Method for non-invasive quantitative analysis of human blood composition based on camera multispectral method

Abstract

The application discloses a method for noninvasive quantitative analysis of human blood components based on a camera multi-spectrum method, and relates to the technical field of cameras and spectrum analysis, and comprises the following steps: taking a camera as a sensor, reducing the bandwidth of a carrier signal through a color channel, and improving the data acquisition accuracy through the superposition of all pixel points in an image; using a sliding high-speed digital phase-locked loop algorithm to demodulate data of each wavelength, and solving the problem that the sensor and the carrier signal are difficult to be synchronized; using dynamic spectrum theory to extract dynamic spectrum, reducing the influence of individual differences, light source instability and other factors in the measurement process, and improving the signal-to-noise ratio of data and the accuracy of modeling. The application adopts the above-mentioned method for noninvasive quantitative analysis of human blood components based on a camera multi-spectrum method, improves the shortcomings of high equipment cost and large size of the existing detection method, solves the problem of low measurement accuracy of the existing detection method, and provides a method for realizing convenient, real-time, high-precision and noninvasive measurement of blood component content.

Description

Technical Field

[0001] This invention relates to the field of biomedical information detection, and in particular to non-invasive quantitative analysis of human blood components based on camera-based multispectral methods. Background Technology

[0002] The detection technology of human blood components plays an important role in monitoring people's health status and early detection of diseases. Currently, most non-invasive quantitative analysis systems for human blood components utilize optical fibers for light guiding and spectrometers as sensors. For example, CN103948393A discloses a near-infrared non-invasive detection method and device for blood component content: using a near-infrared spectrometer to collect near-infrared spectral data from the human skin surface via a Y-shaped reflective optical fiber. This method has the disadvantages of complex instruments, large size, and high cost, limiting the quantitative analysis of blood components to a laboratory environment. Other researchers use photodiodes as sensors to reduce system size and cost. For example, CN117462121A discloses a non-invasive hemoglobin detection method that uses an analog front-end circuit to control four groups of eight LED beads of different wavelengths to illuminate the fingertip in a sequential manner. The light transmitted through the fingertip and a filter is collected by a photodiode and connected to the analog front-end circuit to output a photoplethysmography (PPG) signal. Although the method of invention CN117462121A is low in cost, it has the following drawbacks: First, the method is limited to the detection of human hemoglobin; second, the use of a time-division modulation method in the modulation algorithm leads to a long acquisition time for multi-wavelength data; third, the filter is not an ideal single-wavelength filter, which results in a low signal-to-noise ratio for data of different wavelengths. In the embodiments of this application, LED light sources of different wavelengths adopt a frequency-division multiplexing modulation method, which reduces the acquisition time while improving the signal-to-noise ratio of the acquired data; using a camera as a sensor can greatly improve the measurement accuracy of PPG signals; using a fast digital phase-locked loop algorithm and a sliding method to demodulate data of different wavelengths solves the problem of difficulty in synchronizing the acquired signal and the carrier signal, while the sliding method has an averaging effect, improving the signal-to-noise ratio of the data, thereby improving the accuracy of non-invasive detection of blood component content; finally, the embodiments of this application are applicable to the detection of various blood component contents, and are not limited to a single blood component. Summary of the Invention

[0003] To achieve low-cost, high-precision, and non-invasive detection of human blood components, this invention proposes a method that utilizes a camera to detect multi-wavelength human pulse wave (PPG) signals and perform non-invasive quantitative analysis of blood components. This method not only offers high measurement accuracy but is also convenient for users. Addressing the issues of low camera frame rate and difficulty in synchronizing the sensor and carrier signal phase, this invention proposes an excitation light encoding method and a high-speed digital phase-locked loop demodulation method. The method includes the following steps: S1: A multi-channel adjustable constant current drive circuit generates constant current drive signals with preset carrier frequencies. The above-mentioned multiple constant current drive circuits are used to control multiple light sources of different wavelengths to obtain multiple wavelength light sources; S2: Obtain transmitted light containing changes in blood flow in the human finger at multiple wavelengths when multiple wavelengths of light are used to illuminate the human finger; S3: Use a camera to capture multiple consecutive frames of images after the light from the light source passes through the finger; S4: Each frame of the image captured by the camera is separated into images corresponding to the three primary colors using a three-primary-color channel separation algorithm; that is, each captured frame is separated into three images. Then, the grayscale values ​​of all pixels in each image of each of the three primary color channels are summed up. The summed value is the light intensity value at that moment. The light intensity values ​​obtained from each consecutively captured image are arranged in chronological order of capture time to obtain a digital signal over a period of time, denoted as [missing information]. (in (For R, G, B); S5: The acquired signal of each of the three primary color channels is demodulated by a sliding high-speed digital phase-locked demodulation algorithm to obtain the digital signal at each wavelength, resulting in multiple demodulated digital signals at different wavelengths. Based on the demodulated human pulse wave signal data of all wavelengths, the dynamic spectrum at each wavelength is extracted using the dynamic spectrum extraction method. S6: Use modeling methods to establish a model between dynamic spectra of multiple wavelengths and the content of various human body components, and use the model to predict the component content of unknown samples.

