored, so as to delay the state of an illness.Solution of the wearable ECG detecting equipment of single lead currently risen for the problem is brought
Opportunity.This kind equipment has the characteristics of small volume, cost is cheap, can be encapsulated into the mini-plants such as bracelet, is adapted to long-term
Wear, do not influence the daily routines of patient, be expected to realize a large amount of popularizations in future.However, a large amount of numbers that this kind equipment is gathered
According to being that manual analysis can not bear, it is necessary to analyzed and diagnosed by the technology of automation.In setting for automatic mode
In meter, it is necessary to consider the two major features of this kind of data：
(1) signal may have higher noise jamming；
(2) signal may relate to a variety of arrhythmia cordis symptoms.
The automatic identification method of atrial fibrillation is broadly divided into two classes at present：Method based on atrial activity and based on ventricular response
Method.It whether there is based on the method for atrial activity according to interim P ripples between TQ or F ripples and judged.For noise interference
Less signal, this kind of method can generally obtain higher precision；But for the higher signal of noise, this kind of method
Precision will have significant decline.Method based on ventricular response is mainly according to the phase (i.e. phase between RR) between heartbeat and heartbeat
Predictability judged.Because the phase is derived by the maximum R ripples of amplitude in electrocardiosignal between RR, it is for the anti-of noise
Interference performance is stronger.But by the information contained in the phase between RR is extremely limited, it is more with other for distinguishing atrial fibrillation
The ability of kind arrhythmia status is insufficient.
The content of the invention
It is low and be difficult to atrial fibrillation and other rhythms of the heart the invention aims to solve atrial fibrillation accuracy of detection in the prior art
The problem of not normal situation is distinguished, there is provided a kind of atrial fibrillation detection method based on single lead electrocardiosignal time-frequency characteristics, this method are whole
Phase feature and heartbeat waveform feature between RR are closed, the advantages of having concurrently based on atrial activity and based on two methods of ventricular response.
The present invention judges whether the main body of ratioing signal suffers signal header using the method based on the change of signal energy time domain
By stronger noise jamming, and then the head with strong noise is deleted, follow-up Pan-Tompkins can be significantly improved and calculated
The precision in method identification R ripples site.The distinguishing feature of the present invention aiming at single lead electrocardiosignal with higher noise,
Each heartbeat waveform inside same signal is simply clustered using Pearson came relative coefficient, and then uses the method for average
The noise jamming of heartbeat waveform is eliminated, and then finds out the representative beat waveform of signal, is then based on Matching Pursuits
Algorithm extracts the time-frequency characteristics of representative waveform, the classification for whole piece electrocardiosignal.In addition, the present invention also incorporates the phase between RR
Feature (time-frequency characteristics for combining HRV feature and heartbeat waveform), thus have concurrently based on atrial activity and based on the heart
The characteristics of two methods, is reacted in room, has higher accuracy rate, good robustness.Available for single lead with higher noise
The auxiliary detection of body surface ecg.
To achieve the above object, the technical scheme that the present invention takes is as follows：
A kind of atrial fibrillation detection method based on single lead electrocardiosignal time-frequency characteristics, methods described comprise the following steps：
Step 1：For single lead body surface ecg, pretreatment operation is first carried out, removes the baseline drift in electrocardiosignal
Move and partial noise is disturbed；
Step 2：Method based on sliding window calculates change of the signal energy in time domain, finds first energy and is not higher than
Thereafter the window of the median of each window energy, if the window is present, by the part corresponding to each window before it from letter
Removed in number, carry out step 3；If the window is not present, signal is considered as too noisy, so as to which testing process terminates；
Step 3：Using Pan-Tompkins algorithms, the R ripples site in electrocardiosignal is identified, on the basis of R ripples site,
Intercept each heartbeat waveform in electrocardiosignal；
Step 4：Calculus of differences is done to R ripples site, obtains the RR interval series in electrocardiosignal；
Step 5：Calculate the Pearson came relative coefficient of each heartbeat waveform between any two, Ran Hougen in same electrocardiosignal
Heartbeat waveform is grouped according to the power of correlation, the average waveform of the respective internal heartbeat of maximum two packets is calculated, with this
Representative waveform as the electrocardiosignal；
Step 6：For the representative waveform of each sample in training set or its subset, using Matching Pursuits
Algorithm is deconstructed based on the wavelet packet dictionaries of Symmlet 8 to the representative waveform of electrocardiosignal, and iterations is 30 times, statistics
The 1- norm average values of coefficient of each atom waveform in the decomposition of each electrocardiosignal in wavelet packet dictionary, and be ranked up with this,
Then the atom waveform for choosing ranking preceding 30 forms reference character dictionary；
Step 7：Build the characteristic vector of a sample, including phase standard deviation between phase average value, RR between RR, calculated using MP
Method deconstructs the coefficient and standardization remainder of gained based on the reference character dictionary that step 6 obtains to its signature waveform；
Step 8：The characteristic vector extracted based on above step, it is trained, is obtained on training set using KNN algorithms
To disaggregated model.
