CN113616218A - Epileptic-induced electroencephalogram recognition system based on synchronous compression transformation - Google Patents

Epileptic-induced electroencephalogram recognition system based on synchronous compression transformation Download PDF

Info

Publication number
CN113616218A
CN113616218A CN202110858447.2A CN202110858447A CN113616218A CN 113616218 A CN113616218 A CN 113616218A CN 202110858447 A CN202110858447 A CN 202110858447A CN 113616218 A CN113616218 A CN 113616218A
Authority
CN
China
Prior art keywords
electroencephalogram
synchronous
module
transformation
zone
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
Application number
CN202110858447.2A
Other languages
Chinese (zh)
Inventor
袁琦
杨玉莹
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shandong Normal University
Original Assignee
Shandong Normal University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Shandong Normal University filed Critical Shandong Normal University
Priority to CN202110858447.2A priority Critical patent/CN113616218A/en
Publication of CN113616218A publication Critical patent/CN113616218A/en
Pending legal-status Critical Current

Links

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
    • A61B5/369—Electroencephalography [EEG]
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/40—Detecting, measuring or recording for evaluating the nervous system
    • A61B5/4076—Diagnosing or monitoring particular conditions of the nervous system
    • A61B5/4094—Diagnosing or monitoring seizure diseases, e.g. epilepsy
    • 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
    • A61B5/7225—Details of analog processing, e.g. isolation amplifier, gain or sensitivity adjustment, filtering, baseline or drift compensation
    • 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
    • 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

Landscapes

  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Animal Behavior & Ethology (AREA)
  • General Health & Medical Sciences (AREA)
  • Biophysics (AREA)
  • Pathology (AREA)
  • Artificial Intelligence (AREA)
  • Biomedical Technology (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Medical Informatics (AREA)
  • Molecular Biology (AREA)
  • Surgery (AREA)
  • Veterinary Medicine (AREA)
  • Public Health (AREA)
  • Physiology (AREA)
  • Signal Processing (AREA)
  • Psychiatry (AREA)
  • Neurology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Neurosurgery (AREA)
  • Power Engineering (AREA)
  • Evolutionary Computation (AREA)
  • Fuzzy Systems (AREA)
  • Mathematical Physics (AREA)
  • Psychology (AREA)
  • Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)

Abstract

The invention provides an epileptogenic zone electroencephalogram recognition system based on synchronous compressive transformation, which comprises: the system comprises a data acquisition module, an electroencephalogram feature extraction module and an epileptic zone electroencephalogram identification module; the data acquisition module is used for acquiring an electroencephalogram signal of a patient to be detected; the electroencephalogram feature extraction module is used for carrying out time-frequency analysis on the electroencephalogram signals through synchronous compression transformation to obtain electroencephalogram features under different frequencies; the epileptogenic zone electroencephalogram recognition module is used for inputting electroencephalogram characteristics into a trained classifier to obtain an epileptogenic zone and a non-epileptogenic zone; the problem that a classical time-frequency analysis method often generates fuzzy electroencephalogram time-frequency representation due to the limitation of a Heisenberg uncertainty principle, so that the engineering application is seriously hindered is solved, the concentrated time-frequency representation of a strong time-varying signal can be generated, and the positioning accuracy of an epileptogenic area is improved.

