WO2001070107A1

WO2001070107A1 - Time-frequency method for detecting spike activity of the stomach

Info

Publication number: WO2001070107A1
Application number: PCT/US2001/040282
Authority: WO
Inventors: Ata Akin; Hun H. Sun
Original assignee: Drexel University
Priority date: 2000-03-14
Filing date: 2001-03-13
Publication date: 2001-09-27
Also published as: AU2001253848A1

Abstract

A non-intrusive method and apparatus for detecting spike activity produced by a stomach of a patient. The method includes attaching a pair of electrodes to the skin of a patient for generating electrical signals responsive to the spike activity; frequency filtering the electrical signals generated by the pair of electrodes by suppressing frequency components of the electrical signals outside a pre-determined frequency range; and evaluating the frequency components of the electrical signals within the pre-determined frequency range to determine the presence of the spike activity. The apparatus for performing the method includes first and second electrode pairs (12a, 12b), a polygraph (14), an analog-to-digital converter (16), and a personal computer (18).

Description

TITLE OF THE INVENTION Time-Frequency Method for Detecting Spike Activity of the Stomach

CROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims the benefit of U.S. provisional Patent Application No. 60/189,227, filed March 14, 2000, entitled "Prototype Instrument to Detect Fast Gastromyo electric Activities from EGG Recordings".

BACKGROUND OF THE INVENTION

[0002] The present invention relates generally to medical diagnostic methods and more particularly to a method and apparatus for detecting high frequency action potentials of the stomach.

[0003] Clinical evidence suggests that the cause of most gastrointestinal diseases (i.e. stress-related motility disorders, nausea, vomiting, dyspepsia, impaired gastric emptying, gastroparesis, gastric ulcer, gastric dysrhythmias and the like) is related to some sort of disturbance in the motility of the stomach, i.e. peristaltic contractions. Currently, the rhythm and strength of stomach motility is determined by invasive techniques such as intraluminal manometry, fiuoroscopy or endoscopy. Consequently, a non-invasive method for the detection of motility disorders would have substantial diagnostic value.

[0004] Since the 1950s, scientists have been studying the use of electrogastrogram (EGG) recordings to observe the electrical activity of the stomach. The scientists have discovered that the stomach is electrically active and that slow-wave depolarizations of the electrical activity are generated by a neural network called the interstitial cells of the Cajal (ICC), that resides between the longitudinal and circular muscles and at submucosal borders of the circular muscle. It is generally believed that the observed slow-wave depolarizations are propagated via electrical couplings between neurons of the ICC network and are thus carried throughout the gastrointestinal (GI) tract. [0005] Signals representing a high frequency action potential have also been observed in conjunction with the slow-wave activity in the EGG recordings. The high frequency action potential signals have been observed in the EGGs either superimposed on plateau phases of, or in between, signals representing the slow-wave depolarization activity. The high frequency action potential signals have been termed "spike activity" because of their spiked appearance in the EGG recording. It has been hypothesized by many researchers that the spike activity is responsible for triggering peristaltic contractions.

[0006] The relationship between the spike activity signals revealed by an EGG recording and peristaltic contractions has been widely investigated. When the spike activity is observed, the activity usually entrains the slow-wave signal, and strong stomach contractions are seen. A widely held view among the researchers in this field is that the higher the frequency and the longer the duration of the spike activity signals, the stronger are the peristaltic contractions. [0007] Heretofore, studies of the electrical activity of the stomach have been accomplished by attaching electrodes directly to the serosal wall of the stomach. Preferably, however, the EGG recording would be based on signals generated by electrodes placed on the surface of the abdomen (i.e. cutaneous recordings), thus providing a non-invasive alternative to the aforementioned invasive techniques. However, there are certain complications in detecting spike activity signals using electrodes placed on the surface of the abdomen. One problem in detecting and characterizing the spike activity based on cutaneous EGG recordings is that spike activity does not readily propagate throughout the stomach and thus is more difficult to detect at the abdominal surface than at the serosal wall. Further, it has been observed, through correlation studies, that the slow-wave tends to pass readily through the tissues without being heavily attenuated. Thus, the higher amplitude of the slow-wave signal tends to dominate the spike activity signal when the EGG signals are measured at the abdominal surface. Another major complication is caused by the electrical properties of the layers underlying the skin, including the abdominal muscles, fat, omentum, peritoneum and vessels. These layers may have distorting effects on the original serosal signal resulting in the attenuation, smearing, and smoothing of the spike activity signals. As a consequence of the foregoing, the spike activity signals measured at the surface are generally weak in amplitude and further, may be contaminated by other electrophysiological signals, such as the electrocardiographic (ECG) signals of the heart muscle and the electromyographic (EMG) signals of abdominal or respiratory muscles.

