Blood oxygen saturation measurement method and system based on Gr-WSe2-Pt flexible photoelectric sensor

Abstract

The present invention discloses a blood oxygen saturation measurement method and system based on a Gr-WSe2-Pt flexible photoelectric sensor, involving built-in technology for medical devices. This solution addresses the problems encountered in the prior art. The method primarily utilizes a flexible photoelectric sensor to collect optical signals, then utilizes an FPGA development board to perform median update calculations on 27 sets of electrical signals. The FPGA development board then extracts the extreme values ​​of 49,974 median values ​​updated in real time within 10 seconds. Finally, 250 million clock cycles are used to control the periodic output of two wavelength light sources, and the blood oxygen saturation value is obtained based on the blood oxygen saturation calculation formula. The advantage is that the use of a simple light source and a flexible patch-type sensor allows the detection portion to be attached to the skin. This reduces both measurement errors caused by movement and wearer discomfort. A simple algorithm is also used to calculate and process the measured PPG signal, reducing processor requirements while effectively suppressing noise.

Description

Technical Field

[0001] The present invention relates to a built-in algorithm and system for medical equipment, and in particular to a blood oxygen saturation measurement method and system based on a Gr-WSe2-Pt flexible photoelectric sensor. Background Art

[0002] Currently, there are numerous non-invasive blood oxygen saturation measurement methods available on the market using photoelectric sensors. These can be broadly categorized into two types: multispectral and hyperspectral, and time-sharing detection. The former offers advantages such as high accuracy and real-time performance, but suffers from issues such as bulk, complexity, high cost, and poor adaptability to different skin types and complexions. The latter, while simpler and less expensive, often requires time-sharing exposure to light of different wavelengths, resulting in lower accuracy.

[0003] Public documents CN 116807468A and CN 114711766A, for example, are both typical multispectral and hyperspectral methods. While both methods utilize spectroscopy to obtain more data and introduce numerous error factors in data processing, further improving measurement accuracy, the acquisition and imaging of spectral images require relatively complex optical systems, and the subsequent data processing is also more complex, resulting in significant system size disadvantages.

[0004] Time-sharing detection methods typically use two different wavelengths of light to illuminate the tissue under test at different time intervals. The tissue's light absorption is determined by recording the intensity of the incident and reflected light. Because hemoglobin and oxyhemoglobin absorb light of different wavelengths differently, blood oxygen saturation can be calculated from the absorbance of the two different wavelengths. The main difficulty with time-sharing detection lies in minimizing the noise and errors introduced by the environment and measurement, and ensuring that the selected data points on the PPG waveforms obtained from the two waveforms are consistent in time.

[0005] In recent years, there are many time-sharing detection methods based on photoplethysmography (PPG) signals:

[0006] For example, in the public document CN 118177797A, while using red and near-infrared light as time-sharing illumination sources, they also employed green light for tissue irradiation. Because the green PPG signal has a low penetration depth in the skin, the resulting signal is less susceptible to noise contamination, providing valuable guidance for baseline drift removal and peak detection in red and near-infrared PPG signals. However, the corresponding cost is also significant. The addition of green calibration light lengthens the entire sampling cycle, places higher demands on the functionality of the light source and photosensor, and may also require the addition of additional processing hardware. Furthermore, ensuring the low noise performance of the green PPG signal has become a new challenge.

[0007] Another example is the public document CN 113598761A, which uses a non-contact CCD sensor to sample the light intensity around a light source, thereby obtaining tissue light intensity under different wavelengths of illumination. This method also improves device durability, as different tissues or populations only require changing the distance and angle between the CCD sensor and the tissue, eliminating the need to replace system components. However, this method also presents significant challenges: first, it requires measuring an additional set of tissue absorbance images in the absence of light, increasing the workload; second, the data acquired by the CCD sensor is in the form of images rather than direct data points, requiring additional data processing, which poses significant challenges to processor selection and processing time; third, the CCD does not contact the skin, but is highly sensitive to imaging distance, and the light source must be placed as close to the center of the image as possible, necessitating a relatively precise and secure fixture, which increases the overall system size; finally, when changing the tissue or population to be measured, the CCD-tissue distance and imaging center point must be precisely readjusted, resulting in a high learning curve and making it unsuitable for general use.