Description

Epileptic-induced electroencephalogram recognition system based on synchronous compression transformation
Technical Field
The disclosure belongs to the technical field of signal processing, and particularly relates to an electroencephalogram recognition system for an epileptic zone based on synchronous compression transformation.
Background
The statements in this section merely provide background information related to the present disclosure and may not necessarily constitute prior art.
Epilepsy is a non-infectious neurobiological disease caused by abnormal discharges in the human brain. Has the characteristics of repeatability, burstiness and temporality. Recent epidemiological data show that the overall prevalence rate of epilepsy is 7%, the annual incidence rate is 28.8/10 ten thousand, and the prevalence rate of active epilepsy with seizures within 1 year is 4.6 per thousand. With a new increase of about 40 million epileptic patients each year. Epilepsy is increasingly recognized as a significant public safety problem.
Some epileptic patients need surgical treatment, and electroencephalogram plays a key role in the evaluation of epilepsy before surgery, and is an important basis for diagnosing epileptic seizures and determining the range of focal zone. Meanwhile, the operation treatment often depends on the accurate positioning of an epileptogenic zone, so the accurate identification of the brain electricity of the epileptogenic zone is the key for treating the epilepsy and reducing the side effect. However, the traditional visual identification method is not satisfactory, takes long time and is easily influenced by personal subjective factors. If the positioning of the epileptic region is inaccurate, the diagnosis and treatment of the patient will be affected badly, and even serious medical accidents will be caused.
Electroencephalograms have the characteristics of nonlinearity, non-stationarity and the like, and contain key information of brain states. In early electroencephalogram analysis, time domain and frequency domain analysis were separated from each other. The histogram and template matching isochronal domain analysis method has the advantage of visual intuition, and the Fourier transform as a classical frequency domain analysis method can extract frequency information contained in electroencephalogram signals. However, fourier transform cannot represent seizure frequency information of seizure-causing regions in a certain time period, that is, cannot simultaneously express time domain and frequency domain information in electroencephalogram signals. The single use of the time domain or frequency domain analysis method inevitably causes information loss, resulting in low reliability of the result of electroencephalogram signal analysis. Therefore, researchers have proposed a series of analysis methods combining time domain and frequency domain, and obtain information of signals in time domain and frequency domain. However, due to the limitation of the heisenberg uncertainty principle, the classical time-frequency analysis method often generates fuzzy electroencephalogram time-frequency representation, and the engineering application of the method is seriously hindered. How to generate a centralized time-frequency representation of a strong time-varying signal is a challenging task.
Disclosure of Invention
In order to solve the above problems, the present disclosure provides a seizure-causing area electroencephalogram recognition system based on synchronous compressive transformation, comprising: the system comprises a data acquisition module, an electroencephalogram feature extraction module and an epileptic zone electroencephalogram identification module; the data acquisition module is used for acquiring an electroencephalogram signal of a patient to be detected; the electroencephalogram feature extraction module is used for carrying out time-frequency analysis on the electroencephalogram signals through synchronous compression transformation to obtain electroencephalogram features under different frequencies; the electroencephalogram recognition module for the epilepsy-causing area is used for inputting electroencephalogram characteristics into the trained classifier to obtain the epilepsy-causing area and the non-epilepsy-causing area.
Compared with the prior art, this disclosure possesses following beneficial effect:
1. according to the method, an electroencephalogram feature extraction module is adopted to perform time-frequency analysis on electroencephalogram signals through synchronous compression transformation to obtain electroencephalogram features under different frequencies; the electroencephalogram characteristics are input into a trained classifier through an epileptogenic zone electroencephalogram recognition module to obtain an epileptogenic zone and a non-epileptogenic zone. The problem that a classical time-frequency analysis method often generates fuzzy electroencephalogram time-frequency representation due to the limitation of a Heisenberg uncertainty principle, so that the engineering application is seriously hindered is solved, the concentrated time-frequency representation of a strong time-varying signal can be generated, and the positioning accuracy of an epileptogenic area is improved.
