JPH10216096A - Biological signal analyzing device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a biological signal analyzing device in which wavelet conversion is applied to a biological signal, in which feature extraction to a wavelet conversion coefficient is performed, and in which a change point of the biological signal (position or timing in an electrocardiogram QRS group) can be detected at high precision. SOLUTION: Wavelet conversion coefficients corresponding to five scales in five- octave partition are calculated (S2), the maximum value or minimum value of the wavelet conversion coefficients is determined, a coefficient peak line is determined from the largest scale using the maximum (minimum) value corresponding to each scale (S3), and it is determined it an interval between at least two continuous points in the coefficient peak line nis within a predetermined threshold value or not (S4). In the selected coefficient peak line, the size or mark of at least one wavelet coefficient in the respective scales are compared with each other to determine if the selected peak line is a change point in a biological signal or not (S5), it is stored as the change point in the biological signal, and among the stored living body signal change points, a redundant biological signal change point is eliminated by use of an interval between two continuous points in the coefficient peak line, and the wavelet conversion coefficients in the respective scales.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an analyzer for detecting a change point of a biological signal, and more particularly to a biological signal analyzer for detecting a position or time of an R wave of an electrocardiogram.

[0002]

2. Description of the Related Art In recent years, with the development of electrodes that can be used continuously for a long period of time with the miniaturization and weight reduction of recorders, the behavior of the subject exceeds 24 hours without being restricted, as seen in the Holter monitor. Recorded ECG signals for a long time.

[0003] In this case, the number of recorded electrocardiogram data samples becomes enormous, and it is difficult for a doctor to observe all data. Therefore, as shown in FIG.
In order to efficiently analyze and diagnose electrocardiogram waveforms having waves, Q waves, R waves, S waves, and T waves, a computer detects the ECG QRS complex, and automatic analysis such as arrhythmia and heart disease is widely performed. became.

[0004] However, in the conventional analyzer, since it is performed only by a simple operation combining the linear digital filter and the threshold value processing, it is difficult to correctly detect the QRS group, and erroneous detection and omission of detection have occurred. Was.

For example, in an electrocardiogram measured for a long time, noise due to myoelectric potential and poor contact of electrodes as shown in FIG. 12 is mixed, and baseline fluctuation, fluctuation in QRS amplitude and duration due to respiration and body movement are also observed. Occur. In this case, in the threshold processing in which a part exceeding a certain threshold is set as the QRS group, it is impossible to detect the noise part erroneously if it is erroneously detected as a noise part, or if a baseline fluctuation occurs. Also, 10Hz to 25Hz
Even in the process of extracting only the QRS group which is the frequency band of the above using a linear digital filter, the frequency band of the QRS group varies with each individual and changes with time and cannot be detected accurately. For this reason, an analyzer that can be sufficiently applied to the case where the amplitude of the QRS complex fluctuates or the noise level is high has not been realized yet.

[0006]

SUMMARY OF THE INVENTION Accordingly, the present invention has been made to solve such a problem, and an object of the present invention is to perform a wavelet transform on a biological signal and extract characteristics of a wavelet transform coefficient. To provide a biological signal analyzer capable of detecting a change point (position or time of an ECG QRS complex) of a biological signal with high accuracy.

[0007]

In order to achieve the above object, according to the present invention, a biosignal is subjected to wavelet transform, and a change point of the biosignal is obtained from a coefficient peak sequence obtained by extracting features of wavelet transform coefficients. Is detected.

That is, the biological signal analyzing apparatus according to the first aspect of the present invention comprises an A / D conversion means for taking in a biological signal and converting it into a digital signal, and performing a wavelet transform on the biological signal digitized by the A / D conversion. Applying discrete wavelet transform means for calculating a wavelet transform coefficient corresponding to each scale, and extracting features of the transform coefficient using wavelet transform coefficients of at least one or more scales, and forming a coefficient peak sequence corresponding to each scale. A feature extraction unit to be obtained, an interval determination unit for determining whether an interval between at least two consecutive points in a coefficient peak sequence obtained by the feature extraction unit is within a predetermined threshold, and a selection by the interval determination unit Of the wavelet transform coefficients of at least one or more scales in the coefficient peak train Or a coefficient comparing means for determining one of the coefficient peak strings selected by comparing the signs as a change point of the biological signal, and a coefficient peak string determined by the coefficient comparing means as a change point of the biological signal. The storage means for storing, and among the change points of the biological signal stored in the storage means, the interval between two consecutive points of the coefficient peak sequence and the wavelet transform coefficient of each scale are used to generate a redundant biological signal. It is characterized by comprising a redundancy removing means for removing a change point, and an output display means for outputting and displaying a change point of a biological signal remaining without being removed by the redundancy removing means.