It is of the invention to be relative to the beneficial effect of prior art：The present invention is using the side based on the change of signal energy time domain
Method, judge whether the main body of ratioing signal is subject to stronger noise jamming to signal header, and then deletion had strong noise
Head, hence it is evident that improve the precision in follow-up Pan-Tompkins algorithms identification R ripples site.
The distinguishing feature of the present invention utilizes Pearson came phase aiming at single lead electrocardiosignal with higher noise
Close property coefficient simply to cluster each heartbeat waveform inside same signal, and then heartbeat waveform is eliminated using the method for average
Noise jamming, and then find out the representative beat waveform of signal, be then based on the extraction of Matching Pursuits algorithms and represent
Property waveform time-frequency characteristics, for the classification of whole piece electrocardiosignal, in addition, the present invention also incorporates the feature of phase between RR, thus
The characteristics of having concurrently based on atrial activity and based on two methods of ventricular response, there is higher accuracy rate, good robustness.
Brief description of the drawings
Fig. 1 is the flow chart of technical solution of the present invention；
Fig. 2 is all kinds of electrocardiosignal exemplary plots；
Fig. 3 is that the schematic diagram influenceed on Pan-Tompkins arithmetic results is deleted on strong noise head；
Fig. 4 is characterized the schematic diagram of vector composition.
Embodiment
Technical scheme is further described with reference to the accompanying drawings and examples, but is not limited thereto,
It is every technical solution of the present invention to be modified or equivalent substitution, without departing from the spirit and scope of technical solution of the present invention,
It all should cover in protection scope of the present invention.
Embodiment one：What present embodiment was recorded is a kind of atrial fibrillation based on single lead electrocardiosignal time-frequency characteristics
Detection method, original electro-cardiologic signals are divided into normal sinus rhythm, atrial fibrillation, other arrhythmia cordis or the class of too noisy four, it is described
Method comprises the following steps：
Step 1：For single lead body surface ecg (signal length is in 10~60 seconds sections), first pre-processed
Operation, remove baseline drift and partial noise interference in electrocardiosignal；
Step 2：Method based on sliding window calculates change of the signal energy in time domain, finds first energy and is not higher than
Thereafter the window of the median of each window energy, if the window is present, by the part corresponding to each window before it from letter
Removed in number, carry out step 3；If the window is not present, signal is considered as too noisy, so as to which testing process terminates；
Step 3：Using Pan-Tompkins algorithms, the R ripples site in electrocardiosignal is identified, on the basis of R ripples site,
Intercept each heartbeat waveform in electrocardiosignal；
Step 4：Calculus of differences is done to R ripples site, obtains the RR interval series in electrocardiosignal；
Step 5：Calculate the Pearson came relative coefficient of each heartbeat waveform between any two, Ran Hougen in same electrocardiosignal
Heartbeat waveform is grouped according to the power of correlation, the average waveform of the respective internal heartbeat of maximum two packets is calculated, with this
Representative waveform as the electrocardiosignal；
Step 6：For the representative waveform of each sample in training set or its subset, using Matching Pursuits
(MP) algorithm is deconstructed based on the wavelet packet dictionaries of Symmlet 8 to the representative waveform of electrocardiosignal, and iterations is 30 times,
The 1- norm average values of coefficient of each atom waveform in the decomposition of each electrocardiosignal in wavelet packet dictionary are counted, and are arranged with this
Sequence, the atom waveform for then choosing ranking preceding 30 form reference character dictionary；
Step 7：Build the characteristic vector of a sample, including phase standard deviation between phase average value, RR between RR, calculated using MP
Method deconstructs the coefficient and standardization remainder of gained based on the reference character dictionary that step 6 obtains to its signature waveform；
Step 8：The characteristic vector extracted based on above step, it is trained, is obtained on training set using KNN algorithms
To disaggregated model.
