CN107049239A - Epileptic electroencephalogram (eeg) feature extracting method based on wearable device - Google Patents
Epileptic electroencephalogram (eeg) feature extracting method based on wearable device Download PDFInfo
- Publication number
- CN107049239A CN107049239A CN201611239245.5A CN201611239245A CN107049239A CN 107049239 A CN107049239 A CN 107049239A CN 201611239245 A CN201611239245 A CN 201611239245A CN 107049239 A CN107049239 A CN 107049239A
- Authority
- CN
- China
- Prior art keywords
- feature
- feature extraction
- wearable device
- electroencephalogram
- kernel
- 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
- 230000001037 epileptic effect Effects 0.000 title claims abstract description 22
- 238000000034 method Methods 0.000 title claims abstract description 17
- 238000000605 extraction Methods 0.000 claims abstract description 52
- 230000005012 migration Effects 0.000 claims abstract description 35
- 238000013508 migration Methods 0.000 claims abstract description 35
- 238000012706 support-vector machine Methods 0.000 claims abstract description 20
- 238000004364 calculation method Methods 0.000 claims abstract description 16
- 239000011159 matrix material Substances 0.000 claims description 57
- 206010015037 epilepsy Diseases 0.000 claims description 45
- 238000000513 principal component analysis Methods 0.000 claims description 36
- 238000012545 processing Methods 0.000 claims description 16
- 238000012549 training Methods 0.000 claims description 13
- 238000001914 filtration Methods 0.000 claims description 9
- 230000006870 function Effects 0.000 claims description 9
- 238000007781 pre-processing Methods 0.000 claims description 9
- 238000013507 mapping Methods 0.000 claims description 6
- 230000001131 transforming effect Effects 0.000 claims description 2
- 210000004556 brain Anatomy 0.000 claims 2
- 230000008569 process Effects 0.000 abstract description 3
- 238000001514 detection method Methods 0.000 description 3
- 238000012880 independent component analysis Methods 0.000 description 3
- 238000013526 transfer learning Methods 0.000 description 2
- 208000012902 Nervous system disease Diseases 0.000 description 1
- 230000005856 abnormality Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000002490 cerebral effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 230000001314 paroxysmal effect Effects 0.000 description 1
- 238000012163 sequencing technique Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 230000001052 transient effect Effects 0.000 description 1
Classifications
-
- 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
- A61B5/7203—Signal processing specially adapted for physiological signals or for diagnostic purposes for noise prevention, reduction or removal
-
- 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
- A61B5/369—Electroencephalography [EEG]
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/68—Arrangements of detecting, measuring or recording means, e.g. sensors, in relation to patient
- A61B5/6801—Arrangements of detecting, measuring or recording means, e.g. sensors, in relation to patient specially adapted to be attached to or worn on the body surface
-
- 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
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Biophysics (AREA)
- General Health & Medical Sciences (AREA)
- Veterinary Medicine (AREA)
- Public Health (AREA)
- Animal Behavior & Ethology (AREA)
- Surgery (AREA)
- Molecular Biology (AREA)
- Medical Informatics (AREA)
- Heart & Thoracic Surgery (AREA)
- Pathology (AREA)
- Biomedical Technology (AREA)
- Artificial Intelligence (AREA)
- Signal Processing (AREA)
- Psychiatry (AREA)
- Physiology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Mathematical Physics (AREA)
- Fuzzy Systems (AREA)
- Psychology (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
The present invention discloses a kind of epileptic electroencephalogram (eeg) feature extracting method based on wearable device, comprises the following steps:The unmarked epileptic electroencephalogram (eeg) data of marked epileptic electroencephalogram (eeg) data and the wearable device collection gathered according to hospital, the calculation process calculated by core principle component analysis, multi-kernel support vector machine and migration feature, the grader that output is migrated with feature;The unmarked real-time epileptic electroencephalogram (eeg) data gathered to wearable device carry out feature extraction using core principle component analysis, the unmarked real-time epileptic electroencephalogram (eeg) data after feature extraction are put into grader, output predicts the outcome.The present invention is integrated the image resource of each hospital image section office and image center, is formed an epileptic electroencephalogram (eeg) feature extracting method based on wearable device for serving clinic, is improved clinical position efficiency.The epileptic electroencephalogram (eeg) feature extracting method that the present invention is provided can quickly, accurately be used for feature extraction and the prediction of result of the epileptic electroencephalogram (eeg) of wearable device.