[0008] For example, in public document CN 116725529A, researchers propose a blood oxygen saturation calculation method based on FIR filtering and mean filtering. This method reduces timing errors by selecting the maximum and minimum values ​​in the PPG waveform as data points. A dynamically updated finger touch threshold, Thre, controls the start and end of measurement. While this method is theoretically simple and efficient, its practical implementation requires significant FIR filtering computational complexity and hardware resources. While FPGA processing is possible, integration is difficult.

[0009] Finally, as documented in public document CN 117442199A, they first used a single wavelength of light for measurement. Based on the measured data, they modified the radiation transfer equation, causing the light source group to output corresponding multi-wavelength light based on the radiation transfer equation. Simultaneously, a pre-trained Retinex-LSTM hybrid model was stored within the processing unit to calculate blood oxygen saturation. While the inclusion of a pre-trained processing model can improve the accuracy of blood oxygen saturation analysis, it undoubtedly places high demands on the processing unit's performance and storage space, hindering the portability of the measurement system. Summary of the Invention

[0010] The purpose of the present invention is to provide a blood oxygen saturation measurement method and system based on a Gr-WSe2-Pt flexible photoelectric sensor to solve the problems existing in the above-mentioned prior art.

[0011] The blood oxygen saturation measurement method based on the Gr-WSe2-Pt flexible photoelectric sensor described in the present invention uses two independent light sources with wavelengths of 650nm and 810nm to periodically and staggeredly illuminate the fingertip to be measured, obtain the light signal transmitted by the fingertip to measure the blood oxygen saturation and display it.

[0012] A Gr-WSe2-Pt flexible photoelectric sensor is attached to the fingertip to collect the transmitted light signal. After filtering, amplifying, and converting the collected light signal into analog-to-digital, the following operations are performed using an FPGA development board:

[0013] S1. Count the number of repetitions of the clock cycle of the 50 MHz clock frequency as a;

[0014] S2. Whenever a reaches 10,000, clear a, count the light source switching and blood oxygen calculation enable count value b, and record the input voltage value at that time; if b = 27, execute steps S3 to S4 once; if b > 27, execute steps S5 to S6 once; if b < 27, repeat this step;

[0015] S3. Using the 27 initially recorded voltage values ​​as an initial array;

[0016] S4. First, divide the 27 voltage values ​​into three large groups of 9 data points each. Then, divide each large group into three small groups of 3 data points each. Each group has a maximum value, a minimum value, and a median. Find the minimum value of the three data points of the maximum value of each of the three groups, the maximum value of the three data points of the minimum value of each of the three groups, and the median of the three data points of the median, to form a new small group of three data points. Each large group forms a new small group using the same rule. Each new small group has a maximum value, a minimum value, and a median. Find the minimum value of the three data points of the maximum value of each of the three groups, the maximum value of the three data points of the minimum value of each of the three groups, and the median of the three data points of the median, to obtain the final three-number group. The median of this three-number group is the median of the 27 voltages.

[0017] S5. Every time a new voltage value is input, the earliest input voltage value V b-26 Shift out the array and set the voltage value V 0.2ms ago b Assign values ​​to the previous V in sequence b-1 , assign the new input voltage value to V b ; Perform step S4 again to obtain a new median; obtain the waveform of the voltage value change through median filtering;

[0018] S6. Whenever b reaches 25027, set b = 27 and execute steps S7 to S9 once;

[0019] S7. Invert the light source control value F to change the light source output;

[0020] S8. Find the n maximum points of the waveform within the current 5.0s and take the average value as m avg ;

[0021] S9. Find n-1 low voltage intervals of the current waveform within 5.0s, take 1000 consecutive data points for each low voltage interval and calculate the average value, which is recorded as n. avg ;