According to a second aspect of the present invention, in the biological signal analyzing apparatus according to the first aspect, the discrete wavelet transform means performs 5-octave division.

According to a third aspect of the present invention, in the biological signal analyzing apparatus according to the first aspect, the characteristic extracting means obtains a coefficient peak sequence using wavelet transform coefficients from a larger scale.

According to a fourth aspect of the present invention, in the biological signal analyzer according to the first aspect, the feature extracting means obtains a coefficient peak sequence using a maximum value or a minimum value of the wavelet transform coefficient. .

According to a fifth aspect of the present invention, in the biological signal analyzing apparatus according to the first aspect, the interval determination means uses at least one or more positions (sample points) of a coefficient peak sequence from a smaller scale. And a representative point of the position of the coefficient peak train.

According to a sixth aspect of the present invention, in the biological signal analyzing apparatus according to the first aspect, the interval determining means uses at least one of the coefficient peak trains from a smaller one of the scales. At a representative point of time.

According to a seventh aspect of the present invention, in the biological signal analyzer according to the first aspect, the coefficient comparing means compares whether or not the sign of the wavelet transform coefficient of the selected coefficient peak sequence is a different sign. It is characterized by.

According to an eighth aspect of the present invention, in the biological signal analyzer according to the first aspect, the biological signal is an electrocardiogram signal.

With the above arrangement, according to the first aspect of the present invention, the feature extracting means obtains a coefficient peak sequence using the wavelet transform coefficients, and the interval determining means and the coefficient comparing means determine
The change point of the biological signal is determined from the coefficient peak sequence, and the redundant change point of the biological signal is removed in the redundant removal step, so that the accurate change point of the biological signal can be detected.

In particular, according to the third and fourth aspects of the present invention, the coefficient peak sequence is obtained by using the maximum value or the minimum value of the wavelet transform coefficient from the larger scale by the feature extracting means. The influence of noise is eliminated, the erroneous detection of the coefficient peak sequence is eliminated, and the erroneous detection of the change point of the biological signal and the omission of detection are reduced.

[0018]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the biological signal analyzing apparatus according to the present invention will be described below with reference to the drawings.

(First Embodiment) FIG. 1 shows a first embodiment of the present invention.
1 shows a biological signal analyzer according to the embodiment. In FIG. 1, reference numeral 100 denotes a biological signal, and 101 denotes an A / D converter, which converts the biological signal 100 into a digital signal. 10
Reference numeral 2 denotes a wavelet transform unit, which calculates wavelet transform coefficients corresponding to five scales by five octave divisions. Reference numeral 103 denotes a feature extraction unit that extracts a feature of a transform coefficient using at least one scale among the wavelet transform coefficients, and obtains a coefficient peak sequence corresponding to each scale. Reference numeral 104 denotes an interval determination unit that determines whether the interval between at least two consecutive points in the coefficient peak sequence is within a predetermined threshold. Reference numeral 105 denotes a coefficient comparing unit which compares the coefficient peak sequence selected by comparing the magnitude or sign of the wavelet transform coefficients of at least one or more scales in the coefficient peak sequence selected by the interval determining unit 104 with the biological data. It is determined whether or not it is a signal change point. Reference numeral 106 denotes a storage unit that stores the coefficient peak sequence determined by the coefficient comparison unit 105 as a change point of the biological signal. Reference numeral 107 denotes a redundancy elimination unit, which is one of the consecutive two points of the coefficient peak sequence among the changing points of the biological signal stored in the storage unit 106.
Using the point interval and the wavelet transform coefficient of each scale, redundant biosignal change points are removed. Reference numeral 108 denotes an output display unit which outputs and displays a change point of the biological signal remaining without being removed by the redundancy removing unit 107.