Description
Technical Field
The invention relates to the technical field of electroencephalogram data processing, in particular to an epilepsia electroencephalogram feature extraction method based on wearable equipment.
Background
Epilepsy is a transient cerebral dysfunction caused by paroxysmal abnormality of cerebral neurons, and has the characteristics of complexity and recurrence. EEG (electroencephalogram) is an important means for recognizing epilepsy, and epilepsy electroencephalogram recognition based on a feature extraction technology has recently attracted more and more attention.
The wearable electroencephalogram detection equipment provides great convenience for patients and medical personnel, but due to the particularity and timeliness of the wearable electroencephalogram detection equipment, the feasibility of manual marking of the wearable electroencephalogram detection equipment is low.
Disclosure of Invention
Aiming at the defects in the technology, the invention provides the wearable equipment-based epilepsia electroencephalogram feature extraction method, and the method for extracting the epilepsia electroencephalogram feature for the wearable equipment is rapid and high in accuracy through the methods of kernel principal component analysis, multi-core support vector machine and transfer learning.
To achieve these objects and other advantages in accordance with the present invention, the present invention is implemented by the following solutions:
the invention provides a wearable device-based epilepsia electroencephalogram feature extraction method, which comprises the following steps:
outputting a classifier with feature migration through the operation processing of kernel principal component analysis, a multi-kernel support vector machine and migration feature calculation according to marked epilepsy electroencephalogram data collected by a hospital and unmarked epilepsy electroencephalogram data collected by wearable equipment;
and performing feature extraction on the unmarked real-time epilepsy electroencephalogram data acquired by the wearable equipment by adopting kernel principal component analysis, putting the unmarked real-time epilepsy electroencephalogram data after the feature extraction into the classifier, and outputting a prediction result.
Preferably, the operation processing of the kernel principal component analysis, the multi-kernel support vector machine and the migration feature calculation includes the following steps:
acquiring marked epilepsy electroencephalogram data collected by a hospital and unmarked epilepsy electroencephalogram data collected by wearable equipment in groups, and respectively preprocessing the two groups of data;
performing feature extraction on the preprocessed marked epilepsy electroencephalogram data by adopting kernel principal component analysis, and outputting a first feature parameter matrix;
performing classifier training on the marked epilepsy electroencephalogram data after feature extraction by adopting a multi-core support vector machine, and outputting a first classifier;
performing feature extraction on the preprocessed unmarked epilepsia electroencephalogram data by adopting kernel principal component analysis, and outputting a second feature parameter matrix;
performing space calculation on the first characteristic parameter matrix and the second characteristic parameter matrix, and outputting migration characteristics;
and putting the migration features into the classifier, performing migration learning, and outputting an updated second classifier.
Preferably, the spatial calculation of the first characteristic parameter matrix and the second characteristic parameter matrix and the output of the migration characteristic include the following steps:
defining the Euclidean distance between the n-dimensional principal components of the marked epileptic electroencephalogram data in the first characteristic parameter matrix and the unmarked epileptic electroencephalogram data in the second characteristic parameter matrix as dis (n), and then, if dis (n) ═ PCAlabel-PCAunlabel;
Setting a distance threshold, sorting dis (n), deleting the feature dimension of which the value of dis (n) is smaller than the distance threshold, and finishing feature migration.
Preferably, the method for extracting the characteristics of the unmarked real-time epilepsia electroencephalogram data collected by the wearable device by adopting kernel principal component analysis comprises the following steps:
preprocessing the unmarked real-time epilepsia electroencephalogram data;
and performing feature extraction on the preprocessed unmarked real-time epilepsia electroencephalogram data by adopting kernel principal component analysis, and outputting a third feature parameter matrix.
Preferably, the preprocessing includes sequentially performing a process of removing electro-oculogram, filtering, and baseline calibration on each set of data.
Preferably, the filtering comprises EEMD-ICA based filtering.
Preferably, the method further comprises the steps of:
adding unmarked real-time epilepsia electroencephalogram data acquired by wearable equipment into the unmarked epilepsia electroencephalogram data, and outputting a classifier updated in real time during operation processing together with marked epilepsia electroencephalogram data acquired by a hospital.