Embodiment two：The inspection of the atrial fibrillation based on single lead electrocardiosignal time-frequency characteristics described in embodiment one
Survey method, the step 2 comprise the following steps that：
(1) electrocardiosignal is divided into a series of window, each length of window is 1.8 seconds, is had between adjacent window apertures 0.9 second
Overlapping region；
(2) energy of signal in each window is calculated, its formula is as follows：
Wherein, x is the discrete signal sequence in window, and x [n] represents the value of n-th of sampled point in window, when T is window
Long, N represents the quantity of sampled point in window；
(3) since first window, by its energy compared with the median of each window energy in the signal, until
Find middle position energy of the window amount of enabling it to not higher than each window thereafter；If serial number h of the window in each window,
Then：
H=min i | i ∈ { 1,2 ..., N-1 } ∧ Ei≤median{Ei+1..., EN,
Wherein, N represents window sum, EiWindow i energy is represented, for median to ask median computing, min is to ask minimum
It is worth computing；
(4) if window h is present, by the part corresponding to each window (1,2 ..., h-1) before it from signal
Remove；If window h is not present, signal is considered as too noisy, so as to which testing process terminates.
Embodiment three：The inspection of the atrial fibrillation based on single lead electrocardiosignal time-frequency characteristics described in embodiment one
Survey method, the step 5 comprise the following steps that：
(1) similarity lower limit in definition group, i.e., the minimum Pearson came relative coefficient of heartbeat waveform, value 0.8 in group
Left and right (optical signal noise intensity is adjusted)；
(2) the Pearson came relative coefficient in same signal between each heartbeat waveform is calculated, formula is as follows
Wherein, X and Y represents the sampling point sequence corresponding to two heartbeat waveforms respectively, and E represents mathematic expectaion, μXRepresent X
In each sampled point average value, μYRepresent the average value of each sampled point in Y, σXRepresent the standard deviation of each sampled point in X, σYRepresent Y
In each sampled point standard deviation, so as to obtain a correlation matrix, the length of matrix and wide consistent with the quantity of heartbeat；
(3) first heartbeat waveform not being grouped is found successively, if the heartbeat not being grouped, performs the
(5) step, if the heartbeat not being grouped, then (4) step is performed；
(4) all numerical value are more than similarity lower limit in group in finding the heartbeat and being expert in Pearson came correlation matrix
Element, corresponding heartbeat waveform is subdivided into a new packet, and by current heartbeat in matrix where row and column numerical value
Reset, and return to (3) step；
(5) choose comprising two most packets of heartbeat, waveform of being averaged respectively to each heartbeat waveform of its inside will
Representative waveform of the average waveform of acquisition as the packet, the representative waveform of the two packets is using as the representative of the signal
Property waveform；If packet only one, by its representative two parts of waveform copy, the representative waveform as the signal.
Embodiment four：The room based on single lead electrocardiosignal time-frequency characteristics described in embodiment one or three
Quiver detection method, the step 6 comprises the following steps that：
(1) a subset of training set is chosen, subset should include normal signal, atrial fibrillation signal and other abnormal signals at least
Each 100；
(2) specific steps according to step 5, the representative beat waveform per bars is calculated；
(3) the Matching Pursuits atom ripple dictionaries based on 5 layers of wavelet packets of Symlet 8 are built；
(4) the atom ripple dictionary of structure in (3) is based on to every in above-mentioned subset using Matching Pursuits algorithms
The representative beat waveform of signal is deconstructed, and iterations is 30 times；
(5) the 1- norm average values of each atom ripple coefficient in subset signals destructing are counted, its coefficient 1- norms is chosen and puts down
30 atom ripples of average highest form reference character dictionary.