2. The method adopts a new time-frequency analysis method, and adopts an iterative redistribution method on the basis of synchronous compression and transformation of electroencephalogram data, so that fuzzy time-frequency energy is gradually concentrated while the reconstruction capability of electroencephalogram signals is ensured, and electroencephalogram time-frequency representation with higher quality is obtained. Meanwhile, the complete electroencephalogram recognition system for the epileptogenic zone is established by combining the support vector machine classifier. The support vector machine shows specific advantages in solving small sample, nonlinear and high-dimensional pattern recognition according to a statistical learning theory, has good popularization capability, and has the core idea that a hyperplane capable of dividing all data samples in a space is found, and the distance from all data in a sample set to the hyperplane is shortest.
3. The method adopts a new synchronous compression transformation, and can gradually concentrate fuzzy time-frequency energy so as to obtain electroencephalogram time-frequency representation with higher quality. On the basis, the method is combined with a Support Vector Machine (SVM) to establish a complete electroencephalogram recognition system for the epileptic zone, so that the problems of long time consumption and high misjudgment rate of manual diagnosis are solved.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this application, illustrate embodiments of the application and, together with the description, serve to explain the application and are not intended to limit the application.
FIG. 1 is a schematic diagram of a brain electrical identification system for epileptic regions based on synchronous compression transformation;
fig. 2 is a schematic structural diagram of the epileptogenic zone electroencephalogram recognition system based on synchronous compressive transformation.
Detailed Description
The present disclosure is further described with reference to the following drawings and examples.
It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the disclosure. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.
It is noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments according to the present application. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, and it should be understood that when the terms "comprises" and/or "comprising" are used in this specification, they specify the presence of stated features, steps, operations, devices, components, and/or combinations thereof, unless the context clearly indicates otherwise.
Example 1
Interpretation of terms:
epilepsy-causing area: the cerebral cortex region that causes clinical seizures. The determination of the epileptogenic zone is crucial to epileptogenic resection, and to completely eliminate the epileptic seizure, enough cortex of the epileptogenic zone must be removed. The number of epileptogenic regions of an epileptic patient can be 1 or more than 1. The epileptogenic zone may be within or adjacent to or distant from the epileptogenic lesion.
As shown in fig. 1, a seizure-causing electroencephalogram recognition system based on synchronous compressive transformation includes: the system comprises a data acquisition module, an electroencephalogram feature extraction module and an epileptic zone electroencephalogram identification module;
the data acquisition module is used for acquiring an electroencephalogram signal of a patient to be detected;
the electroencephalogram feature extraction module is used for carrying out time-frequency analysis on the electroencephalogram signals through synchronous compression transformation to obtain electroencephalogram features under different frequencies;
the electroencephalogram recognition module for the epilepsy-causing area is used for inputting electroencephalogram characteristics into the trained classifier to obtain the epilepsy-causing area and the non-epilepsy-causing area.
The classifier divides the acquired training set into hyperplanes for training; the training set can be obtained from a Bonn electroencephalogram database and a Bern _ Barcelona electroencephalogram database. The classifier can adopt an SVM classifier, and a divided hyperplane is found based on a training set, so that different types of samples are separated to obtain an epileptic inducing region and a non-epileptic inducing region.