Preferably, the feature extraction by using the kernel principal component analysis comprises the following steps:
the mapping of the input space X to the feature space F, i.e. X, is realized by transforming hl=>h(xl);
Principal component analysis is used in the feature space, namely, a feature solution is solved: lambda [ alpha ]iui=Cui(ii) a 1, 2.. ·, l; wherein,is a sample covariance matrix, λ, in the feature spaceiIs a non-zero eigenvalue of C, uiIs λiThe corresponding feature vector;
and arranging the eigenvectors into a matrix from top to bottom according to the size of the corresponding eigenvalue, and taking the first k rows to form a matrix P, wherein P is the obtained eigenvector matrix.
Preferably, the classifier training is performed by using a multi-core support vector machine, and comprises the following steps:
a kernel matrix is established for each feature using radial basis functions: based on n training samples, the feature vector of the ith sample is defined asThe label corresponding to each feature vector is yi-1,1 }; wherein M represents M types of features;
using a weighting factor βmAnd forming a mixed kernel matrix by the multi-kernel support vector machine, wherein the mixed kernel matrix is defined as:
the decision function is
Wherein,when 0 is less than or equal to αiWhen the content is less than or equal to C,phi (-) represents a kernel-function-guided mapping function; k is a radical of(m)(xi (m),xj (m)) Representing a training sample xi (m)And xj (m)In the mth characteristic, the kernel matrix is a Lagrange multiplier, and represents the inner product operation, βmAnd more than or equal to 0 represents the weight factor of the mth characteristic, and C represents the number of the constraint conditions in the model parameter.
The invention at least comprises the following beneficial effects:
according to the epilepsy electroencephalogram feature extraction method based on the wearable device, a classifier with feature migration is output through operation processing of kernel principal component analysis, a multi-core support vector machine and migration feature calculation according to marked epilepsy electroencephalogram data collected by a hospital and unmarked epilepsy electroencephalogram data collected by the wearable device, unmarked real-time epilepsy electroencephalogram data collected by the wearable device after feature extraction is put into the classifier, a prediction result is output, and the method can be used for feature extraction and result prediction of the epilepsy electroencephalogram of the wearable device quickly and accurately.
Additional advantages, objects, and features of the invention will be set forth in part in the description which follows and in part will become apparent to those having ordinary skill in the art upon examination of the following or may be learned from practice of the invention.
Drawings
Fig. 1 is a schematic diagram of an epilepsia electroencephalogram feature extraction method based on wearable equipment;
FIG. 2 is a flow chart of an epilepsia electroencephalogram feature extraction method based on a wearable device according to the present invention;
FIG. 3 is a flowchart of the computational processing of the kernel principal component analysis, the multi-kernel support vector machine, and the migration feature calculation of the present invention;
FIG. 4 is a flow chart of a method for feature extraction using kernel principal component analysis according to the present invention.
Detailed Description
The present invention is further described in detail below with reference to the attached drawings so that those skilled in the art can implement the invention by referring to the description text.
It will be understood that terms such as "having," "including," and "comprising," as used herein, do not preclude the presence or addition of one or more other elements or groups thereof.
As shown in fig. 1 and fig. 2, the invention provides a wearable device-based epilepsia electroencephalogram feature extraction method, which includes the following steps:
s10, outputting a classifier with feature migration through operation processing of kernel principal component analysis, a multi-kernel support vector machine and migration feature calculation according to marked epilepsy electroencephalogram data acquired by a hospital and unmarked epilepsy electroencephalogram data acquired by wearable equipment;
and S20, performing feature extraction on the unmarked real-time epilepsy electroencephalogram data acquired by the wearable device by adopting kernel principal component analysis, putting the unmarked real-time epilepsy electroencephalogram data after feature extraction into a classifier, and outputting a prediction result.
