KR20190126604A - A device for diagnosis of peripheral neuropathy using wavelet transform of needle electromyography signal - Google Patents

Abstract

The present invention relates to a device for the diagnosis of peripheral neuropathy using a needle electromyography signal, and more specifically, to a technology for a device for the diagnosis of peripheral neuropathy, which suggests a wavelet transform method capable of statistically processing multiple needle electromyogram signals from one person and performing time-frequency analysis of the needle electromyography signals to have increased reliability and reduced errors in diagnosis and to diagnose peripheral neuropathy immediately after conducting electromyography. The device for the diagnosis of peripheral neuropathy using a needle electromyography signal includes an electromyography detecting unit, a signal processing unit, and a discriminating unit, wherein the signal processing unit is configured to perform wavelet transform on an action potential signal to enable time-frequency analysis.

Description

A device for diagnosis of peripheral neuropathy using wavelet transform of needle electromyography signal

The present invention relates to a device for discriminating peripheral neuropathy using acupuncture EMG signals, and more particularly, to a wavelet transform method capable of statistically processing a plurality of acupuncture EMG signals extracted from one person and time-frequency analysis of the needle EMG signals. It is a technique for peripheral neuropathy discrimination device that can be diagnosed with peripheral neuropathy immediately after EMG.

Action potentials occurring in muscles are characteristic of peripheral neuropathy. Accordingly, as a method widely used in the evaluation of neuropathy, there is an acupuncture conduction test. The purpose of the examination is to determine the presence of the lesion, whether the lesion is invading the peripheral nervous system or the central nervous system, whether the lesion is local or systemic, what is the pathophysiology and severity of the lesion, and the lesion is improving. Or is it deteriorating? In both adults and children, the agreement between clinical and electrical diagnostic tests is known to have a high diagnosis rate of 73-98%.

The EMG signal extracted by EMG is extracted from the features of diagnosing neuropathy through the shape of a motor unit. Herein, the motor unit refers to an anterior horn cell, nerve fibers, and all muscle fibers connected thereto, and refers to a potential generated in a motor unit that contracts muscles. It is detected by an EMG signal. As shown in FIG. 1, the EMG signal may analyze an action potential pattern of an exercise unit by magnitude, duration, pole, number of zero points, and the like.

Looking at the research literature 'Research Study of stability of time-domain features for electromyographic pattern recognition', which Tkach, a conventional technique for processing EMG signals to diagnose peripheral neuropathy, identifies the characteristics of EMG signals. Multiple features can be extracted, where the five key components for diagnosis are duration (wavelength, WL, waveform length), zero crossings (ZC), slope sign changes (SSC). , Magnitude (WA, Willison Amplitude), root mean square (RMS).

Diagnosis of 10 normal sets and 10 sets of patients using the five features is shown in FIG. 2. Figure 2 is a graph showing the number of past zero for the duration, the change in the slope for the past zero, the average cubic root for the change of the slope, the average cubic root for the size, the analysis does not distinguish between normal data and neuropathy data It can be seen that the case occurs.

As described above, the diagnosis of neuropathy through electromyography is difficult to quantitatively describe and describe the characteristics of the exercise unit.

 Tkach et al., “Research Study of stability of time-domain features for electromyographic pattern recognition,” Journal of NeuroEngineering and Rehabilitation 2010, 7:21

Until now, studies on extracting features for diagnosing peripheral neuropathy using EMG signals have been made, but studies to improve the ease of diagnosis and reduce errors by quantifying the results of EMG tests have been lacking. In the present invention, the electromyography signal for diagnosing peripheral neuropathy is a wavelet transform using a basis function suitable for the characteristics of the signal, and then extracts a feature and compares it with a normal reference value. To provide a peripheral neuropathy discrimination apparatus using.

In addition, the present invention is to provide a peripheral neuropathy discrimination apparatus using a needle electromyography signal with high reliability can reduce the error of the diagnosis by statistical processing a plurality of EMG signals extracted from one person.

