CN116350233A - Electrocardiosignal quality assessment method based on self-encoder - Google Patents
Electrocardiosignal quality assessment method based on self-encoder Download PDFInfo
- Publication number
- CN116350233A CN116350233A CN202310313430.8A CN202310313430A CN116350233A CN 116350233 A CN116350233 A CN 116350233A CN 202310313430 A CN202310313430 A CN 202310313430A CN 116350233 A CN116350233 A CN 116350233A
- Authority
- CN
- China
- Prior art keywords
- signal
- self
- quality
- electrocardiosignal
- encoder
- 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
- 238000001303 quality assessment method Methods 0.000 title description 6
- 238000000034 method Methods 0.000 claims abstract description 33
- 238000012216 screening Methods 0.000 claims abstract description 20
- 238000013441 quality evaluation Methods 0.000 claims abstract description 16
- 238000001914 filtration Methods 0.000 claims abstract description 13
- 238000012549 training Methods 0.000 claims abstract description 10
- 238000011156 evaluation Methods 0.000 claims abstract description 6
- 239000012634 fragment Substances 0.000 claims description 13
- 239000013598 vector Substances 0.000 claims description 8
- 230000008569 process Effects 0.000 claims description 7
- 238000012545 processing Methods 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000013527 convolutional neural network Methods 0.000 claims description 3
- 238000012163 sequencing technique Methods 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 2
- 230000009467 reduction Effects 0.000 claims description 2
- 230000006870 function Effects 0.000 description 12
- 238000013461 design Methods 0.000 description 3
- 238000002372 labelling Methods 0.000 description 3
- 238000013135 deep learning Methods 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 230000000877 morphologic effect Effects 0.000 description 2
- 230000004913 activation Effects 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000010485 coping Effects 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000004880 explosion Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 231100000915 pathological change Toxicity 0.000 description 1
- 230000036285 pathological change Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000011002 quantification Methods 0.000 description 1
- 230000000284 resting effect Effects 0.000 description 1
- 238000011282 treatment Methods 0.000 description 1
- 210000000707 wrist Anatomy 0.000 description 1
Images
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
-
- 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
- A61B5/7221—Determining signal validity, reliability or quality
-
- 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
- A61B5/725—Details of waveform analysis using specific filters therefor, e.g. Kalman or adaptive filters
-
- 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
-
- 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
- A61B5/7267—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems involving training the classification device
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
- G06N3/0455—Auto-encoder networks; Encoder-decoder networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/088—Non-supervised learning, e.g. competitive learning
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Health & Medical Sciences (AREA)
- Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Heart & Thoracic Surgery (AREA)
- Medical Informatics (AREA)
- Surgery (AREA)
- Animal Behavior & Ethology (AREA)
- Pathology (AREA)
- Public Health (AREA)
- Veterinary Medicine (AREA)
- Theoretical Computer Science (AREA)
- Mathematical Physics (AREA)
- Evolutionary Computation (AREA)
- Signal Processing (AREA)
- Psychiatry (AREA)
- Physiology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- Cardiology (AREA)
- Fuzzy Systems (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
The invention relates to the technical field of signal quality evaluation, in particular to an electrocardiosignal quality evaluation method based on a self-encoder, which comprises the following steps: acquiring an original electrocardiosignal, and performing multilayer filtering pretreatment; dividing the preprocessed signals into a plurality of segments with fixed lengths, and screening signal points of each signal segment to obtain a batch of primary screening electrocardiosignals; training a pre-constructed self-encoder model by using the primary screening electrocardiosignal to enable the self-encoder model to learn the general rule of the electrocardiosignal; inputting an original signal to be evaluated into a trained self-encoder model for signal reconstruction, and calculating the degree of uncertainty; and calculating the quality score of the original signal to be evaluated based on the reconstructed signal and the uncertainty. The invention can perform self-adaptive continuous quality evaluation on the signals, and has high evaluation accuracy.
Description
Technical Field
The invention relates to the technical field of signal quality evaluation, in particular to an electrocardiosignal quality evaluation method based on a self-encoder.