Embodiment five：The inspection of the atrial fibrillation based on single lead electrocardiosignal time-frequency characteristics described in embodiment one
Survey method, in step 7, the composition of the characteristic vector：
(a) between RR the phase average value；
(b) between RR the phase standard deviation；
(c) coefficient and standardization remainder of gained are deconstructed to its signature waveform based on reference character dictionary using MP algorithms, wherein
The calculation formula for standardizing remainder is as follows：
β=| | r | |2/||x||2,
R represents remainder vector, and x represents the heartbeat waveform vector deconstructed.
Embodiment 1：
A kind of atrial fibrillation detection method based on single lead electrocardiosignal time-frequency characteristics proposed by the present invention, its main flow is such as
Shown in Fig. 1, the extraction step of phase and signal representativeness waveform is as follows between wherein RR：
(1) single lead electrocardiosignal (as shown in Figure 2) is directed to, baseline drift is first removed, is disappeared using bandpass filter
Except pretreatment operations such as noises, to reduce influence of the noise to signal waveform.
(2) electrocardiosignal is divided into a series of window, each length of window is 1.8 seconds, is had between adjacent window apertures 0.9 second
Overlapping region, and calculate the energy of each window, its formula is as follows：
Wherein, x is the discrete signal sequence in window, and T is window duration.
(3) since first window, by its energy compared with the median of each window energy in the signal, until
Find middle position energy of the window amount of enabling it to not higher than each window thereafter；If serial number h of the window in each window,
Then：
H=min i | i ∈ { 1,2 ..., N-1 } ∧ Ei≤median{Ei+1..., EN,
Wherein, N represents window sum, EiWindow i energy is represented, for median to ask median computing, min is to ask minimum
It is worth computing.
(4) signal section before the window that will be found in (3) is deleted from primary signal.The step can be lifted subsequently
The precision in Pan-Tompkins algorithms identification R ripples site, as shown in Figure 3.
(5) Pan-Tompkins algorithms are used, identify the R ripples site in electrocardiosignal, on the basis of R ripples site, interception
Each heartbeat waveform in signal, heartbeat waveform centered on R ripples site, total time span be 0.6 second, due to sample frequency may
Difference, the quantity for the sampled point that heartbeat waveform is included also can be accordingly different.
(6) calculus of differences is done to R ripples site, obtains the RR interval series in signal.
(7) the Pearson came relative coefficient in same signal between each heartbeat waveform is calculated, obtains a correlation matrix,
The length and width of matrix are consistent with the quantity of heartbeat.
(8) first heartbeat waveform not being grouped is found successively；If there is ungrouped heartbeat waveform, then sequentially hold
Row is in next step；If the heartbeat not being grouped, jumps to (10) step.
(9) find during the heartbeat is expert in similarity matrix all numerical value be more than setting critical value (such as 0.8, can
Be adjusted depending on signal noise intensity) element, corresponding heartbeat waveform is subdivided into one.Individual new packet, and front center will be worked as
The numerical value of row and column is reset where jumping in matrix, returns to (8) step.
(10) choose comprising two most packets of heartbeat, each heartbeat waveform of its inside is averaged respectively, will be obtained
Representative waveform of the average waveform obtained as the packet, the representative waveform of the two packets is using as the representativeness of the signal
Waveform；If packet only one, by its representative two parts of waveform copy, the representative waveform as the signal.
The construction method of Matching Pursuits atom ripple reference character dictionaries is as follows：
(1) a subset of training set (for example, the databases of PhysioNet/Cinc Challenge 2017), son are chosen
Collection should be at least each 100 comprising normal signal, atrial fibrillation signal and other abnormal signals, and the length per bars was in 10~60 seconds areas
In, calculate the representative beat waveform per bars.
(2) the Matching Pursuits atom ripple dictionaries based on 5 layers of wavelet packets of Symlet 8 are utilized, are then utilized
Atom ripple dictionary of the Matching Pursuits algorithms based on structure is to the representative beat waveform in above-mentioned subset per bars
Deconstructed, iterations is 30 times.
(3) the 1- norm average values of each atom ripple coefficient in subset signals destructing are counted, its coefficient 1- norms is chosen and puts down
30 atom ripples of average highest form reference character dictionary.
The extraction step of signal characteristic is as follows：
(1) phase average value between the RR of calculating signal；
(2) phase standard deviation between the RR of calculating signal；
(3) coefficient and standardization remainder of gained are deconstructed to its signature waveform based on reference character dictionary using MP algorithms, wherein
The calculation formula for standardizing remainder is as follows：
β=| | r | |2/||x||2,
R represents remainder vector, and x represents the heartbeat waveform vector deconstructed.