[0008] In order to establish a reliable non-invasive method for determining the rhythm and the strength of gastric motility in a patient, it would be desirable to have a method for distinguishing the spike activity signals from the slow -wave activity signals of the stomach and from the various interfering signals generated by ECG and EMG potentials. Further, it would be desirable for the non-invasive technique to provide a diagnostic indication having a high correlation with one or more of the invasive techniques. Also, it would be desirable for the method to provide a high probability of detecting the presence of a gastric disorder simultaneously with providing a low probability of a false indication of the gastric disorder.

BRIEF SUMMARY OF THE INVENTION

[0009] Briefly stated the present invention comprises a non-intrusive method for detecting spike activity produced by a stomach of a patient, the method includes the steps of: attaching a pair of electrodes to a skin of a patient for generating electrical signals responsive to the spike activity; frequency filtering the electrical signals generated by the pair of electrodes, the frequency filtering suppressing frequency components of the electrical signals outside a predetermined frequency range; and evaluating the frequency components of the electrical signals within the pre-determined frequency range to determine the presence of the spike activity. [0010] The present invention further comprises an apparatus for detecting spike activity produced by a stomach of a patient, the apparatus including: a pair of electrodes connected to skin of the patient, the pair of electrodes for generating electrical signals responsive to the spike activity; a frequency filter passing frequency components of the electrical signals within a predetermined frequency range to an output of the filter and suppressing frequency components of the electrical signals outside the pre-determined frequency range from appearing at the output of the filter; and a detector connected to the output of the frequency filter for evaluating a signal at the output of the frequency filter.

BRIEF DESCRIPTION OF THE SEVERAL VIEWS OF THE DRAWINGS [0011] The foregoing summary, as well as the following detailed description of preferred embodiments of the invention, will be better understood when read in conjunction with the appended drawings. For the purpose of illustrating the invention, there are shown in the drawings embodiments which are presently preferred. It should be understood, however, that the invention is not limited to the precise arrangements and instrumentalities shown. [0012] In the drawings:

[0013] Fig. 1 is a schematic block diagram of a preferred data acquisition system for acquiring simultaneous electrogastrogram (EGG) data from a serosal wall of a stomach of a patient and from an abdominal surface of a patient in accordance with the present invention. [0014] Fig. 2 is a time domain plot of a spike activity signal generated by an electrode pair attached to the serosal surface of the stomach of the patient;

[0015] Fig. 3 is a time domain plot of the spike activity generated by an electrode pair attached to the abdominal surface of the patient; [0016] Fig. 4 is two plots of a power spectrum estimate (PSE) of an EGG signal generated by the electrode pair attached to the abdominal surface of an anesthetized canine , where plot "a" shows the PSE when the stomach is not in contraction and plot "b" shows the

PSE when the stomach in contraction;

[0017] Fig. 5 is two plots of the PSE of the EGG signal from an electrode pair attached to the abdominal surface of an un-anesthetized canine, where plot "a" shows the PSE when the stomach is not in contraction and plot "b" shows the PSE when the stomach is in contraction;

[0018] Fig. 6 is a schematic block diagram of an adaptive wavelet Morlet filter;

[0019] Fig. 7 is a plot of the frequency response of each sub-filter of the adaptive wavelet Morlet filter shown in Fig. 6; [0020] Fig. 8 is a plot of the time domain response of the adaptive wavelet Morlet filter shown in Fig. 6;

[0021] Fig. 9 is a plot of the frequency domain response of the adaptive wavelet Morlet filter shown in Fig. 6;

[0022] Fig. 10 is schematic flow diagram of a preferred detection process for determining an electrogastrogram motility index; and

[0023] Fig. 11 is two plots of an output of the detection circuit shown in Fig. 10, wherein plot "a" shows the output when spike activity is present and plot "b" shows the output when spike activity is not present.

