CN105997055A - Automatic classification method, system and device of electrocardiosignal ST band - Google Patents
Automatic classification method, system and device of electrocardiosignal ST band Download PDFInfo
- Publication number
- CN105997055A CN105997055A CN201610541616.9A CN201610541616A CN105997055A CN 105997055 A CN105997055 A CN 105997055A CN 201610541616 A CN201610541616 A CN 201610541616A CN 105997055 A CN105997055 A CN 105997055A
- Authority
- CN
- China
- Prior art keywords
- electrocardiosignal
- wave
- section
- ripple
- automatic classification
- 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.)
- Pending
Links
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
- A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
- A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
- A61B5/346—Analysis of electrocardiograms
- A61B5/349—Detecting specific parameters of the electrocardiograph cycle
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/7253—Details of waveform analysis characterised by using transforms
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
Abstract
The invention discloses an automatic classification method of an electrocardiosignal ST band. The automatic classification method is characterized by comprising the following steps: S1. acquiring an electrocardiosignal wave form of a human body and pretreating the electrocardiosignal wave form; S2. performing characteristic point detection to the pretreated electrocardiosignal wave form; S3. based on the characteristic point detection in the step S2, determining the wave form of the electrocardiosignal ST band, and acquiring the characteristic parameters on the wave form of the electrocardiosignal ST band so as to establish a to-be-classified characteristic input matrix; S4. classifying the wave form of the electrocardiosignal ST band into a training sample and a testing sample, and establishing a classifier model based on the training sample; and S5. inputting the testing sample into the classifier model for testing, and completing the final classification by combining decision fusion. The invention further discloses an automatic classification system and device of the electrocardiosignal ST band. By establishing the classifier model and decision fusion by using a nerve network method, calculation can be effectively reduced, the time cost can be decreased, the classification precision of the ST band can be improved, and the classification is easier.
Description
Technical field
The present invention relates to medical signals process field, in particular it relates to the side of classification automatically of a kind of electrocardiosignal ST section
Method, system and device.
Background technology
In recent years, quickly grow for Electrocardiographic auxiliary diagnostic device, along with the scientific and technological progress of message area, in particular with
The progress of mode identification technology, the function of ecg equipment is no longer only to obtain electrocardiosignal, printing electrocardiogram, but towards
Excavate the valid data in electrocardiogram and identification, tagsort direction are developed automatically.Analytical equipment energy with tagsort function
Enough provide effective ECG information more directly perceived for doctor, effectively save Diagnostic Time, improve the diagnosis efficiency of doctor, be important
One of auxiliary medical equipment.
Work automatic classification system on the computing device is the core of this kind equipment, and technological approaches is by extracting sign electrocardiogram
The characteristic vector of effective information, is input to classifier algorithm and obtains the classification of ecg characteristics.Electro-cardiologic signal waveforms mainly includes QRS
Wave group, P ripple and T ripple, wherein QRS complex is mainly made up of Q ripple, R ripple and three ripples of S ripple.Current electrocardio
Figure tagsort field often point on the basis of the R ripple of location, then extracts QRS complex morphological feature, and extracts its parameter work
It is characterized.The further feature constitutive characteristic vector that this morphological feature is aided with on electrocardiogram is input to grader, the most defeated
Go out classification results.
But, this kind of from aroused in interest bat identification system along with characteristic vector dimension raise, grader operand increases rapidly so that
Generally there are dimension catastrophic phenomena in grader.ST section is the waveform in electrocardiosignal between S ripple and T ripple, due to the amplitude of ST section
Less, frequency is relatively low, and the dimension as feature can be greatly lowered.Meanwhile, electrocardiosignal ST section is the weight of ecg wave form
Wanting ingredient, the generation of many heart diseases is often attended by the change of ST section waveform.Therefore, detection and location ST section exactly,
And this section of waveform is carried out to extraction and the analysis of characteristic parameter, to diagnosing, corresponding heart disease is significant.
Summary of the invention
The present invention solves the problem that electrocardiogram form feature learning rule difficulty is limited with class condition, it is provided that a kind of electrocardio letter
Number automatic classification method based on ST section, system and device, by extracting three kinds of spies of the ST section that amplitude is less in electrocardiosignal
The parameter of reference breath, as input, is set up sorter model, and classification results is carried out Decision fusion.It is achieved thereby that study rule
Then simple, automatic classification method, system and the device of the electrocardiosignal ST section feature that operand is little.
According to an aspect of the invention, it is provided the automatic classification method of a kind of electrocardiosignal ST section, it is characterised in that bag
Include step:
S1, the electro-cardiologic signal waveforms of acquisition human body, and carry out pretreatment；
S2, carry out feature point detection to through pretreated electro-cardiologic signal waveforms；
S3, based on the feature point detection in step S2, determine electrocardiosignal ST section waveform, and obtain described electrocardiosignal ST
The characteristic parameter of section waveform, to set up feature input matrix to be sorted；
S4, electrocardiosignal ST section waveform is divided it is set to training sample and test sample, and set up point based on training sample
Class device model；
S5, test sample is inputted in sorter model and tests, and combine Decision fusion and complete finally to classify.
Preferably, described pretreatment include method based on wavelet transformation remove the baseline drift of low frequency and the Hz noise of high frequency and
Dynamo-electric interference, including step:
S11, use Quadric Spline small echo carry out 8 layers of decomposition to electro-cardiologic signal waveforms；
S12, by the high-frequency wavelet coefficient zero setting of the 2nd layer, to remove the Hz noise of high frequency and dynamo-electric interference；
S13, by the low-frequency wavelet coefficients zero setting of the 8th layer, to remove the baseline drift of low frequency, and carry out wavelet inverse transformation reconstruct with
Obtain pure electro-cardiologic signal waveforms.
Preferably, described feature point detection includes the detection of R wave-wave peak position, the detection of S wave-wave peak position, T wave-wave peak position, T
Ripple start position and the detection of T ripple final position.