The problem to be solved by the present invention is not limited to the above-mentioned problem, another problem to be solved by the present invention not mentioned herein is to those skilled in the art from the following description. It will be clearly understood.

Peripheral neuropathy discrimination apparatus using the acupuncture electromyography signal according to the present invention, after inserting a needle electrode into the muscle to be examined, it is repeated so that only the action potential signals for one exercise unit is generated when the muscle is contracted EMG detection unit for detecting a plurality of action potential signals for one exercise unit; A signal processor for processing the activity potential signal to derive a quantified value so as to visualize characteristics of a time-frequency signal; And a discriminating unit configured to compare the quantified value with a reference value representing a normal person stored in advance, and to generate diagnostic information about whether the activity potential signal corresponds to peripheral neuropathy or is normal. Wavelet transform the action potential signal to enable time-frequency analysis.

In the peripheral neuropathy discrimination apparatus using the acupuncture electromyogram signal according to the present invention, The signal processing unit, Pre-processing unit for dividing the detected plurality of action potential signals into a single action potential signal; A wavelet transform unit configured to perform a wavelet transform on the divided action potential signals by setting a molet as a basis function; From the wavelet transformed action potential signals, only signals having a time sample in a predetermined range are separated, and a wavelet coefficient sum obtained by adding a wavelet coefficient for the predetermined time sample range with respect to the scale is calculated, and the diagnosis is performed from the wavelet coefficient sum. Feature extraction unit for extracting a feature that can discriminate information; And a data selecting unit for selecting the representative value of the feature for the same exercise unit as the quantified value by averaging the features extracted for each of the plurality of action potential signals.

In the peripheral neuropathy discrimination apparatus using the acupuncture electromyogram signal according to the present invention, the feature extraction unit extracts a first feature and a second feature from the wavelet coefficient summation, and then performs a comparative analysis of the wavelet coefficient summation and normal data. And extracting the third feature.

In the peripheral neuropathy discrimination apparatus using the acupuncture electromyogram signal according to the present invention, the first feature is calculated as an effective value of the wavelet coefficient sum using Equation 4, and the second feature is the wavelet using Equation 5. The third feature is calculated from a correlation analysis between the normalized representative wavelet coefficient sum of normalized data and the wavelet coefficient sum of normal data using Equation 6.

In the peripheral neuropathy discrimination apparatus using the acupuncture electromyogram signal according to the present invention, the discriminating unit may include: a first condition in which the first feature is 0 to 3000, a second condition in which the second feature is 15 to 23, and the third Differentiate whether the feature satisfies the third condition of 0.8 to 1, and if all of the first to third conditions are satisfied, the result is 'normal', and any one of the first to third conditions If it is not satisfactory, it is characterized by generating diagnostic information including the results of 'peripheral neuropathy'.

By means of solving the above problems, peripheral neuropathy discrimination apparatus using acupuncture electromyography signal of the present invention, after diagnosing peripheral neuropathy by wavelet transforming the electromyography signal using a basis function suitable for the characteristics of the signal and extracting features It is effective to make a diagnosis immediately after electromyography in a way that is compared with a normal reference value.

In addition, the peripheral neuropathy discrimination apparatus using the acupuncture EMG signal of the present invention can reduce the error of diagnosis by statistically processing a plurality of EMG signals extracted from a person has the effect of increasing the reliability of the test.