As an implementation mode, the data acquisition module identifies an intracranial electroencephalogram signal of a patient to be detected, wherein the electroencephalogram signal is a multi-lead intracranial electroencephalogram signal. The intracranial electroencephalogram is mainly characterized in that an electrode plate is placed in the intracranial space to detect the discharge information of neurons in a micro-operation mode, and the electroencephalogram information detected by the method is relatively pure, has no noise interference and is not influenced by the muscle activity of a human body.
As another embodiment, the electroencephalogram feature extraction module specifically comprises a preprocessing module, a synchronous compression transformation module and a feature extraction module; the preprocessing module is used for preprocessing the electroencephalogram data and transmitting the electroencephalogram data to the synchronous compression transformation module; the synchronous compression transformation module is used for carrying out synchronous compression transformation of multiple iterations on the preprocessed electroencephalogram data; the feature extraction module is used for extracting the electroencephalogram features of the electroencephalogram data after synchronous compression and transformation.
The device comprises a preprocessing module, a data processing module and a data processing module, wherein the preprocessing module is used for preprocessing electroencephalogram data, and the preprocessing step comprises filtering and segmentation processing; the synchronous compression transformation module is used for carrying out synchronous compression transformation on the preprocessed electroencephalogram data and carrying out multiple iterations to obtain compressed data; a feature extraction module: and constructing a time-frequency image model, and transmitting the compressed data to the time-frequency image model to perform time-frequency multi-level feature extraction to obtain electroencephalogram features, wherein the electroencephalogram features comprise feature indexes and relative mean values of wave amplitudes.
In particular, the basic theory of synchronous compression transformation (SST).
Signal
Figure RE-GDA0003302251680000051
The corresponding s transform is as follows:
Figure RE-GDA0003302251680000052
where t represents the time axis displacement parameter, ω represents frequency, and the unit Hz, i represents the imaginary unit, the above equation is transformed into:
Figure RE-GDA0003302251680000053
order to
Figure RE-GDA0003302251680000054
Can be transformed into:
Figure RE-GDA0003302251680000055
in the formula:
Figure RE-GDA0003302251680000056
represents the Fourier transform of s (t),
Figure RE-GDA0003302251680000057
to represent
Figure RE-GDA0003302251680000058
The complex conjugate of the fourier transform of (a).
The condition of harmonic signals is mathematically analyzed, and s (t) is equal to Acos (2 pi omega)0t) then:
Figure RE-GDA0003302251680000061
the original formula is carried into:
Figure RE-GDA0003302251680000062
bonding of
Figure RE-GDA0003302251680000063
The instantaneous frequency estimate of s (t) can be found as:
Figure RE-GDA0003302251680000064
thus, for forms such as s (t) Acos (2 pi ω)0t), from the above calculation:
Figure RE-GDA0003302251680000065
thus, for a conventional multi-component signal
Figure RE-GDA0003302251680000066
Simultaneously satisfy An(t),φ’(t)＞0,
Figure RE-GDA0003302251680000067
Here phi'n(t) represents phinThe first derivative of (t). It is known that the S time-frequency variation process is linear, so that the S transformation of a signal S (t) consisting of multiple components can be equivalent to N components Sn(t) sum of S-transforms, i.e.
Figure RE-GDA0003302251680000068
Wherein:
Figure RE-GDA0003302251680000069
wherein each component snThe instantaneous estimator of (t) is:
Figure RE-GDA00033022516800000610
the instantaneous frequency estimate of the final multi-component signal s (t) is:
Figure RE-GDA0003302251680000071
in the formula: δ (·) denotes a dirac function. Since the actually acquired brain electrical signal is in discrete form, the integration over the frequency period is changed to discrete summation form, thereby defining the synchronous compression S transformation of the discrete signal S (t) as:
ssst(t,ω1)＝Lω -1∑|ST(t,ωk)|ωkΔωk
Figure RE-GDA0003302251680000072
wherein: omega1Is the instantaneous frequency, L, of the EEG signal after S transformationωAt instantaneous frequency ω in S-conversion1Frequency interval half length, omega, centeredkFor L in S transformωDiscretized frequency sample points within the interval, and Δ ω ═ ωk-ωk-1. The electroencephalogram synchronous compression transformation is to superpose short-time Fourier transform coefficients in a certain range, and because the performance of S Transformation (ST) is superior to that of short-time Fourier transformation (STFT) and Wavelet Transformation (WT) in time-frequency analysis of high-frequency low-amplitude component signals, the performance of synchronous compression S transformation (SSST) is superior to that of synchronous compression transformation (SST).