DETAILED DESCRIPTION OF THE INVENTION

[0024] Referring to the drawings, wherein like numerals are used to indicate like elements throughout the several figures and the use of the indefinite article "a" may indicate a quantity of one, or more than one, of an element, there is shown in Fig. 1 a data acquisition system 10 for acquiring simultaneous electrogastrogram (EGG) data from a serosal wall of a stomach and from an abdominal surface of a canine 20. The data acquisition system comprises first and second electrode pairs 12a, 12b; a Sandhill Data Acquisition System, Model 7, Grass Polygraph, 14 manufactured by Sandhill Science Inc.; an analog-to-digital converter 16 manufactured by National Instruments; and a computer 18 commonly referred to as a personal computer (PC). Commercially available data acquisition software (i.e. Bioview and Labview) manufactured by Sandhill Science Inc., executing in the PC 18, is used to record the EGG data in a memory of the PC 18. Preferably, signals generated by the electrode pairs 12a, 12b are sampled at 100 Hz. The sampled signals from the electrode pairs 12a, 12b are digitized by the analog-to-digital converter 16. Preferably, the digitized samples are then decimated to a 4 Hz. sample rate to eliminate artifacts and interference from other electrophysiological sources. As would be clear to those skilled in the art, the data acquisition system 10 is not limited to the specific components described above or to acquiring the EGG from canines. Further, other sample rates and decimation factors could be used and still be within the spirit and scope of the invention.

[0025] Figs. 2 and 3 are time domain plots of simultaneously recorded EGG serosal and abdominal surface signals generated by the electrode pairs 12a, 12b attached to the canine 20 undergoing induced stomach contractions. The EGG signal shown in Fig. 2, was acquired from the electrode pair 12a attached to a serosal region close to the caudad corpus. The EGG signal shown in Fig. 3 was acquired from the electrode pair 12b attached to the surface of the abdomen. As would be clear to those skilled in the art, while there is an indication of a high frequency spike activity signal in the EGG signals shown in Figs 2 and 3, the magnitude of the spike activity signal is small in comparison to the magnitude of the slow-wave signal, thus making reliable detection of the spike activity signal difficult.

[0026] As known to those skilled in the art, it is fundamental to the detection of signals in the presence of noise to optimize the pre-detection bandwidth of a detection circuit by tailoring the pre-detection bandwidth to the frequency range of the signal of interest, in order to maximize the signal-to-noise ratio of the signal of interest within the pre-detection bandwidth (see for example, Zadeh, L. A. and Ragazzini, J. R., "Optimum Filters for the Detection of Signals in Noise", Proc. IRE 40, (10), 1123-1131, Oct. 1952, which is hereby incorporated by reference in its entirety). However, in the case of selecting a pre-detection bandwidth for the spike activity of the EGG signals, the exact frequency range of the spike activity signals has not been known up to the present time. Conventionally, the frequency range of a stationary signal can be determined by Fourier transform techniques, wherein the integration time is selected to achieve an acceptable signal-to-noise ratio. However, the determination of the spike activity frequency range is non-trivial, since the amplitude of the spike activity signal waveform is small compared to the amplitude of the slow -wave signal waveform in which the spike activity waveform is embedded. Further, both the EGG signal and the slow-wave signal (noise) are non-stationary, limiting the integration time which can be applied to the frequency range determination process. [0027] The frequency range of a signal is preferably characterized by the frequency spectrum function of the signal |H( w)\ . A bispectrum-based method of spectral analysis is applicable to non- stationary signals and thus is suitable for determining the frequency spectrum of the spike activity of the serosal and cutaneous EGG signals analysis (see Nikias, L. C. and Petropulu, A. P. "Higher Order Spectra Analysis: a Nonlinear Signal Processing Framework", Prentice Hall, 1993).

[0028] The bispectrum-based method depends on the third-order statistics of the signal.

The assumption is that the probability distribution function of the excitation (EGG) signal is non-Gaussian. Consequently, the measured signal x(n) can be expressed by a convolutional model, i.e.:

M x(n) = ∑h(k) w(n-k) + g(n), (1) k = N where h(ή) represents a linear time invariant, non-minimum phase system, w(n) is stationary, zero-mean, non-Gaussian, independently identical distributed white noise, and g(n) is additive zero-mean Gaussian white noise, independent of w(n). The goal of the bispectrum-based analysis is to reconstruct the impulse response h(n) from the bispectrum B_x(wj, w-y). The power

spectrum oϊh(n), |H(w)| , is computed from h(n) by using a standard periodogram method.

