Design and implementation of electronic sphygmomanometer based on upper airway measurement

    As the problem of population aging worsens, home medical monitoring has an increasingly broad market. Blood pressure, as an important physiological parameter reflecting the cardiovascular system and heart status, has become an important part of home medical monitoring. Although the traditional mercury sphygmomanometer is considered the gold standard for measuring blood pressure, it requires training and is inconvenient to carry, so electronic sphygmomanometers have received widespread attention.

　　The measurement methods of electronic sphygmomanometers include: oscillometric method, constant volume method, Korotkoff sound method, double cuff method, ultrasonic method, tension method, pulse wave velocity method, multi-parameter regression method, etc. [1]. Among all the methods, the oscillometric method has become the first choice of sphygmomanometer manufacturers because of its mature theory and easy implementation.

　　The methods of measuring blood pressure based on the oscillometric principle can be divided into two categories: one is called the waveform feature method, which determines blood pressure by analyzing the waveform features of the pulse wave envelope; the other is called the amplitude coefficient method, which determines blood pressure by analyzing the proportional relationship between the amplitudes of the pulse wave. Since waveform features are difficult to identify [2], the second method is easier to implement than the first method.

　　To date, oscillometric blood pressure monitors have undergone three generations of technological innovation. In the first generation, the air pump quickly pressurized the cuff to a certain pressure value, and the air was deflated at a rate of 3-5 mmHg/s through the air core. Blood pressure was measured during the deflation process. In the second generation, compared with the first generation, intelligent pressurization was performed according to the blood pressure of the subject, and the air core was replaced with an electronic constant-speed exhaust valve, which made the deflation speed more accurate. In the third generation, the air pump pressurized at a uniform speed, and blood pressure was measured during the pressurization process. The first two generations of technology are called the lower air method, and the third generation is called the upper air method. The mainstream practice in China is to use the first generation of technology. The disadvantage is that the air core deflates unevenly, affecting the accuracy of the measurement results. The mainstream practice in the world, mainly in Japan, is to use the third generation of technology. Its advantages are short measurement time and comfortable wrist feeling. Its disadvantages are interference from the motor operation and cuff expansion during measurement, which places high requirements on the filtering algorithm [3].

　　Based on the consideration of calculation amount and measurement accuracy, this system adopts the amplitude coefficient method and upper air measurement technology. Because pulse waves are generated during blood pressure measurement, in addition to systolic and diastolic pressure, heart rate is also an important measurement parameter of this system.

　　1. The principle of measuring blood pressure by breathing

　　As mentioned above, blood pressure measurement is performed during the process of the motor pressurizing the cuff. In this process, the corresponding cuff pressure is collected by the sensor, and this cuff pressure is considered to be a mixed signal of blood pressure and pulse wave. The pulse wave is obtained by bandpass filtering the mixed signal. Since the amplitude of the pulse wave is much smaller than the amplitude of the blood pressure, the mixed signal can be estimated as a blood pressure signal. The characteristic points corresponding to the blood pressure signal are systolic pressure and diastolic pressure, which are obtained by using the proportional relationship of the pulse wave amplitude. The detailed description is shown in Figure 1, where the horizontal axis represents time in seconds and the vertical axis represents cuff pressure in mmHg. Since it is difficult to filter the pulse wave when the cuff pressure is very small, the pressurization process can be divided into two stages. In the first stage (0~t1 time period in the figure), the motor is controlled to quickly pressurize the cuff to about 40mmHg, and the pressurization speed is generally 10mmHg/s, and the cuff pressure represented by the OA straight line in the figure is obtained. In the second stage, the motor is controlled to slowly pressurize the cuff, and the pressurization speed is generally 2~3mmHg/s, and the cuff pressure represented by the AB curve segment in the figure is obtained. The corresponding pulse wave is obtained by filtering, which is the blue fluctuation curve in the figure. The cuff pressure corresponding to the maximum value Am of the pulse wave peak is recorded as the mean pressure pm. To the left of the maximum value of the pulse wave peak, the cuff pressure pd corresponding to the pulse wave with a peak amplitude of Ad=Am*kd (kd is the diastolic pressure characteristic coefficient) is recorded as the diastolic pressure. To the right of the maximum value of the pulse wave peak, the cuff pressure ps corresponding to the pulse wave with a peak amplitude of As=Am*ks (ks is the diastolic pressure characteristic coefficient) is recorded as the systolic pressure. The heart rate is defined as the number of pulse waves obtained per minute. The heart rate is calculated from the number of pulse waves obtained in this stage and the time required. In the third stage, after calculating the systolic pressure, diastolic pressure and heart rate, the air is quickly deflated [4].