1 is a diagram illustrating parameters that can be distinguished for aspects of exercise unit activity potential.
Figure 2 is a graph showing the results when the normal and peripheral neuropathy patients undergoing electromyography according to the prior art.
Figure 3 is a block diagram of a peripheral neuropathy discrimination apparatus using acupuncture EMG signal according to the present invention.
4 is a diagram illustrating a differentiation process of a peripheral neuropathy discrimination apparatus using acupuncture electromyogram signal according to the present invention.
5 is a graph showing normal data (NR) and neuropathy data (NR) for each exercise unit in the peripheral neuropathy discrimination process using the acupuncture EMG signal according to the present invention.
FIG. 6 is a graph showing the fundamental waveforms of the Db2, Db4, Meyer, and Morlet basis functions, wavelet transform graphs, and features for normal data and peripheral neuropathy data.
7 is a graph showing the basis function of the wavelet transform used to well characterize the EMG signal according to the present invention.
8 is a diagram showing the results of wavelet transforming the EMG signal divided in the peripheral neuropathy discrimination process using the needle electromyogram signal according to the present invention.
9 is a view for explaining that the wavelet coefficient where the center of the motion unit and the wavelet function coincide best represents the characteristics of the signal.
FIG. 10 is a diagram illustrating a process of deriving wavelet coefficients in a peripheral neuropathy discrimination process using acupuncture EMG signals according to the present invention.
11 is a graph depicting features of normal and peripheral neuropathy patient signals, respectively, for a defined time range in accordance with the present invention.
12 is a graph in which representative normal data is selected from a plurality of normal data in order to extract a correlation coefficient with normal data in a peripheral neuropathy discrimination process using the acupuncture EMG signal according to the present invention.
FIG. 13 is a graph in which representative values are selected for each feature of a plurality of action potentials for the same exercise unit repeated during electromyography of a neuropathy patient in the peripheral neuropathy discrimination process using the acupuncture electromyography signal according to the present invention.
14 is a graph for determining a reference value indicating a normal person in the peripheral neuropathy discrimination process using the acupuncture EMG signal according to the present invention.

Specific matters including the problem to be solved, the solution to the problem, and the effects of the present invention as described above are included in the embodiments and drawings to be described below. Advantages and features of the present invention and methods for achieving them will be apparent with reference to the embodiments described below in detail with the accompanying drawings.

In the peripheral peripheral neuropathy discrimination apparatus using the acupuncture EMG signal according to an embodiment of the present invention, as shown in Figure 3, after inserting the needle electrode to the test muscles, when the test muscles contract one movement An EMG detector 10 which detects a plurality of action potential signals for one exercise unit by repeating only activity potential signals for a unit; A signal processor 20 for processing the activity potential signal to derive a quantified value so as to visualize the characteristics of the time-frequency signal; And a discrimination unit 30 that compares the quantified value with a reference value representing a normal stored person and generates diagnostic information on whether the activity potential signal corresponds to peripheral neuropathy or is normal.

First, the EMG detection unit 10, using a commonly known method of acupuncture electromyography, inserts a needle electrode to the test muscle to check the electrical stimulation delivered, and exercise unit activity for one exercise unit Detect repeatedly so that only the potential comes out. Here, the exercise unit activity potential is an electric signal generated when one exercise unit is contracted.

Next, the signal processing unit 20 is characterized in that the wavelet transform the action potential signal to enable time-frequency analysis and extract features for diagnosing peripheral neuropathy, the peripheral neuropathy shown in Figure 4 Perform the sifting process.

In addition, the signal processor 20 may include a preprocessor 21 for dividing the plurality of detected action potential signals into one action potential signal; A wavelet transform unit 22 for performing a wavelet transform in which a molet is set as a basis function for the divided action potential signals; From the wavelet transformed action potential signals, only signals having a time sample in a predetermined range are separated, and a wavelet coefficient sum obtained by adding a wavelet coefficient for the predetermined time sample range with respect to the scale is calculated, and the diagnosis is performed from the wavelet coefficient sum. Feature extraction unit 23 for extracting a feature that can discriminate information; And a data selecting unit 24 for selecting the representative value of the feature for the same exercise unit as the quantified value by averaging the features respectively extracted for the plurality of action potential signals.