[0004] The multi-channel adjustable constant current drive circuit operates as follows: it is driven by multiple preset signals with different carrier frequencies, and then the light source of each wavelength is controlled according to the operating mode of the constant current drive module.

[0005] The camera's specific operation involves: Its chip contains multiple CMOS photoelectric conversion sensors; when the camera is working, it transmits data at a preset frame rate (…). The data collection period is... A video of seconds, containing The frame image is used to obtain the modulated image.

[0006] The method for setting the preset frame rate is as follows: preset frame rate ( At least the maximum carrier frequency times, of which It is a constant.

[0007] Specifically, the high-speed digital phase-locked loop demodulation algorithm involves: when demodulating the modulation signal at each wavelength, the acquired digital signal... The carrier frequencies of the corresponding wavelengths are sampled down to respectively. Four times the sample size, then divided into four groups according to the sampling order, each group... Each group of data points is then summed to obtain the digital signal sampled for each wavelength. The fast digital phase-locked loop (PLL) algorithm distinguishes between signal amplitude and phase by multiplying the acquired signal with two orthogonal reference signals. The sine and cosine reference signal sequences for one period are as follows: , (1) Let four consecutive points constitute one period. The cross-correlation signal after low-pass filtering for each period is: (2) (3) The amplitude and phase of the demodulated signal in each cycle are then calculated as follows: (4) (5) Based on the above addition and subtraction operations, the demodulated digital signal at each wavelength is obtained. .

[0008] Specifically, the sliding fast digital phase-locked demodulation algorithm is as follows: [The algorithm involves] acquiring the digital signal... The carrier frequencies sampled to the corresponding wavelengths are respectively ( 4 times After demodulation in four groups according to the sampling order, each group is used as a sliding window with one group as the sliding step. Each time the window slides, the light intensity value at a given moment is calculated using formulas (1), (2), (3), (4), and (5) until the data acquisition slide ends, thus obtaining the PPG signal at each wavelength. .

[0009] The calculation process for calculating the dynamic spectrum of human blood at multiple wavelengths using the dynamic spectral extraction method is as follows: right Logarithmic operation of the signal: (6) Then, dynamic spectral extraction method was used to extract the wavelength at each wavelength. Dynamic spectrum of human blood in the signal The absorbance values ​​of components in human blood are obtained.

[0010] The modeling method described herein specifically involves establishing the relationship between feature parameters and the content of different blood components.

[0011] The beneficial effects of the technical solution provided by the present invention are: the method provided by the present invention makes full use of the collected data, improves the signal-to-noise ratio of the data, and realizes non-invasive detection of human blood components using a camera. Attached Figure Description

[0012] Figure 1 A flowchart for non-invasive quantitative analysis of human blood components based on camera-based multispectral method provided in an embodiment of the present invention.

[0013] Figure 2 This is a diagram of an experimental apparatus for a non-invasive method of measuring human blood component content using a camera, as provided in an embodiment of the present invention.

[0014] Figure 3 This is a flowchart illustrating the acquisition process of a multi-wavelength modulated light source provided in an embodiment of the present invention.

[0015] Figure 4 This is a flowchart for obtaining human pulse wave signals modulated by all wavelengths from images captured by a camera, as provided in an embodiment of the present invention.