Preferably, described R wave-wave peak position is detected and is included step:
S211, discrete wavelet transformation will be carried out through pretreated electrocardiosignal, according to modulus maximum principle, detection labelling
The zero crossing position of modulus maximum pair；
S212, near the zero crossing position of modulus maximum pair, search for the maximum of points of amplitude absolute value, if the single order of described point is left
Right difference value is stationary value, then described point is set to R wave-wave peak position；
S213, spacing average Ra calculated between adjacent two R wave-wave peaks, in the R wave-wave peak detected, if existing two
It is smaller than 0.4Ra between individual R wave-wave peak, then it is determined that the presence of flase drop, and increase the threshold value of R ripple；If there are two R ripples
Spacing between crest is more than 1.5Ra, then it is determined that the presence of missing inspection, and reduce the threshold value of R ripple；
S214, again carry out modulus maximum search, to detect R wave-wave peak position.
Preferably, described S wave-wave peak position is detected and is included step:
S221, it is set to starting point with detected R wave-wave peak position, chooses the first Preset Time window scope to the right；
S222, in described first Preset Time window scope, detect first modulus maximum point after R ripple, be set to S
Wave-wave peak position.
Preferably, described T wave-wave peak position, T ripple start position and T ripple final position are detected and are included step:
S231, it is set to starting point with detected S wave-wave peak position, chooses the second Preset Time window scope to the right；
S232, in described second Preset Time window scope, detect first modulus maximum point after S ripple, be set to T
Wave-wave peak position；
S233, it is set to starting point with described T wave-wave peak position, detects first scope discontinuity of the left and right sides respectively, and will detection
To first scope discontinuity of the left and right sides be respectively set to T ripple start position and final position.
Preferably, the characteristic parameter of described electrocardiosignal ST section waveform includes the relative offset level of ST section waveform, ST section ripple
The relative area that the first-order difference value of shape and ST section waveform surround with datum line.
Preferably, the characteristic sequence of the first-order difference value of described electrocardiosignal ST section waveform is carried out resampling, and by resampling
After characteristic sequence set up feature input matrix to be sorted.
Preferably, described sorter model of setting up includes carrying out training sample two-dimensional interpolation process, and based on the training after processing
Radial base neural net built by sample.
According to a further aspect in the invention, it is provided that the automatic classification system of a kind of electrocardiosignal ST section, including:
First module, for obtaining the electro-cardiologic signal waveforms of human body, and carries out pretreatment；
Second module, carries out feature point detection to through pretreated electro-cardiologic signal waveforms；
Three module, based on the feature point detection in the second module, determines electrocardiosignal ST section waveform, and obtains described electrocardio
The characteristic parameter of signal ST section waveform, to set up feature input matrix to be sorted；
4th module, divides electrocardiosignal ST section waveform and is set to training sample and test sample, and based on training sample
Set up sorter model；
5th module, inputs test sample in sorter model and tests, and combines Decision fusion and complete finally to classify.
In accordance with a further aspect of the present invention, it is provided that the apparatus for automatically sorting of a kind of electrocardiosignal ST section, including:
Memorizer, for storage one application program；
Processor, is used for running described program and performs following steps:
S1, the electro-cardiologic signal waveforms of acquisition human body, and carry out pretreatment；
S2, carry out feature point detection to through pretreated electro-cardiologic signal waveforms；
S3, based on the feature point detection in step S2, determine electrocardiosignal ST section waveform, and obtain described electrocardiosignal ST
The characteristic parameter of section waveform, to set up feature input matrix to be sorted；
S4, electrocardiosignal ST section waveform is divided it is set to training sample and test sample, and set up point based on training sample
Class device model；
S5, test sample is inputted in sorter model and tests, and combine Decision fusion and complete finally to classify.
Beneficial effects of the present invention:
The invention provides automatic classification method, system and the device of a kind of electrocardiosignal ST section, compared with prior art,
On the premise of ensureing the effectiveness of classification, owing to the amplitude of ST section is less, frequency is relatively low, is consequently belonging to low-dimensional input, effectively
Decrease computing and reduce time cost；Choose the sorting technique of the ST section feature having merged three kinds of parameters simultaneously, compare
In the extraction of traditional ST section feature, it is effectively improved the nicety of grading of ST section；And due to ST section waveform amplitude without
Hard requirement, the requirement to grader link is relatively low, and common existing grader all can effectively be classified so that realizes classification and more holds
Easily.
Accompanying drawing explanation
The invention will be further described with example below in conjunction with the accompanying drawings:
Fig. 1 is the automatic classification method schematic flow sheet of a kind of electrocardiosignal ST section according to the present invention；
Fig. 2 is electro-cardiologic signal waveforms comparison diagram and the noise of removal before and after denoising according to embodiments of the present invention；
Fig. 3 is the feature point detection figure of the automatic classification method of a kind of electrocardiosignal ST section according to the present invention；
Fig. 4 is the neural network classifier Organization Chart of the automatic classification method of a kind of electrocardiosignal ST section according to the present invention；
Fig. 5 is the classification results figure of the automatic classification method of a kind of electrocardiosignal ST section according to the present invention；
Fig. 6 is the structured flowchart of the apparatus for automatically sorting of a kind of electrocardiosignal ST section according to the present invention.
Detailed description of the invention
Automatic classification method, system and the device of a kind of electrocardiosignal ST section of the present invention, its core is, by " carrying
The characteristic parameter taking three kinds of ST sections is classified by Decision-level fusion model " step obtain a kind of fusion depth S T feature make
Morphological characteristic for electrocardiosignal.Three kinds of characteristic parameters are respectively from three sides of tendency degree of waveforms amplitude, waveform tendency and waveform
Face the most deeply describes the form of ST section, and is automatically identified by effective feature input grader.From self property of ST section
For matter, reach dimensionality reduction, thus reduced the purpose of operand；For the angle that the three kinds of parameters carrying out extracted merge,
Achieve the accuracy rate purpose improving classification.
Being described further inventive feature and principle below in conjunction with the accompanying drawings, example is served only for illustrating the present invention, but
It is not limited to the scope of the present invention.
According to a preferred embodiment of the invention, it is provided that plant the automatic classification method of electrocardiosignal ST section, with reference to Fig. 1, it illustrates
The overall flow block diagram of the automatic classification method according to the electrocardiosignal ST section described in the preferred embodiment of the present invention, including step:
S1, the electro-cardiologic signal waveforms of acquisition human body, and carry out pretreatment.