Next, the use procedure and operation of the biological signal analyzer according to the present invention will be described. FIG. 2 is a diagram for explaining the overall use procedure and operation of the biological signal analyzer according to the present invention. First, in step S1, the captured biological signal is converted into a digital signal, and in step S2, a wavelet transform is performed by five-octave division to calculate a wavelet transform coefficient. Next, in step S3, a maximum value or a minimum value of the wavelet transform coefficient is obtained, and a coefficient peak sequence is obtained using a maximum (small) value from a larger scale. Then, in step S4, at least the coefficient peak sequence is continuous. It is determined whether or not the interval between the two points is within a predetermined threshold value. In step S5, the magnitude or sign of at least one or more wavelet transform coefficients of each scale is compared in the coefficient peak sequence selected in step S4. Then, it is determined whether or not the selected coefficient peak sequence is a change point of the biological signal. In step S6, the coefficient peak sequence is stored as a change point of the biological signal. Finally, step S
In step 7, among the changing points of the stored biological signal, the redundant changing point of the biological signal is removed by using the interval between two consecutive points of the coefficient peak sequence and the wavelet transform coefficient of each scale. At S8, the change point of the remaining biological signal is output and displayed.

First, the conversion processing in step S2 will be described with reference to equations and FIG. FIG. 3 is a configuration diagram showing a discrete wavelet transform filter bank that performs 5-octave division. The wavelet transform is a functional system (Equation 1) obtained by a power-of-two scale transform of the basic wavelet transform Ψ (x),

[0022]

(Equation 1)

And the inner product of the signal (Equation 2),

[0024]

(Equation 2)

[0025] For the Fourier transform フ ー i (ω) of Ψi (x), (Equation 3)

[0026]

(Equation 3)

A scaling function Φ (x) that satisfies is defined. Also, a smoothed signal obtained by convolution of the scaling function Φ (x) and the signal f (x) scaled by 2 ス ケ ー ル i is defined as si (t). A signal obtained by discretizing si (t) is defined as si (n). Discrete time signal s0
(N) can be reconstructed by an M-octave discrete wavelet transform and a discrete-time signal sM (n). Further, the discrete wavelet transform ωi (n) is expressed by (Equation 4) (Equation 5) (Equation 6) from the discrete time signal si (n),

[0028]

(Equation 4)

[0029]

(Equation 5)

[0030]

(Equation 6)

The calculation can be performed by the discrete-time low-pass filter G (ω) and the discrete-time high-pass filter H (ω) satisfying the following relationship. FIG. 3 shows the configuration of a filter bank that represents a discrete wavelet transform that performs 5-octave division using G (ω) and H (ω). Discrete-time high-pass filter H (ω) 310 and discrete-time low-pass filter G (ω) for digital (discrete) input signal 300
The wavelet transform coefficient ω1 (n) 330 and the smoothed signal s1 (n) are obtained by performing the transformation process using the signal 320.
Similarly, a high-pass filter 312 and a low-pass filter 322 are applied to s1 (n) to obtain a wavelet transform coefficient ω
2 (n) 332 and a smoothed signal s2 (n) are obtained. Further, the same processing is executed up to i = 3, 4, 5, and the high-pass filters 314, 316, 318 and the low-pass filter 32
4, 326, and 328, respectively, to obtain wavelet transform coefficients ωi (n) 334, 336, and 338 and a smoothed signal s5 (n) 340. Here, each ωi (n) is a wavelet transform coefficient. Next, a process of obtaining a coefficient peak sequence in feature extraction in step S3 will be described in detail. FIG. 4 is a diagram for explaining a flow of a process of obtaining a coefficient peak sequence, and FIG. 5 is a diagram showing a typical example of a coefficient peak sequence extracted from an electrocardiogram waveform. First, in step S31, a maximum value or a minimum value of the wavelet transform coefficient ωi (n) corresponding to each scale obtained in step S2 is obtained. If ωi (n)> 0 and (Equation 7) is satisfied, ω
Let i (n) be a local maximum.

[0032]

(Equation 7)

On the other hand, if ωi (n) <0 and (Equation 8) is satisfied, ωi (n) is set to a minimum value.

[0034]

(Equation 8)

Next, in step S32, the most scale 2
The maximum (small) value of each scale corresponding to the maximum (small) value is sequentially determined in steps S33 to S36 from the larger value of ＾ 5.
Detect from. In step S33, the position of the maximum (small) value of the largest scale or the local maximum (small) closest to the time.
The value is detected from the next smaller scale 2 ＾ 4. In the following step S34, the local maximum (small) detected in step S33
One smaller scale 2 ＾ 3 closest to the position or time of the value
The maximum (small) value at is detected. Step S35, S36
The same is true for In step S37, a set of maximum (small) values obtained in each scale is used as a coefficient peak sequence, and in step S32, a coefficient peak sequence is obtained until there is no maximum (small) value in the largest scale.