[0022] S10. If F = 0, then m avg Assign to m 650nm , n avg Assign to n 650nm ; If F = 1, then m avg Assign to m 810nm , n avg Assign to n 810nm ; m 650nm , n 650nm , m 810nm , n 810nm Substitute the blood oxygen saturation calculation formula to obtain the blood oxygen saturation value and output it to the display screen;

[0023] in,

[0024] m 650nm is the average maximum value of the PPG signal corresponding to the wavelength of 650nm;

[0025] n 650nm is the average minimum value of the PPG signal corresponding to the wavelength of 650nm;

[0026] m 810nm is the average maximum value of the PPG signal corresponding to the wavelength of 810nm;

[0027] n 810nm It is the average minimum value of the PPG signal corresponding to the wavelength of 810nm.

[0028] The blood oxygen saturation measurement system based on the Gr-WSe2-Pt flexible photoelectric sensor described in the present invention uses the method to measure blood oxygen saturation.

[0029] The blood oxygen saturation measurement method and system based on the Gr-WSe2-Pt flexible photoelectric sensor described in this invention have the advantage of using a simple light source and a flexible patch-type sensor, allowing the detection portion to be attached to the skin. This reduces measurement errors caused by movement and reduces wearer discomfort. Furthermore, a simple algorithm is used to calculate and process the measured PPG signal, reducing processor requirements while effectively suppressing noise. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 It is a schematic flow chart of the method described in the present invention.

[0031] Figure 2 It is a schematic diagram of the voltage curve of the method described in the present invention in the low voltage small amplitude oscillation area. DETAILED DESCRIPTION

[0032] The blood oxygen saturation measurement system based on the Gr-WSe2-Pt flexible photoelectric sensor described in the present invention includes an FPGA development board, a 650nm light source, an 810nm light source, a Gr-WSe2-Pt flexible photoelectric sensor, a filter amplifier, an ADC module, and a display screen, all connected to the FPGA development board. A high-frequency AC light source transmits light through the fingertip being measured. The photoelectric sensor converts the transmitted light signal into an electrical signal, which is demodulated and amplified by the filter amplifier. The ADC module then converts the analog signal into a digital signal and passes it to the FPGA development board. The FPGA development board's built-in algorithm controls the two wavelength light sources to alternately switch in a 5-second cycle. The blood oxygen saturation is calculated using data from the first two switching cycles, and the resulting blood oxygen saturation data is transmitted to the display screen, which displays the blood oxygen saturation.

[0033] Using a simple light source and a flexible patch-type sensor allows the detection part to be attached to the skin, effectively reducing the impact of ambient light on the measured light intensity. This further reduces noise errors caused by motion displacement and optimizes wearing comfort.

[0034] When the wavelength is greater than 600nm, the molar absorption coefficients of both reduced hemoglobin and oxygenated hemoglobin are greatly reduced, that is, the light waves in this band have better penetrability, which can ensure that there are still relatively good detection results under low light intensity. Moreover, if one of the two wavelengths used for detection is selected to be below 720 nanometers and the other wavelength is greater than 730 nanometers, the crosstalk is low, the separation is high, and accurate concentration changes can be obtained. Therefore, the present invention uses two wavelengths of LED light-emitting tubes with wavelengths of 650nm and 810nm as light sources to detect human blood oxygen saturation. The two light sources are controlled by an FPGA development board and work alternately with a cycle of 5.0s.

[0035] After the FPGA development board works with either a 650nm wavelength light source or an 810nm wavelength light source, it first samples at a 0.2ms period to filter out a large amount of repeated or approximate data to reduce the subsequent workload. Then, the median filter method is used to filter out the instantaneous extreme abnormal data in the original waveform to obtain a smoother waveform data set. With a 5.0s period, the switching of the two light sources is controlled, and the data within this 5s is processed to obtain a new m avg and n avg After assigning the corresponding variables, the blood oxygen saturation is calculated according to the formula.