[0016] Figure 5 The flowchart illustrates the calculation of demodulating human pulse wave signals at each wavelength, as provided in an embodiment of the present invention. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments of the present invention will be described in further detail below.

[0018] To achieve convenient and real-time non-invasive detection of human blood component content and to solve the problem of synchronization between carrier signals and acquisition signals, embodiments of this invention provide a camera-based multispectral method for non-invasive quantitative analysis of human blood components. (See [link to relevant documentation]). Figure 1 and Figure 2 See the description below for details.

[0019] 101: A multi-channel adjustable constant current drive circuit generates constant current drive signals with different carrier frequencies. This constant current drive circuit controls multiple light sources of different wavelengths, resulting in multi-wavelength modulated light sources. When the object under test is illuminated by light from these multi-wavelength modulated light sources, multi-wavelength transmitted light containing information about human fingers is obtained. This step specifically includes steps 1011-1013, see [link / reference] Figure 3 See the description below for details: 1011: Multiple adjustable constant current drive circuits generate multiple preset constant current drive signals with different carrier frequencies; The constant current driving signal can be a square wave, a sine wave, or other waveforms. In specific implementation, the embodiments of the present invention do not limit this. Among them, multiple different frequencies can be arranged in equal proportions or set in other ways. In specific implementation, the embodiments of the present invention do not limit this. 1012: Multiple constant current circuits of different frequencies drive multiple light sources of different wavelengths to generate multi-wavelength modulated optical signals; The light source can be a laser, a light-emitting diode, or other types of light source, and the embodiments of the present invention do not limit this; Among them, there are multiple different wavelengths, the number of which can be arbitrary, and the wavelengths can be any different wavelengths. The embodiments of the present invention do not impose any restrictions on this. 1013: Multi-wavelength modulated light signals are transmitted through the human finger to obtain multi-wavelength transmitted light containing information about the arterial volume and pulse of the human finger. Among them, human fingers (tissues) have physical properties such as absorption and scattering of light; when a multi-wavelength modulated light source is perpendicularly irradiated onto the test object, after absorption and scattering by the human fingers and the blood inside the fingers, multi-wavelength transmitted light containing information about the human fingers is obtained.

[0020] 102: A camera captures multiple consecutive frames of images of light passing through a finger. Each frame is then separated into three images corresponding to the three primary colors using a three-primary-color channel separation algorithm. The grayscale values ​​of all pixels in each image for each primary color channel are then summed; this sum represents the light intensity value at that moment. The light intensity values ​​from each consecutively captured image are arranged chronologically to obtain a digital signal over a given time period, denoted as [image 102]. (in (For R, G, B); This step specifically includes steps 1021-1022, see [link / reference] Figure 4 See the description below for details: 1021: The camera's chip has multiple CMOS photoelectric conversion sensors. When the camera is working, it operates at a preset frame rate ( The data collection period is... A video of seconds, containing From the frame image, the modulated image is obtained; 1022: Each frame of image captured by the camera is separated into images corresponding to the three primary colors using a three-primary-color channel separation algorithm, that is, each frame of image captured is separated into three images; 1023: The grayscale values ​​of all pixels in each image of each of the three primary color channels are summed up. The summed value is the light intensity value at that moment. The light intensity values ​​obtained from each continuously acquired image are arranged in chronological order of acquisition time to obtain a digital signal over a period of time, denoted as . (in For R, G, B).

[0021] 103: The acquired signals of each of the three primary color channels are demodulated using a sliding high-speed digital phase-locked loop demodulation algorithm to obtain the digital signal at each wavelength. After obtaining demodulation The signal contains pulse wave spectral information of human fingers at each wavelength, and the dynamic spectrum of human blood at each wavelength is calculated using Lambert-Beer's law and dynamic spectral theory. This step specifically includes steps 1031-1032, see [link / reference] Figure 5 See the description below for details: 1031: Demodulates the digital signal of each of the three primary color channels using a sliding high-speed digital phase-locked loop demodulation algorithm. ( =R, G, B), respectively obtain the information of the human finger contained in the transmitted light of each wavelength: Among them, demodulation operations are performed using formulas (1), (2), (3), (4), and (5) to obtain light intensity information at each wavelength; 1032: For each wavelength Take the logarithm of the signal: Then, dynamic spectral extraction method is used to extract the wavelength at each wavelength. Dynamic spectra of human blood in the data The absorbance value of human blood is positively correlated with the concentration of its components; therefore, the concentration of blood components can be indirectly obtained by calculating the absorbance value. However, individual differences, light sources, and the human body inevitably cause fluctuations during the measurement process, which can affect the results. The measurement results show that the dynamic spectral extraction method can effectively reduce the influence of the factors mentioned above. The dynamic spectral extraction method can be the difference extraction method, the single-line extraction method, or other dynamic spectral extraction methods; this embodiment of the invention does not limit this.