Specifically, the electrocardio primary signal collected is carried out pretreatment, predominantly to the noise such as baseline drift and High-frequency Interference
Remove.By electro-cardiologic signal waveforms is carried out spectrum analysis, the frequency of electrocardiosignal is mainly distributed on 0.05～100Hz, and
And 95% energy concentrate on 0.25～35Hz in, belong to the infrasonic wave that frequency is relatively low.Weak output signal, is easily subject in gatherer process
To interference.Hz noise is concentrated mainly on the 2nd yardstick, and myoelectricity interference is concentrated the most the 1st, on 2 yardsticks, and baseline drift
It is the frequency low-frequency disturbance that is less than 1Hz, has overlapping with the frequency spectrum of electrocardiosignal.Owing to ST section is the emphasis studied herein,
In preprocessing process, it is necessary to assure ST section shape information complete.Therefore method based on wavelet transformation is used to be removed,
Utilize Quadric Spline small echo (bior2,8 small echos) that pulse wave signal carries out 8 layers of decomposition.Ai represents the chi decomposing each layer
Degree coefficient, Di represents the detail coefficients decomposing each layer.It is concentrated mainly on the of wavelet decomposition owing to Hz noise and myoelectricity disturb
On 2 yardsticks, utilize threshold method to select zero setting to process the wavelet coefficient D2 of corresponding the 2nd layer of frequency band.Baseline drift simultaneously is made an uproar
Sound is concentrated mainly on the decomposed signal of the 8th yardstick, therefore by the low-frequency wavelet coefficients A8 zero setting of the 8th layer, does for such two kinds
Disturb noise substantially to remove, and then reconstruct obtains the electrocardiosignal after removing baseline drift.As in figure 2 it is shown, be electrocardio before and after denoising
Signal waveform comparison diagram and the noise of removal.
S2, carry out feature point detection to through pretreated electro-cardiologic signal waveforms；
Specifically, need pretreated pure electrocardiosignal electrocardiosignal ST section waveform is carried out detection and localization.But the heart
In electrical waveform, QRS complex amplitude is relatively big, and energy is concentrated, and its feature is obvious, easily it is distinguished from other waveforms and noise,
First R wave-wave peak is detected, secondly go out starting point and the terminal, i.e. S of ST section waveform according to R ripple location positioning
Ripple position and T ripple initial point position.The electrocardiosignal completing denoising is carried out discrete wavelet transformation, and carries out modulus maximum detection
Process, according to modulus maximum principle, obtain the positional information at R wave-wave peak；Next step judges whether missing inspection or flase drop, order
In electrocardiosignal, R--R interval is designated as Ri, i is the label that signal center claps.Making Ra is the average of R-R spacing in signal, public
Formula is as follows:
Wherein, the number in cycle during N is every electrocardiosignal.Next the relation between checking Ri and Ra, if Ri≤ 0.4Ra,
Then this spacing internal memory is at flase drop, detects R wave-wave peak more, needs to re-start difference sudden change and judges to screen with this；If
Ri>=1.5Ra, then this spacing internal memory is in missing inspection, and some R wave-wave peaks that leaks down do not detect, and need again to detect modulus maximum and carry out
Search R ripple.Use the positive and negative threshold method of adjustment to judge, if there is missing inspection, then need to reduce threshold value, if there is flase drop,
Then increasing threshold value, detection modulus maximum scans for R ripple the most again.R wave-wave peak position after the adjustment is set to starting point, to the right
Side arranges first modulus maximum of 30ms time window search thus obtains S ripple position, in like manner can position and obtain rising of Q ripple
Initial point.For T ripple, typically last for 100ms-250ms according to T ripple in electrocardiosignal, and by the feelings to actual electrocardiogram (ECG) data
Condition experiment can obtain, and time window is set to the crest location that 150ms is used for searching for difference catastrophe point as T ripple, this time window distance
R wave-wave peak shift 200ms.Finally utilize rising of partial transformation location T ripple respectively in left side and right side for T wave-wave peak position
Point and final position, T ripple starting point is first scope discontinuity of T ripple left end, and T ripple terminal is that first slope of T ripple right-hand member is dashed forward
Height, the most i.e. completes the detection of all characteristic points.As it is shown on figure 3, be the Q ripple of pure electrocardiosignal, R ripple and
S ripple position and T wave-wave peak, starting point and the detection and localization figure of terminal.
S3, based on the feature point detection in step S2, determine electrocardiosignal ST section waveform, and obtain described electrocardiosignal ST
The characteristic parameter of section waveform, to set up feature input matrix to be sorted.
Specifically, the S ripple obtained according to previous step location and the material time point of T ripple, intercept and obtain ST section shape information,
Extraction can represent the characteristic information of ST section waveform, i.e. obtains the characteristic parameter of each cardiac electrical cycle ST section waveform, including relatively
Average voltage X, average difference K surrounds area S with relative.
1) computing formula of relative offset level average X is:
In formulaRepresent the average voltage of m moment ST section waveform, x (QRSs) represent QRS complex starting point voltage
Meansigma methods, selects x (ST) as the average level value of skew relatively, it is therefore an objective to avoid the crowd in all ages and classes stage due to individuality
Differ greatly the difference of the absolute ST section level value brought.
2) ST section difference value K can see the angle of QRS-T in electrocardiosignal as.Owing to the form of ST section is more changeable, one
As use fitting process, the function obtained according to matching describes the waveform characteristic of ST section.First order difference equation is used to obtain difference:
kn=yn+1-yn
Wherein n is the coordinate of discrete time, and y is corresponding ST section range value.If knJust it is all or is all negative, now can regard as
For linear type ST section.IfJust it is not all or is all negative, when i.e. changing irregular, then needing to carry out second order difference equation and carry out point
Analysis.
3) relative area of ST section is the area that the waveform of ST section surrounds with datum line.Calculate the area at ABL respectively
S1With the area S below datum line2, according to following criterion, it is possible to obtain the degree level that ST raises or reduces.If
S1=0 and S2≠ 0, then understand ST section below datum line；If S1≠ 0 and S2=0, then understand ST section at datum line
Above；If S1≠ 0 and S2≠ 0, then the area formula of ST section is as follows:
S=| S1-S2|
S4, electrocardiosignal ST section waveform is divided it is set to training sample and test sample, and set up point based on training sample
Class device model.