Figure 1 Schematic diagram of blood pressure measurement by upper air

　　It should be emphasized here that in the second stage, as long as the pressure is increased to the systolic blood pressure of the individual to be tested, the third stage of deflation can be entered. However, the pressure of the upper air pressure measurement method is generally increased to more than 160mmHg, which is much higher than the normal systolic blood pressure range. Therefore, compared with the lower air pressure measurement method, the upper air measurement method takes less time and enhances the comfort of the person being measured.

　　2. Overall design of hardware and software

　　2.1 Overall hardware design

　　The overall hardware design is shown in Figure 2. Due to the rich on-chip and peripheral resources of the MCU ATmega128, this system connects the air pump, air valve, cuff, and pressure sensor to it to realize the collection of cuff pressure. The measurement algorithm is implemented by the MCU, and the display device and buttons are connected to the MCU to increase the human-computer interaction function, and the serial port is connected to realize communication with the PC.

　　Figure 2 Overall block diagram of the sphygmomanometer hardware

    2.2 Overall software design

　　The key algorithm flow chart of the software blood pressure measurement is shown in Figure 3. It should be noted that: First, initialization includes the initialization of the timer, display, motor, and pressure sensor. During initialization, the motor is controlled to inflate quickly, and it is only switched to slow inflation when the rapid inflation reaches 40mmHg. Second, it is determined whether the sampling time has been reached, which can be obtained by setting the initial value of the corresponding TCNTn register and the corresponding timer overflow interrupt. Third, after switching to slow inflation, the pulse wave error obtained by filtering the first 50 cuff pressure sampling values ​​is large, so it is set to 0. Fourth, instead of saving all the cuff pressures and pulse waves obtained by sampling and filtering, only the cuff pressures and pulse waves corresponding to the pulse peak after filtering are saved. The subscripts of these two arrays are the positions of the pulse peaks in all peaks. According to these two arrays, the systolic and diastolic pressures can be obtained by applying the upper gas pressure measurement principle. Fifth, the heart rate is obtained by calculating the number of cuff pressure samples bp_count between the first pulse wave peak to the 21st peak, and then applying the definition of heart rate. Through experiments, we choose a sampling period of 100ms, so the heart rate calculation formula is: heart rate/60=20/(0.1*bp_count), that is, heart rate=1200/bp_count.

　　Figure 3 Overall flow chart of blood pressure monitor software

　　3 Implementation of core modules

　　3.1 Implementation of intelligent control pressurization module

　　For the upper air method to measure blood pressure, a key technology is how to control the motor to pressurize and inflate at the required speed. Here, the second function of the corresponding pin of the ATMEGA128 microcontroller and the fast PWM working mode of the microcontroller are used. As shown in Figure 4, the second function of pin PB4 is the output comparison and PWM output of T/C0. When working in fast PWM mode, the comparison unit can output a PWM waveform on the PB4 pin. The specific mechanism of generating the waveform can be shown in Figure 5. The double-buffered output compare register OCR0 is always compared with the value of T/C (TCNT0). The OC0 register is cleared when the comparison matches and set when the counter is cleared. The PB4 pin is controlled according to OC0 to generate a PWM wave. This waveform controls the average voltage of the motor to pressurize and inflate at the required speed. Therefore, the so-called fast and slow inflation is to set different OCR0 register values.

　　Figure 4 ATmega128 pinout

　　Figure 5 Fast PWM mode timing diagram

    3.2 Implementation of filtering module

　　Compared with the lower air method, the upper air method collects the corresponding cuff pressure while the motor is pressurized. Therefore, the disturbance caused by the motor operation and the friction caused by the expansion of the cuff have a greater impact on the collected data. This requires stricter filtering algorithms. Due to the good linear phase characteristics, flat amplitude-frequency response characteristics, high stability, and fast response characteristics of the low-order Butterworth filter, this system uses a second-order Butterworth filter. For the selection of filter parameters, the function of the filter is to filter the pulse wave signal from the mixed signal of the cuff pressure, pulse wave, and some motor disturbances. The cuff pressure is a low-frequency signal, and the frequency range of the pulse wave is 0.5~60HZ, but more than 90% of the pulse wave frequency is below 5HZ. Therefore, a second-order Butterworth bandpass filter of 0.5~3.5HZ is selected to filter the pulse wave signal from the mixed signal.

　　The filter can be implemented by hardware or software. For simplicity, this system chooses software implementation. As described in 2.2, under the premise of satisfying the number of samples, the sampling frequency fs is selected as 10HZ. And because the second-order Butterworth filter with a passband of 0.5~3.5HZ is selected. Therefore, the filter can be designed in matlab as follows: n=2; wn=[0.53.5]/(fs/2); [b, a]=butter(n, wn); [y, t]=impz(b, a); Result=conv(y, bp). Among them, bp is the cuff pressure collected by the pressure sensor, and Result is the pulse wave obtained by convolution of the cuff pressure and the coefficient y of the designed Butterworth filter. Matlab demonstrates the cuff pressure and the pulse wave obtained by the corresponding filtering shown in Figure 6. The Butterworth filter coefficient y obtained from the MATLAB experiment is saved and applied to the program of this system. It is convolved with the cuff pressure collected by the sensor to achieve filtering, obtain the corresponding pulse wave signal, and then calculate the systolic and diastolic pressures.