In the above embodiment, as can be seen from fig. 1 and 3, the operation processing of the kernel principal component analysis, the multi-kernel support vector machine, and the migration feature calculation in step S10 includes the following steps:
s11, acquiring marked epilepsy electroencephalogram data acquired by a hospital and unmarked epilepsy electroencephalogram data acquired by wearable equipment in groups, and preprocessing the two groups of data respectively;
s12, performing feature extraction on the preprocessed marked epilepsy electroencephalogram data by adopting kernel principal component analysis, and outputting a first feature parameter matrix;
s13, performing classifier training on the marked epilepsy electroencephalogram data after feature extraction by adopting a multi-core support vector machine, and outputting a first classifier; the method for training the classifier by adopting the multi-core support vector machine specifically comprises the following steps:
s131, establishing a kernel matrix for each feature by using the radial basis function: based on n training samples, the feature vector of the ith sample is defined as xi={xi (1),...,xi (M)And y is the label corresponding to each feature vectori-1,1 }; wherein M represents M types of features;
s132, applying a weighting factor βmAnd forming a mixed kernel matrix by the multi-kernel support vector machine, wherein the mixed kernel matrix is defined as:
the decision function is
Wherein,when 0 is less than or equal to αiWhen the content is less than or equal to C,phi (-) represents a kernel-function-guided mapping function; k is a radical of(m)(xi (m),xj (m)) Representing a training sample xi (m)And xj (m)In the mth characteristic, the kernel matrix is a Lagrange multiplier, and represents the inner product operation, βmAnd more than or equal to 0 represents the weight factor of the mth characteristic, and C represents the number of the constraint conditions in the model parameter.
S14, performing feature extraction on the preprocessed unmarked epilepsia electroencephalogram data by adopting kernel principal component analysis, and outputting a second feature parameter matrix;
s15, performing spatial calculation on the first characteristic parameter matrix and the second characteristic parameter matrix, and outputting the migration characteristic, specifically including the following steps:
s151, defining an euclidean distance between the n-th dimensional principal components of the labeled epileptic electroencephalogram data in the first characteristic parameter matrix and the unlabeled epileptic electroencephalogram data in the second characteristic parameter matrix as dis (n), and then, dis (n) ═ PCAlabel-PCAunlabel;
S152, setting a distance threshold, sequencing dis (n), deleting the feature dimension of which the value of dis (n) is smaller than the distance threshold, and finishing feature migration.
And S16, putting the migration features into the classifier, performing migration learning, and outputting an updated second classifier. In the above embodiment, the feature extraction by using kernel principal component analysis includes the following steps:
s111, mapping the input space X to the feature space F is realized through transformation h, namely Xl=>h(xl);
And S112, using principal component analysis in the feature space, namely solving a feature solution: lambda [ alpha ]iui=Cui(ii) a 1, 2.. ·, l; wherein,is specially designed forSample covariance matrix in eigenspace, λiIs a non-zero eigenvalue of C, uiIs λiThe corresponding feature vector;
s113, arranging the eigenvectors into a matrix from top to bottom according to the size of the corresponding eigenvalue, and taking the first k rows to form a matrix P, wherein P is the obtained eigenvector. In this embodiment, since a large number of complex nonlinear relations exist in the unlabeled epileptic electroencephalogram data acquired by the wearable device, compared with the Principal Component Analysis (PCA) in the prior art, the Kernel Principal Component Analysis (KPCA) adopted in the present application is more suitable for feature extraction of high-dimensional nonlinear data, and has higher processing accuracy.
Transfer learning can identify knowledge learned from a previous domain and apply it to another domain. Therefore, in the step S15 and the step S16, the step of putting the output migration features and the migration features into a classifier for migration learning means that the first feature parameter matrix is used as a source field, the second feature parameter matrix is used as a target field, and the first feature parameter matrix and the second feature parameter matrix are respectively subjected to feature extraction by using kernel principal component analysis, so that data of the first feature parameter matrix and data of the second feature parameter matrix belong to one feature space, and only feature distribution is different, so that the first feature parameter matrix and the second feature parameter matrix are subjected to spatial calculation, and after the migration features are output, the migration features are put into the classifier for migration learning, so that the second classifier suitable for classifying unmarked epileptic electroencephalogram data acquired by wearable equipment corresponding to the second feature parameter matrix can be output. The second classifier is used as an initial value, and can be used for training the unmarked real-time epilepsia electroencephalogram data after the feature extraction in the step S20, so that a prediction result is obtained.