Background
With the wide application of wearable and handheld products embedded with sensors, quality assessment is an indispensable important link in the sensor signal data processing process. For wearable and holding ECG acquisition equipment/products, the long-term dynamic electrocardiosignals can be recorded, the defect that the traditional electrocardiograms can only be acquired in a short-term and resting state is overcome, but complicated noise is unavoidably existed because the acquisition environment cannot reach the medical requirement level, and the electrocardiosignals are complicated and weak signals and are easily influenced by various noises, such as the electrocardiosignals are easily influenced by noises such as myoelectricity and motion under the hand-held or wrist wearing scene, so that the accuracy of subsequent function detection based on the electrocardiosignals is reduced.
At present, many electrocardiosignal quality evaluation algorithms finish calculation and judgment based on electrocardiosignal characteristics, but when a human body is physiologically changed, the electrocardiosignal characteristics of the electrocardiosignal quality evaluation algorithms are changed, and the electrocardiosignal data under pathological changes are easily treated as noise or heterogeneous filtration based on the characteristic signal quality evaluation, so that loss of signal fragments with important diagnostic value and the like is caused, and accurate evaluation and estimation are difficult to obtain. In addition, some environmental noise (e.g., baseline drift, etc.) that is not easily avoided but is within an acceptable range may also cause significant variations in the characteristics of the electrocardiographic signals. It follows that it is difficult to address these problems by merely extracting the electrocardiographic features.
In addition, in the field of electrocardiographic signal quality assessment, many methods have made this technique a task of classification or multiple classification. The algorithm is essentially to learn electrocardiosignal distribution with different quality, and cannot explore the general rule of electrocardiosignals and noise signals due to the scale of training data and a tag set. The method adopts a deep learning method to carry out multi-classification on the quality of the electrocardiosignals, relies on labels for labeling the quality of the electrocardiosignals, and labeling standards of the labels are different from person to person, and requirements on the quality of the electrocardiosignals required in different scenes are different, wherein some detection scenes need clear and accurate waves (clusters) such as P waves, ST waves, QRS waves and the like, and some scenes only require R-peak clear and accurate, so that a set of stable and high-universality electrocardiosignal quality evaluation model is difficult to construct.
Therefore, how to perform adaptive continuous quality evaluation on signals, and to improve accuracy are a problem that needs to be solved by those skilled in the art.
Disclosure of Invention
In view of the above, the invention provides an electrocardiosignal quality evaluation method based on a self-encoder, which can perform self-adaptive continuous quality evaluation on signals and has high evaluation accuracy.
In order to achieve the above purpose, the present invention adopts the following technical scheme:
an electrocardiosignal quality assessment method based on a self-encoder comprises the following steps:
acquiring an original electrocardiosignal, and performing multilayer filtering pretreatment;
dividing the preprocessed signals into a plurality of segments with fixed lengths, and screening signal points of each signal segment to obtain a batch of primary screening electrocardiosignals;
training a pre-constructed self-encoder model by using the primary screening electrocardiosignal to enable the self-encoder model to learn the general rule of the electrocardiosignal;
inputting an original signal to be evaluated into a trained self-encoder model for signal reconstruction, and calculating the uncertainty of the signal;
and calculating the quality score of the original signal to be evaluated based on the reconstructed signal and the uncertainty.
Further, the preprocessing of the original electrocardiosignal at least comprises: baseline drift, power frequency signal filtering and high-frequency signal noise reduction are removed.
Further, the screening basis for the signal fragments is as follows: and judging the signal point duty ratio of each signal segment, wherein the amplitude of the signal point duty ratio exceeds a threshold value, and filtering out the signal points under the duty ratio if the signal point duty ratio exceeds a preset ratio.
Further, in the process of screening the signal points, marking the signal points with the amplitudes exceeding the whole signal amplitude average value by 2 x std in each signal segment as oversized signal points, and marking the oversized signal points under the duty ratio as filtered signals and filtering if the duty ratio of the oversized signal points in the whole signal segment exceeds 0.5; where std denotes the standard deviation of the signal segment.
Further, before screening the signal points in each signal segment, the method further comprises: and carrying out data standardization processing on each segmented signal segment.