Specifically, different people and different lead under mode, the amplitude difference of electrocardiosignal is very big, the Different Individual heart in addition
Rate also has bigger difference, therefore ST section needs to be normalized before classification, & apos.The ST section feature that will obtain in previous step
Parameter all normalizes between [-1,1], inputs as feature.Electrocardiosignal ST section normalization obtained is divided into training sample
Trainx, trainy and test sample testx, testy, use two-dimensional interpolation function interp2 to carry out two-dimensional interpolation training sample.
Then set up initial radial base neural net (RBF) ST section disaggregated model, be divided into input layer, hidden layer and output layer three
Layer, wherein Artificial neural network ensemble carries from Matlab platform.Kernel function chooses Radial basis kernel function, i.e. uses newrb function
Creating sorter model, setting up at model needs the selection of some parameters, including step-up error tolerance limit er, invasin spread
With hidden layer maximum neuron number N.As shown in Figure 4, for the neural network classifier model built in Matlab software interface.
Being separately input in sorter model by tri-characteristic parameters of training sample trainx, call newrb function, system can increase one by one
Add hidden layer neuron number, make training error be gradually reduced so that classification results is output as trainy, until error σ reaches
Error margin er, so that it is determined that each parameter of Artificial Neural Network Structures.Utilize corresponding with electrocardiosignal ST section classification results
Value (is selected numeral 1,2,3 to represent respectively to force down, normally, raise), sets up training set target variable matrix.Each sample
Corresponding target output, utilizes the sorter model that training obtains, the ST section characteristic parameter testx of input test collection.
S5, test sample is inputted in sorter model and tests, and combine Decision fusion and complete finally to classify.
Specifically, it is input to test sample testx in the radial basis neural network trained test, is classified
Result testy1.By classification results testy1It is input in the model of Decision-level fusion draw final result of determination testy2, it is prediction
Export the classification results that ST section is corresponding, and result of determination testy be given with expert does contrast verification.Obtain with actual value contrast
Test accuracy rate, the accuracy of verification algorithm.The Decision-level fusion model used is as follows:
F=α1f1+α2f2+α3f3
Wherein, f is reliability function value, and bigger that class of functional value is as last classification results, α1, α2, α3Point
Not Wei weights coefficient, f1, f2, f3It is respectively the correlation coefficient of characteristic parameter.The rule of Decision-level fusion foundation is as follows: if three
The classification planting parameter judgement is consistent, then can need not be input in Decision-level fusion model, directly export result；If three seed ginsengs
Count to have and judge inconsistent classification, then characteristic parameter is input in Fusion Model obtain the result of classification.As it is shown in figure 5, be
The final classification results of ST section classification.Finally according to the classification in each cycle in every electrocardiosignal, according to Voting principle, obtain
The ST section classification of whole signal.
According to another preferred embodiment of the invention, it is provided that the automatic classification system of a kind of electrocardiosignal ST section, including:
First module, for obtaining the electro-cardiologic signal waveforms of human body, and carries out pretreatment；Second module, to after pretreatment
Electro-cardiologic signal waveforms carry out feature point detection；Three module, based on the feature point detection in the second module, determines electrocardiosignal
ST section waveform, and obtain the characteristic parameter of described electrocardiosignal ST section waveform, to set up feature input matrix to be sorted；4th
Module, divides electrocardiosignal ST section waveform and is set to training sample and test sample, and set up classification based on training sample
Device model.5th module, inputs test sample in sorter model and tests, and combines Decision fusion and complete finally to classify.
Further embodiment according to the present invention, it is provided that the apparatus for automatically sorting of a kind of electrocardiosignal ST section, including:
Memorizer, for storage one application program；Processor, is used for running described program and performs following steps: S1, obtains people
The electro-cardiologic signal waveforms of body, and carry out pretreatment；S2, carry out feature point detection to through pretreated electro-cardiologic signal waveforms；
S3, based on the feature point detection in step S2, determine electrocardiosignal ST section waveform, and obtain described electrocardiosignal ST section ripple
The characteristic parameter of shape, to set up feature input matrix to be sorted；S4, by electrocardiosignal ST section waveform divide be set to train sample
Basis and test sample, and set up sorter model based on training sample.S5, test sample is inputted in sorter model carry out
Test, and combine Decision fusion and complete finally to classify.
Table 1 utilizes the above preferred embodiment of the present invention to predict the outcome the data file randomly selected in CSE data base, passes through
Accuracy rate obtains the effect of classification.By last judgement classification contrast Expert opinion result as recruitment evaluation according to being able to verify that
That classifies is whether accurate.
Table 1 classification results table
| Filename | Test accuracy rate | Judge classification | Expert opinion result |
| MA_001 | 97.09% | Normal type | Upper the most normal |
| MA_002 | 94.06% | Force down type | Recessed force down |
| MA_007 | 97.67% | Elevation | On tiltedly raise |
| MA_011 | 79.80% | Elevation | Fovea superior is raised |
| MA_016 | 73.89% | Normal type | Straight line is normal |
| MA_026 | 71.01% | Force down type | Declivity is forced down |
In this example, current international practice ECG data storehouse CSE is used to test, by work software system on computers
Matlab simulated environment well known in system and industry realizes.
As shown in table 1, for the random electrocardiosignal chosen in storehouse, the ST section tagsort method using the present invention to provide obtains
The classification results arrived, and actual value compares, accuracy rate minimum more than 71%, can reach more than 97%, average statistics is accurate
Really rate is more than 85%, more preferable than the method and system of most existing existence.
The electrocardiosignal ST section tagsort method of the present invention by carrying out denoising, feature point detection to electrocardiosignal, and then obtains
To ST section waveform feature parameter, Feature Selection process is simple.ST section form is complex, and single parameter cannot show it completely
Feature, so the present invention is considered as multiple parameter and describes its feature simultaneously.Sorter model is set up by neural net method,
And by Decision-level fusion, model structure is succinct, can overcome the too high problem of electrocardiosignal dimension well, and time delay is little, is not subject to
The impact that ST section amplitude is less than normal, can be that identification and the clinical diagnosis of electrocardiosignal provides important theoretical foundation.
