A method for automatically detecting pregnancy based on pulse signals at Cun, Guan, and Chi positions

Abstract

A method for automatically detecting pregnancy based on pulse signals at the Cun, Guan, and Chi positions, characterized in that it includes the following steps: (1) reading three pulse wave signals synchronously detected from the Cun, Guan, and Chi positions of a woman, and performing the same preprocessing on them to remove baseline drift and high-frequency noise therein; (2) performing the same Fourier transform on the three preprocessed pulse wave signals to produce their respective amplitude spectra or power spectra, and calculating their respective comprehensive spectral entropy indicators based on the above-mentioned Fourier transforms, and selects the signal with the smallest comprehensive spectral entropy indicator among the three signals, that is, the signal with the best and cleanest quality; (3) for the selected signal with the smallest comprehensive spectral entropy indicator, a frequency point is selected on its amplitude spectrum or power spectrum, and the ratio of the sum of the spectral line amplitudes greater than the frequency point to the sum of the spectral line amplitudes less than the frequency point is calculated. If the ratio is greater than a certain threshold, the woman can be determined to be pregnant; if the ratio is less than a certain threshold, the woman can be determined to be non-pregnant.

Description

Technical Field

[0001] The present application relates to a method for automatically performing pregnancy detection based on Cun, Guan, and Chi pulse signals.

[0002] Traditional Chinese medicine has a history of more than 2,000 years, and its modernization has received more and more attention. Pulse diagnosis is one of the key components of the four diagnostic methods of "looking, smelling, asking and touching". The industry is launching a variety of pulse signal detection devices that imitate traditional Chinese medicine pulse diagnosis, which use pressure sensors to pick up pulse wave signals from the wrist at the inch, guan and chi positions for analysis and diagnosis.

[0003] According to TCM theory, after pregnancy, women's physiological state changes, and their pulse will also change accordingly. However, traditional Chinese medicine judges whether a woman is pregnant by feeling the pulse characteristics through finger touch, which relies heavily on the accumulation of subjective experience. Even experienced TCM practitioners may not succeed. A typical application of the TCM pulse-taking instrument is to pick up the pulse wave signal from the wrist inch, guan, and chi positions based on the pressure sensor for analysis, feature extraction, and auxiliary diagnosis of whether a woman is pregnant. Background Art

[0004] To automatically detect pregnancy based on the Cun, Guan, and Chi pulse signals, it is first necessary to collect and preprocess the Cun, Guan, and Chi pulse signals to remove baseline drift and high-frequency noise, and then automatically select or synthesize a relatively clean pulse signal from the three Cun, Guan, and Chi signals. Finally, based on the selected or synthesized enhanced pulse wave signal, effective features are extracted to automatically detect pregnancy.

[0005] The whole process can usually be divided into three stages: data acquisition preprocessing, feature extraction, and pattern recognition. The first stage of data acquisition preprocessing mainly includes data preprocessing processes such as collecting raw data and removing baseline drift and high-frequency noise. To remove baseline drift and high-frequency noise, bandpass filtering, HHT transform, wavelet transform, etc. can be used. To automatically pick out or synthesize a relatively clean pulse signal, it is necessary to define the signal quality index as the objective function to achieve unsupervised learning. The second stage is feature extraction, which extracts parameters that can be used to distinguish different patterns from complex physiological signals. Pattern recognition is the last stage of computer-aided diagnosis. It uses statistical methods to analyze the numerical differences of different patterns based on the extracted feature vectors, or generates prediction results through machine learning algorithms.

[0006] Many methods mentioned above are common, but a complete method for automatic pregnancy detection based on Cun, Guan, and Chi pulse signals has not yet been reported in the literature. There is still a lot of room for research on automatic pregnancy detection based on Cun, Guan, and Chi pulse signals. Summary of the invention

[0007] Purpose of the Invention

[0008] A complete method for automatic pregnancy detection based on Cun, Guan, and Chi pulse signals is proposed. It can automatically detect pregnancy stably and robustly, distinguish pregnant and non-pregnant women, and has high computational efficiency.

[0009] Technical Solution

[0010] A method for automatically detecting pregnancy based on pulse signals at the Cun, Guan, and Chi positions, characterized in that it includes the following steps: (1) reading three pulse wave signals synchronously detected from the Cun, Guan, and Chi positions of a woman, and performing the same preprocessing on them to remove baseline drift and high-frequency noise therein; (2) performing the same Fourier transform on the three preprocessed pulse wave signals to produce their respective amplitude spectra or power spectra, and calculating their respective comprehensive spectral entropy indicators based on the above-mentioned Fourier transforms, and selects the signal with the smallest comprehensive spectral entropy indicator among the three signals, that is, the signal with the best and cleanest quality; (3) for the selected signal with the smallest comprehensive spectral entropy indicator, a frequency point is selected on its amplitude spectrum or power spectrum, and the ratio of the sum of the spectral line amplitudes greater than the frequency point to the sum of the spectral line amplitudes less than the frequency point is calculated. If the ratio is greater than a certain threshold, the woman can be determined to be pregnant; if the ratio is less than a certain threshold, the woman can be determined to be non-pregnant.