The signal characteristic vector finally obtained is as shown in Figure 4.
In the model training stage, can generally have higher precision using KNN graders, while the model has instruction
It is short to practice the time, the characteristics of being easy to the extension of training set and constantly update.In addition, can be with by being superimposed principal component analysis (PCA)
Avoid producing over-fitting, the versatility of lift scheme.
Claims (5)
- A kind of 1. atrial fibrillation detection method based on single lead electrocardiosignal time-frequency characteristics, it is characterised in that：Methods described is included such as Lower step：Step 1：For single lead body surface ecg, first carry out pretreatment operation, remove baseline drift in electrocardiosignal and Partial noise is disturbed；Step 2：Method based on sliding window calculates change of the signal energy in time domain, finds first energy not higher than thereafter The window of the median of each window energy, if the window is present, by the part corresponding to each window before it from signal Remove, carry out step 3；If the window is not present, signal is considered as too noisy, so as to which testing process terminates；Step 3：Using Pan-Tompkins algorithms, the R ripples site in electrocardiosignal is identified, on the basis of R ripples site, interception Each heartbeat waveform in electrocardiosignal；Step 4：Calculus of differences is done to R ripples site, obtains the RR interval series in electrocardiosignal；Step 5：The Pearson came relative coefficient of each heartbeat waveform between any two in same electrocardiosignal is calculated, then according to phase The power of closing property is grouped to heartbeat waveform, calculates maximum two average waveforms for being grouped each internal heartbeat, in this, as The representative waveform of the electrocardiosignal；Step 6：For the representative waveform of each sample in training set or its subset, using Matching Pursuits algorithms The representative waveform of electrocardiosignal is deconstructed based on the wavelet packet dictionaries of Symmlet 8, iterations is 30 times, counts small echo The 1- norm average values of coefficient of each atom waveform in the decomposition of each electrocardiosignal in bag dictionary, and be ranked up with this, then The atom waveform that ranking is chosen preceding 30 forms reference character dictionary；Step 7：Build the characteristic vector of a sample, including phase standard deviation between phase average value, RR between RR, utilize MP algorithm bases The coefficient and standardization remainder of gained are deconstructed to its signature waveform in the reference character dictionary that step 6 obtains；Step 8：Based on the characteristic vector constructed by above step, it is trained, is divided on training set using KNN algorithms Class model.
- 2. the atrial fibrillation detection method according to claim 1 based on single lead electrocardiosignal time-frequency characteristics, it is characterised in that： The step 2 comprises the following steps that：(1) electrocardiosignal is divided into a series of window, each length of window is 1.8 seconds, the weight for having 0.9 second between adjacent window apertures Folded region；(2) energy of signal in each window is calculated, its formula is as follows：<mrow> <mi>E</mi> <mo>=</mo> <msubsup> <mi>T&Sigma;</mi> <mrow> <mi>n</mi> <mo>=</mo> <mn>0</mn> </mrow> <mrow> <mi>N</mi> <mo>-</mo> <mn>1</mn> </mrow> </msubsup> <msup> <mi>x</mi> <mn>2</mn> </msup> <mo>&lsqb;</mo> <mi>n</mi> <mo>&rsqb;</mo> <mo>,</mo> </mrow>Wherein, x is the discrete signal sequence in window, and x [n] represents the value of n-th of sampled point in window, and T is window duration, N Represent the quantity of sampled point in window；(3) since first window, by its energy compared with the median of each window energy in the signal, until finding Middle position energy of one window amount of enabling it to not higher than each window thereafter；If serial number h of the window in each window, then：H=min i | i ∈ { 1,2 ..., N-1 } ∧ Ei≤median{Ei+1..., EN,Wherein, N represents window sum, EiWindow i energy is represented, for median to ask median computing, min is fortune of minimizing Calculate；(4) if window h is present, the part corresponding to each window (1,2 ..., h-1) before it is removed from signal；If Window h is not present, then signal is considered as too noisy, so as to which testing process terminates.