In order to further sharpen the result of electroencephalogram synchronous compression transformation (SST) and obtain time-frequency expression with better focusing, the electroencephalogram data is continuously subjected to a Multi-time Multi-synchronous compression (MSST) algorithm, the algorithm is continuously subjected to Multi-time iteration after synchronous compression transformation (SST), and the obtained time-frequency estimator can be closer to the real frequency of electroencephalogram signals in an epileptogenic area. First, the basic short-time fourier transform is as follows:
Figure RE-GDA0003302251680000073
in the formula: g (-) is a short-time Fourier window function.
Then, carrying out multiple iterative compression transformation on the electroencephalogram data on the basis of short-time Fourier, as follows:
Figure RE-GDA0003302251680000081
Figure RE-GDA0003302251680000082
…
Figure RE-GDA0003302251680000083
in the formula:
Figure RE-GDA0003302251680000084
is an instantaneous frequency estimator, MSST[m](t, ω) represents the result of iterating m times using a simultaneous compression operator based on SST.
After the electroencephalogram data are iterated for multiple times by using the synchronous compression operator, the instantaneous frequency estimator is equivalently iterated for multiple times to obtain new distribution:
Figure RE-GDA0003302251680000085
by analogy, Ts can be obtained[m](t, ω), and thus the estimated amount of the instantaneous frequency after m iterations can be obtained:
Figure RE-GDA0003302251680000086
Figure RE-GDA0003302251680000087
in the formula:
Figure RE-GDA0003302251680000088
after MSST transformation is carried out on the electroencephalogram data, the relative mean values of the characteristic indexes and the wave amplitudes are input into an SVM classifier, and automatic classification of seizure-causing areas and non-seizure-causing areas is achieved.
As an embodiment, the seizure-causing electroencephalogram recognition module specifically includes: a feature input module and an SVM classifier;
the characteristic input module is used for inputting the electroencephalogram characteristics to the trained SVM classifier;
the SVM classifier is used for carrying out linear separable processing and classification processing on the electroencephalogram data to obtain an epilepsy-causing area and a non-epilepsy-causing area; the linear separable processing includes performing linear separable on a hyperplane basis by mapping the linear inseparable electroencephalographic data samples to a high dimensional space. Specifically, the basic principle of the SVM classifier is that linear inseparable samples are subjected to nonlinear transformation and are mapped into another high-dimensional space, and an optimal interface (hyperplane) is searched in the transformed space so as to be linearly separable.
Let training sample set D { (x)1,y1),(x2,y2),...,(xm,ym)},yiE {1, -1}, the SVM finds a partitioned hyperplane based on the training set D, thereby separating the different classes of samples, i.e. there is a partitioned hyperplane as follows: omegaTx+b＝0；
Where ω is (ω)1；ω2；...；ωd) Is the normal vector and b is the bias term. The maximum interval for classification at this time is:
Figure RE-GDA0003302251680000091
to better solve this convex quadratic programming problem, the Lagrange multiplier method (a) is introducediNot less than 0) to get its dual problem.
Figure RE-GDA0003302251680000092
Figure RE-GDA0003302251680000101
If it is
Figure RE-GDA0003302251680000102
To an optimal solution, then
Figure RE-GDA0003302251680000103
I.e. the weight coefficient vector of the optimal classification face is a linear combination of the training sample vectors. Then, b can be obtained from any support vector.
By solving the above problem, the optimal classification function can be obtained as
Figure RE-GDA0003302251680000104
In the practical application of classifying the seizure-causing area and the non-seizure-causing area, a large number of linear inseparable problems exist due to the nonlinear characteristics of electroencephalogram data, and a reasonable hyperplane does not exist in the original sample space so as to ensure the correct division of the samples. In order to solve the problem, a kernel function concept is introduced in the electroencephalogram classification process of the seizure-causing areas to replace the dot product operation after the dual problem and the nonlinear mapping. The SVM is a nuclear learning method for classifying test samples by constructing a hyperplane by using a training set. The SVM algorithm is unique in that an inner product kernel function is used. The final decision function depends on a few support vectors, so that the effect of the algorithm can be prevented from being influenced by a large number of redundant samples, and the SVM algorithm is endowed with more outstanding robustness.