The preprocessor 21 divides a plurality of action potential signals for one exercise unit into one data. The data obtained by measuring the EMG signal once is defined as one set, and one set includes a plurality of action potential data for one exercise unit. For example, if there are 10 sets of normal people and 10 sets of peripheral neuropathy patients, about 30 to 70 activity potential signals for each exercise unit exist in each set, and thus about 1000 data may be generated. 5 is a graph showing two sets of normal data (NR) and neuropathy data (NP) for each exercise unit.

The wavelet transform unit 22 may wavelet transform the active potential signal to be simultaneously classified into a shape, a magnitude, and a frequency characteristic.

Wavelet transform is a signal processing technique that expresses a signal as a small wave called a mother wavelet function. The wavelet transform represents a signal by moving the scale and position of the expansion and contraction of the basis function.

The equation of the wavelet transform is shown in Equation 1 below.

[Equation 1]

Where C is the wavelet coefficient.

The basic functions of the wavelet transform include Db2, Db4, Meyer, and Morlet. Basal functions should be well-expressed in the basic form of action potentials and should be selected so that the distinction between normal and peripheral neuropathy data is clear.

FIG. 6 shows a basic waveform of each basis function, a wavelet transform graph, and a graph showing features of normal data and peripheral neuropathy data. Referring to this, Morlet simulates the basic shape of the action potential. It can be seen that it is a basic function that can be well expressed.

In other words, considering the difference between the basic form of EMG signal and peripheral neuropathy form by exercise unit, the basis function that can distinguish the difference was Morlet.

In the present invention, the basis function for the wavelet transform is set to a molet capable of expressing the basic form of the action potential of the exercise unit. The formula of the molet basis function is shown in Equation 2 below, and the waveform graph is shown in FIG. 7.

[Equation 2]

In wavelet transform, it is very important to properly select the basis function in order to represent the characteristics of the signal well. If the signal is similar to the shape to be analyzed, the characteristics of the signal can be observed in the wavelet transform results. The Morlet wavelet transform uses a basis function that has a similar shape to an impulse signal. Human signals such as fault signals, electrocardiograms, and EMG signals of mechanical systems are impulse signals.

The wavelet transform unit 22 may perform a Morlet wavelet transform of EMG signals of a normal person and a peripheral neuropathy patient, as shown in FIG. 8. The graph on the left is an EMG time signal graph, and the graph on the right, which is the result of wavelet transform, is a graph showing wavelet coefficients with respect to time-scale (frequency) in color. Here, the frequency means the pseudo frequency, and the pseudo frequency is replaced by a scale that is a variable proportional to the inverse of the pseudo frequency by the center frequency determined for each basis function, and is shown in a wavelet transform graph as shown in FIG. 8. The relationship between scales is shown in Equation 3 below.

[Equation 3]

는 스케일

에 해당하는 슈도 주파수(Hz)이고,here,

Scale

Is the pseudo frequency in Hz,

는 웨이블릿의 중심 주파수(Hz)이고,

Is the center frequency of the wavelet (Hz),

는 스케일이며,

Is the scale,

는 샘플링 시간(

)을 의미한다.

Is the sampling time (

).

Table 1 shows a value obtained by converting the pseudo frequency into a scale. In the present invention, since the pseudo frequency is converted to the scale range 1 to 70, the range of the frequency that can be analyzed is 58 Hz to 4060 Hz.

scale Pseudofrequency (Hz) 10 406 20 203 40 102 70 58

The feature extractor 23 is provided to extract a feature capable of discriminating the diagnostic information from the Morlet wavelet transformed action potential signals.

As shown in FIG. 9, the activity potential signal of the exercise unit has a characteristic that a time sample having the largest value is zero. Thus, as shown in FIG. 10, it would be appropriate to compare when the center of the wavelet function is near zero, and a predetermined range of time samples for separating signals to set the wavelet coefficient sum will have to be set.

The wavelet coefficient sum is computed by a time sample summing signal wavelet coefficients within the predetermined range with respect to the scale.