In the foregoing embodiment, as shown in fig. 4, the performing, in step S20, feature extraction on the unmarked real-time epilepsy electroencephalogram data acquired by the wearable device by using kernel principal component analysis includes the following steps:
s21, preprocessing unmarked real-time epilepsia electroencephalogram data;
and S22, performing feature extraction on the preprocessed unmarked real-time epilepsia electroencephalogram data by adopting kernel principal component analysis, and outputting a third feature parameter matrix. In this embodiment, the preprocessing includes performing a process of removing electro-oculogram, filtering, and baseline calibration on each set of data in sequence. Further, the filtering includes filtering based on EEMD (noise aided data analysis method) -ICA (Independent Component analysis).
According to the wearable device-based epilepsy electroencephalogram feature extraction method provided by the embodiment, firstly, according to marked epilepsy electroencephalogram data collected by a hospital and unmarked epilepsy electroencephalogram data collected by a wearable device, through operation processing of kernel principal component analysis, a multi-core support vector machine and migration feature calculation, after a classifier with feature migration is output, namely, the second classifier in the embodiment is adopted, a large amount of unmarked real-time epilepsy electroencephalogram data collected subsequently by the wearable device can be classified and predicted through the second classifier, and a wearable device holder can conveniently monitor the unmarked real-time epilepsy electroencephalogram data in real time and obtain a high-precision classification result.
As another embodiment of the present invention, the wearable device-based epileptic electroencephalogram feature extraction method further includes the steps of:
and S30, adding the unmarked real-time epilepsy electroencephalogram data acquired by the wearable device into the unmarked epilepsy electroencephalogram data, and outputting a classifier updated in real time during the operation processing together with the marked epilepsy electroencephalogram data acquired by the hospital.
In the embodiment, the unmarked real-time epilepsy electroencephalogram data collected by the wearable device is added into the unmarked epilepsy electroencephalogram data, and the unmarked real-time epilepsy electroencephalogram data and the marked epilepsy electroencephalogram data collected by the hospital are subjected to operation processing together to update the classifier in real time, namely update the second classifier in real time. Along with the continuous change and the update of unmarked real-time epilepsia electroencephalogram data collected by the wearable device, the model of the second classifier can be gradually optimized, and the classification accuracy is improved.
While embodiments of the invention have been disclosed above, it is not intended to be limited to the uses set forth in the specification and examples. It can be applied to all kinds of fields suitable for the present invention. Additional modifications will readily occur to those skilled in the art. It is therefore intended that the invention not be limited to the exact details and illustrations described and illustrated herein, but fall within the scope of the appended claims and equivalents thereof.
Claims (9)
1. A epilepsia electroencephalogram feature extraction method based on wearable equipment is characterized by comprising the following steps:
outputting a classifier with feature migration through the operation processing of kernel principal component analysis, a multi-kernel support vector machine and migration feature calculation according to marked epilepsy electroencephalogram data collected by a hospital and unmarked epilepsy electroencephalogram data collected by wearable equipment;
and performing feature extraction on the unmarked real-time epilepsy electroencephalogram data acquired by the wearable equipment by adopting kernel principal component analysis, putting the unmarked real-time epilepsy electroencephalogram data after the feature extraction into the classifier, and outputting a prediction result.
2. The wearable device-based epilepsia electroencephalogram feature extraction method as claimed in claim 1, wherein the computational processing of kernel principal component analysis, multi-kernel support vector machine and migration feature calculation comprises the following steps:
acquiring marked epilepsy electroencephalogram data collected by a hospital and unmarked epilepsy electroencephalogram data collected by wearable equipment in groups, and respectively preprocessing the two groups of data;
performing feature extraction on the preprocessed marked epilepsy electroencephalogram data by adopting kernel principal component analysis, and outputting a first feature parameter matrix;
performing classifier training on the marked epilepsy electroencephalogram data after feature extraction by adopting a multi-core support vector machine, and outputting a first classifier;
performing feature extraction on the preprocessed unmarked epilepsia electroencephalogram data by adopting kernel principal component analysis, and outputting a second feature parameter matrix;
performing space calculation on the first characteristic parameter matrix and the second characteristic parameter matrix, and outputting migration characteristics;
and putting the migration features into the classifier, performing migration learning, and outputting an updated second classifier.