FIG. 5 shows a typical example in which the processing in step S3 is applied to an actual electrocardiogram waveform to obtain a coefficient peak sequence. In FIG. 5, the coefficient peak sequence represents Rk. R1 is a coefficient peak sequence (solid arrow line) combining the minimum values, and R2 is a coefficient peak sequence (thick solid line arrow) combining the maximum values. Further, R3 is a coefficient peak sequence (dotted line of arrow) obtained by combining the minimum values. Since the small value of the maximum (small) value in the largest scale is unlikely to be a change point of the biological signal, it is compared with a predetermined threshold value, and the maximum (small) value below the threshold value is ignored, and the coefficient peak sequence is ignored. Is not detected.

Next, the processing in step S4 for determining the interval between the coefficient peak trains will be described in detail. FIG. 6 is a diagram for explaining the flow of processing for determining the interval between the coefficient peak strings. First, in step S41, a position of a coefficient peak train or a representative point of time is set. At this time, at least one position or time from the smaller scale is used. For example, when the scale 2-1 and the scale 2-2 are used, the average of the positions (sample points) is set as the representative point of the position of the coefficient peak row. Alternatively, when the sampling frequency in the A / D conversion means 101 is F (Hz), the position (sample point) is divided by F and converted into time, so that it can be set as a representative point of the time of the coefficient peak train. That is, since the position and the time represent the same meaning, the following description will be made using the representative point of the position of the coefficient peak string. Next, in step S42, step S
Step S is repeated until the coefficient peak sequence Rk obtained in step 3 is reached.
At 43, it is determined whether the interval between two consecutive positions of the coefficient peak sequence is within a predetermined threshold TH1. For example, as shown in FIG. 5, when the interval between the positions of the coefficient peak arrays R1 and R2 is | R1−R2 | ≦ TH1, the fact is output to step S5. And | R1-R2 |> TH
In the case of 1, the process returns to step S42, and the same processing is performed for the next coefficient peak sequence.

Next, the processing in the coefficient comparison of the coefficient peak sequence in step S5 will be described in detail. FIG. 7 is a diagram for explaining the flow of processing for comparing coefficients of a coefficient peak sequence. First, in step S51, the signs of the two coefficient peak strings selected in step S4 are compared. At that time, the codes are compared using at least one or more wavelet transform coefficients (maximum value or minimum value) of each scale. Then, for example, the sign of the largest scale 2 ＾ 5 is compared,
Only in the case of a different sign, the fact is output to step S52.
In the case of the same code, the selected two coefficient peak trains are ignored. Next, in step S52, the sizes of the two coefficient peak arrays are compared. At this time, the magnitudes are compared using the absolute values of the wavelet transform coefficients (maximum value or minimum value) of at least one or more scales. Then, for example, the magnitudes of the sums of the maximum values (small values) of the scale 2 ＾ 3 and the scale 2 ＾ 4 are compared, and the fact that the larger coefficient peak sequence is the change point of the biological signal is output to step S6. . And
In step S6, a change point of the biological signal is set as step S5
Save the position of the coefficient peak sequence input from. Here, when an electrocardiogram waveform is used, a change point of a biological signal is called an R wave point.

Next, the process of removing redundant change points (R wave points) in step S7 will be described in detail. FIG. 8 shows a redundant R
It is a figure for explaining the flow of processing which removes a wave point.
First, in step S71, the interval between two points of the R wave point stored in step S6 is checked. Then, for example, only when the two-point interval is within the predetermined threshold value TH2, the fact is output to step S72. If the difference is equal to or larger than the threshold value TH2, both R wave points are output to step S8 as they are. Next, in step S72, the magnitudes of the two R wave points are compared.
Then, for example, the magnitudes of the absolute value sums of the maximum (small) values of the R wave points of the scale 2 ＾ 3 and the scale 2 ＾ 4 are compared, and the larger absolute value sum is determined in advance as the smaller absolute value sum. If it is equal to or more than the constant TH3 times, that fact is output to step S73. On the other hand, if less than TH3 times, both R wave points are output to step S8 as they are. Step S73
Then, the smaller R wave point is removed, and the larger R wave point is output to step S8. Then, in step S8, the R wave point input from step S7 is output and displayed.

Therefore, in the present embodiment, a coefficient peak sequence is obtained using the wavelet transform coefficients, a change point of the biological signal is determined from the coefficient peak sequence, and a redundant change point of the biological signal is removed. Therefore, it is possible to accurately detect a change point of the biological signal.