Further, the self-encoder model is composed of an encoder and a decoder; the encoder adopts a convolutional neural network to extract features from an original one-dimensional electrocardiosignal and performs multi-layer convolutional expansion, so that the original electrocardiosignal is abstracted into low-dimensional dense vector electrocardiosignal features; the decoder reconstructs the low-dimensional dense vector electrocardiographic features by adopting transpose convolution, and calculates the unreliability of the reconstructed signals.
Further, a loss function AE-LLH of the self-encoder model is constructed by adopting a log-likelihood function, and the expression is as follows:
wherein,,
representing a log-likelihood function; x represents an original electrocardiosignal; μ represents the reconstructed signal; sigma represents the degree of uncertainty; l represents the total number of signal segments in the batch, and L represents the first signal segment.
Further, an evaluation formula of the quality score of the original signal to be evaluated is:
where Score represents the quality fraction of the signal to be evaluated, lambda represents the penalty factor,
representing an uncertainty bias term.
Further, the method further comprises the following steps: dividing an original signal to be evaluated into signal segments with the same signal length as a training stage, and carrying out reconstruction and uncertainty calculation on each signal segment based on a trained self-encoder model.
Further, the method further comprises the following steps: and sequencing the quality scores of the signal fragments under the original signal to be evaluated, and taking the signal fragments with the top n ranks or the quality score meeting the score threshold as the high-quality signal.
Compared with the prior art, the invention discloses an electrocardiosignal quality evaluation method based on a self-encoder, which learns the general rule of the electrocardiosignal by encoding and decoding the electrocardiosignal through the self-encoder, and further judges the signal noise ratio through the reconstruction error of an output signal and an input signal because the noise signal cannot be reconstructed due to lack of regularity, so that the noise ratio of the signal can be effectively identified, the noise-mixed electrocardiosignal without analytical value can be effectively identified, more abundant and detailed quality feedback information can be provided, the whole process can be completed without other instruments and equipment and manual labeling, and the feasibility is strong, and the accuracy and the efficiency are high. Meanwhile, the invention can realize self-adaptive continuous quality assessment of electrocardiosignal segments, and avoids the problem that the signal quality can only be classified into two categories or three categories in the traditional deep learning.
The method has the advantages that the receptive field and the unreliability (namely the standard deviation of the signal fragments) are increased to the network design, the data distribution is wider, the model performance is improved, and the robustness is enhanced.
Based on the learning target of the network loss function, the learning target of the network is designed to be the result of quantification of the electrocardiosignal quality, namely the output value of the model is positively correlated with the signal quality. Meanwhile, uncertainty bias is introduced, so that quality fraction deviation is further reduced, and confidence is improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are required to be used in the embodiments or the description of the prior art will be briefly described below, and it is obvious that the drawings in the following description are only embodiments of the present invention, and that other drawings can be obtained according to the provided drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of an electrocardiosignal quality evaluation method based on a self-encoder provided by the invention;
FIG. 2 is a schematic diagram of a self-encoder model codec process provided by the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
As shown in fig. 1, the embodiment of the invention discloses an electrocardiosignal quality evaluation method based on a self-encoder, which comprises the following steps:
step 1, acquiring an original electrocardiosignal, and performing multilayer filtering pretreatment;
step 2, dividing the preprocessed signals into a plurality of segments with fixed lengths, and screening signal points of each signal segment to obtain a batch of primary screening electrocardiosignals;
step 3, training a pre-constructed self-encoder model by utilizing the primary screening electrocardiosignal so that the self-encoder model learns the general rule of the electrocardiosignal;
step 4, inputting the original signal to be evaluated into a trained self-encoder model for signal reconstruction, and calculating the uncertainty of the signal;
and 5, calculating the quality score of the original signal to be evaluated based on the reconstructed signal and the uncertainty.
The steps described above are further described below.
In the step 1, after the original electrocardiosignal is obtained, the operations of removing baseline drift, filtering power frequency signals, reducing noise of high-frequency signals and the like are carried out on the signals.
In step 2, the electrocardiograph signal obtained in step 1 is segmented and screened, and a section of continuously acquired electrocardiograph signal is segmented into a plurality of signal segments with fixed length (such as 5 s) and data standardization processing is performed. Wherein, the screening basis for the signal fragments is as follows: and judging the signal point duty ratio of each signal segment, wherein the amplitude of the signal point duty ratio exceeds a threshold value, and filtering out the signal points under the duty ratio if the signal point duty ratio exceeds a preset ratio.