- 3. the atrial fibrillation detection method according to claim 1 based on single lead electrocardiosignal time-frequency characteristics, it is characterised in that： The step 5 comprises the following steps that：(1) similarity lower limit in definition group, i.e., the minimum Pearson came relative coefficient of heartbeat waveform in group, value are 0.8 or so；(2) the Pearson came relative coefficient in same signal between each heartbeat waveform is calculated, formula is as follows<mrow> <msub> <mi>&rho;</mi> <mrow> <mi>X</mi> <mo>,</mo> <mi>Y</mi> </mrow> </msub> <mo>=</mo> <mfrac> <mrow> <mi>E</mi> <mo>&lsqb;</mo> <mrow> <mo>(</mo> <mi>X</mi> <mo>-</mo> <msub> <mi>&mu;</mi> <mi>X</mi> </msub> <mo>)</mo> </mrow> <mrow> <mo>(</mo> <mi>Y</mi> <mo>-</mo> <msub> <mi>&mu;</mi> <mi>Y</mi> </msub> <mo>)</mo> </mrow> <mo>&rsqb;</mo> </mrow> <mrow> <msub> <mi>&sigma;</mi> <mi>X</mi> </msub> <msub> <mi>&sigma;</mi> <mi>Y</mi> </msub> </mrow> </mfrac> </mrow>Wherein, X and Y represents the sampling point sequence corresponding to two heartbeat waveforms respectively, and E represents mathematic expectaion, μXRepresent each in X The average value of sampled point, μYRepresent the average value of each sampled point in Y, σXRepresent the standard deviation of each sampled point in X, σYRepresent each in Y The standard deviation of sampled point, so as to obtain a correlation matrix, the length and width of matrix are consistent with the quantity of heartbeat；(3) first heartbeat waveform not being grouped is found successively, if the heartbeat not being grouped, performs (5) step, If the heartbeat not being grouped, then (4) step is performed；(4) all numerical value are more than the member of similarity lower limit in group in finding the heartbeat and being expert in Pearson came correlation matrix Element, corresponding heartbeat waveform is subdivided into a new packet, and the numerical value of current heartbeat place row and column in matrix is clear Zero, and return to (3) step；(5) choose comprising two most packets of heartbeat, waveform of being averaged respectively to each heartbeat waveform of its inside, will obtain Representative waveform of the average waveform as the packet, the representative waveform of the two packets is using as the representative ripple of the signal Shape；If packet only one, by its representative two parts of waveform copy, the representative waveform as the signal.
- 4. the atrial fibrillation detection method based on single lead electrocardiosignal time-frequency characteristics according to claim 1 or 3, its feature exist In：The step 6 comprises the following steps that：(1) a subset of training set is chosen, it is at least each that subset should include normal signal, atrial fibrillation signal and other abnormal signals 100；(2) specific steps according to step 5, the representative beat waveform per bars is calculated；(3) the Matching Pursuits atom ripple dictionaries based on 5 layers of wavelet packets of Symlet 8 are built；(4) the atom ripple dictionary of structure in (3) is based on using Matching Pursuits algorithms to every bars in above-mentioned subset Representative beat waveform deconstructed, iterations be 30 times；(5) the 1- norm average values of each atom ripple coefficient in subset signals destructing are counted, choose its coefficient 1- norm average values 30 atom ripples of highest form reference character dictionary.