(Second Embodiment) FIGS. 9 and 10 show a biological signal analyzer according to a second embodiment of the present invention. In the present embodiment, the overall configuration of the biological signal analyzer is the same as the configuration of FIG. 1 showing the first embodiment of the present invention. The procedure and operation are the same, and only the steps S4 and S5 are different.

First, the difference from step S4 will be described. FIG. 9 shows the flow of the process of step S40 in the present embodiment. Step S41 is the same as that of FIG. 6, and sets the position of the coefficient peak train or a representative point of time. Step S402
Is repeated up to the coefficient peak sequence Rk in step S403.
It is determined whether the interval between three consecutive positions of the coefficient peak train is within a predetermined threshold value TH1. For example, as shown in FIG. 5, three consecutive coefficient peak trains R1
If | R1−R3 | ≦ TH1, the information is output to step S50. And | R1-R3
If |> TH1, the process returns to step S402, and the same process is performed on the next coefficient peak sequence.

Next, the difference from step S5 will be described. FIG. 10 shows the flow of the process of step S50 in the present embodiment. First, in step S501, the signs of the three coefficient peak strings selected in step S40 are compared. At that time, the codes are compared using at least one or more wavelet transform coefficients (maximum value or minimum value) of each scale. Then, for example, the sign of the largest scale 2 ＾ 5 is compared,
Only when R1 and R2 and R2 and R3 are different signs, that is, alternately different signs, the fact is output to step S502.
If one or both have the same sign, select 3
The two coefficient peak trains are ignored. Next, in step S502, the sizes of the three coefficient peak strings are compared. At this time, the magnitudes are compared using the absolute values of the wavelet transform coefficients (maximum value or minimum value) of at least one or more scales. Then, for example, scale 2 ＾ 3 and scale 2 ＾ 4
The magnitude of the sum of the absolute values of the maximum (small) values is compared, and the fact that the largest coefficient peak sequence is the change point of the biological signal is output to step S6.

Therefore, in the present embodiment, a coefficient peak sequence is obtained using the wavelet transform coefficients, a change point of the biological signal is determined from the coefficient peak sequence, and a redundant change point of the biological signal is removed. Therefore, it is possible to accurately detect a change point of the biological signal.

The present invention is not limited to the above embodiment. For example, in the first and second embodiments, the wavelet transform means 102 (step S2) performs the five-octave division shown in FIG. 3, but it goes without saying that the division may be other than five octaves. In the first and second embodiments, when the feature extracting unit 103 (step S3) obtains the coefficient peak sequence from the local maximum (small) value of each scale, the characteristic extracting unit 103 determines the position of the local maximum (small) value of the current scale. Has detected the local maximum (small) value closest to the time from the next smaller scale, but may detect the local maximum (small) value having the largest absolute value within a predetermined range. Similarly, when comparing the size of the selected coefficient peak sequence, the coefficient comparing means 105 (step S5) uses the sum of absolute values of the wavelet transform coefficients (maximum value or minimum value) of each scale to determine the magnitude. Although the comparison has been made, the present invention is not limited to the sum of absolute values or the like, and other methods may be used.

[0046]

As described above, according to the biological signal analyzing apparatus of the present invention, the biological signal is subjected to the wavelet transform, the coefficient peak train is obtained by using the wavelet transform coefficients, and the biological signal is obtained from the coefficient peak train. Determine the change point of
Further, since the redundant biological signal is removed, it is possible to accurately detect a change point of the biological signal.

In addition, since the maximum value or the minimum value of the wavelet transform coefficient from the larger scale is used when obtaining the coefficient peak sequence, the influence of the noise of the high frequency component is eliminated and the erroneous detection of the coefficient peak sequence is eliminated. In addition, erroneous detection of a change point of a biological signal and omission of detection can be reduced.

Further, since the generation position or time of the R wave, which is the changing point of the electrocardiogram signal, can be correctly detected, the accuracy of automatic analysis of arrhythmia, heart disease, etc. can be improved. In addition, R required for analysis of heart rate and autonomic nervous system
The accuracy of the R interval can be improved.

[Brief description of the drawings]

FIG. 1 is a diagram showing a biological signal analyzer according to a first embodiment of the present invention.

FIG. 2 is a diagram showing an overall use procedure and operation of the biological signal analyzing apparatus according to the first embodiment of the present invention.