The support vector machine maps the input space to a high-dimensional feature space by some non-linear transformation phi (x). The dimension of the feature space may be very high. If the solution of the support vector machine only uses the inner product operation, and there is some function K (x, x ') in the low-dimensional input space, it is exactly equal to this inner product in the high-dimensional input space, i.e., K (x, x ') < phi (x) · phi (x ') >. The support vector machine does not need to calculate complex nonlinear transformation, and the inner product of the nonlinear transformation is directly obtained by the function K (x, x'), so that the calculation is greatly simplified. Such a function K (x, x') is called a kernel function.
In particular, a kernel function K (x, x') is used instead
Figure RE-GDA0003302251680000111
Figure RE-GDA0003302251680000112
Dot product in the formula, i.e.
Figure RE-GDA0003302251680000113
A commonly used kernel function K (x, x') is of the form: polynomial kernel function, radial basis function, Sigmoid function.
The method selects a polynomial kernel function, the polynomial kernel is a nonstandard kernel function, and the method is very suitable for data after orthonormal and has the following specific form:
k(x,x’)＝[a(x·x’)+c]d
wherein d is the order of the polynomial.
The polynomial kernel function can map a low-dimensional input space to a high-dimensional feature space, and is suitable for the condition of electroencephalogram feature classification (vector orthogonality and mode 1).
Classification experiments are respectively carried out on a Bonn electroencephalogram database and a Bern _ Barcelona electroencephalogram database so as to evaluate the classification capability of the method.
The Bonn data set used is a public data set that contains brain wave recordings of healthy individuals and epileptic patients. The data set contains five subsets, each subset containing 100 pieces of single-channel electroencephalogram data, the duration is 23.3 seconds, the sampling frequency is 173.61Hz, and each piece of data contains 4097 sample points. The two subsets of F, N that this patent used are the brain electrical activity of epileptic pathogenic region and the brain region's of other one side symmetry brain activity respectively when epileptic does not catch a disease. 80 groups were selected as training set and 20 groups were selected as test set for this experiment.
The second database used an open-acquired Bern _ Barcelona electroencephalographic dataset containing electroencephalographic recordings of 5 patients with long-standing drug-resistant epilepsy. All electroencephalogram signals are filtered in a forward and backward filtering mode through a fourth-order Butterworth filter so as to reduce phase distortion. The signal sampling frequency is 512 or 1024Hz, depending on whether the number of acquisition channels exceeds 64. These intracranial brain electrical signals both localize the brain region of the seizure and do not risk neurological deficits. In the experiment, 30 groups were selected as training sets and 20 groups were selected as test sets.
The performance of the system is evaluated by three indexes of sensitivity, specificity and accuracy in the experiment, which are defined as follows: wherein TP represents the brain electrical sample marked as seizure-causing zone by both the patent program and the physician. TN represents the brain electrical sample that was simultaneously marked by the patent program and physician as a non-epileptic zone.
Sensitivity (Sensitivity): the ratio of the number of TPs to the number of seizure-inducing brain electrical samples marked by the physician.
Specificity (Specificity): the ratio of the number of TNs to the number of non-epileptogenic brain electrical samples marked by the physician.
Accuracy (Accuracy): the ratio of the number of electroencephalogram samples correctly marked by the patent program to the total number of electroencephalogram samples.
The verification results of the method on two electroencephalogram databases are as follows:
Figure RE-GDA0003302251680000121
although the present disclosure has been described with reference to specific embodiments, it should be understood that the scope of the present disclosure is not limited thereto, and those skilled in the art will appreciate that various modifications and changes can be made without departing from the spirit and scope of the present disclosure.