The predetermined range of the present invention may be set to -1 to 1 according to one embodiment, -2 to 2 according to another embodiment, and -3 to 3 according to another embodiment.

FIG. 11 shows a graph depicting features of normal and peripheral neuropathy patient signals for the three ranges described above (-1 to 1, or -2 to 2, or -3 to 3), respectively. It can be seen that all can be clearly distinguished.

In this case, the feature extraction unit 23 may extract the first feature and the second feature from the calculated wavelet coefficient sum.

In addition, the feature extraction unit 23, in order to more clearly distinguish the peripheral neuropathy and the signal of the normal person, as shown in Figure 12, the wavelet coefficient sum for a plurality of signals of the normal person wavelet transformed the EMG signal of the normal person The normalized representative wavelet coefficient sum is derived.

In this case, the feature extractor 23 may extract a third feature by performing a comparative analysis, specifically, a correlation analysis between the derived representative wavelet coefficient sum and the wavelet coefficient sum.

The first feature may be calculated as an effective value of the wavelet coefficient sum using Equation 4 below.

[Equation 4]

Where n is the number of scales,

(According to the above, in the present invention, since the range of the scale is set to 1 to 70, n is 70).

C is the wavelet coefficient sum of each signal.

The second feature may be calculated as a center scale for the wavelet coefficient sum using Equation 5 below.

[Equation 5]

Where s is the scale,

C (s) is the wavelet coefficient sum for each scale.

The third feature may be calculated from a correlation analysis between the normal representative wavelet coefficient sum and the wavelet coefficient sum using Equation 6 below.

[Equation 6]

Where x is the wavelet coefficient sum,

는 x(웨이블릿계수합)의 평균이고,

Is the average of x (the wavelet coefficients),

y is the normal representative wavelet coefficient,

는 y(정상인 대표 웨이블릿계수합)의 평균이다.

Is the average of y (normal representative wavelet coefficients).

Since the data extracting unit 24 does not have all of the motor unit signals of peripheral neuropathy patients, all of them are characterized by peripheral neuropathy. As shown in FIG. 13, the data extracting unit 24 is extracted from a plurality of action potential signals for the same motor unit. By averaging the features, a representative value of the features for the same exercise unit is selected, so that the representative value enables the diagnosis of peripheral neuropathy.

On the other hand, for a quick and efficient diagnosis of peripheral neuropathy, it would be desirable to compare the representative value of the quantified feature with a normal reference value that can be set as a threshold. To this end, in the present invention, the normal person and peripheral neuropathy data are shown together in FIG. As a result, the normal signal satisfies the effective value of the first feature 0 to 3000, the central scale of the second feature is 15 to 23, and the correlation coefficient of the third feature is 0.8 to 1 compared to the peripheral neuropathy signal. Could conclude. Therefore, the above-described numerical values can be used as a normal discrimination criterion for distinguishing whether the subject is normal or peripheral neuropathy.

Next, the discriminating unit 30 is a first condition of the first feature is 0 to 3000 in accordance with the above-described normal criteria for diagnosing peripheral neuropathy quickly and reliably immediately after the EMG test of the subject, the second condition The second condition having the features 15 to 23 and the third condition having the first features 0.8 to 1 are discriminated. After that, the discriminating unit 30 satisfies a 'normal' result when all of the first to third conditions are satisfied, and 'peripheral neuropathy' if any of the first to third conditions are not satisfied. Diagnostic information including the results may be generated.

Here, the diagnostic information may not only discriminate whether it is 'normal' or 'peripheral neuropathy', but also easily observe the relief of the treatment including the change over time during the last test, and also the 'normal' 'The severity of the disease may also be included by setting levels for the numbers depending on how far from the data.

As described above, the peripheral neuropathy discrimination apparatus using the acupuncture electromyography signal according to the present invention is highly reliable and can reduce the error of diagnosis even for an inexperienced person to diagnose peripheral neuropathy. .