Specifically, in the process of screening signal points, marking the signal points with the amplitudes exceeding the whole signal amplitude average value plus 2 x std in each signal segment as oversized signal points, and if the duty ratio of the oversized signal points in the whole signal segment exceeds 0.5, marking the oversized signal points under the duty ratio as filtered signals and filtering; where std denotes the standard deviation of the signal segments. The amplitude exceeds the whole amplitude mean value +2 std of the signal segment, and the signal point with the duty ratio exceeding 0.5 is mostly the signal seriously polluted by noise and drift in the production environment, and has no analysis value.
In step 3, the processed electrocardiosignal segments are input into a self-encoder model, and the model is trained. The self-encoder model is an unsupervised deep learning model, and the process is that electrocardiosignal fragments are encoded and then decoded through a deep neural network. A self-encoder is composed of an encoder and a decoder. The encoder is responsible for abstracting the input one-dimensional electrocardiosignals into multidimensional features, and the decoder is responsible for recovering feature vectors into original electrocardiosignals.
Specifically, as shown in fig. 2, the encoder adopts a convolutional neural network to extract features from an original one-dimensional electrocardiosignal and performs multi-layer convolutional expansion, can set expansion rate parameters which are more than 2 times, expands the range of a convolutional kernel, and abstracts the original electrocardiosignal into low-dimensional dense vector electrocardio features, namely the final flat layer output (mostly 1*n-dimensional dense vectors) of the convolutional network. Compared with the traditional convolution kernel, the expansion convolution kernel can expand the scope of the receptive field and promote the fusion of long-time span features in the time sequence.
The decoder reconstructs the low-dimensional dense vector electrocardiographic features by adopting transpose convolution, and calculates the unreliability of the reconstructed signals. Compared with forward convolution, the embodiment of the invention adopts transposed convolution, and can output the input characteristics as tensors with larger scales.
The uncertainty is essentially the standard deviation of the signal fragments, the fluctuation of the whole signal (deviating from the mean value) is measured, the parameter is introduced into the network design, the model can learn the morphological knowledge of the signal waveform, the larger the signal set in the sample is, the more the uncertainty diversity is, the more the distribution is confidence, and the model also has the capability of coping with the wide signal morphological distribution. For example, if a signal segment has a greater degree of uncertainty (greater standard deviation), this signal is morphologically distorted from the standard segment waveform morphology.
The transposed convolution of the reconstructed electrocardiosignal and the unreliability only makes a difference on the last layer, and the parameters of other layers are shared, so that the difference and the correlation of the two are ensured.
In a specific embodiment, the loss function AE-LLH of the self-encoder model is constructed using a log-likelihood function, expressed as follows:
wherein,,
representing a maximum log likelihood function after residual Gaussian mapping of the reconstructed signal and the original signal; x represents an original electrocardiosignal; μ represents the reconstructed signal; sigma represents the degree of uncertainty; l represents the total number of signal segments in the batch, and L represents the first signal segment.
The model training objective is to maximize the log-likelihood function to ensure that the reconstructed signal is close to the original signal. Between the convolutions, normalization and activation layer treatments were also performed in order to avoid gradient explosions and gradient vanishing.
In step 4, after the network training is completed, a self-encoder model which learns the general law of electrocardiosignals can be obtained. Dividing an original signal to be evaluated into signal segments with the same signal length as a training stage, inputting the signal segments into a self-encoder model, and reconstructing and calculating the uncertainty of each signal segment to be evaluated.
In step 5, according to the unreliability of the signal segment to be evaluated and the reconstructed signal calculated from the encoder model, a log-likelihood function is calculated, and a penalty term is added to the value of the unreliability. The final designed mass fraction (Score) is the mean value of the log-likelihood function minus the mean value of the penalty factor multiplied by the uncertainty, and the evaluation formula of the mass fraction is:
where Score represents the quality fraction of the signal to be evaluated, the signal quality can be measured and represented, lambda represents the penalty factor,
representing an uncertainty bias term.