　　Figure 6 0.5~3.5HZ second-order Butterworth bandpass filter

　　3.3 Implementation of Systolic and Diastolic Blood Pressure Calculation Module

　　Because the selected blood pressure calculation method is the amplitude coefficient method, the key to the calculation module is the selection of characteristic coefficients ks and kd. Different institutions have different choices. Bao Xuhe from Fudan University chose ks=0.5,

　　kd=0.8[5]; Shanghai Medical Instrument Factory selected ks=0.58, kd=0.77[6]; Dr. Chin-TengLin of National Chiao Tung University used an analyzer to generate a standard average pressure MAP=90mmHg, SP=120mmHg, DP=80mmHg, and the corresponding pulse wave-cuff pressure (OA-PC) curve, and calibrated the PC curve corresponding to the OA maximum value OA_MAX as MAP(90mmHg). Based on this calibration scale, SP and DP are found on the PC curve in proportion, and the corresponding pulse wave amplitudes OAS and OAD on the OA curve corresponding to SP and DP are recorded, and ks=OAS/OA_MAX and kd=OAD/OA_MAX are recorded. Repeating this process several times, the average characteristic coefficients are obtained as ks=0.55 and kd=0.7[7]; JMoraes proposed to determine ks and kd according to the size of the average pressure, as shown in the following table[8]. In the oscillometric measurement based on upper air flow, Takashi Usuda gave ks=0.5, 0.55<=kd<=0.69[9]. It can be seen that the selection of coefficients is a summary of experience. We chose upper air flow measurement. Therefore, based on Takashi Usuda's research, we adjusted the coefficients according to the measurement results and then calculated the systolic and diastolic blood pressures.

　　Table 1 Distribution of ks and kd average pressures summarized by JMoraes

　　3.4 Implementation of communication module between MCU and PC

　　In order to analyze the cuff pressure measured in the experiment and the pulse wave obtained by filtering, it is necessary to realize the communication between the microcontroller and the PC. Here, serial communication is selected. Because the output voltage of the PC serial port is RS232 level (up to 15V), and the microcontroller uses TTL level (within 5V), the two are directly connected, which will burn the microcontroller. Therefore, the MAX232 chip is used to provide level conversion. The specific schematic diagram is shown in Figure 7. The MAX232 chip only needs four external capacitors and a +5V power supply to provide two (only one in the figure) TTL to RS232 level conversion.

Figure 7 Schematic diagram of communication between MCU and PC

　　After the hardware connection is established, the software only needs to implement the functions of serial port initialization and data transmission, and then the corresponding data can be received on the PC using the serial port assistant. The receiving effect diagram is shown in Figure 8. It should be noted that the baud rate, check bit, data bit, and stop bit set by the serial port assistant must be consistent with the corresponding settings in the serial port initialization function, otherwise garbled characters will appear.

　　Figure 8 Serial port receiving effect diagram

    4. Result Verification

　　Finally, we used this breathing measurement method and auscultation method to measure blood pressure on 20 patients at the same time. The measurement results are shown in Table 2. The data in Table 2 are all expressed in the form of mean ± standard deviation. As can be seen from the table, the mean and standard deviation of the errors of systolic blood pressure, diastolic blood pressure, and heart rate of the two measurement methods are (3.3 ± 4.40) mmHg, (2.1 ± 4.9) mmHg, and (-3.8 ± 4.1) beats/min, respectively. It can be seen that the mean error is <5 mmHg, and the standard deviation of the error is <8 mmHg, which meets the AAMI standard [10].

　　Table 2 Comparison of measurement results

　　5 Conclusion

　　In order to solve the shortcomings of too high maximum value of pressurization and too long pressurization time in the lower air measurement, a blood pressure meter based on upper air measurement was designed and implemented. It has been verified that the measurement time of the system can be shortened to 30s, and the maximum value of pressurization only needs to be added to the individual's systolic blood pressure range (about 120mmHg), while the traditional lower air measurement generally takes up to 1 minute and the maximum value of pressurization is up to 160mmHg. The measurement accuracy meets the AAMI standard. The next step will be to specifically study how to eliminate the pseudo pulse waves generated by some disturbances in the measurement to improve the system's anti-interference ability.

　　Author: Wang Weiwei, Pu Baoming, He Baoyue, Li Shengjin

　　(Graduate School of the Chinese Academy of Sciences, Beijing 100049)

　　(Institute of Computing Technology, Chinese Academy of Sciences, Shenyang 110168)