In addition, since peripheral neuropathy and myopathy have different characteristics of signals, the peripheral neuropathy discrimination apparatus using the needle electromyography signal of the present invention can be used to visualize and quantify not only peripheral neuropathy but also characteristic action potentials in muscle disease. It can be used more clinically.

The technical configuration of the present invention described above will be understood by those skilled in the art that the present invention can be implemented in other specific forms without changing the technical spirit or essential features of the present invention.

Therefore, the above-described embodiments are to be understood as illustrative and not restrictive in all respects, and the scope of the present invention is indicated by the following claims rather than the detailed description, and the meaning and scope of the claims and their All changes or modifications derived from an equivalent concept should be construed as being included in the scope of the present invention.

10: EMG detector
20: signal processing unit
21: preprocessing unit
22: wavelet transform unit
23: feature extraction unit
24: data selection
30: discrimination unit

Claims (5)

After inserting the needle electrode into the muscle to be examined, the EMG detector detects a plurality of activity potential signals for one exercise unit by repeatedly generating only the action potential signals for one exercise unit when the muscle to be examined contracts. ;
A signal processor for processing the activity potential signal to derive a quantified value so as to visualize characteristics of a time-frequency signal; And
And a discriminating unit configured to compare the quantified value with a reference value representing a normal person, which is stored in advance, to generate diagnostic information on whether the activity potential signal corresponds to peripheral neuropathy or is normal.
The signal processor unit peripheral neuropathy discrimination apparatus using acupuncture EMG signal, characterized in that the wavelet transform the action potential signal to enable time-frequency analysis. The method of claim 1,
The signal processing unit,
A preprocessor dividing the detected plurality of action potential signals into one action potential signal;
A wavelet transform unit configured to perform a wavelet transform on the divided action potential signals by setting a molet as a basis function;
From the wavelet transformed action potential signals, only signals having a time sample within a predetermined range are separated, and a wavelet coefficient sum obtained by adding a wavelet coefficient for the predetermined time sample range with respect to a scale is calculated, and the diagnosis is performed from the wavelet coefficient sum. Feature extraction unit for extracting a feature that can discriminate information; And
And a data preliminary unit for selecting the representative value of the feature for the same exercise unit as the quantified value by averaging the features extracted for each of the plurality of action potential signals. Neuropathy differentiation device. The method of claim 2,
The feature extraction unit,
After extracting the first feature and the second feature from the wavelet coefficient sum,
Peripheral neuropathy discrimination apparatus using acupuncture EMG signal, characterized in that the third feature is extracted by performing a comparative analysis of the wavelet coefficient sum and normal data. The method of claim 3,
The first feature is calculated as the effective value of the wavelet coefficient sum using Equation 4 below,
[Equation 4]

Where n is the number of scales,
C is the wavelet coefficient sum,
The second feature is calculated as the center scale for the wavelet coefficient sum using Equation 5 below.
[Equation 5]

Where s is the scale,
C (s) is the wavelet coefficient sum for the scale s,
The third feature is calculated from the correlation analysis between the normalized representative wavelet coefficient sum and the wavelet coefficient sum of normalized data using Equation 6 below.
[Equation 6]

Where x is the wavelet coefficient sum,

Is the mean of x,
y is the normal representative wavelet coefficient,

Peripheral neuropathy discrimination apparatus using acupuncture EMG signal, characterized in that the average of y. The method of claim 4, wherein
The discriminating unit,
A first condition that the first feature is 0 to 3000,
A second condition that the second feature is 15 to 23,
Discriminating whether or not the third feature satisfies a third condition of 0.8 to 1;
If all of the first to third conditions are satisfied, a 'normal' result is obtained.
Peripheral neuropathy discrimination apparatus using acupuncture EMG signal, characterized in that for generating any diagnostic information including a 'peripheral neuropathy' results if any one of the first to third conditions are not satisfied.

Images