Claims (10)

1. An epileptogenic electroencephalogram recognition system based on synchronous compressive transformation is characterized by comprising: the system comprises a data acquisition module, an electroencephalogram feature extraction module and an epileptic zone electroencephalogram identification module;
the data acquisition module is used for acquiring an electroencephalogram signal of a patient to be detected;
the electroencephalogram feature extraction module is used for carrying out time-frequency analysis on the electroencephalogram signals through synchronous compression transformation to obtain electroencephalogram features under different frequencies;
the electroencephalogram recognition module for the epilepsy-causing area is used for inputting electroencephalogram characteristics into the trained classifier to obtain the epilepsy-causing area and the non-epilepsy-causing area.
2. The epileptic zone electroencephalogram recognition system based on synchronous compressive transformation as claimed in claim 1, wherein the electroencephalogram feature extraction module comprises a preprocessing module, a synchronous compressive transformation module and a feature extraction module;
the preprocessing module is used for preprocessing the electroencephalogram data and transmitting the electroencephalogram data to the synchronous compression transformation module;
the synchronous compression transformation module is used for carrying out synchronous compression transformation of multiple iterations on the preprocessed electroencephalogram data;
the feature extraction module is used for extracting the electroencephalogram features of the electroencephalogram data after synchronous compression and transformation.
3. The synchronous compressive transformation-based epileptic zone electroencephalogram recognition system of claim 2, wherein the preprocessing step comprises filtering and segmentation processing.
4. The epileptic zone electroencephalogram recognition system based on synchronous compressive transformation as claimed in claim 3, wherein the synchronous compressive transformation module is used for carrying out synchronous compressive transformation on the preprocessed electroencephalogram data and carrying out multiple iterations to obtain compressed data.
5. The epileptic zone electroencephalogram recognition system based on synchronous compressive transformation as claimed in claim 4, wherein the feature extraction module is used for constructing a time-frequency image model, transmitting the compressed data to the time-frequency image model for time-frequency multi-level feature extraction to obtain electroencephalogram features, and the electroencephalogram features comprise feature indexes and relative mean values of wave amplitudes.
6. The epileptogenic zone electroencephalogram recognition system based on synchronous compressive transformation as claimed in claim 1, wherein the epileptogenic zone electroencephalogram recognition module specifically comprises: the SVM classifier comprises a feature input module and a trained SVM classifier; the characteristic input module is used for inputting the electroencephalogram characteristics to the trained SVM classifier.
7. The epileptogenic zone electroencephalogram recognition system based on synchronous compressive transformation as claimed in claim 6, wherein the SVM classifier is used for performing linear separable processing and classification processing on electroencephalogram data to obtain an epileptogenic zone and a non-epileptogenic zone.
8. The epileptic zone electroencephalogram recognition system based on synchronous compressive transformation of claim 7, wherein the linear separable processing comprises hyperplane-based realization of linear separability by mapping linearly inseparable electroencephalogram data samples into a high-dimensional space.
9. The epileptic zone electroencephalogram recognition system based on synchronous compressive transformation of claim 1, wherein the classifier is trained by partitioning the acquired training set into hyperplanes.
10. The epileptogenic zone electroencephalogram recognition system based on synchronous compressive transformation as claimed in claim 1, wherein the data acquisition module recognizes intracranial electroencephalogram signals of a patient to be detected, and the electroencephalogram signals are multi-lead scalp electroencephalogram signals.
CN202110858447.2A 2021-07-28 2021-07-28 Epileptic-induced electroencephalogram recognition system based on synchronous compression transformation Pending CN113616218A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110858447.2A CN113616218A (en) 2021-07-28 2021-07-28 Epileptic-induced electroencephalogram recognition system based on synchronous compression transformation