3. The wearable device-based epilepsia electroencephalogram feature extraction method as claimed in claim 2, wherein the spatial calculation is performed on the first feature parameter matrix and the second feature parameter matrix, and the migration feature is output, comprising the following steps:
defining the Euclidean distance between the n-dimensional principal components of the marked epileptic electroencephalogram data in the first characteristic parameter matrix and the unmarked epileptic electroencephalogram data in the second characteristic parameter matrix as dis (n), and then, if dis (n) ═ PCAlabel-PCAunlabel;
Setting a distance threshold, sorting dis (n), deleting the feature dimension of which the value of dis (n) is smaller than the distance threshold, and finishing feature migration.
4. The wearable device-based epilepsia electroencephalogram feature extraction method as claimed in claim 1, wherein feature extraction is performed on unlabeled real-time epilepsia electroencephalogram data collected by the wearable device by adopting kernel principal component analysis, and the method comprises the following steps:
preprocessing the unmarked real-time epilepsia electroencephalogram data;
and performing feature extraction on the preprocessed unmarked real-time epilepsia electroencephalogram data by adopting kernel principal component analysis, and outputting a third feature parameter matrix.
5. The wearable device based epileptic brain electrical characteristic extraction method as claimed in claim 2 or 3, wherein the preprocessing comprises sequentially processing each set of data for eye electrical removal, filtering and baseline calibration.
6. The wearable device based epileptic brain electrical feature extraction method of claim 4, wherein the filtering comprises EEMD-ICA based filtering.
7. The wearable device-based epileptic electroencephalogram feature extraction method according to any one of claims 1-6, further comprising the steps of:
adding unmarked real-time epilepsia electroencephalogram data acquired by wearable equipment into the unmarked epilepsia electroencephalogram data, and outputting a classifier updated in real time during operation processing together with marked epilepsia electroencephalogram data acquired by a hospital.
8. The wearable device based epilepsia electroencephalogram feature extraction method as claimed in any one of claims 1-6, wherein feature extraction is performed by using the kernel principal component analysis, and the method comprises the following steps:
the mapping of the input space X to the feature space F, i.e. X, is realized by transforming hl=>h(xl);
Principal component analysis is used in the feature space, namely, a feature solution is solved: lambda [ alpha ]iui=Cui(ii) a 1, 2.. ·, l; wherein,is a sample covariance matrix, λ, in the feature spaceiIs a non-zero eigenvalue of C, uiIs λiThe corresponding feature vector;
and arranging the eigenvectors into a matrix from top to bottom according to the size of the corresponding eigenvalue, and taking the first k rows to form a matrix P, wherein P is the obtained eigenvector matrix.
9. The wearable device-based epileptic electroencephalogram feature extraction method as claimed in any one of claims 1-6, wherein a multi-core support vector machine is adopted for classifier training, comprising the following steps:
a kernel matrix is established for each feature using radial basis functions: based on n training samples, the feature vector of the ith sample is defined as xi={xi (1),...,xi (M)And y is the label corresponding to each feature vectori-1,1 }; wherein M represents M types of features;
using a weighting factor βmAnd forming a mixed kernel matrix by the multi-kernel support vector machine, wherein the mixed kernel matrix is defined as:
the decision function is
Wherein k is(m)(xi (m),xj (m))=<φ(xi (m)),φ(xj (m)) When 0 is not more than αiWhen the content is less than or equal to C,phi (-) represents a kernel-function-guided mapping function; k is a radical of(m)(xi (m),xj (m)) Representing a training sample xi (m)And xj (m)In the mth characteristic, the kernel matrix is a Lagrange multiplier, and represents the inner product operation, βmAnd more than or equal to 0 represents the weight factor of the mth characteristic, and C represents the number of the constraint conditions in the model parameter.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611239245.5A CN107049239A (en) | 2016-12-28 | 2016-12-28 | Epileptic electroencephalogram (eeg) feature extracting method based on wearable device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611239245.5A CN107049239A (en) | 2016-12-28 | 2016-12-28 | Epileptic electroencephalogram (eeg) feature extracting method based on wearable device |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107049239A true CN107049239A (en) | 2017-08-18 |