It is above the preferably enforcement of the present invention is illustrated, but the invention is not limited to described embodiment, is familiar with
Those skilled in the art also can make all equivalent variations or replacement on the premise of spirit of the present invention, these equivalents
Modification or replacement be all contained in the application claim limited range.
Unless a required step needs to be inputted by produced by preceding step, the particular order of step the most described herein is only
For exemplary illustration, and unrestricted.
Claims (11)
1. the automatic classification method of an electrocardiosignal ST section, it is characterised in that include step:
S1, the electro-cardiologic signal waveforms of acquisition human body, and carry out pretreatment；
S2, carry out feature point detection to through pretreated electro-cardiologic signal waveforms；
S3, distinguished point based detect, and determine electrocardiosignal ST section waveform, and obtain the characteristic parameter of described electrocardiosignal ST section waveform, to set up feature input matrix to be sorted；
S4, electrocardiosignal ST section waveform is divided it is set to training sample and test sample, and set up sorter model based on training sample；
S5, test sample is inputted in sorter model and tests, and combine Decision fusion and complete classification.
The automatic classification method of electrocardiosignal ST section the most according to claim 1, it is characterised in that described pretreatment includes that method based on wavelet transformation removes the baseline drift of low frequency and the Hz noise of high frequency and dynamo-electric interference, further includes steps of
S11, use Quadric Spline small echo carry out 8 layers of decomposition to electro-cardiologic signal waveforms；
S12, by the high-frequency wavelet coefficient zero setting of the 2nd layer, to remove the Hz noise of high frequency and dynamo-electric interference；
S13, by the low-frequency wavelet coefficients zero setting of the 8th layer, and carry out wavelet inverse transformation reconstruct to obtain pure electro-cardiologic signal waveforms.
The automatic classification method of electrocardiosignal ST section the most according to claim 1, it is characterised in that described feature point detection includes the detection of R wave-wave peak position, the detection of S wave-wave peak position, T wave-wave peak position, T ripple start position and the detection of T ripple final position.
The automatic classification method of electrocardiosignal ST section the most according to claim 3, it is characterised in that described R wave-wave peak position is detected and included step:
S211, discrete wavelet transformation will be carried out through pretreated electrocardiosignal, according to modulus maximum principle, detection the zero crossing position of labelling modulus maximum pair；
S212, near the zero crossing position of modulus maximum pair search for amplitude absolute value maximum of points, if the single order left and right difference score value of described point is stationary value, then described point is set to R wave-wave peak position；
S213, spacing average Ra calculated between adjacent two R wave-wave peaks, in the R wave-wave peak detected, be smaller than 0.4Ra if existing between two R wave-wave peaks, then it is determined that the presence of flase drop, and increase the threshold value of R ripple；If the spacing existed between two R wave-wave peaks is more than 1.5Ra, then it is determined that the presence of missing inspection, and reduce the threshold value of R ripple；
S214, again carry out modulus maximum search, to detect R wave-wave peak position.
The automatic classification method of electrocardiosignal ST section the most according to claim 3, it is characterised in that described S wave-wave peak position is detected and included step:
S221, it is set to starting point with detected R wave-wave peak position, chooses the first Preset Time window scope to the right；
S222, in described first Preset Time window scope, detect first modulus maximum point after R ripple, be set to S wave-wave peak position.
The automatic classification method of electrocardiosignal ST section the most according to claim 3, it is characterised in that described T wave-wave peak position, T ripple start position and T ripple final position are detected and included step:
S231, it is set to starting point with detected S wave-wave peak position, chooses the second Preset Time window scope to the right；
S232, in described second Preset Time window scope, detect first modulus maximum point after S ripple, be set to T wave-wave peak position；
S233, it is set to starting point with described T wave-wave peak position, detects first scope discontinuity of the left and right sides respectively, and first scope discontinuity of the left and right sides detected is respectively set to T ripple start position and final position.
The automatic classification method of electrocardiosignal ST section the most according to claim 1, it is characterized in that, the characteristic parameter of described electrocardiosignal ST section waveform includes the relative area that the relative offset level of ST section waveform, the first-order difference value of ST section waveform and ST section waveform surround with datum line.
The automatic classification method of electrocardiosignal ST section the most according to claim 7, it is characterised in that the characteristic sequence of the first-order difference value of described electrocardiosignal ST section waveform is carried out resampling, and the characteristic sequence after resampling is set up feature input matrix to be sorted.
The automatic classification method of electrocardiosignal ST section the most according to claim 1, it is characterised in that described step S4 also includes: training sample is carried out two-dimensional interpolation process, and builds radial base neural net based on the training sample after processing.
10. the automatic classification system of an electrocardiosignal ST section, it is characterised in that including:
First module, for obtaining the electro-cardiologic signal waveforms of human body, and carries out pretreatment；
Second module, carries out feature point detection to through pretreated electro-cardiologic signal waveforms；
Three module, based on the feature point detection in the second module, determines electrocardiosignal ST section waveform, and obtains the characteristic parameter of described electrocardiosignal ST section waveform, to set up feature input matrix to be sorted；
4th module, divides electrocardiosignal ST section waveform and is set to training sample and test sample, and set up sorter model based on training sample；
5th module, inputs test sample in sorter model and tests, and combines Decision fusion and complete finally to classify.