Among many available methods for determining h(n) from the bispectrum, the method of Alshebeile and Cetin is preferred (see Alshebeile, S.A. and Cetin, A. E. "A phase reconstruction algorithm from bispectrum", IEEE trans. Geosci., Rem. Sensing, 28(2), 1990, pp. 166-170, which is hereby incorporated by reference in its entirety). [0029] The third-order cumulants of the signal x(n) are given by: r (n n ) = E[x(n)x(n + n. )x(n + «_ )] (2)

where E[-] is the expected value operator. By taking the Fourier transform of the cumulant sequence in equation 2, the bispectrum B_x(wj, M>2) ofx(n) is obtained as:

« = -oo n = -∞

[0030] The substitution of equation 2 into equation 3 yields:

^Bx^(wY^W2^{) =} β ^H(^wl) H(w₂) H'(w_l + w₂) + B (w_r w₂) (4)

3 where: β = E[w(n) ] and Bg(wj, >2) is the bispectrum of (n). The advantage of the bispectrum-based method, in theory, is that the bispectrum of the Gaussian noise term becomes null. Hence, the bispectrum operator actually eliminates the effects of any additive Gaussian noise to the signal.

[0031] The results of the determination of |H(w)| for the EGG signals can be seen in Figs.

4 and 5. Fig. 4 shows two plots of a power spectrum estimate (PSE) of the EGG signal from an electrode pair attached to the abdominal surface of an anesthetized canine, where plot "a" shows the PSE of the electrode signals with the stomach not in contraction and plot "b" shows the PSE of the electrode signals with the stomach in contraction. Fig. 5 shows two plots of the PSE from the electrode signals attached to the abdominal surface of an un-anesthetized canine, where plot "a" shows the PSE of the electrode signals, with the stomach not in contraction and plot "b" shows the PSE of the electrode signals with the stomach in contraction. A distinct energy increase is observed in the spectrum in the range 50 to 80 cycles per minute (cpm) when stomach contractions are present. Accordingly, a pre-detection frequency range of about 50-80 cpm is selected as being optimum for analyzing EGG signals in the presence of spike activity. [0032] Preferably, the power dynamics method suggested by Mintchev is used to determine whether the energy variations of the abdominal surface EGG signals are correlated to the serosal signal's energy variations within certain frequency bands (see Mintchev, M.P., Stickel, A. and Bowes, K.L. "Comparative assessment of power dynamics of gastric electrical activity" Dig. Dis. Sci. 42(6), pp. 1154-1157, 1997, and Mintchev, M.P., Stickel, A. and Bowes, K.L. "Reliability of percent distribution of power of electrogastrogram in recognizing gastric electric uncoupling", IEEE Trans. Biomed. Eng. 44(12) pp. 1281-1291, 1997, which is hereby incorporated by reference in its entirety). The power dynamics method is based on calculating the total energy of the serosal and cutaneous EGG signals within pre-determined frequency bands based on periodograms of the serosal and cutaneous EGG signals, and computing the correlation coefficient, p, between the serosal signals and the cutaneous EGG signals in each of the pre-determined frequency bands. The correlation coefficient p is calculated as:

₌ Cov jX ] ^χy -JVarX * VarY where is the X is the progression of the total energy of the serosal signal and Fis the progression of the total energy of the cutaneous signal.

[0033] The periodograms of the EGG signals are determined by first taking the Fourier transform X(w) of the EGG signal x(n), over L points, utilizing a rectangular window function w(n):

LA X(w) = ∑χ(n) w(n) e^'JWn (6)

« = 0 [0034] The periodogram estimate of X(w) is computed as:

P (w) = — \ X(w) \² , (1)

where U is a constant to normalize the bias in the spectral estimate. Equation 7 is called the periodogram ofX(w) if the window, w(n), used is rectangular (see Oppenheim and Shafer,

1989). The total energy, e_x(wj, w ), in the interval [wj, w?] is calculated according to the formula below:

^W2 ^ex^(wV^W2^{) =} \^{P W}) ^dw (⁸>

^Wl [0035] Preferably, the duration of the analysis window w(n) (see equation 6) for the EGG signals is 1 minute with no overlap; and the frequency bands, within which the total energy is determined, are 1-5, 5-20, 20-50, 50-80 and 80-110 cycles per minute (cpm) as dictated by the results of the power spectrum analyses (see Figs. 4 and 5).