[0011] According to the above-mentioned method for automatically detecting pregnancy based on Cun, Guan, and Chi pulse signals, in step (2), the same Fourier transform is performed on the three pre-processed pulse wave signals to produce their respective amplitude spectra or power spectra, and their respective comprehensive spectral entropy indicators are calculated based on the above-mentioned method, characterized in that the pulse signal sequence of Cun, Guan, or Chi is x(n), n=0, 1, ..., N-1, N is the number of sampling points, and the sampling rate is f s , the peak-to-peak value is PP, and its amplitude spectrum or power spectrum based on discrete Fourier transform is |X(k)|k=0,1,…,N-1, and the probability distribution corresponding to each spectral line is The probability distribution of the spectral lines of all the peaks on the spectrum is recorded as p(k) k = peakIndex0, peakIndex1, ..., peakIndex n-1 , then the comprehensive spectral entropy index is defined as Or in order to normalize the comprehensive spectral entropy index to a good-looking range, it can be multiplied by a constant C, defined as

[0012] According to the method for automatically detecting pregnancy based on Cun, Guan, and Chi pulse signals as described above, in step (3), for the selected signal with the smallest comprehensive spectral entropy index, a frequency point is selected on its amplitude spectrum or power spectrum, and the ratio of the sum of the spectral line amplitudes greater than the frequency point to the sum of the spectral line amplitudes less than the frequency point is calculated. The method is characterized in that the pulse signal is subjected to wavelet decomposition, and each layer further subdivides the low-frequency part of the decomposition result of the previous layer, and the signal of the corresponding frequency band is reconstructed by the coefficient of each frequency band, and the frequency range where the main energy of the pulse signal is located is determined, and 1 / 3 of the maximum frequency in the range is taken as the demarcation frequency point for calculating the ratio, i.e., the so-called selected frequency point.

[0013] According to the above-mentioned method for automatically detecting pregnancy based on Cun, Guan, and Chi pulse signals, in step (3), for the selected signal with the smallest comprehensive spectral entropy index, a frequency point is selected on its amplitude spectrum or power spectrum, and the ratio of the sum of the spectral line amplitudes greater than the frequency point to the sum of the spectral line amplitudes less than the frequency point is calculated. If it is greater than a certain threshold, it can be determined that the person is pregnant. If it is less than a certain threshold, it can be determined that the person is a non-pregnant woman. The characteristic of the method is that the threshold is about 0.36 for the amplitude spectrum and about 0.13 for the power spectrum.

[0014] Beneficial effects.

[0015] At present, there is no literature reporting a complete method for automatic pregnancy detection based on Cun, Guan, and Chi pulse signals.

[0016] The method of this application proposes in step (2) to define a comprehensive spectral entropy index in the frequency domain to evaluate the signal quality. Since FFT calculation can be used, its computational complexity is N log2 N, which is much smaller than the sample entropy complexity N proposed in the literature. 2 . At the same time, due to the periodic characteristics of the pulse signal, it is more reasonable to define the signal measurement index from the frequency domain in signal theory. The larger the defined comprehensive spectral entropy index is, the richer the signal components are, the more noisy the detected pulse signal is, and the lower the quality is; conversely, the detected pulse signal is cleaner. At the same time, it was found that the recorded clear pulse signal has the characteristic of a relatively large peak-to-peak value. The proposal of the comprehensive spectral entropy index in this application is based on the recognition and research of the above characteristics of the recorded pulse signal.

[0017] In step (3) of the method of the present application, for the selected signal with the smallest comprehensive spectral entropy index, the sum of the spectrum lines before and after a selected frequency is calculated on its amplitude spectrum or power spectrum, and the ratio of the sum of the high-frequency part to the sum of the low-frequency part is used as the basis for automatic pregnancy detection. This scheme has a physiological basis. Because, for pregnant women, due to the presence of the fetus, the blood flow in their veins is naturally more complex, that is, it has the characteristic of richer high-frequency components.

[0018] After dozens of experiments, the method of the present application can 100% stably, reliably and quickly select high-quality channels in the pulse signals at the Cun, Guan and Chi positions, and 100% correctly and automatically determine whether the woman is pregnant. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 , the present application is a block diagram of a method for automatically performing pregnancy detection based on Cun, Guan, and Chi pulse signals.

[0020] Figure 2 , schematic diagram of the Cun, Guan, and Chi pulse signal detection device and software interface.

[0021] Figure 3 , schematic diagram of the pre-processed Cun, Guan, and Chi pulse signals in the embodiment. From top to bottom are the waveforms of the Cun, Guan, and Chi signals after filtering.