The apparatus for automatically sorting of 11. 1 kinds of electrocardiosignal ST sections, including:
Memorizer, for storage one application program；
Processor, is used for running described program and performs following steps:
S1, the electro-cardiologic signal waveforms of acquisition human body, and carry out pretreatment；
S2, carry out feature point detection to through pretreated electro-cardiologic signal waveforms；
S3, based on the feature point detection in step S2, determine electrocardiosignal ST section waveform, and obtain the characteristic parameter of described electrocardiosignal ST section waveform, to set up feature input matrix to be sorted；
S4, electrocardiosignal ST section waveform is divided it is set to training sample and test sample, and set up sorter model based on training sample；
S5, test sample is inputted in sorter model and tests, and combine Decision fusion and complete finally to classify.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610541616.9A CN105997055A (en) | 2016-07-11 | 2016-07-11 | Automatic classification method, system and device of electrocardiosignal ST band |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610541616.9A CN105997055A (en) | 2016-07-11 | 2016-07-11 | Automatic classification method, system and device of electrocardiosignal ST band |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN105997055A true CN105997055A (en) | 2016-10-12 |
Family
ID=57109521
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610541616.9A Pending CN105997055A (en) | 2016-07-11 | 2016-07-11 | Automatic classification method, system and device of electrocardiosignal ST band |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105997055A (en) |
Cited By (32)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107361753A (en) * | 2017-08-29 | 2017-11-21 | 哈尔滨理工大学 | Health state monitoring method based on pulse wave characteristic point |
| CN107622259A (en) * | 2017-10-20 | 2018-01-23 | 山东大学 | A kind of T wave detecting methods, electrocardiogram (ECG) data analysis method and device |
| CN107837082A (en) * | 2017-11-27 | 2018-03-27 | 乐普（北京）医疗器械股份有限公司 | Electrocardiogram automatic analysis method and device based on artificial intelligence self study |
| WO2018082190A1 (en) * | 2016-11-03 | 2018-05-11 | 华为技术有限公司 | Ecg signal processing method and apparatus |
| CN108537100A (en) * | 2017-11-17 | 2018-09-14 | 吉林大学 | A kind of electrocardiosignal personal identification method and system based on PCA and LDA analyses |
| CN108597336A (en) * | 2018-02-28 | 2018-09-28 | 天津天堰科技股份有限公司 | Ecg wave form emulation mode |
| CN108647614A (en) * | 2018-04-28 | 2018-10-12 | 吉林大学 | The recognition methods of electrocardiogram beat classification and system |
| CN108714026A (en) * | 2018-03-27 | 2018-10-30 | 杭州电子科技大学 | The fine granularity electrocardiosignal sorting technique merged based on depth convolutional neural networks and on-line decision |
| CN109044347A (en) * | 2018-07-11 | 2018-12-21 | 上海夏先机电科技发展有限公司 | Identify electrocardio wave image junctional escape beat method, apparatus, system and electronic equipment |
| CN109464140A (en) * | 2019-01-09 | 2019-03-15 | 浙江强脑科技有限公司 | Bioelectrical signals detection method, equipment and storage medium |
| CN109620263A (en) * | 2018-12-13 | 2019-04-16 | 湖南仪峰安安网络科技股份有限公司 | A kind of enterprises and institutions are on duty personnel's sign safety analytical method |
| CN109620214A (en) * | 2018-12-07 | 2019-04-16 | 上海数创医疗科技有限公司 | ST sections of automatic judging methods of electrocardiosignal and device based on artificial intelligence technology |
| CN109770858A (en) * | 2019-03-28 | 2019-05-21 | 广州视源电子科技股份有限公司 | A kind of method for detecting abnormality of electrocardiogram, device, equipment and storage medium |
| CN109770859A (en) * | 2019-03-28 | 2019-05-21 | 广州视源电子科技股份有限公司 | The treating method and apparatus of electrocardiosignal, storage medium, processor |
| CN109948396A (en) * | 2017-12-20 | 2019-06-28 | 深圳市理邦精密仪器股份有限公司 | A kind of beat classification method, beat classification device and electronic equipment |
| CN110101383A (en) * | 2019-04-19 | 2019-08-09 | 长沙理工大学 | A kind of Denoising Algorithm of ECG Signals based on wavelet energy |
| CN110226917A (en) * | 2019-06-26 | 2019-09-13 | 广州视源电子科技股份有限公司 | Electrocardiosignal type detection method, apparatus, computer equipment and storage medium |
| CN110236518A (en) * | 2019-04-02 | 2019-09-17 | 武汉大学 | The method and device of electrocardio and the heart shake combined signal classification neural network based |
| CN110547786A (en) * | 2018-06-01 | 2019-12-10 | 山东省计算中心（国家超级计算济南中心） | form recognition method of electrocardiogram waveform based on multi-scale wavelet transform |
| WO2020006939A1 (en) * | 2018-07-06 | 2020-01-09 | 苏州大学张家港工业技术研究院 | Electrocardiogram generation and classification method based on generative adversarial network |
| CN110772248A (en) * | 2019-10-17 | 2020-02-11 | 深圳邦健生物医疗设备股份有限公司 | Wearable device electrocardio real-time monitoring system and method |
| CN110897627A (en) * | 2018-09-14 | 2020-03-24 | 杭州脉流科技有限公司 | Electrocardiogram signal feature extraction method, device, equipment, system and storage medium |
| CN110974211A (en) * | 2019-12-09 | 2020-04-10 | 上海数创医疗科技有限公司 | ST-segment classification neural network of high-order polynomial activation function and application thereof |
| CN111407261A (en) * | 2020-03-31 | 2020-07-14 | 京东方科技集团股份有限公司 | Method and device for measuring periodic information of biological signal and electronic equipment |
| CN111460956A (en) * | 2020-03-26 | 2020-07-28 | 山东科技大学 | Unbalanced electrocardiogram sample classification method based on data enhancement and loss weighting |
| CN111956213A (en) * | 2020-07-29 | 2020-11-20 | 鲁东大学 | QRS point detection method of electrocardiosignal |
| CN112200232A (en) * | 2020-09-29 | 2021-01-08 | 上海移视网络科技有限公司 | QRS identification method and electronic equipment |
| CN112438731A (en) * | 2019-08-27 | 2021-03-05 | 深圳市理邦精密仪器股份有限公司 | Electrocardiosignal analysis method, electrocardiosignal analysis device and storage medium |