[0036] Table 1 presents the cumulative results of correlating experimental data from five separate experiments on canines with eight phases in each experiment. Data that have a correlation coefficient of less that 0.6 is excluded from Table 1. The results show that in each of the frequency ranges, the correlation between the serosal and surface data is relatively high. Table 1

[0037] After determining the frequency range of the spike activity (see above), a time- frequency analysis tool is preferably selected that provides the highest frequency-time resolution in the 50 to 80 cpm frequency range. The continuous wavelet transform (CWT) is such a method of for representing the frequency contents of a non-stationary signal as a function of time. The CWT is expressed as:

[0038] Here * denotes the complex conjugate of the window function ψ(t) , a is a scaling

1 factor, b is a translation factor and _a is added for energy conservation. Equation 9 is defined

on the open 'time scale' half-plane (b 6 R , a>0). Consequently, equation 9 is called a 'time- scale' analysis, since the window ψ(t) is scaled by the constant a in time. Equation 9, if analyzed in detail, is equivalent to a 'constant-Q filter bank'. (Netterli and Barley, "Wavelets and filter banks: theory and design", IEEE Trans. Sig. Process. 40(9), pp. 2207-2232, 1992, which is hereby incorporated by reference in its entirety).

[0039] Alternatively, equation 9 can be written in the Fourier domain with the help of Parseveal's Identity:

CWT (b, a) = e-' x(w) dw (10)

[0040] Here ^Λ denotes Fourier transform. In equation 10, the Fourier transform of the signal x(t) is multiplied by the Fourier transform of a scaled version of the window function ψ(aw) , termed the mother wavelet. By scaling the argument of the mother wavelet a times the frequency response of the wavelet is changed.

[0041] Conventionally, a Morlet wavelet is used as the mother wavelet in the CWT analysis. The Morlet wavelet is a modulated Gaussian having a time domain and Fourier transform pair as follows:

[0042] A value for w_n = fl . = 5.336, is conventionally chosen, such that the second

In2 maximum of the real part of the wavelet, R (ψ(t)), t > 0, is half the first maximum at t = 0. (see Grossmann, et al., "Reading and understand the continuous wavelet transform", in Combes, J.M. and Grossman, A. (Eds.) Springer Verlag, 1987, which is hereby incorporated by

Λ reference in its entirety). With the aforementioned value of ψ(w) , the band pass filter center frequency is localized at the center of the frequency range. [0043] For the discrete case, equation 11 is written as:

CWT_χ(iT_s, a) = T_s (12)

where Ts is the sampling interval and / is the integer sample number.

[0044] In the preferred embodiment of the present invention, a modified form of the Morlet wavelet is used to analyze the EGG signals, as shown in equation 13:

[0045] A proof of the validity of equation 13 is found in Akin, et al., "Comparison of methods to analyze the antropylloric electrical activity", Proceedings of the 19th IEEE EMBS Conference., pp. 1602-1605, 1997, which reference is incorporated herein in its entirety. [0046] In the preferred embodiment, the CWT calculation (equation 9) forms a pre- detection filter having a predetermined frequency range which passes frequency components of the EGG signal generated by the electrode pairs 12a, 12b laying within a predetermined frequency range to an output of the pre-detection filter and which suppresses frequency components of the electrical signal laying outside the pre-determined frequency range from appearing at the output of the pre-detection filter. Preferably, the pre-detection filter is a CWT employing the modified form of the Morlet wavelet (equation 13) to form an aggregated adaptive Morlet wavelet filter (AAMW). The mathematical form of the AAMW is given in equation 14 as:

[0047] In the preferred embodiment, the AAMW filter sub-band filter structure resulting from scaling the parameter a over the frequency band 50 cpm to 80 cpm is shown in Fig 6. The m scaling is such that w_s - w/a where α = α₍ ^ ,m = 0,2,...M - 1, a_π = 2 and M = 10 . The

spectral characteristics of each sub-band filter of the AAMW are shown in Fig. 7, the time domain response of the AAMW is shown in Fig. 8 and the and frequency domain response of the AAMW is shown in 9. While the AAMW is believed to be optimal for detecting the EGG signals, it would be apparent to one skilled in the art that other forms of filters could be used as a pre-detection filter. For instance a fourth order Butterworth filter could be used as the pre- detection filter and still be within the spirit and scope of the invention. [0048] Referring now to Fig. 10, there is shown a flow diagram of a preferred method for detecting the EGG signal. In block 32, the EGG signal from each one of the electrode pairs 12a, 12b is normalized and filtered with the AAMW filter, where the normalized EEG signal is found as:

[0049] The output of the AAMW filter is:

x _f (n) = _jx_kψ 5₀ _ ₈₀(" - ) (16)

^J k

[0050] At Block 34, the output signal from the AAMW filter is squared. At block 36, the squared signal is filtered by a fourth order low pass filter having a bandwidth of 20 cpm. At block 38, the output of the low pass filter is integrated. Preferably, the integration period is 1 minute. The output of the integrator is a signal which is called the EGG motility index (EMI). The EMI is a measure for evaluating the energy of the spike activity within the pre-determined frequency range. Block 40 is a comparator in which the EMI signal is compared with a predetermined threshold. The pre-determined threshold is selected to provide a high probability of detecting the presence of a gastric disorder simultaneously with providing a low probability of a false indication of the gastric disorder.

[0051] Fig. 11 shows two sonogram plots of an output of the detection method shown in Fig. 10, wherein plot "a" shows the averaged EMI when spike activity is present and plot "b" shows the averaged EMI when spike activity is not present.

[0052] The foregoing description has disclosed a method and apparatus by which spike activity, believed to be indicative of peristaltic contractions, can be detected by a non- invasive diagnostic procedure. A bispectrum analysis method is described by which the frequency range of the spike activity in EGG signals is determined. It is then established that EGG signals measured the surface of the abdomen correlate with the EGG signals measured at the serosal surface of the stomach. A continuous wave transform (CWT) is than applied to filtering the EGG signals. The parameters of the CWT transform are selected to optimize the signal-to- noise ratio of the spike activity components of the EGG signals. The filtered EGG signals are detected by squaring and averaging the output of the CWT filter. The result of the preferred filtering and detection process is a signal representing an EGG motility index which is shown to have a high correlation with the spike activity in the EGG signal.

[0053] It will be appreciated by those skilled in the art that changes could be made to the embodiments described above without departing from the broad inventive concept thereof. It is understood, therefore, that this invention is not limited to the particular embodiments disclosed, but it is intended to cover modifications within the spirit and scope of the present invention as defined by the appended claims.

Claims

CLAIMS We claim:

1. A non-intrusive method for detecting spike activity produced by a stomach of a patient, the method comprising the steps of: attaching a pair of electrodes to a skin of a patient for generating electrical signals responsive to the spike activity; frequency filtering the electrical signals generated by the pair of electrodes, the frequency filtering for suppressing frequency components of the electrical signal outside of a pre-determined frequency range; and evaluating the frequency components of the electrical signals within the pre-determined frequency range to determine the presence of the spike activity.

2. The method according to claim 1 , wherein the pre-determined frequency range is about 50 cycles per minute to about 80 cycles per minute.

3. The method according to claim 1, wherein the frequency filtering is performed by an aggregated adaptive Morlet wavelet filter (AAMW).

4. The method according to claim 3, wherein the AAMW has a value of w_c in the range of about 4.3.

5. The method according to claim 1, further including the step of measuring a magnitude of the frequency components within the predetermined frequency range to determine an electrogastrogram motility index of the electrical signals.

6. The method according to claim 5, further including the step of squaring the frequency components within the predetermined frequency range prior to measuring the magnitude of the frequency components.

7. The method according to claim 1, further including the step of evaluating the frequency components of the electrical signals within the pre-determined frequency range by recording said frequency components on a sonogram and observing a magnitude and a time of occurrence of the frequency components displayed by the sonogram.

8. An apparatus for detecting spike activity produced by a stomach of a patient, the apparatus comprising: a pair of electrodes connected to skin of the patient, the pair of electrodes for generating electrical signals responsive to the spike activity; a frequency filter, the input of which is connected to the pair of electrodes, the frequency filter passing frequency components of the electrical signals within a predetermined frequency range to an output of the filter and suppressing frequency components of the electrical signal outside the pre-determined frequency range from appearing at the output of the filter; and a detector connected to the output of the frequency filter for evaluating a signal at the output of the frequency filter.

9. The apparatus according to claim 8, wherein the pre-determined frequency range is about 50 cycles per minute to about 80 cycles per minute.

10. The apparatus according to claim 1, wherein the frequency filter is an aggregated adaptive Morlet wavelet filter (AAMW).

11. The apparatus according to claim 10, wherein the AAMW has a value of w_c in the range of about 4.3.

12. The apparatus according to claim 8, wherein the detector comprises a square law device.

13. The apparatus according to claim 8, wherein the detector comprises a sonograph.