FIG. 3 is a diagram showing a configuration of a discrete wavelet transform filter bank that performs 5-octave division.

FIG. 4 is a diagram showing a flow of a process of obtaining a coefficient peak sequence in the biological signal analyzer according to the first embodiment of the present invention.

FIG. 5 is a diagram showing a typical example of a coefficient peak sequence extracted from an electrocardiogram waveform.

FIG. 6 is a diagram showing a flow of a process for determining an interval between coefficient peak strings in the biological signal analyzing apparatus according to the first embodiment of the present invention.

FIG. 7 is a diagram showing a flow of processing for comparing coefficients of a coefficient peak sequence in the biological signal analyzing apparatus according to the first embodiment of the present invention;

FIG. 8 is a diagram showing a flow of processing for removing redundant R wave points in the biological signal analyzer according to the first embodiment of the present invention.

FIG. 9 is a diagram showing a flow of a process of obtaining a coefficient peak sequence in the biological signal analyzer according to the second embodiment of the present invention.

FIG. 10 is a diagram showing a flow of a process of determining an interval between coefficient peak arrays in the biological signal analyzer according to the second embodiment of the present invention.

FIG. 11 is a diagram showing a typical example of an electrocardiogram waveform.

FIG. 12 is a diagram illustrating a general example of noise and artifacts mixed in an electrocardiogram waveform.

[Explanation of symbols]

REFERENCE SIGNS LIST 100 biological signal 101 A / D conversion means 102 wavelet conversion means 103 feature extraction means 104 interval determination means 105 coefficient comparison means 106 storage means 107 redundancy removal means 108 output display means 300 digital input signals 310, 312, 314, 316, 318 discrete Time high-pass filters 320, 322, 324, 326, 328 Discrete-time low-pass filters 330, 332, 334, 336, 338 Wavelet transform coefficients 340 Smoothed signal

 ──────────────────────────────────────────────────続 き Continued on the front page (72) Inventor Takushi Katsura 1006 Kazuma Kadoma, Kadoma, Osaka Matsushita Electric Industrial Co., Ltd.

Claims (8)

[Claims] 1. A / D conversion means for taking in a biological signal and converting it into a digital signal, performing a wavelet transform on the biological signal digitized by the A / D conversion means, and applying a wavelet transform coefficient corresponding to each scale. , And at least one or more scales of the wavelet transform coefficients are used to extract features of the transform coefficients, and a wavelet transform coefficient corresponding to each scale (hereinafter referred to as a coefficient peak sequence). , And at least two consecutive peaks in the coefficient peak sequence obtained by the feature extraction means.
An interval determination unit that determines whether the interval between the points is within a predetermined threshold, and a magnitude or sign of at least one or more wavelet transform coefficients of each scale in a coefficient peak sequence selected by the interval determination unit. Coefficient comparing means for determining whether or not the coefficient peak sequence selected by comparison is a change point of the biological signal, and storage means for storing the coefficient peak sequence determined by the coefficient comparing means as a change point of the biological signal And, among the changing points of the biological signal stored in the storage means, removing the redundant changing point of the biological signal using the interval between two consecutive points of the coefficient peak sequence and the wavelet transform coefficient of each scale. Means for removing redundancy,
A biological signal analyzer, comprising: output display means for outputting and displaying a change point of the biological signal remaining without being removed by the redundancy removing means. 2. The discrete wavelet transform means according to claim 2, wherein
2. The biological signal analyzer according to claim 1, wherein octave division is performed. 3. The biological signal analyzing apparatus according to claim 1, wherein said characteristic extracting means obtains a coefficient peak sequence using wavelet transform coefficients from a larger scale. 4. The biological signal analyzer according to claim 1, wherein said characteristic extracting means obtains a coefficient peak sequence using a maximum value or a minimum value of a wavelet transform coefficient. 5. The method according to claim 1, wherein the interval determining means uses at least one or more positions (sample points) of the coefficient peak train from a smaller scale as a representative point of the position of the coefficient peak train. Item 2. The biological signal analyzer according to Item 1. 6. The method according to claim 1, wherein said interval determination means uses a time from at least one of the smaller scales of the coefficient peak train as a representative point of the time of the coefficient peak train. Biological signal analyzer. 7. The biological signal analyzer according to claim 1, wherein said coefficient comparing means compares whether or not the sign of the wavelet transform coefficient of the selected coefficient peak sequence is a different sign. 8. The biological signal analyzer according to claim 1, wherein the biological signal is an electrocardiogram signal.