| CN112656393A (en) * | 2020-12-08 | 2021-04-16 | 山东中科先进技术研究院有限公司 | Heart rate variability detection method and system |
| CN112869753A (en) * | 2021-02-10 | 2021-06-01 | 上海交通大学医学院附属瑞金医院 | Analysis method, equipment, medium and electrocardiograph for QRST waveform of electrocardiogram |
| CN112932498A (en) * | 2021-01-29 | 2021-06-11 | 山东大学 | T wave morphology classification system with strong generalization capability based on deep learning |
| CN113012807A (en) * | 2021-03-04 | 2021-06-22 | 武汉中旗生物医疗电子有限公司 | ECG signal marking method, system, device and computer readable storage medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6171256B1 (en) * | 1998-04-30 | 2001-01-09 | Physio-Control Manufacturing Corporation | Method and apparatus for detecting a condition associated with acute cardiac ischemia |
| CN101002682A (en) * | 2007-01-19 | 2007-07-25 | 哈尔滨工程大学 | Method for retrieval and matching of hand back vein characteristic used for identification of status |
| US20110040200A1 (en) * | 2009-08-12 | 2011-02-17 | Antony Louis Piriyakumar Douglas | System and method for analyzing an electrocardiogram signal |
| CN105054926A (en) * | 2015-04-13 | 2015-11-18 | 深圳市飞马与星月科技研究有限公司 | Electrocardiosignal feature information extraction method and device |
| CN105212922A (en) * | 2014-06-11 | 2016-01-06 | 吉林大学 | The method and system that R wave of electrocardiosignal detects automatically are realized towards FPGA |
-
2016
- 2016-07-11 CN CN201610541616.9A patent/CN105997055A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6171256B1 (en) * | 1998-04-30 | 2001-01-09 | Physio-Control Manufacturing Corporation | Method and apparatus for detecting a condition associated with acute cardiac ischemia |
| CN101002682A (en) * | 2007-01-19 | 2007-07-25 | 哈尔滨工程大学 | Method for retrieval and matching of hand back vein characteristic used for identification of status |
| US20110040200A1 (en) * | 2009-08-12 | 2011-02-17 | Antony Louis Piriyakumar Douglas | System and method for analyzing an electrocardiogram signal |
| CN105212922A (en) * | 2014-06-11 | 2016-01-06 | 吉林大学 | The method and system that R wave of electrocardiosignal detects automatically are realized towards FPGA |
| CN105054926A (en) * | 2015-04-13 | 2015-11-18 | 深圳市飞马与星月科技研究有限公司 | Electrocardiosignal feature information extraction method and device |
Non-Patent Citations (1)
| Title |
|---|
| 杨宇 等: "基于小波变换和支持向量机的心电信号ST段分类", 《吉林大学学报（信息科学版）》 * |
Cited By (44)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11337638B2 (en) | 2016-11-03 | 2022-05-24 | Huawei Technologies Co., Ltd. | ECG signal processing method and apparatus |
| WO2018082190A1 (en) * | 2016-11-03 | 2018-05-11 | 华为技术有限公司 | Ecg signal processing method and apparatus |
| CN107361753A (en) * | 2017-08-29 | 2017-11-21 | 哈尔滨理工大学 | Health state monitoring method based on pulse wave characteristic point |
| CN107622259A (en) * | 2017-10-20 | 2018-01-23 | 山东大学 | A kind of T wave detecting methods, electrocardiogram (ECG) data analysis method and device |
| CN107622259B (en) * | 2017-10-20 | 2019-12-24 | 山东大学 | T wave detection method, electrocardiogram data analysis method and device |
| CN108537100A (en) * | 2017-11-17 | 2018-09-14 | 吉林大学 | A kind of electrocardiosignal personal identification method and system based on PCA and LDA analyses |
| CN107837082A (en) * | 2017-11-27 | 2018-03-27 | 乐普（北京）医疗器械股份有限公司 | Electrocardiogram automatic analysis method and device based on artificial intelligence self study |
| CN109948396B (en) * | 2017-12-20 | 2021-07-23 | 深圳市理邦精密仪器股份有限公司 | Heart beat classification method, heart beat classification device and electronic equipment |
| CN109948396A (en) * | 2017-12-20 | 2019-06-28 | 深圳市理邦精密仪器股份有限公司 | A kind of beat classification method, beat classification device and electronic equipment |
| CN108597336A (en) * | 2018-02-28 | 2018-09-28 | 天津天堰科技股份有限公司 | Ecg wave form emulation mode |
| CN108597336B (en) * | 2018-02-28 | 2021-11-05 | 天津天堰科技股份有限公司 | Electrocardiogram waveform simulation method |
| CN108714026B (en) * | 2018-03-27 | 2021-09-03 | 杭州电子科技大学 | Fine-grained electrocardiosignal classification method based on deep convolutional neural network and online decision fusion |
| CN108714026A (en) * | 2018-03-27 | 2018-10-30 | 杭州电子科技大学 | The fine granularity electrocardiosignal sorting technique merged based on depth convolutional neural networks and on-line decision |
| CN108647614A (en) * | 2018-04-28 | 2018-10-12 | 吉林大学 | The recognition methods of electrocardiogram beat classification and system |
| CN110547786B (en) * | 2018-06-01 | 2020-05-05 | 山东省计算中心（国家超级计算济南中心） | Form recognition method of electrocardiogram waveform based on multi-scale wavelet transform |
| CN110547786A (en) * | 2018-06-01 | 2019-12-10 | 山东省计算中心（国家超级计算济南中心） | form recognition method of electrocardiogram waveform based on multi-scale wavelet transform |
| WO2020006939A1 (en) * | 2018-07-06 | 2020-01-09 | 苏州大学张家港工业技术研究院 | Electrocardiogram generation and classification method based on generative adversarial network |
| CN109044347A (en) * | 2018-07-11 | 2018-12-21 | 上海夏先机电科技发展有限公司 | Identify electrocardio wave image junctional escape beat method, apparatus, system and electronic equipment |
| CN109044347B (en) * | 2018-07-11 | 2021-08-27 | 上海夏先机电科技发展有限公司 | Method, device and system for identifying junctional escape of electrocardiowave image and electronic equipment |
| CN110897627A (en) * | 2018-09-14 | 2020-03-24 | 杭州脉流科技有限公司 | Electrocardiogram signal feature extraction method, device, equipment, system and storage medium |
| CN109620214A (en) * | 2018-12-07 | 2019-04-16 | 上海数创医疗科技有限公司 | ST sections of automatic judging methods of electrocardiosignal and device based on artificial intelligence technology |
| CN109620263A (en) * | 2018-12-13 | 2019-04-16 | 湖南仪峰安安网络科技股份有限公司 | A kind of enterprises and institutions are on duty personnel's sign safety analytical method |
| CN109464140A (en) * | 2019-01-09 | 2019-03-15 | 浙江强脑科技有限公司 | Bioelectrical signals detection method, equipment and storage medium |
| CN109770858B (en) * | 2019-03-28 | 2021-07-13 | 广州视源电子科技股份有限公司 | Method, device and equipment for detecting abnormality of electrocardiogram and storage medium |