In the Score loss function design, an uncertainty bias is introduced, the bias caused by the waveform form of the signal can be constrained as a regular penalty term, and for the signal form with a more biased standard form, the greater the term value is, the quality calibration of the signal is adjusted, so that the quality Score bias is further reduced, and the confidence is improved.
In other embodiments, further comprising:
step six, sequencing the quality scores of all the signal fragments under the original signal to be evaluated, and taking the signal fragments with the top n ranking or the quality score meeting the score threshold as high-quality signals. The proportion and the fraction threshold of the selected signal fragments are determined by the general quality of the collected electrocardiosignals and the discrimination requirements of the electrocardiosignals.
In the present specification, each embodiment is described in a progressive manner, and each embodiment is mainly described in a different point from other embodiments, and identical and similar parts between the embodiments are all enough to refer to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant points refer to the description of the method section.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (10)
1. An electrocardiosignal quality evaluation method based on a self-encoder is characterized by comprising the following steps of:
acquiring an original electrocardiosignal, and performing multilayer filtering pretreatment;
dividing the preprocessed signals into a plurality of segments with fixed lengths, and screening signal points of each signal segment to obtain a batch of primary screening electrocardiosignals;
training a pre-constructed self-encoder model by using the primary screening electrocardiosignal to enable the self-encoder model to learn the general rule of the electrocardiosignal;
inputting an original signal to be evaluated into a trained self-encoder model for signal reconstruction, and calculating the uncertainty of the signal;
and calculating the quality score of the original signal to be evaluated based on the reconstructed signal and the uncertainty.
2. The method for evaluating the quality of an electrocardiographic signal based on a self-encoder according to claim 1, wherein the preprocessing of the original electrocardiographic signal comprises at least: baseline drift, power frequency signal filtering and high-frequency signal noise reduction are removed.
3. The method for evaluating the quality of an electrocardiograph signal based on a self-encoder according to claim 1, wherein the screening basis of the signal segments is: and judging the signal point duty ratio of each signal segment, wherein the amplitude of the signal point duty ratio exceeds a threshold value, and filtering out the signal points under the duty ratio if the signal point duty ratio exceeds a preset ratio.
4. The method for evaluating the quality of an electrocardiograph signal based on a self-encoder according to claim 3, wherein in the process of screening signal points, signal points with the amplitude exceeding the whole signal amplitude average value plus 2 x std in each signal segment are marked as oversized signal points, and if the duty ratio of the oversized signal points in the whole signal segment exceeds 0.5, the oversized signal points under the duty ratio are marked as filtered signals and are filtered; where std denotes the standard deviation of the signal segment.
5. The method for evaluating the quality of an electrocardiographic signal based on a self-encoder according to claim 1, further comprising, before screening the signal points in each signal segment: and carrying out data standardization processing on each segmented signal segment.
6. The method for evaluating the quality of an electrocardiographic signal based on a self-encoder according to claim 1, wherein the self-encoder model is composed of an encoder and a decoder; the encoder adopts a convolutional neural network to extract features from an original one-dimensional electrocardiosignal and performs multi-layer convolutional expansion, so that the original electrocardiosignal is abstracted into low-dimensional dense vector electrocardiosignal features; the decoder reconstructs the low-dimensional dense vector electrocardiographic features by adopting transpose convolution, and calculates the unreliability of the reconstructed signals.
7. The method for evaluating the quality of an electrocardiographic signal based on a self-encoder according to claim 1, wherein a loss function AE-LLH of the self-encoder model is constructed by using a log-likelihood function, and the expression is as follows:
wherein,,
representing a log-likelihood function; x represents an original electrocardiosignal; μ represents the reconstructed signal; sigma represents the degree of uncertainty; l represents the total number of signal segments in the batch, and L represents the first signal segment.
8. The method for evaluating the quality of an electrocardiographic signal based on a self-encoder according to claim 7, wherein the evaluation formula of the quality score of the original signal to be evaluated is:
where Score represents the quality fraction of the signal to be evaluated, lambda represents the penalty factor,
representing an uncertainty bias term.