[0036] The formula is:

[0037]

[0038] The operation parameters are: Hb650nm =3750.12cm -1 M -1 , ε Hb810nm =717.08cm -1 M -1 ,

[0039] Specifically, the system uses the FPGA development board as the core and uses the following method to measure blood oxygen saturation.

[0040] like Figure 1 As shown, the blood oxygen saturation measurement method based on the Gr-WSe2-Pt flexible photoelectric sensor described in the present invention includes the following steps:

[0041] S1. Count the number of repetitions of the clock cycle of the 50 MHz clock frequency as a.

[0042] S2. Whenever a reaches 10,000, i.e., after 0.2 ms, a is cleared, and the light source switching and blood oxygen calculation enable count value b is counted once, while the input voltage value at this time is recorded; if b = 27, execute steps S3 to S4 once; if b > 27, execute steps S5 to S6 once; if b < 27, repeat this step.

[0043] S3. After repeating step S2 27 times, b = 27, and the first 27 voltage values ​​are stored as the initial array, that is, the data recorded in the first 5.4 ms.

[0044] S4. First, divide the 27 voltage values ​​into three large groups of 9 data points each. Each group is further divided into three small groups of 3 data points each. Each group has a maximum value, a minimum value, and a median. Find the minimum value of the three data points with the maximum value of each of the three groups, the maximum value of the three data points with the minimum value of each group, and the median of the three data points with the median value of each group, to form a new group of three data points. Each large group will form a new group like this, and each new group has a maximum value, a minimum value, and a median. Find the minimum value of the three data points with the maximum value of each of the three groups, the maximum value of the three data points with the minimum value of each group, and the median of the three data points with the median value of each group, to obtain the final three-number group. The median of these three data points is the median of the 27 voltages. The median filter filters out most of the instantaneous (within 0.2ms) extreme abnormal data, eliminating the interference of some erroneous data.

[0045] S5. After that, every 0.2ms, the enable count value b is calculated and counted once. At this time, b>27. At this time, each time a new voltage value is input, the voltage value V that enters the array earliest will be b-26 Move out of the array and assign Vb 0.2ms ago to V b-1 , V b-1 Assign to V b-2 , V b-2 Assign to V b-3 ..., assign the new input voltage value to V b , and then perform step iv again to obtain a new median. After median filtering, a waveform with relatively smooth voltage value changes is obtained.

[0046] S6. Whenever b reaches 25027, that is, after 5.0 seconds, set b = 27 and execute steps S7 to S9 once.

[0047] S7. Invert the light source control value F to change the light source type. In this embodiment, if F=0, only the light source with a wavelength of 650nm works, and if F=1, only the light source with a wavelength of 810nm works.

[0048] S8. Find the n maximum points of the waveform within 5.0s and record them as m1, m2…m n , take the average value and record it as m avg .

[0049] S9. Find n-1 low voltage intervals of the waveform within 5.0s, take 1000 (0.2s) continuous data of each interval, and take the average value of 1000 (n-1) data as n avg The principle and beneficial effects of this step are as follows: The photovoltages caused by the two wavelengths cannot be measured at the same time, so in the low voltage and small oscillation area, such as Figure 2 As shown in the yellow box, when taking the photovoltage of a single data point, the difference between the values ​​of the two wavelengths may be large. When measuring the transmittance of light with a wavelength of 650nm, the voltage value is taken at the high point of a small oscillation, that is, the voltage V corresponding to point A in the figure A When measuring the transmittance of 810nm wavelength light, the voltage value is taken at the low point of a small oscillation, that is, point B in the figure corresponds to the voltage V B Therefore, 1000 data points within 200ms are taken in the low voltage and small oscillation area, such as the 1000 data points within 200ms in the α and β boxes in the low voltage and small oscillation area diagram, and the voltage average value V is taken for these data points. α 、V β . Comparison | V A -V B |with|V α -V β |, we can find |V α -V β |<<|VA -V B This error is significantly reduced.