[0022] Figure 4 , schematic diagram of the time domain and frequency domain waveforms of the selected off-position signal in the embodiment. DETAILED DESCRIPTION

[0023] Embodiment. A real-time detection device system and software system imitating the traditional Chinese medicine pulse signal of Cun, Guan, and Chi positions were built. Figure 2 As shown, the pulse signals at the Cun, Guan, and Chi positions are synchronously detected at a sampling rate of 500 Hz. This application relates to the software part of the system. As described above, the implementation steps are as follows:

[0024] (1) Read the three pulse wave signals detected synchronously from the Cun, Guan, and Chi positions, and perform the same preprocessing on them to remove the baseline drift and high-frequency noise in them. Specifically, a 0.8Hz-60Hz FIR bandpass filter is used, with a tap number of about 2001. The preprocessing results are shown in the figure below. Figure 3 shown.

[0025] (2) The peak-to-peak values ​​of the three preprocessed pulse wave signals are calculated respectively, and Fourier transform is performed on them respectively to produce their respective amplitude spectra, and their respective comprehensive spectral entropy indices are calculated accordingly. After calculation, the comprehensive spectral entropy indices of the Cun, Guan, and Chi positions of the signals analyzed in this implementation are 0.2372, 0.1155, and 0.1543 respectively. The comprehensive spectral entropy index of the Guan position signal is the smallest, so it is selected as a signal for automatic pregnancy detection.

[0026] (3) For the signal with the smallest spectral entropy index, that is, the closed position signal here, wavelet decomposition is performed on this signal. Each layer continues to subdivide the low-frequency part of the decomposition result of the previous layer, and reconstructs the signal of the corresponding frequency band through the coefficient of each frequency band. It is judged that the pulse signal is mainly a low-frequency biological signal of 0-30Hz. 1 / 3 of the maximum frequency in this range, that is, 10Hz, is taken as the dividing frequency point for calculating the sum of amplitudes. 10-30Hz contains high-frequency detail information of the pulse signal. In terms of its amplitude spectrum or power spectrum, the amplitude spectrum is used here, and the time domain and frequency domain waveforms are as follows: Figure 4 As shown, the sum of the amplitudes corresponding to the frequency points of 0-10Hz and 10-30Hz is calculated respectively, and the latter sum is compared with the former sum. If the ratio is greater than 0.36, it can be determined that the person is pregnant; if it is less than 0.36, it can be determined that the person is a non-pregnant woman. In the implementation, the ratios calculated for two pregnant women were 0.3630 and 0.5594, respectively, and the ratios calculated for five non-pregnant women were 0.2908, 0.3491, 0.2656, 0.3344, and 0.1405, respectively. It can be seen that the method of the present application can stably, reliably and quickly select high-quality channels in the pulse signals of the Cun, Guan, and Chi positions, and automatically perform pregnancy detection.

Claims

1. A method for automatically detecting pregnancy based on pulse signals at Cun, Guan, and Chi positions, characterized in that: The method comprises the following steps: (1) reading three pulse wave signals synchronously detected from the Cun, Guan and Chi positions of a female, and performing the same preprocessing on them to remove baseline drift and high-frequency noise therein; (2) performing the same Fourier transform on the three preprocessed pulse wave signals to produce their respective amplitude spectra or power spectra, and calculating their respective comprehensive spectral entropy indices accordingly, and selecting the signal with the smallest comprehensive spectral entropy index among the three signals, that is, the signal with the best and cleanest quality. Regarding the definition of the comprehensive spectral entropy index, the sequence of the pulse signal at the Cun, Guan or Chi positions is assumed to be x(n), where n=0, 1, ..., N-1, N is the number of sampling points, and the sampling rate is f. s , the peak-to-peak value is PP, and its amplitude spectrum or power spectrum based on discrete Fourier transform is |X(k)|k=0,1,…,N-1, and the probability distribution corresponding to each spectral line is k = 0, 1, ..., N-1, where the probability distribution of the spectral lines of all the peaks on the spectrum is recorded as p(k)k = peakIndex0, peakIndex1, ..., peakIndex n-1 , then the comprehensive spectral entropy index is defined as Or in order to normalize the comprehensive spectral entropy index to a good-looking range, multiply it by a constant C, defined as (3) For the signal with the smallest comprehensive spectral entropy index, a frequency point is selected on its amplitude spectrum or power spectrum, and the ratio of the sum of the spectral line amplitudes greater than the frequency point to the sum of the spectral line amplitudes less than the frequency point is calculated. If this ratio is greater than a certain threshold, the woman can be determined to be pregnant. If it is less than a certain threshold, the woman can be determined to be non-pregnant. When selecting the frequency point, the frequency range where the main energy of the pulse signal is located is determined, and 1 / 3 of the maximum frequency in the range is taken as the demarcation frequency point for calculating the ratio, which is the so-called selected frequency point.

Images