- 5. the atrial fibrillation detection method according to claim 1 based on single lead electrocardiosignal time-frequency characteristics, it is characterised in that： In step 7, the composition of the characteristic vector：(a) between RR the phase average value；(b) between RR the phase standard deviation；(c) coefficient and standardization remainder of gained, its Plays are deconstructed to its signature waveform based on reference character dictionary using MP algorithms The calculation formula for changing remainder is as follows：β=| | r | |2/||x||2,R represents remainder vector, and x represents the heartbeat waveform vector deconstructed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710725679.4A CN107595276B (en) | 2017-08-22 | 2017-08-22 | Atrial fibrillation detection method based on single-lead electrocardiosignal time-frequency characteristics |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710725679.4A CN107595276B (en) | 2017-08-22 | 2017-08-22 | Atrial fibrillation detection method based on single-lead electrocardiosignal time-frequency characteristics |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107595276A true CN107595276A (en) | 2018-01-19 |
| CN107595276B CN107595276B (en) | 2020-06-05 |
Family
ID=61065514
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710725679.4A Expired - Fee Related CN107595276B (en) | 2017-08-22 | 2017-08-22 | Atrial fibrillation detection method based on single-lead electrocardiosignal time-frequency characteristics |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107595276B (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108968941A (en) * | 2018-05-25 | 2018-12-11 | 深圳市太空科技南方研究院 | A kind of arrhythmia detection method, apparatus and terminal |
| CN109124620A (en) * | 2018-06-07 | 2019-01-04 | 深圳市太空科技南方研究院 | A kind of atrial fibrillation detection method, device and equipment |
| CN109171712A (en) * | 2018-09-28 | 2019-01-11 | 东软集团股份有限公司 | Auricular fibrillation recognition methods, device, equipment and computer readable storage medium |
| CN109303560A (en) * | 2018-11-01 | 2019-02-05 | 杭州质子科技有限公司 | A kind of atrial fibrillation recognition methods of electrocardiosignal in short-term based on convolution residual error network and transfer learning |
| CN109864736A (en) * | 2019-03-22 | 2019-06-11 | 深圳市理邦精密仪器股份有限公司 | Processing method, device, terminal device and the medium of electrocardiosignal |
| CN110575152A (en) * | 2018-06-08 | 2019-12-17 | 原相科技股份有限公司 | Atrial fibrillation detection device and operation method thereof |
| CN111096740A (en) * | 2018-10-25 | 2020-05-05 | 上海微创电生理医疗科技股份有限公司 | Electrocardiosignal analysis method and device, signal recorder and three-dimensional mapping system |
| CN112244861A (en) * | 2020-10-09 | 2021-01-22 | 广东工业大学 | Single-lead electrocardiosignal f-wave extraction method |
| CN115137316A (en) * | 2022-06-10 | 2022-10-04 | 佳禾智能科技股份有限公司 | Multi-physiological-parameter monitoring watch and monitoring method thereof |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103479349A (en) * | 2013-09-25 | 2014-01-01 | 深圳市理邦精密仪器股份有限公司 | Electrocardiosignal data acquisition and processing method and system |
| CN106073755A (en) * | 2016-05-27 | 2016-11-09 | 成都信汇聚源科技有限公司 | The implementation method that in a kind of miniature holter devices, atrial fibrillation identifies automatically |
-
2017
- 2017-08-22 CN CN201710725679.4A patent/CN107595276B/en not_active Expired - Fee Related
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103479349A (en) * | 2013-09-25 | 2014-01-01 | 深圳市理邦精密仪器股份有限公司 | Electrocardiosignal data acquisition and processing method and system |
| CN106073755A (en) * | 2016-05-27 | 2016-11-09 | 成都信汇聚源科技有限公司 | The implementation method that in a kind of miniature holter devices, atrial fibrillation identifies automatically |
Non-Patent Citations (1)
| Title |
|---|
| IVAYLO CHRISTOV 等: "Comparative study of morphological and time-frequency ECG descriptors for heartbeat classification", 《MEDICAL ENGINEERING & PHYSICS》 * |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108968941A (en) * | 2018-05-25 | 2018-12-11 | 深圳市太空科技南方研究院 | A kind of arrhythmia detection method, apparatus and terminal |
| CN108968941B (en) * | 2018-05-25 | 2022-03-25 | 深圳市太空科技南方研究院 | Arrhythmia detection method, device and terminal |
| CN109124620A (en) * | 2018-06-07 | 2019-01-04 | 深圳市太空科技南方研究院 | A kind of atrial fibrillation detection method, device and equipment |
| CN110575152A (en) * | 2018-06-08 | 2019-12-17 | 原相科技股份有限公司 | Atrial fibrillation detection device and operation method thereof |
| CN109171712A (en) * | 2018-09-28 | 2019-01-11 | 东软集团股份有限公司 | Auricular fibrillation recognition methods, device, equipment and computer readable storage medium |