9. The method for evaluating the quality of an electrocardiographic signal based on a self-encoder according to claim 1, further comprising: dividing an original signal to be evaluated into signal segments with the same signal length as a training stage, and carrying out reconstruction and uncertainty calculation on each signal segment based on a trained self-encoder model.
10. The method for evaluating the quality of an electrocardiographic signal based on a self-encoder according to claim 9, further comprising: and sequencing the quality scores of the signal fragments under the original signal to be evaluated, and taking the signal fragments with the top n ranks or the quality score meeting the score threshold as the high-quality signal.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310313430.8A CN116350233A (en) | 2023-03-28 | 2023-03-28 | Electrocardiosignal quality assessment method based on self-encoder |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310313430.8A CN116350233A (en) | 2023-03-28 | 2023-03-28 | Electrocardiosignal quality assessment method based on self-encoder |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116350233A true CN116350233A (en) | 2023-06-30 |
Family
ID=86917281
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310313430.8A Pending CN116350233A (en) | 2023-03-28 | 2023-03-28 | Electrocardiosignal quality assessment method based on self-encoder |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116350233A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116919414A (en) * | 2023-07-06 | 2023-10-24 | 齐鲁工业大学(山东省科学院) | Electrocardiosignal quality assessment method based on multi-scale convolution and dense connection network |
-
2023
- 2023-03-28 CN CN202310313430.8A patent/CN116350233A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116919414A (en) * | 2023-07-06 | 2023-10-24 | 齐鲁工业大学(山东省科学院) | Electrocardiosignal quality assessment method based on multi-scale convolution and dense connection network |
| CN116919414B (en) * | 2023-07-06 | 2024-02-13 | 齐鲁工业大学(山东省科学院) | Electrocardiosignal quality assessment method based on multi-scale convolution and dense connection network |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107822622B (en) | Electrocardiogram diagnosis method and system based on deep convolutional neural network | |
| Hsieh et al. | Feature selection framework for XGBoost based on electrodermal activity in stress detection | |
| CN111407243B (en) | Pulse signal pressure identification method based on deep learning | |
| CN112200162B (en) | Non-contact heart rate measuring method, system and device based on end-to-end network | |
| CN109875543B (en) | Heart rate estimation method and device for wearable heart rate monitoring equipment under various fitness exercise states | |
| CN111967354B (en) | Depression tendency identification method based on multi-mode characteristics of limbs and micro-expressions | |
| CN113349752A (en) | Wearable device real-time heart rate monitoring method based on sensing fusion | |
| Azar et al. | Deep recurrent neural network-based autoencoder for photoplethysmogram artifacts filtering | |
| CN108567418A (en) | A kind of pulse signal inferior health detection method and detecting system based on PCANet | |
| CN112370015A (en) | Physiological signal quality evaluation method based on gram angular field | |
| CN116350233A (en) | Electrocardiosignal quality assessment method based on self-encoder | |
| CN106599903B (en) | Signal reconstruction method for weighted least square dictionary learning based on correlation | |
| CN111387975A (en) | Electroencephalogram signal identification method based on machine learning | |
| CN115221926A (en) | Heart beat signal classification method based on CNN-GRU network model | |
| Chatterjee et al. | Signal quality assessment of photoplethysmogram signals using quantum pattern recognition technique and lightweight cnn module | |
| CN114595725B (en) | Electroencephalogram signal classification method based on addition network and supervised contrast learning | |
| CN115530788A (en) | Arrhythmia classification method based on self-attention mechanism | |
| CN113576472B (en) | Blood oxygen signal segmentation method based on full convolution neural network | |
| CN114403897A (en) | Human body fatigue detection method and system based on electroencephalogram signals | |
| Wang et al. | Ppg signal reconstruction using deep convolutional generative adversarial network | |
| CN112215196B (en) | Electrocardiogram identity recognition method | |
| Zhang et al. | Heart sound classification and recognition based on EEMD and correlation dimension | |
| CN115281676A (en) | Fatigue detection method based on GRU neural network and ECG signal | |
| CN115293210A (en) | Instruction prediction output control method based on brain waves | |
| Kumble et al. | Evalutionary STBD model for bio-signal compression provisioning in wire-less sensor network |
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 |