[0022] 104: A model is established using modeling methods to establish a relationship between dynamic spectra of multiple wavelengths and the content of human blood components, and the model is used to predict the content of human blood components. Wherein: human blood components refer to a specific component within human blood; this embodiment of the invention does not limit the specific component.

[0023] The constant current drive circuit and other technologies used in the methods of this invention are all well-known technologies in data processing methods and are known to those skilled in the art.

[0024] In summary, this invention provides a camera-based multispectral method for non-invasive quantitative analysis of human blood components. Compared with existing technologies, this method uses multiple constant current drive circuits to modulate multiple light sources of different wavelengths using frequency division multiplexing, eliminating the need for mechanical structures such as choppers required in traditional technologies, while increasing the amount of information about target components at multiple wavelengths in the measurement system. Using a mobile phone camera as a sensor makes the detection of blood component content more convenient and lays the foundation for real-time monitoring of human blood component content. Using a camera to acquire image data not only reduces the frequency bandwidth of the carrier signal through color channel separation, but also uses the accumulated value obtained by adding all pixels in each image as the light intensity value at each moment, improving the accuracy and sensitivity of the data by using a space-for-precision method. Using a high-speed digital phase-locked loop algorithm to demodulate the digital signal of each wavelength solves the problem of difficulty in synchronizing the sensor and the carrier signal. The dynamic spectrum extraction method reduces the influence of individual differences and light source instability on the absorbance value during the measurement process from the measurement principle, improves the signal-to-noise ratio of the dynamic spectrum, and thus improves the accuracy of modeling.

[0025] Those skilled in the art will understand that the accompanying drawings are merely schematic diagrams of a preferred embodiment, and the above-described embodiment numbers are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments. The above descriptions are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for non-invasive quantitative analysis of blood constituents in a human body based on camera multispectral method, characterized in that, The method includes the following steps: S1: The multi-channel adjustable constant current drive circuit generates a constant current drive signal with a preset carrier frequency. The above-mentioned multiple constant current drive circuits are used to control multiple light sources with different wavelengths to obtain multiple wavelength light sources. S2: Obtain transmitted light containing changes in blood flow in the human finger at multiple wavelengths when multiple wavelengths of light are used to illuminate the human finger; S3: Use a camera to capture multiple consecutive frames of images after the light from the light source passes through the finger; S4: each frame of image collected by the camera is separated into pictures corresponding to three primary colors by using a three-primary-color color channel separation algorithm, i.e. each frame of collected image is separated into three images. Then the gray scale values of all pixel points in each image of each three-primary-color channel obtained are accumulated, and the accumulated value is the light intensity value at this moment. The light intensity values obtained from each image collected in succession are arranged in the order of collection time to obtain a digital signal in a period of time, denoted as (wherein, R, G, B); S5: The acquired signal of each of the three primary color channels is demodulated by a sliding high-speed digital phase-locked demodulation algorithm to obtain the digital signal at each wavelength, resulting in multiple demodulated digital signals at different wavelengths. Based on the demodulated human pulse wave signal data of all wavelengths, the dynamic spectrum at each wavelength is extracted using the dynamic spectrum extraction method. S6: Use modeling methods to establish a model between dynamic spectra of multiple wavelengths and the content of various human body components, and use the model to predict the component content of unknown samples.

2. The method for noninvasive quantitative analysis of blood constituents of a human being by camera-based multispectral method according to claim 1, characterized in that, The frequency division modulated multi-wavelength light is generated as follows: Multiple adjustable constant current drive circuits are driven by signals of multiple preset different carrier frequencies. Then, the light source of each wavelength is controlled according to the working mode of the constant current drive module. The frequency division multiplexing method and encoding method of the multiple preset different carrier frequency signals can be a proportional method or other methods, and the embodiments of the present invention do not limit this.