[0050] S10. If F = 0, then m avg Assign to m 650nm , n avg Assign to n 650nm If F=1, then m avg Assign to m 810nm , n avg Assign to n 810nm 。 650nm , n 650nm , m 810nm , n 810nm Substitute the blood oxygen saturation calculation formula to obtain the blood oxygen saturation value and output it to the display screen.

[0051] Those skilled in the art can make various other corresponding changes and deformations based on the technical solutions and concepts described above, and all of these changes and deformations should fall within the scope of protection of the claims of the present invention.

Claims

1. A blood oxygen saturation measurement method based on a Gr-WSe2-Pt flexible photoelectric sensor uses two independent light sources with wavelengths of 650nm and 810nm to periodically and alternately illuminate the fingertip to be measured. The transmitted light signal from the fingertip is obtained to measure and display the blood oxygen saturation. It is characterized by: A Gr-WSe2-Pt flexible photoelectric sensor is attached to the fingertip to collect the transmitted light signal. After filtering, amplifying, and converting the collected light signal into analog-to-digital, the following operations are performed using an FPGA development board: S1. Count the number of repetitions of the clock cycle of the 50 MHz clock frequency as a; S2. Whenever a reaches 10,000, clear a, count the light source switching and blood oxygen calculation enable count value b, and record the input voltage value at that time; if b = 27, execute steps S3 to S4 once; if b > 27, execute steps S5 to S6 once; if b < 27, repeat this step; S3. Using the 27 initially recorded voltage values ​​as an initial array; S4. First, divide the 27 voltage values ​​into three large groups of 9 data points each. Then, divide each large group into three small groups of 3 data points each. Each group has a maximum value, a minimum value, and a median. Find the minimum value of the three data points of the maximum value of each of the three groups, the maximum value of the three data points of the minimum value of each of the three groups, and the median of the three data points of the median, to form a new small group of three data points. Each large group forms a new small group using the same rule. Each new small group has a maximum value, a minimum value, and a median. Find the minimum value of the three data points of the maximum value of each of the three groups, the maximum value of the three data points of the minimum value of each of the three groups, and the median of the three data points of the median, to obtain the final three-number group. The median of this three-number group is the median of the 27 voltages. S5. Every time a new voltage value is input, the earliest input voltage value V b-26 Shift out the array and set the voltage value V 0.2ms ago b Assign values ​​to the previous V in sequence b-1 , assign the new input voltage value to V b ; Perform step S4 again to obtain a new median; obtain the waveform of the voltage value change through median filtering; S6. Whenever b reaches 25027, set b = 27 and execute steps S7 to S9 once; S7. Invert the light source control value F to change the light source output; S8. Find the n maximum points of the waveform within the current 5.0s and take the average value as m avg ; S9. Find n-1 low voltage intervals of the current waveform within 5.0s, take 1000 consecutive data points for each low voltage interval and calculate the average value, which is recorded as n. avg ; S10. If F = 0, then m avg Assign to m 650nm , n avg Assign to n 650nm ; If F = 1, then m avg Assign to m 810nm , n avg Assign to n 810nm ; m 650nm , n 650nm , m 810nm , n 810nm Substitute the blood oxygen saturation calculation formula to obtain the blood oxygen saturation value and output it to the display screen; in, m 650nm is the average maximum value of the PPG signal corresponding to the wavelength of 650nm; n 650nm is the average minimum value of the PPG signal corresponding to the wavelength of 650nm; m 810nm is the average maximum value of the PPG signal corresponding to the wavelength of 810nm; n 810nm is the average minimum value of the PPG signal corresponding to the wavelength of 810nm; The blood oxygen saturation calculation formula is: The operation parameters are: Hb650nm =3750.12cm -1 M -1 , ε Hb810nm =717.08cm -1 M -1 , 2. A blood oxygen saturation measurement system based on a Gr-WSe2-Pt flexible photoelectric sensor, characterized in that: Blood oxygen saturation is measured using the method as claimed in claim 1.

Images