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110858447.2A CN113616218A (en) 2021-07-28 2021-07-28 Epileptic-induced electroencephalogram recognition system based on synchronous compression transformation

Publications (1)

Publication Number Publication Date
CN113616218A true CN113616218A (en) 2021-11-09

Family

ID=78381362

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110858447.2A Pending CN113616218A (en) 2021-07-28 2021-07-28 Epileptic-induced electroencephalogram recognition system based on synchronous compression transformation

Country Status (1)

Country Link
CN (1) CN113616218A (en)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110130797A1 (en) * 2008-04-30 2011-06-02 Talathi Sachin S Method and system for detecting epileptogenesis
WO2013043157A2 (en) * 2011-09-20 2013-03-28 Vivaquant Llc Physiological signal denoising
CN107957783A (en) * 2017-12-21 2018-04-24 北京航天测控技术有限公司 A kind of Multimode Intelligent control system and method based on brain electricity with myoelectric information
CN108681391A (en) * 2018-03-19 2018-10-19 南京邮电大学 A kind of EEG signals dummy keyboard design method based on multi-mode
CN108710895A (en) * 2018-04-23 2018-10-26 安徽大学 Mental imagery Method of EEG signals classification based on independent component analysis

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110130797A1 (en) * 2008-04-30 2011-06-02 Talathi Sachin S Method and system for detecting epileptogenesis
WO2013043157A2 (en) * 2011-09-20 2013-03-28 Vivaquant Llc Physiological signal denoising
CN107957783A (en) * 2017-12-21 2018-04-24 北京航天测控技术有限公司 A kind of Multimode Intelligent control system and method based on brain electricity with myoelectric information
CN108681391A (en) * 2018-03-19 2018-10-19 南京邮电大学 A kind of EEG signals dummy keyboard design method based on multi-mode
CN108710895A (en) * 2018-04-23 2018-10-26 安徽大学 Mental imagery Method of EEG signals classification based on independent component analysis

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
CURA, OK 等: "《Epileptic EEG Classification Using Synchrosqueezing Transform with Machine and Deep Learning Techniques》", no. 28, pages 1210 - 1214 *
MADHAVAN, S等: "《Time-Frequency Domain Deep Convolutional Neural Network for the Classification of Focal and Non-Focal EEG Signals》", 《IEEE SENSORS JOURNAL》, vol. 20, no. 6, pages 3078 - 3086 *
OZEL, P等: "《Emotion Detection Using Multivariate Synchrosqueezing Transform via 2D Circumplex Model》", 《 2018 MEDICAL TECHNOLOGIES NATIONAL CONGRESS》, pages 4 *
VILLAZANA, SERGIO等: "《Epileptic Signal Detection Using Quilted Synchrosqueezing Transform Based Convolutional Neural Networks》", 《COMPUTACION Y SISTEMAS》, vol. 25, no. 2, pages 269 - 286 *

Similar Documents

Publication Publication Date Title
Sharma et al. Automated detection of abnormal EEG signals using localized wavelet filter banks
George et al. Classification of epileptic EEG signals using PSO based artificial neural network and tunable-Q wavelet transform
García-Martínez et al. A review on nonlinear methods using electroencephalographic recordings for emotion recognition
Cho et al. EEG-based prediction of epileptic seizures using phase synchronization elicited from noise-assisted multivariate empirical mode decomposition
Dwi Saputro et al. Seizure type classification on EEG signal using support vector machine
Hamad et al. Feature extraction of epilepsy EEG using discrete wavelet transform
EP2498676B1 (en) Brain activity as a marker of disease
AlSharabi et al. EEG signal processing for Alzheimer’s disorders using discrete wavelet transform and machine learning approaches
Lasefr et al. Epilepsy seizure detection using EEG signals
Alturki et al. Common spatial pattern technique with EEG signals for diagnosis of autism and epilepsy disorders
Kumari et al. Seizure detection in EEG using time frequency analysis and SVM
Chaovalitwongse et al. Pattern recognition approaches for identifying subcortical targets during deep brain stimulation surgery
Sharma et al. Automated classification of focal and non-focal EEG signals based on bivariate empirical mode decomposition
Bashar et al. Epileptic seizure classification from intracranial EEG signals: A comparative study EEG-based seizure classification
Thanaraj et al. Multichannel interictal spike activity detection using time–frequency entropy measure
Jose et al. Robust classification of intramuscular EMG signals to aid the diagnosis of neuromuscular disorders
Noor et al. Improving outcome prediction for traumatic brain injury from imbalanced datasets using RUSBoosted trees on electroencephalography spectral power
Höller et al. Age, sex, and pathology effects on stability of electroencephalographic biometric features based on measures of interaction
Sriraam et al. Multichannel EEG based inter-ictal seizures detection using Teager energy with backpropagation neural network classifier
Hussain et al. An intelligent system to classify epileptic and non-epileptic EEG signals
Awang et al. Analysis of EEG signals by eigenvector methods
Zhang et al. Identification of detailed time-frequency components in somatosensory evoked potentials
Krishnaprasanna et al. Focal and non-focal EEG signal classification by computing area of 2D-PSR obtained for IMF
CN113616218A (en) Epileptic-induced electroencephalogram recognition system based on synchronous compression transformation
Parui et al. An improved cognitive approach for automated epileptic seizure detection from multichannel eeg

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