3. The method for noninvasive quantitative analysis of blood constituents of a human being by camera-based multispectral method according to claim 1, characterized in that, The specific operation of the camera is as follows: The camera's chip has multiple CMOS photoelectric conversion sensors, which, when the camera is working, transmit data at a preset frame rate. The data collection period is... A video of seconds, containing A frame image is obtained by modulating and acquiring the image. The preset frame rate (...) At least the maximum carrier frequency times, of which It is a constant.

4. The method for noninvasive quantitative analysis of blood constituents of a human being by camera-based multispectral method according to claim 1, characterized in that, The specific algorithm for separating the three primary color channels is as follows: The camera has three channels of red, green and blue three primary colors, when a picture is collected by the camera, a picture corresponding to each of the three primary color channels can be obtained by using a three primary color separation algorithm. Therefore, each frame of image collected by the camera is separated into three images corresponding to the three primary colors by using the three primary color channel separation algorithm. Then, the gray values of all the pixel points in each image of each three primary color channel are accumulated, and the accumulated value is the light intensity value at this moment. The light intensity values obtained from each image collected in succession are arranged in the order of collection time to obtain a digital signal in a period of time, that is, the digital signals corresponding to the three primary color channels are finally obtained, denoted as (wherein, R, G, B).

5. The method for noninvasive quantitative analysis of blood constituents of a human being by camera-based multispectral method according to claim 1, characterized in that, The high-speed digital phase-locked demodulation algorithm is specifically as follows: When demodulating the modulated signal at each wavelength, the acquired digital signal The carrier frequencies of the corresponding wavelengths are sampled down to respectively. Four times the sample size, then divided into four groups according to the sampling order, each group... Each group of data points is then summed to obtain the digital signal sampled for each wavelength. The fast digital phase-locked loop (PLL) algorithm compares the acquired signal with two orthogonal reference signals (one period of sine and one period of cosine reference signal sequences, respectively): , Multiplication enables the differentiation of signal amplitude and phase: Let four consecutive points constitute one period. The cross-correlation signal after low-pass filtering for each period is: and Then, the amplitude and phase of the demodulated signal in each cycle are calculated as follows: and After each cycle of calculation, the demodulated digital signal at each wavelength is obtained. .

6. The method for noninvasive quantitative analysis of blood constituents of a human being by camera-based multispectral method according to claim 1, characterized in that, The fast digital phase-locked loop demodulation algorithm for sliding is specifically as follows: The digital signal collected in claim 4 is respectively down-sampled to 4 times of the carrier frequency corresponding to the wavelength is demodulated into 4 groups in the order of sampling, and one PPG signal is calculated by using formulas (1), (2), (3), (4) and (5) every time the data is slid until the sliding of the data is finished, so that the PPG signal under each wavelength is obtained .​ 7. The method for noninvasive quantitative analysis of blood constituents of a human being by camera-based multispectral method according to claim 1, characterized in that, The specific calculation process for calculating the dynamic spectrum of human blood at each wavelength using the dynamic spectral extraction method, based on Beer-Lambert's law and dynamic spectral theory, is as follows: right Logarithmic operation of the signal: Then, the dynamic spectral extraction method is used to calculate the wavelength at each wavelength. Dynamic spectrum of human blood in the signal The dynamic spectral extraction method can be the difference extraction method, the single-line extraction method, or other dynamic spectral extraction methods, and the embodiments of the present invention do not limit this.

8. The method for noninvasive quantitative analysis of blood constituents of a human being by camera-based multispectral method according to claim 1, characterized in that, The specific calculation process for establishing a model relating dynamic spectra across multiple wavelengths to the content of a specific blood component using modeling methods is as follows: A modeling method is used to establish a model between the dynamic spectra of samples at all wavelengths in the modeling set and the content of a certain blood component in the human body. Then, the model is used to predict the blood component content value of samples with unknown blood component content. Wherein, the certain blood component in the human body can be hemoglobin, blood glucose, or other blood components; this embodiment of the invention is not limited to this. Wherein, the modeling method can be partial least squares or other modeling methods; this embodiment of the invention is not limited to this.

Images