| CN109171712B (en) * | 2018-09-28 | 2022-03-08 | 东软集团股份有限公司 | Atrial fibrillation identification method, atrial fibrillation identification device, atrial fibrillation identification equipment and computer readable storage medium |
| CN111096740A (en) * | 2018-10-25 | 2020-05-05 | 上海微创电生理医疗科技股份有限公司 | Electrocardiosignal analysis method and device, signal recorder and three-dimensional mapping system |
| CN111096740B (en) * | 2018-10-25 | 2022-04-01 | 上海微创电生理医疗科技股份有限公司 | Electrocardiosignal analysis method and device, signal recorder and three-dimensional mapping system |
| CN109303560A (en) * | 2018-11-01 | 2019-02-05 | 杭州质子科技有限公司 | A kind of atrial fibrillation recognition methods of electrocardiosignal in short-term based on convolution residual error network and transfer learning |
| CN109864736A (en) * | 2019-03-22 | 2019-06-11 | 深圳市理邦精密仪器股份有限公司 | Processing method, device, terminal device and the medium of electrocardiosignal |
| CN112244861A (en) * | 2020-10-09 | 2021-01-22 | 广东工业大学 | Single-lead electrocardiosignal f-wave extraction method |
| CN112244861B (en) * | 2020-10-09 | 2021-08-10 | 广东工业大学 | Single-lead electrocardiosignal f-wave extraction method |
| CN115137316A (en) * | 2022-06-10 | 2022-10-04 | 佳禾智能科技股份有限公司 | Multi-physiological-parameter monitoring watch and monitoring method thereof |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107595276B (en) | 2020-06-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107595276A (en) | A kind of atrial fibrillation detection method based on single lead electrocardiosignal time-frequency characteristics | |
| CN109171712B (en) | Atrial fibrillation identification method, atrial fibrillation identification device, atrial fibrillation identification equipment and computer readable storage medium | |
| CN107951485B (en) | Ambulatory ECG analysis method and apparatus based on artificial intelligence self study | |
| CN108714026B (en) | Fine-grained electrocardiosignal classification method based on deep convolutional neural network and online decision fusion | |
| De Chazal et al. | Automatic classification of heartbeats using ECG morphology and heartbeat interval features | |
| De Chazal et al. | A patient-adapting heartbeat classifier using ECG morphology and heartbeat interval features | |
| Ince et al. | A generic and robust system for automated patient-specific classification of ECG signals | |
| Anuradha et al. | CARDIAC ARRHYTHMIA CLASSIFICATION USING FUZZY CLASSIFIERS. | |
| Alquran et al. | ECG classification using higher order spectral estimation and deep learning techniques | |
| Ye et al. | Combining general multi-class and specific two-class classifiers for improved customized ECG heartbeat classification | |
| de Chazal et al. | Automatic classification of ECG beats using waveform shape and heart beat interval features | |
| Debnath et al. | Analysis of ECG signal and classification of heart abnormalities using Artificial Neural Network | |
| Dhyani et al. | Analysis of ECG-based arrhythmia detection system using machine learning | |
| Murugappan et al. | Time domain features based sudden cardiac arrest prediction using machine learning algorithms | |
| CN115486855A (en) | Electrocardiogram heart beat classification method based on QRS wave group indefinite cycle superposition | |
| Ebrahimzadeh et al. | Classification of ECG signals using Hermite functions and MLP neural networks | |
| Qu et al. | ECG signal classification based on BPNN | |
| Rao et al. | Performance identification of different heart diseases based on neural network classification | |
| Jiang et al. | Heartbeat classification system based on modified stacked denoising autoencoders and neural networks | |
| Sathawane et al. | Inception based GAN for ECG arrhythmia classification | |
| Mohseni et al. | Heart arrhythmias classification via a sequential classifier using neural network, principal component analysis and heart rate variation | |
| Raghav et al. | Fractal feature based ECG arrhythmia classification | |
| Nahak et al. | Ensembled Feature based Multi-Label ECG Arrhythmia Classification | |
| CN110755065A (en) | Electrocardiosignal classification method and system based on PSO-ELM algorithm | |
| Elamin et al. | Wavelet-based ECG signal analysis for human recognition |
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 | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20201112 Address after: Room 1404, No. 2-C, Wenhua West Road, Huancui District, Weihai City, Shandong Province Patentee after: Xia Yong Address before: High tech Zone Huidalu Pukou District of Nanjing City, Jiangsu province 210000 No. 9 State Power South Industrial Park Rd East Building 5 floor 517-7 room Patentee before: NANJING YIHA TECHNOLOGY Co.,Ltd. |
|
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20200605 |