Family
ID=59624353
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201611239245.5A Pending CN107049239A (en) | 2016-12-28 | 2016-12-28 | Epileptic electroencephalogram (eeg) feature extracting method based on wearable device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107049239A (en) |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107684424A (en) * | 2017-10-09 | 2018-02-13 | 南京大学 | It is a kind of to remove the method that redundant signals disturb in high density electroencephalogram |
CN110013249A (en) * | 2019-03-19 | 2019-07-16 | 西北大学 | A kind of Portable adjustable wears seizure monitoring instrument |
WO2019211676A1 (en) * | 2018-05-01 | 2019-11-07 | International Business Machines Corporation | Epilepsy seizure detection and prediction using techniques such as deep learning methods |
CN110448273A (en) * | 2019-08-29 | 2019-11-15 | 江南大学 | A kind of low-power consumption epileptic prediction circuit based on support vector machines |
WO2020061971A1 (en) * | 2018-09-27 | 2020-04-02 | 电子科技大学 | Epilepsy brain wave state detection method based on machine learning |
CN111387975A (en) * | 2020-03-20 | 2020-07-10 | 徐州市健康研究院有限公司 | Electroencephalogram signal identification method based on machine learning |
CN111603135A (en) * | 2020-05-11 | 2020-09-01 | 江南大学 | Low-power-consumption epilepsy detection circuit based on master-slave support vector machine |
CN113261974A (en) * | 2021-06-07 | 2021-08-17 | 吉林大学 | Sports fatigue monitoring method based on multiple physiological signals |
CN113951898A (en) * | 2021-10-15 | 2022-01-21 | 浙江大学 | P300 electroencephalogram signal detection method and device for data migration, electronic device and medium |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102291392A (en) * | 2011-07-22 | 2011-12-21 | 中国电力科学研究院 | Hybrid intrusion detection method based on bagging algorithm |
CN104523269A (en) * | 2015-01-15 | 2015-04-22 | 江南大学 | Self-adaptive recognition method orienting epilepsy electroencephalogram transfer environment |
CN104997507A (en) * | 2015-07-09 | 2015-10-28 | 郑以山 | Three-in-one hat type intelligent monitoring early warning system and control method thereof |
-
2016
- 2016-12-28 CN CN201611239245.5A patent/CN107049239A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102291392A (en) * | 2011-07-22 | 2011-12-21 | 中国电力科学研究院 | Hybrid intrusion detection method based on bagging algorithm |
CN104523269A (en) * | 2015-01-15 | 2015-04-22 | 江南大学 | Self-adaptive recognition method orienting epilepsy electroencephalogram transfer environment |
CN104997507A (en) * | 2015-07-09 | 2015-10-28 | 郑以山 | Three-in-one hat type intelligent monitoring early warning system and control method thereof |
Non-Patent Citations (6)
Title |
---|
XIAOOU LI等: "Classification of EEG Signals Using a Multiple Kernel Learning Support Vector Machine", 《SENSORS》 * |
刘桂峰 等: "一种改进的多源域多视角学习算法", 《青岛大学学报(自然科学版)》 * |
杨昌健 等: "基于迁移学习的癫痫EEG信号自适应识别", 《计算机科学与探索》 * |
杨昌健 等: "引入迁移学习的癫痫EEG信号自适应识别", 《计算机工程》 * |
蔡静颖 著: "《模糊聚类算法及应用》", 31 August 2015 * |
赵强利: "基于选择性集成的在线机器学习关键技术研究", 《中国博士学位论文全文数据库 信息科技辑》 * |
Cited By (17)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107684424A (en) * | 2017-10-09 | 2018-02-13 | 南京大学 | It is a kind of to remove the method that redundant signals disturb in high density electroencephalogram |
CN112106074A (en) * | 2018-05-01 | 2020-12-18 | 国际商业机器公司 | Seizure detection and prediction using techniques such as deep learning methods |
GB2588523B (en) * | 2018-05-01 | 2022-03-09 | Ibm | Epilepsy seizure detection and prediction using techniques such as deep learning methods |
WO2019211676A1 (en) * | 2018-05-01 | 2019-11-07 | International Business Machines Corporation | Epilepsy seizure detection and prediction using techniques such as deep learning methods |
US11219405B2 (en) | 2018-05-01 | 2022-01-11 | International Business Machines Corporation | Epilepsy seizure detection and prediction using techniques such as deep learning methods |
GB2588523A (en) * | 2018-05-01 | 2021-04-28 | Ibm | Epilepsy seizure detection and prediction using techniques such as deep learning methods |