| CN109770858A (en) * | 2019-03-28 | 2019-05-21 | 广州视源电子科技股份有限公司 | A kind of method for detecting abnormality of electrocardiogram, device, equipment and storage medium |
| CN109770859A (en) * | 2019-03-28 | 2019-05-21 | 广州视源电子科技股份有限公司 | The treating method and apparatus of electrocardiosignal, storage medium, processor |
| CN110236518A (en) * | 2019-04-02 | 2019-09-17 | 武汉大学 | The method and device of electrocardio and the heart shake combined signal classification neural network based |
| CN110101383A (en) * | 2019-04-19 | 2019-08-09 | 长沙理工大学 | A kind of Denoising Algorithm of ECG Signals based on wavelet energy |
| CN110226917A (en) * | 2019-06-26 | 2019-09-13 | 广州视源电子科技股份有限公司 | Electrocardiosignal type detection method, apparatus, computer equipment and storage medium |
| CN112438731A (en) * | 2019-08-27 | 2021-03-05 | 深圳市理邦精密仪器股份有限公司 | Electrocardiosignal analysis method, electrocardiosignal analysis device and storage medium |
| CN110772248A (en) * | 2019-10-17 | 2020-02-11 | 深圳邦健生物医疗设备股份有限公司 | Wearable device electrocardio real-time monitoring system and method |
| CN110974211A (en) * | 2019-12-09 | 2020-04-10 | 上海数创医疗科技有限公司 | ST-segment classification neural network of high-order polynomial activation function and application thereof |
| CN111460956A (en) * | 2020-03-26 | 2020-07-28 | 山东科技大学 | Unbalanced electrocardiogram sample classification method based on data enhancement and loss weighting |
| CN111407261A (en) * | 2020-03-31 | 2020-07-14 | 京东方科技集团股份有限公司 | Method and device for measuring periodic information of biological signal and electronic equipment |
| CN111956213A (en) * | 2020-07-29 | 2020-11-20 | 鲁东大学 | QRS point detection method of electrocardiosignal |
| CN112200232A (en) * | 2020-09-29 | 2021-01-08 | 上海移视网络科技有限公司 | QRS identification method and electronic equipment |
| CN112200232B (en) * | 2020-09-29 | 2024-03-22 | 上海移视网络科技有限公司 | QRS (QRS) recognition method and electronic equipment |
| CN112656393A (en) * | 2020-12-08 | 2021-04-16 | 山东中科先进技术研究院有限公司 | Heart rate variability detection method and system |
| CN112656393B (en) * | 2020-12-08 | 2023-06-27 | 山东中科先进技术有限公司 | Method and system for detecting heart rate variability |
| CN112932498A (en) * | 2021-01-29 | 2021-06-11 | 山东大学 | T wave morphology classification system with strong generalization capability based on deep learning |
| CN112869753A (en) * | 2021-02-10 | 2021-06-01 | 上海交通大学医学院附属瑞金医院 | Analysis method, equipment, medium and electrocardiograph for QRST waveform of electrocardiogram |
| CN112869753B (en) * | 2021-02-10 | 2023-03-07 | 上海交通大学医学院附属瑞金医院 | Analysis method, equipment, medium and electrocardiograph for QRST waveform of electrocardiogram |
| CN113012807A (en) * | 2021-03-04 | 2021-06-22 | 武汉中旗生物医疗电子有限公司 | ECG signal marking method, system, device and computer readable storage medium |
| CN113012807B (en) * | 2021-03-04 | 2023-06-09 | 武汉中旗生物医疗电子有限公司 | ECG signal marking method, system, device and computer readable storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105997055A (en) | Automatic classification method, system and device of electrocardiosignal ST band | |
| Koh et al. | Diagnosis of retinal health in digital fundus images using continuous wavelet transform (CWT) and entropies | |
| CN104523266B (en) | A kind of electrocardiosignal automatic classification method | |
| US6745156B2 (en) | Petroleum exploration and prediction apparatus and method | |
| CN108511055B (en) | Ventricular premature beat recognition system and method based on classifier fusion and diagnosis rules | |
| CN108932452A (en) | Arrhythmia cordis beat classification method based on multiple dimensioned convolutional neural networks | |
| CN103083013A (en) | Electrocardio signal QRS complex wave detection method based on morphology and wavelet transform | |
| CN103034837B (en) | Characteristic parameter is associated with pulse condition key element | |
| CN109117730A (en) | Electrocardiogram auricular fibrillation real-time judge method, apparatus, system and storage medium | |
| CN104769578A (en) | Automated feature analysis, comparison, and anomaly detection | |
| CN105411565A (en) | Heart rate variability feature classification method based on generalized scale wavelet entropy | |
| CN104809480A (en) | Retinal vessel segmentation method of fundus image based on classification and regression tree and AdaBoost | |
| CN105320969A (en) | A heart rate variability feature classification method based on multi-scale Renyi entropy | |
| CN111449644A (en) | Bioelectricity signal classification method based on time-frequency transformation and data enhancement technology | |
| CN106725420A (en) | VPB recognition methods and VPB identifying system | |
| CN110495877A (en) | A kind of Multi resolution feature extraction method and device based on ECG | |
| CN112472048B (en) | Method for realizing neural network for identifying pulse condition of cardiovascular disease patient | |
| CN111000553A (en) | Intelligent classification method for electrocardiogram data based on voting ensemble learning | |
| CN107239684A (en) | A kind of feature learning method and system for ECG identifications | |
| CN108256453A (en) | A kind of method based on one-dimensional ECG signal extraction two dimension CNN features | |
| KR20190105180A (en) | Apparatus for Lesion Diagnosis Based on Convolutional Neural Network and Method thereof | |
| Doshi et al. | (Retracted) Deep belief network-based image processing for local directional segmentation in brain tumor detection | |
| KR102632272B1 (en) | Classification method of atrial fibrillation and congestive heart failure using a convolutional artificial neural network | |
| Saraswat et al. | Classification of ECG signals using cross-recurrence quantification analysis and probabilistic neural network classifier for ventricular tachycardia patients | |
| Nath et al. | Quantum annealing for automated feature selection in stress detection |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20161012 |
|
| RJ01 | Rejection of invention patent application after publication |