WO2020061971A1 (en) * | 2018-09-27 | 2020-04-02 | 电子科技大学 | Epilepsy brain wave state detection method based on machine learning |
CN110013249B (en) * | 2019-03-19 | 2022-02-18 | 西北大学 | Portable adjustable head-mounted epilepsy monitor |
CN110013249A (en) * | 2019-03-19 | 2019-07-16 | 西北大学 | A kind of Portable adjustable wears seizure monitoring instrument |
CN110448273A (en) * | 2019-08-29 | 2019-11-15 | 江南大学 | A kind of low-power consumption epileptic prediction circuit based on support vector machines |
CN111387975A (en) * | 2020-03-20 | 2020-07-10 | 徐州市健康研究院有限公司 | Electroencephalogram signal identification method based on machine learning |
CN111387975B (en) * | 2020-03-20 | 2022-06-17 | 徐州市健康研究院有限公司 | Electroencephalogram signal identification method based on machine learning |
CN111603135A (en) * | 2020-05-11 | 2020-09-01 | 江南大学 | Low-power-consumption epilepsy detection circuit based on master-slave support vector machine |
CN111603135B (en) * | 2020-05-11 | 2021-09-28 | 江南大学 | Low-power-consumption epilepsy detection circuit based on master-slave support vector machine |
CN113261974A (en) * | 2021-06-07 | 2021-08-17 | 吉林大学 | Sports fatigue monitoring method based on multiple physiological signals |
CN113951898A (en) * | 2021-10-15 | 2022-01-21 | 浙江大学 | P300 electroencephalogram signal detection method and device for data migration, electronic device and medium |
CN113951898B (en) * | 2021-10-15 | 2023-03-10 | 浙江大学 | P300 electroencephalogram signal detection method and device for data migration, electronic equipment and medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107049239A (en) | Epileptic electroencephalogram (eeg) feature extracting method based on wearable device | |
Hemanth et al. | RETRACTED ARTICLE: An enhanced diabetic retinopathy detection and classification approach using deep convolutional neural network | |
Kesav et al. | Efficient and low complex architecture for detection and classification of Brain Tumor using RCNN with Two Channel CNN | |
Pan et al. | Accurate segmentation of nuclei in pathological images via sparse reconstruction and deep convolutional networks | |
Sharma et al. | Brain tumor detection based on machine learning algorithms | |
Akram et al. | Identification and classification of microaneurysms for early detection of diabetic retinopathy | |
Jambhekar | Red blood cells classification using image processing | |
Keivani et al. | Automated analysis of leaf shape, texture, and color features for plant classification. | |
Fabietti et al. | Adaptation of convolutional neural networks for multi-channel artifact detection in chronically recorded local field potentials | |
Goel et al. | Prediction model for automated leaf disease detection & analysis | |
Chakraborty et al. | Automated breast cancer identification by analyzing histology slides using metaheuristic supported supervised classification coupled with bag-of-features | |
Aswathi et al. | Transfer learning approach for grading of diabetic retinopathy | |
Jasm et al. | Deep image mining for convolution neural network | |
Al-Hadidi et al. | Glioblastomas brain tumour segmentation based on convolutional neural networks. | |
Nigam et al. | 12 Wheat rust disease identification using deep learning | |
Al Jannat et al. | Detection of multiple sclerosis using deep learning | |
Khan et al. | Intelligent Systems for early malaria disease detection in patient cells using transfer learning approaches | |
Ghosh et al. | BBBC-U-Net: Optimizing U-Net for automated plant phenotyping using big bang big crunch global optimization algorithm | |
Quellec et al. | Weakly supervised classification of medical images | |
Saranya Rubini et al. | Deep convolutional neural network-based diabetic retinopathy detection in digital fundus images | |
Basavaraju et al. | Early Detection of Diabetic Retinopathy Using K-means Clustering Algorithm and Ensemble Classification Approach. | |
Cornforth et al. | Automated classification reveals morphological factors associated with dementia | |
Kalpana Devi et al. | Classification of Diabetic Retinopathy Using Ensemble of Machine Learning Classifiers with IDRiD Dataset | |
Cuevas et al. | Blood vessel segmentation using differential evolution algorithm | |
Salesi et al. | A hybrid model for classification of biomedical data using feature filtering and a convolutional neural 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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170818 |